Novel Processing of Ceramics and Composites: Ceramic Transactions (Ceramic Transactions Series, Vol. 195)

Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites

Ceramic Transactions Series, Volume 195 Proceedings of the 6th Pacific Rim Conference on Ceramic and Glass Technology (PacRim6); September 11-16, 2005; Maui, Hawaii Edited by

Narottam P. Bansal J. P. Singh James E. Smay Tatsuki Ohji

l^INTERSCIENCE A JOHN WILEY & SONS, INC., PUBLICATION

Copyright © 2006 by the American Ceramics Society. All rights reserved. Published by John Wiley & Sons, Inc., Hoboken, New Jersey Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923,978-750-8400, fax 978-646-8600, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representation or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services please contact our Customer Care Department within the U.S. at 877-762-2974, outside the U.S. at 317-572-3993 or fax 317-572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print, however, may not be available in electronic format. Library of Congress Cataloging-in-Publication Data is available.

ISBN-13 978-0-470-08389-5 ISBN-10 0-470-08389-1 Printed in the United States of America. 10 9 8 7 6 5 4 3 2 1

Contents

Preface

ix

Chemical Vapor Deposition High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

3

Teiichi Kimura and Takashi Goto

Preparation of Ru-C Nano-Composite Films and Their Electrode Properties for Oxygen Sensors

13

Teiichi Kimura and Takashi Goto

Combustion Synthesis Synthesis of Sm0.5Sr0i5CoO3_x and La0.6Sr0i4CoO3_x Nanopowders by Solution Combustion Process

23

Narottam P. Bansal and Zhimin Zhong

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis Processes

33

Wang Weimin, Fu Zhengyi, and Wang Hao

Reaction Forming Fabrication of Silicon Carbide From Bamboo Carbon Templates Yung-Jen Lin and Yi-Hsiang Chiu

45

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

57

Toru Wakihara, Junichi Tatami, Katsutoshi Komeya, Meguro Takeshi, Hideki Kita, Naoki Kondo, and Kiyoshi Hirao

Polymer Processing Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers and Their Magnetic Properties

67

Yukiko Uchida, Makoto Nakanishi, Tatsuo Fujii, Jun Takada, Akinori Muto, and Yusaku Sakata

Electochemical Deposition Fabrication of YSZ Thin Films in an Aqueous Solution by ElectroChemical Deposition

77

Atsushi Saiki, Hiroki Uno, Satoka Ui, Takashi Hashizume, and Kiyoshi Terayama

Plasma Synthesis Preparation and Characterization of Epitaxial Fe2_xTix03 Solid Solution Films

87

Tatsuo Fujii, Hideki Hashimoto, Yusuke Takada, Makoto Nakanishi, and Jun Takada

Solid Freeform Fabrication Microtomography of Solid Freeform Fabrication

Jay C. Hanan, James E. Smay, Francesco DeCarlo, and Yong Chu

97

Floc-Casting Fabrication and Evaluation of Transparent Amorphous Si0 2 Sintered Body Through Floc-Casting

107

D. Hiratsuka, J. Tatami, T. Wakihara, K. Komeya, T. Meguro, and M. Ibukiyama

Solution Deposition Yttria Stabilized Zirconia Thin Films Formation From an Aqueous Solution by Mist Deposition Atsushi Saiki, Yukimine Fujisawa, Takashi Hashizume, and Kiyoshi Terayama

vi

• Novel Processing of Ceramics and Composites

115

Nanopowders and Nanorods Synthesis and Structural Characterization of Nanoapatite Ceramics Powders for Biomédical Applications

125

Kanae Ando, Mizuki Ohkubo, Satoshi Hayakawa, Kanji Tsuru, Akiyoshi Osaka, Eiji Fujii, Koji Kawabata, Christian Bonhomme, and Florence Babonneau

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

133

Kazuya Okamoto, Satoshi Hayakawa, Kanji Tsuru, and Akiyoshi Osaka

Coatings and Films Novel Process for Surface Treatment of AIN - Characterization and Application

141

Takehiko Yoneda, Motonobu Teramoto, Kazuya Takada, and Hiroyuki Fukuyama

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

149

Hiroyuki Fukuyama, Tetsuharu Tanoue, and Kazuhiro Nagata

MYCRONID™ Based Long-Lasting BN Hardcoating as Release Agent and Protection Against Corrosion for Aluminum Foundry Applications

159

Jochen Greim, Martin Engler, Krishna Uibel, and Christoph Lesniak

Composites A Study into Epoxy Composites for High-Voltage Device Encapsulation

169

Ammer K. Jadoon, John C. Fothergill, and Andy Wilb

Qeopolymers

y

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials*

187

Kenneth J.D. MacKenzie, Dan Brew, Ross Fletcher, Catherine Nicholson, Raymond Vagana, and Martin Schmücker

Author Index

201

"Paper presented at the 107th Annual Meeting of The American Ceramic Society, April 10-13, 2005, Baltimore, Maryland

Novel Processing of Ceramics and Composites

• vii

Preface

An international symposium, "Novel Processing of Ceramics and Composites" was held during the 6th Pacific Rim (PacRim-6) Conference on Ceramic and Glass Technology in Kapalua, Maui, Hawaii, during September 11-16, 2005. This symposium provided an international forum for scientists, engineers, and technologists to discuss and exchange state-of-the-art ideas, information, and technology on advanced methods and approaches for processing and synthesis of ceramics, glasses, and composites. A total of 56 papers, including four invited talks, were presented in the form of oral and poster presentations indicating continued interest in the scientifically and technologically important field of ceramic processing. Authors from 15 countries (Australia, Brazil, Canada, China, France, Germany, India, Italy, Japan, Korea, Spain, Taiwan, Turkey, United Kingdom, and the United States) participated. The speakers represented universities, industries, and government research laboratories. These proceedings contain contributions on various aspects of synthesis and processing of ceramics, glasses, and composites that were discussed at the symposium. Eighteen papers describing the latest developments in the areas of combustion synthesis, reaction forming, polymer processing, solid freeform fabrication, chemical vapor deposition, electrochemical and solution depositions, plasma synthesis and floc-casting for fabrication of nanopowders, nanorods, electronic ceramics, composites, thin films, coatings, etc. are included in this volume. Each manuscript was peer-reviewed using the American Ceramic Society review process. The editors wish to extend their gratitude and appreciation to all the authors for their cooperation and contributions, to all the participants and session chairs for their time and efforts, and to all the reviewers for their useful comments and suggestions. Financial support from the American Ceramic Society is gratefully acknowledged. Thanks are due to the staff of the meetings and publications departments of the American Ceramic Society for their invaluable assistance.

ix

It is our earnest hope that this volume will serve as a valuable reference for the researchers as well as the technologists interested in innovative approaches for synthesis and processing of ceramics, composites, nanopowders, nanorods, thin films, coatings, etc. NAROTTAM P. BANSAL J. P. SINGH JAMES E. SMAY TATSUKI OHJI

x

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Chemical Vapor Deposition

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

HIGH SPEED DEPOSITION OF YSZ FILMS BY LASER CHEMICAL VAPOR DEPOSITION Teiichi Kimura and Takashi Goto Institute for Materials Research, Tohoku University 2-1-1 Katahira, Aoba Sendai, Miyagi, Japan 980-8577 ABSTRACT Partially yttria-stabilized zirconia (YSZ) films were prepared by laser chemical vapor deposition (LCVD). The assistance of laser increased the deposition rate for YSZ films up to 660 u.m/h. The increase in the deposition rate was accompanied by plasma formation around the deposition zone, and the plasma was observed over critical laser power and substrate pre-heating temperature. A wide variety of morphologies of films from feather-like columnar to dense textures were obtained depending on deposition conditions. The columnar texture contained a large amount of nano-pores at columnar boundary and inside grains. These columnar texture and nano-pores were advantageous for applying YSZ films to thermal barrier coatings.

INTRODUCTION Laser chemical vapor deposition (LCVD) has been utilized to fabricate mainly thin films in semiconductor devise applications '. In general, LCVD can be categorized into two types; one is photolytic LCVD where laser is used as a high-energy source for photochemical reactions and the other is pyrolytic LCVD where laser is used as a heat-source for thermal reactions. Photolytic LCVD would often adopt ultra-violet laser with energy of several eV. The chemical reactions for the film deposition would proceed by high energy photon energy even without substrate heating. In pyrolytic LCVD, on the other hand, significantly high deposition rates have been achieved by focusing laser beam. However, the volume deposition rate (deposition rate in thickness multiplied by area) has been very small ranging around 10"12 to 10"8 m h"', where thin films, small dots and thin wires have been prepared 2. A CO2 laser with high power about several 100 W was employed in LCVD to prepare relatively thick materials such as TiN and TÍB2 several 10 urn in thickness 2. However, pyrolytic LCVD using the CO2 laser has not been widely utilized due to several difficulties such as absorption by window material; ZnSe could be commonly chosen to avoid absorption of an infra-red light. We have found that many oxide thick films can be prepared at high deposition rates more than several 100 um/h by using LCVD 3'4. This paper focuses on the preparation of yttriastabilized zirconia (YSZ) films by LCVD using high power Nd:YAG laser. Since YSZ films are chemically stable at high temperatures having a low thermal conductivity and good compatibility with Ni-base superalloy, they have been intensively investigated as thermal barrier coatings (TBCs) 5. The thickness of TBCs should be more than several 100 urn, and therefore high-speed deposition processes commonly atmospheric plasma spray (APS) 6 and electron-beam physical vapor deposition (EBPVD) 7 have been employed. However, another route for high-speed deposition should be pursued to develop higher

3

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

Fig. 1 A schematic diagram of LCVD apparatus. performance YSZ coatings. We have reported high-speed deposition of YSZ films at 108 u,m/h by using conventional thermal MOCVD 8. However, the CVD process with much higher deposition rates would be required for practical applications. In this paper, we report the highspeed deposition of YSZ films by LCVD, and describes the effect of deposition conditions mainly on deposition rates, morphology and nano-structure. EXPERIMENTAL Fig. 1 demonstrates a schematic diagram of LCVD apparatus that was made of stainless steel with a hemispherical cold-wall type chamber. The laser light (Nd:YAG, continuous mode, X = 1063 nm) was introduced into the chamber through a quartz window. The laser beam expanded to about 25 mm in diameter was emitted to the whole AI2O3 substrate (polycrystalline, 15x 15x2 mm). The laser power (PL) was changed from 0 to 250 W. ß-diketone complexes, Zr (dpm)4 (dpm: dipivaloylmethanate) and Y (dpm)î were used as precursors. Although we have changed the Y2O3 content in YSZ films from 1 to 8 mol% by controlling the precursor temperature, the results of 4 mol% Y2O3 are described hereafter. O2 gas was separately introduced by a double tube nozzle and mixed with the precursor vapors around the substrate. The substrate temperature (Tsub) was measured by a R-type thermocouple attached underneath the substrate surface. The total pressure (PIot) was kept at 0.93 kPa. Surface and cross-sectional morphologies were observed by scanning electron microscopy (SEM). Transmission electron microscopy (TEM) was employed to investigate the nano-structure of films. The crystal structure and preferred orientation were determined by Xray diffraction (XRD), and the composition was estimated by electron probe X-ray microanalysis (EPMA).

4

• Novel Processing of Ceramics and Composites

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

300

500

700

900

1100

Pre-heating Temperature, Tpre/ K

Fig. 2 Effects of laser power (PL) and substrate pre-heating temperature (Tprc) on the deposition rate of YSZ films. RESULTS AND DISCUSSION Fig. 2 demonstrates the effects of laser power (Pi ) and substrate pre-heating temperature (Tpre) on the deposition rates. While almost no deposition occurred below PL = 50 W, significant increase in deposition rates were observed above PL = 100W. The deposition rates of YSZ films by thermal MOCVD have been commonly reported as few to several 10 pm/h; however, we have achieved a deposition rate of 108 pm/h by using cold-wall type CVD and the ß-diketone precursors [8]. LCVD, on the other hand, has attained a deposition rate more than several 100 pm/h. The deposition rate increased with increasing Tprc and Pi., and showed maximum at Pi. = 200 W and Tpre = 823 K. The decrease in the deposition rate at higher Tpre could be resulted from the premature powder formation in a gas phase. A strong plasma emission was appeared and accompanied with the increase in deposition rates above a critical PL. According to our plasma diagnosis, the plasma had an electron temperature of 4000 K with a continuous spectrum similar to the plank distribution 9. The substrate temperature was significantly increased accompanying the plasma formation. Fig. 3 demonstrates the time dependence of substrate temperature (Tsuh) after the laser emission and introduction of precursor vapors. Increases in Tsub of 150 to 200 K were identified after the laser emission. Since the laser power would have more capability to increase the Tsub, the laser might be partially reflected from the AI2O3 substrate surface resulting to rather small increase in the T^b- After the T^b was stabilized, the precursor vapors and O2 gas were introduced, and after a minute an abrupt temperature increase accompanying the plasma formation was identified. Fig. 4 demonstrates the effect of Ts„b on the deposition rate of YSZ films comparing with literature data of conventional MOCVD l0"13, where the results of relatively high-speed deposition of YSZ films were chosen. YSZ films have been widely prepared by MOCVD owing to their useful applications as oxide ion conducting solid electrolyte 14"16 and buffer layers for high-temperature superconducting oxides 17' 18. The deposition rate was generally several pm/h in literatures; meanwhile high-speed deposition of

Novel Processing of Ceramics and Composites

• 5

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

200 W

1100 Plasma formation -

900

3

700

K

a. E

500

oo w

Laser emission

300

50

100

150

200

Time, f/s Fig. 3 Time dependence of substrate temperature (T^i,) after the laser emission and introduction of precursor vapors.

Substrate temperature, 7SUb / K 13001200 1100 1000 900 800

0.9

1.0

1.1

1.3

Tsub"1/10-3K-1 Fig. 4 Effect of substrate temperature (T5Ub) on the deposition rate of YSZ films comparing with literature data of conventional MOCVD (PL> 100 W).

6

• Novel Processing of Ceramics and Composites

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

YSZ films by MOCVD has been recently reported due to strong requirements for the application to TBCs. The deposition rate generally increases with increasing Ts„b in MOCVD. The activation energy in a low temperature region could be more than several 10 kj/mol, suggesting a chemical reaction limited process 19. The activation energy decreases to several kJ/mol with further increase in Tsub, suggesting a mass transfer (mainly source gas supply) limited process |l). The deposition rates might be decreased at further higher temperatures due to premature powder formation in a gas phase. In the present LCVD, the activation energy was 9 kj/mol in the temperature range between TSUb = 800 and 1300 K, implying the mass transfer limited process. It can be assumed that the plasma would enhance the reactivity of precursor vapors and the surface mobility of chemical species could be also accelerated by the laser emission. The highest deposition rate increased to 660 um/h by increasing the source flux, corresponding to the mass transfer limited process of the present LCVD. Figs 5 to 8 depict the cross-sectional texture of YSZ films. The YSZ film prepared at PL = 100 W and Tsub - 893 K (Fig. 5) had a fine grained dense texture with insignificant preferred orientation. Fig. 6 shows the YSZ film prepared at PL = 150 W and Tsub = 953 K with (200) oriented cauliflower-like texture. Well-developed columnar texture with strongly (200) oriented YSZ films were obtained at PL = 100 W and Tsllb = 1123 K (Fig. 7). The YSZ films prepared at higher temperature (PL = 250 W, Tsl,b = 1213 K) had wider columnar grains (Fig. 8). The cross-section of columnar texture for the YSZ film prepared at Pi = 200 W and T5ub = 1173 K was observed by TEM as shown in Fig. 9. The gaps of about 100 nm in width and a feather-like texture were observed near the surface (Fig. 9(a)). The feather-like texture has been commonly observed in YSZ films prepared by EB-PVD 20 and plasma-enhanced CVD (PECVD) 2I . Fig. 9 (b) demonstrates the nano-structure of middle of columnar texture, where voids about 10 nm in size and a large amount of nano-pores about a few nm in size were observed at the columnar boundary and inside the grains, respectively. Fig. 9 (c) represents the nano-pores

Fig. 5 Cross-section of YSZ films prepared at P,.=100 W and Tsub=893 K

Novel Processing of Ceramics and Composites

• 7

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

Fig. 6 Cross-section of YSZ films prepared at P|.= 150 W and Tsul,=953 K

Fig. 7 Cross-section of YSZ films prepared at P[ =100 W and Ts,,i,=l 123 K

8

• Novel Processing of Ceramics and Composites

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

Fig. 8 Cross-section of YSZ films prepared at PL=250 W and T„b=1213 K around the YSZ/substrate interface. Fine grained poly-crystalline grains with nano-pores were observed around the interface. It is generally known that the te.xture in CVD would change from fine grains with no specific orientation to significantly oriented columns with increasing the thickness of films ", as typically depicted in Fig. 9. The nano-pores inside grains were effective to improve the performance of YSZ films for the application to TBCs 22. Fig. 10 shows the effect of deposition rates on the thermal

Fig. 9 Nano-structure of YSZ film prepared at Pi=200 W and Ts„b=1200 K, (a): near thefilmsurface, (b): middle of the film, (c): near the substrate

Novel Processing of Ceramics and Composites

• 9

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

¿

2.0

?

1.5

.> tí

1 0

1

TJ

§

0.5

to

2 o.o

•-

0

100

200

300

400

500

Deposition rate, RI Jim h"1 Fig. 10 Effect of deposition rates on the thermal conductivity of YSZ films prepared by LCVD. conductivity of YSZ films prepared by LCVD. The thermal conductivity of YSZ films decreased with increasing deposition rate. The thermal conductivity of the YSZ iilm prepared at R=50 um/h (PL=I00 W and 1^=893 K) was 1.3 W/m K, which is almost a half of that of bulk cubic YSZ, while that prepared at R=450 um/h (PL=200 W and T5ub=1200 K) was 0.7 W/m K. This value is almost the same level as those of practical TBCs fabricated by APS and EBPVD. The nano-pores could be the main reason for the low conductivity by phonon scattering as reported in YSZ coatings fabricated by EBPVD The nano-pores at the YSZ/substrate interface combined with the columnar texture would yield excellent adherence of YSZ coatings on Nibase super-alloy substrates surviving 1200 heat-cycles between 773 and 1673 K 23. CONCLUSIONS As an alternate route for the high-speed deposition process of YSZ coatings, we have proposed a laser CVD process, where the highest deposition rate of 660 u.m/h was attained. This speed is almost competitive to those of practical APS and EBPVD. A plasma formation and an exothermic reaction in LCVD have caused significant increase in deposition rates. The high deposition rates have yielded a large amount of nano-pores in columnar grains resulting in the significantly lower thermal conductivity of about 0.7 W/m K. ACKNOWLEDGEMENT This work was performed as a part of Nano-Coating Project sponsored by New Energy and Industrial Technology Development Organization (NEDO), Japan.

10 • Novel Processing of Ceramics and Composites

High Speed Deposition of YS2 Films by Laser Chemical Vapor Deposition

REFERENCES 1 Duty, C. Jean, D., Lackey, W. J., "Laser chemical vapor deposition: materials, modeling, and process control", Int. Mater. Rev., 46, 271-283 (2001). Goto, T. "Thermal barrier coatings deposited by laser CVD ", Surf. Coat. Tech. 198, 367-371 (2005). 3 Kimura, T. and Goto, T., "Rapid Synthesis of Yttria-Stabilized Zirconia Films by Laser Chemical Vapor Deposition", Mater. Trans., 44,421 -424 (2003). 4 Goto, T., "High-speed deposition of zirconia films by laser-induced plasma CVD ", Solid State Ionics, 172, 225-229(2004). 5 Evans, A. G., Mumm, D. R., Hutchinson, W. J., Meier G. H. and Pettit, F. S., "Mechanisms controlling the durability of thermal barrier coatings". Progress in Mater. Sei., 46, 505553(2001). 6 Glocker, D. A."Handbook of Thin Film Process Technology", Inst. Phys. (1995). Czech, N., Fietzek, H., Juez-Lorenzo, M., Kolarik V. and Stamm, W. '"Studies of the bond-coat oxidation and phase structure of TBCs ", Surf. Coat. Tech., 113, 157-164 (1999). 8 Tu, R., Kimura T. and Goto, T., "Rapid Synthesis of Yttria-Partially-Stabilized Zirconia Films by Metal-Organic Chemical Vapor Deposition", Mater. Trans., 43, 2354-2356 (2002),. 9 Miyazaki, H., Goto, T. and Kimura, T., "Acceleration of Deposition Rates in a Chemical Vapor Deposition Process by Laser Irradiation". Jpn../. Appl. Pltys., 42, L316-L318 (2003). 10 Bourhila, N., Feiten, F., SenateurJ. P., Schuster, F., Madar, R. and Abrutis, A., : "Deposition and Characterization of Zr0 2 And Yttria-Stabilized Zr0 2 Films using injection-LPCVD", Proc. 14th Intern. Conf. and EUROCVD-11, Electrochem. Soc. Proc. Vol. 97-25, M. D. Allendorf C. Bernard (Eds.), 1997, p. 417-424. " Whal, G., Nemetz, W., Giannozzi, M., Rushworth, S, Baxter, D., Archer, N., Cernuschi, F. and Boyle, N.,"Chemicai Vapor Deposition of TBC: an Alternative Process for Gas Turbine Components", Trans. Am. Soc. Mech. Eng., 123, 520-524 (2001). 12 Akiyama, Y., Sato, T. and Imaishi, N.,"Reaction Analysis for ZrCh and Y2O3 Thin Film Growth by Low-Pressure Metalorganic Chemical Vapor Deposition Using ß-Diketonate Complexes",./. Crys. Growth, 147, 130-146 (1995). 13 Pulver, M., Nemetz, W. and Wahl, G., "CVD of Zr0 2 , AI203 and Y203 from Metalorganic Compounds in Different Reactors", Surf. Coatings Tech., 125,400-406 (2000). 14 Bonanos, N , Slotwinski, R. K., Steele B. C. H., Butler, E. P., "High Ionic-conductivity in Polycrystalline Tetragonal Y 2 0 3 Zr02",7. Mater. Sei. Lett., 3, 245-248(1984). 15 Brown, J. T.,"Solid oxide fuel cell technology", IEEE Trans. Energy Conversion, 3, 193198(1988). 16 Asada, A., Yamamoto, H., Nakazawa, M. and Osanai, H.,"Limiting current type of oxygen sensor with high performance ", Sensors Actuators B, 1, 312-318 (1990). 17 Tiwari, P., Kanetkar, S. M., Sharan, S. and Narayan, }.,"In situ single chamber laser processing of YBa2Cu307 ¿ superconducting thin films on Si (100) with yttria-stabilized zirconia buffer layers", Appl. Phys. Lett., 57, 1578-1580(1990). 18 Ogale, S. B., Vispute, R. D., Rao, R. R.,"Pulsed excimer laser deposition YiBa2Cu307_x superconductor films on silicon with laser-deposited Y-Zr02 buffer layer". Appl. Phys. Lett., 57, 1805-1807(1990). 19 Bryant, W. A.. "Fundamentals of Chemical Vapor-deposition", J. Mater. Sei., 12. 1285-1306 (1977).

Novel Processing of Ceramics and Composites

• 11

High Speed Deposition of YSZ Films by Laser Chemical Vapor Deposition

Lu, T. L., Levi, C. G., Wadley, H. N. G. and Evans, A. G., "Distributed Porosity as a Control Parameter for Oxide Thermal Barriers Made by Physical Vapor Deposition", J. Am. Ceram. Soc, 84, 2937-2976(2001). 21 Préauchat, B, and Drawin S., "Properties of PECVD-Deposited Thermal Barrier Coatings", Surf. Coatings Tech., 142-144, 835-842(2001). 22 Clarke, D.R. and Levi C. G., "Materials Design for the Next Generation Thermal Barrier Coatings", Annu. Rev. Mater. Res., 33, 383-417 (2003). 23 Tu, R. and Goto, T., "Thermal Cycle Resistance of Yttria Stabilized Zirconia Coatings Prepared by MO-CVD", Mater. Trans., 46, 1318-1323(2005).

12

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

PREPARATION OF Ru-C NANO-COMPOSITE FILMS AND THEIR ELECTRODE PROPERTIES FOR OXYGEN SENSORS Teiichi Kimura and Takashi Goto Institute for Materials Research, Tohoku University 2-1-1 Katahira, Aoba Sendai, Miyagi, Japan 980-8577 ABSTRACT Ru-C nano-composite films containing about 73 vol% of carbon were prepared by MOCVD, and their microstructures and electrode properties were investigated. Ru particles of 5-20 nm in diameter were dispersed in amorphous C matrix. The AC conductivities associating to the interface charge transfer between Ru-C composite electrode and YSZ electrolyte were 100-1000 times higher than that of Pt electrodes. The emf values of the oxygen gas concentration cell constructed from the nano-composite electrodes and YSZ electrolyte showed the Nernstian theoretical values at low temperatures around 500 K. The response time of the concentration cell was 900 s at 500 K. INTRODUCTION Solid-electrolyte type oxygen sensors are widely used for monitoring oxygen concentration in exhaust gas of automobiles and chemical plants because of their relatively simple configuration and direct indication of oxygen content in ambient atmosphere. This type of oxygen sensors is mainly constructed from a solid electrolyte and electrodes. Yttria-stabilized zirconia (YSZ) is commonly used as a solid electrolyte due to high ionic conductivity and mechanical strength. Electrodes should have high electronic conductivity, high chemical/thermal stability and catalytic activity for the dissociation of oxygen molecules "3. Since platinum group metals, particularly Pt, could satisfy these requirements, Pt electrodes have been generally applied to the oxygen sensors. The operation temperature of usual Pt/YSZ/Pt sensor is above 1000 K due to low catalytic activity of Pt and slow charge transfer at electrode/electrolyte/gas triple points at low temperatures. Since the low temperature operation of oxygen sensors is strongly required, a new electrode material with high catalytic activity at low temperatures should be developed. Metal-organic chemical vapor deposition (MOCVD) can be suitable for preparing electrodes because of its controllability of microstructure of films by changing deposition conditions. Many kinds of metal films have been prepared by MOCVD, in which impurity C has been often contained degrading the electrical conductivity 4 ' 5 . On the other hand, the carbon phase has an advantage to hinder the grain growth of metals, and to form metal nano-particle dispersed composite films. The co-deposited C would often enhance the catalytic activity as reported in Pt-C catalysts. Thus, metal-C nano-composite electrodes having high catalytic activity can be prepared by MOCVD. In this study, Ru-C composite electrodes were prepared by MOCVD, and their microstructures and electrode properties were investigated.

13

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

Table 1 Deposition conditions Precursor Vaporize temperature [K] Total pressure [kPa] Substrate temperature [K] Ar gas flow rate [10 8 m V ] 0 2 gas flow rate flO"8 m3s"'l

Ru-C Ru(dpm)3 473 0.93 673 33 6.8

EXPERIMENTAL Ru-C films were prepared on silica glass and YSZ(8 mol%Y203-Zr02) substrates using a horizontal hot-wall type MOCVD apparatus . Ru(dpm)3 (dpm: dipivaloylmethanato) was used as precursors. Deposition conditions are summarized in Table 1. Compositions and crystalline phases of films were analyzed by X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD). Microstructures were investigated with a scanning electron microscope (SEM) and a transmission electron microscope (TEM). The electrical properties were studied by AC impedance spectroscopy with a two-probe method in the frequency range between 0.1 Hz and 10 Hz. The oxygen concentration cell was constructed with the Ru-C nano-composite film electrodes and YSZ electrolyte. The electro-motive-force (emf) values were measured at temperatures from 500 to 773 K by changing the oxygen partial pressure ratio from 1 to 5. RESULTS AND DISCUSSION Microstructure Fig.l demonstrates XRD patterns of Ru-C composite films. There are a few narrow peaks assigned to Ru and a broad peak around 26=20°. Average crystalline size of Ru estimated from full width at half maximum of (100), (101) and (110) diffraction peak using the Scherrer's equation was about 8 nm. The AES spectra of the composite films after surface etching by Ar ions for 600 s indicated a significant amount of C in the films. The C contents in the films were estimated by XPS analysis and was 73 vol%. Ru-C composite films consisted of spherical grains of 50 nm in diameter as shown in Fig.2. Fig.3 shows TEM images of Ru-C composite films. Dark particles of 5-20 nm in size were dispersed in an amorphous matrix without pore or gap at the boundary. Hereafter, these films are mentioned as Ru-C nano-composite. Electrode properties Fig.4 depicts the AC impedance spectrum of YSZ with Ru-C nano-composite electrodes at 773 K. Two semicircles near the original point could be assigned to bulk and grain boundary responses of YSZ substrate, because they were independent of electrodes. The associated capacitances were 5.8 pF and 0.11 nF, respectively, close to reported values 7. The third

14

• Novel Processing of Ceramics and Composites

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

10

20

30

40

50

60

70

80

29 (CuKa) / deg.

Fig.l XRD pattern of Ru-C composite film

Fig.2 Surface SEM image of Ru-C composite film.

Fig.3 TEM images of Ru-C composite films, (b) is higher magnification of (a).

Novel Processing of Ceramics and Composites

• 15

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

200

E 5.8x10"12F g 100 1.1x10-10F Kl 100

1.3x10"7F

1

o ° ° °

200

300

400

Z'/Qm

Fig.4 AC impedance spectrum for Ru-C electrode deposited on YSZ electrolyte at 773 K. I¿

1 10 ^

8

■f

fi

.

• 473 K o 513 K

y

▼ 533 K V 563 K

¿s

y

CO

c

(D ■o

4 _ •*-* c
ü

0 - 4&

0

i

i

i

i

10 20 30 40 Applied voltage, Vapp/ kV nr 1

50

Fig.5 Current-voltage characteristics of Ru-C composite electrode at various temperatures. semicircle in low frequency region was assigned to the response of charge transfer at the electrode/YSZ interface due to a large capacitance of 0.1 u.F. The interfacial semicircle was partially drawn below 600 K, because the electrical resistivity of YSZ becomes too high and the frequency range was not enough to obtain the whole semicircles. The current-voltage characteristics (Fig.5) were investigated to measure the whole resistivity, and the interfacial conductivity was estimated by subtracting the bulk and grain boundary resistivities of substrate from the whole resistivity. Fig.6 summarizes temperature dependence of the interfacial conductivity. The interfacial conductivity of Ru-C nano-composite electrode was 1000-10000 times higher than that of reported Pt electrode 8. The high interfacial conductivities of the nano-composite electrodes

16

• Novel Processing of Ceramics and Composites

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

Temperature, 77 °C 500

-3 i

E

ä

400

1

300

r-

200

O Ru-C(th¡s work) ▲ Pt (Badwalle, 1979)

-5 -6

o

-A_

-7 -8

*^ f

1.2

1.4

- * K

1.6 1.8 7-i/10-3K-i

2.0

2.2

Fig.6 Temperature dependence of Ru-C/YSZ intertacial conductivity.

>

30 - ' " '

E

UJ U_~

20

. ' ' '

O

O

UJ 10

n

400

o ■

500

Ru-C(This work) Theoretical <

600

Temperature, TI K

700

Fig.7 EMF values of the oxygen concentration cell using the Ru-C nano-composite electrodes at 500 K.

Novel Processing of Ceramics and Composites • 17

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

20

> E

$ H" 10 LU

0

0

1

2

3

4

5

Time, f/ks Fig.8 Time response of oxygen concentration cell using Ru-C nano-composite electrodes at 500 K. could suggest the high catalytic activities of Ru nano-particles. The high interfacial conductivity of Ru-C nano-composite could be mainly caused of the large effective surface area of Ru particles in the nano-composite film without aggregation as shown in Fig, 2. Fig.7 shows the emf values of the oxygen gas concentration cells using the nanocomposite electrodes. The Ru-C nano-composite electrodes showed the theoretical values even at 500 K. Fig.8 demonstrates the time response of the oxygen gas concentration cells using Ru-C nano-composite electrodes. The response time of Ru-C electrode was 900 s at 500 K. CONCLUSION Ru-C nano-composite films containing about 73 vol% of carbon were prepared by MOCVD. Ru particles of 5-20 run in diameter were dispersed in amorphous C matrix. The AC interface electrical conductivities for Ru-C nano-composite electrodes were 1000-10000 times higher than that of reported Pt electrode. The emf values of the oxygen gas concentration cell constructed from Ru-C nano-composite electrodes showed the Nernstian theoretical values even at 500 K. The response time of the concentration cell was 900 s at 500 K for Ru-C nano-composite electrodes.

18

• Novel Processing of Ceramics and Composites

Preparation of Ru-C Nano-Composite Films and their Electrode Properties

ACKNOWLEDGEMENT This work has been financially supported by Japan Atomic Energy Research Institute, Furuya metal co., ltd., Japan, and Lonmin PLC, UK. REFERENCES 1

Green, M. L., Gross, M. E., Papa, L. E., Schnoes, K. J. and Brasen, D.," Chemical Vapor Deposition of Ruthenium and Ruthenium Dioxide Films" J. Electrochem. Soc, 132, 26772684(1985). 2 So, F. C. T., Kolawa, E., Zhao, X. -A., Pan, E. T. -S. and. Nicolet, M. -A., "Reactively sputtered Ru0 2 and Mo-0 diffusion barriers",/ Vac. Sei. Technol, B5 , 1748-1749(1987). 3 Kolawa, E., So, F. C. T., Pan, E. T. -S. and Nicolet, M. -A., " Reactively sputtered Ru02 diffusion barriers", Appl. Phys. Lett., 50, 854-855(1987). 4 Rand, M. J.," Plasma-promoted deposition of thin inorganic films", J. Electrochem. Soc, 16, 420-427(1979). 5 Zhen, W., Vargas, R., Goto, T., Someno, Y. and Hirai, T." Preparation of epitaxial A1N films by electron cyclotron resonance plasma-assisted chemical vapor deposition on Ir- and Pt-coated sapphire substrates", Appl. Phys. Lett., 64, 1359-1361(1994). 6 Goto, T., Ono, T. and Hirai, T., "Electrochemical Properties of Amorphous Carbon/Nanogranular Iridium Films Prepared by MOCVD", J. Jpn. Soc. Powder and powder Metallurgy, 47, 386-390(2000). 7 Irvine, J. T. S., Sinclair, D. C. and West, A. R., " Electroceramics: Characterization by Impedance Spectroscopy", Adv. Mater., 2 , 132-138(1990). 8 Badwal, S. P. S. and Bruin, H. J. de, " Electrode Kinetics at the Pt/Yttria-Stabilized Zirconia Interface by Complex Impedance Dispersion Analysis", Phys. Stat. Sol., (a)54,261-270(1979).

Novel Processing of Ceramics and Composites

• 19

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Combustion Synthesis

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

SYNTHESIS OF Sitio 5Sr0 5Co03.x AND Lao 6Sr0 4Co03-x NANOPOWDERS BY SOLUTION COMBUSTION PROCESS Narottam P. Bansal National Aeronautics and Space Administration Glenn Research Center Cleveland, OH 44135 Zhimin Zhong QSS Group, Inc. NASA Glenn Research Center Group Cleveland, OH 44135 ABSTRACT Nanopowders of Smo.sSro.sCoOs-x (SSC) and Lao 6Sr0 4Co03.x (LSC) compositions, which are being investigated as cathode materials for intermediate temperature solid oxide fuel cells, were synthesized by a solution-combustion method using metal nitrates and glycine as fuel. Development of crystalline phases in the as-synthesized powders after heat treatments at various temperatures was monitored by x-ray diffraction. Perovskite phase in LSC formed more readily than in SSC. Single phase perovskites were obtained after heat treatment of the combustion synthesized LSC and SSC powders at 1000 °C and 1200 °C, respectively. The as-synthesized powders had an average particle size of -12 nm as determined from x-ray line broadening analysis using the Scherrer equation. Average grain size of the powders increased with increase in calcination temperature. Morphological analysis of the powders calcined at various temperatures was done by scanning electron microscopy. 1. INTRODUCTION Solid oxide fuel cells (SOFC) are being considered1 as the premium power generation devices in the future as they have demonstrated high energy conversion efficiency, high power density, extremely low pollution, in addition to flexibility in using hydrocarbon fuel. A major obstacle for commercial applications of SOFC still is high cost, both in terms of materials and processing. Intermediate Temperature Solid Oxide Fuel Cell (IT-SOFC) operated between 500~800°C, which allows utilization of available and inexpensive interconnects and sealing materials, can significantly reduce the cost of SOFC. The IT-SOFC also will have better reliability and portability. To keep up with the performance of traditional SOFC that operates between 900-1000°C, new materials with improved performance have to be used2'3. To enhance the oxygen ion conductivity of the electrolyte at the reduced temperature, Lai.xSrxGai.yMgyOz (LSGM), scandium stabilized zirconia or lanthanum (gadolinium, samarium) doped ceria can be used to replace the yttrium stabilized zirconia. Similarly, cathode materials with higher performance at the lower temperature such as Smo.5Sro.5Co03.x (SSC), Lao.6Sr0.4Co03_x (LSC), Lao gSr0 2Coo.2Feo.803.x (LSCF) will be used to substitute La^ySryMnOs.x (LSM), the performance of which decreases rapidly when the operating temperature is below 800°C. The primary objective of this study was to synthesize fine powders of SSC and LSC compositions for applications as SOFC cathodes. A number of approaches such as, solid state reaction, sol-gel, hydrothermal, spray-drying, freeze-drying, co-precipitation, and solution

23

Synthesis of Sm0 5Sr0 5Co03_x and La0.6Sr0 4Co03_x Nanopowders by Solution Combustion

combustion have been used for ceramic powders processing. The solution-combustion method is particularly useful in the production of ultrafine ceramic powders of complex oxide compositions in a relatively short time. This approach has been utilized4"10 for the synthesis of various oxide powders such as ferrites, chromites, manganites, Ni-YSZ cermet, zirconates, doped ceria, hexaaluminates, pyrochlores, oxide phosphors, spinels, etc. An amino acid such as glycine is commonly used as the fuel in the combustion process. However, urea, citric acid, oxylydihydrazide, and sucrose have also been recently utilized6-10 as complexing agents and fuel in the combustion synthesis. In the present study, SSC and LSC cathode powders were synthesized using the glycinenitrate solution-combustion technique4"6 because of its high energy efficiency, fast heating rates, short reaction times, and high reaction temperatures. This process is also unique as all the reactants are mixed in solution at the molecular level resulting in homogeneous reaction products and faster reaction rates. Development of crystalline phases in the powders, on heat treatments at various temperatures, was followed by powder x-ray diffraction. Morphology of the powders was characterized by scanning electron microscopy (SEM). 2. EXPERIMENTAL METHODS 2.1. Powder Synthesis: The starting materials used in the synthesis were metal nitrates Sm(N03)3.6H20 (99.9 % purity), La(N03)3.6H20 (99.9% purity), Sr(N03)2 (98 % purity), Co(N03)2.6H20 (97.7 % purity) and glycine (NH2CH2COOH, 99.5 % purity), all from Alfa Aesar. A flow chart showing the various steps involved in the synthesis of powders by the solution-combustion process is shown in Fig. 1. Metal nitrates are employed both as metal precursors and oxidizing agents. Stoichiometric amounts of the metal nitrates, to yield 10g of the final SSC or LSC oxide powder, were dissolved in deionized water. A calculated amount of the amino acid glycine (0.7 mole per mole of NO3") was also dissolved in deionized water. The glycine solution was slowly added to the metal nitrate aqueous solution under constant stirring. Glycine acts as a complexing agent for metal cations of varying sizes as it has a carboxylic group at one end and an amino group at the other end. The complexation process increases the solubility of metal ions and helps to maintain homogeneity by preventing their selective precipitation. The resulting clear and transparent red colored solution was heated on a hot plate until concentrated to about 2 mole/liter on metal nitrate basis. While the solution was still hot, it was added drop wise to a 2 liter glass beaker that was preheated between 300~400°C. The water in the solution quickly evaporated, the resulting viscous liquid swelled, auto-ignited and initiated a highly exothermic self-contained combustion process, converting the precursor materials into fine powder of the complex oxides. Glycine acts as a fuel during the combustion reaction, being oxidized by the nitrate ions. Oxygen from air does not play an important role during the combustion process. The overall combustion reactions can be represented as: 0.6 La(N03)3 + 0.4 Sr(N03)2 + Co(N03)2 + 3.2 H2NCH2COOH + (1.8 - x/2) 0 2 — Lao 6Sr0 4Co03.x + 6.4 C0 2 + 8 H 2 0 + 3.9 N2

(1)

0.5 Sm(N03)3 + 0.5 Sr(N03)2 + Co(N03)2 + 3.2 H2NCH2COOH + (1.95 - x/2) 0 2 -+ Smo5Sro5Co03.x + 6.4 C0 2 + 8 H 2 0 + 3.85 N2

(2)

24

• Novel Processing of Ceramics and Composites

Synthesis of Sm06Sr05Co03_x and La0.6Sr0 ^CoCv,, Nanopowders by Solution Combustion

Nitrates of La, Sm, Sr, Co + water

Glycine + water

Mix metal nitrates and glycine solutions under stirring

D

G

Clear red solution; heat at -80 °C; concentrate to ~2M metal nitrate basis

Add above solution dropwise to a beaker preheated to 300-400 °C

Black powder; heat treat 700-1300 °C, 2 h each, in air

I

f XRD, SEM ) Figure 1 —Flow chart for solution-combustion synthesis of Lan.eSrn 4CoC>3.x and Smrj.sSrn 5CoC>3-x nanopowders

indicating the formation of CO2, N2, and H2O as the gaseous products. The evolution of gases during the combustion process helps in the formation of fine ceramic powder by limiting the inter-particle contact. The resulting black powder contained some carbon residue and was further calcined to convert to the desired product. Small portions (~1 g) of this powder were heat treated in air at various temperatures between 700 and 1300°C for two hours to study the development of crystalline phases. 2. 2. Characterization Thermal gravimetric analysis (TGA) of the powders was carried out using a PerkinElmer Thermogravimetric Analyzer 7 system which was interfaced with computerized data acquisition and analysis system at a heating rate of 10 °C/min. Air at 40 ml/min was used as a purge gas. X-ray diffraction (XRD) analysis was carried out on powders heat treated at various temperatures for crystalline phase identification and crystallite size determination. Powder XRD patterns were recorded at room temperature using a step scan procedure (0.02720 step, time per step 0.5 or 1 s) in the 20 range 10-70" on a Philips ADP-3600 automated diffractometer equipped with a crystal monochromator employing Cu K„ radiation. Microstructural analysis was carried out using a JEOL JSM-840A scanning electron microscope (SEM). Prior to analysis, a thin layer of Pt or carbon was evaporated onto the SEM specimens for electrical conductivity.

Novel Processing of Ceramics and Composites

•

25

Synthesis of Smo.5Sr05Co03_x and Lao.eSro.iCoCvx Nanopowders by Solution Combustion

3. RESULTS AND DISCUSSION 3.1. Thermogravimetric Analysis Figure 2 shows the TGA curves recorded at a heating rate of 10'C/min in air from room temperature to 1200°C for the as-synthesized LSC and SSC powders using the solutioncombustion method. For both precursors, about 6% weight loss was observed between 600 to 850°C that was likely due to loss of carbon residue by oxidation and also from decomposition of SrCOi. For SSC, there was additional 1% weight loss between 850 to 1000°C for which there is no simple explanation based on the x-ray diffraction results of Figure 4.

102 c o

100

I 98 8

S> 96

I

1

O) to

94

I 92 Q.

90 88

0

200

400

600 800 Temperature, "C

1000

1200

Figure 2—TGA curves of as-synthesized precursor powders by solutioncombustion method for Lao 6Sro.4Co03_x and Smo.sSro.sCoOs.,, at a heating rate of 10 °C/min in air.

3.2. Phase Formation and Microstructure Both the LSC and SSC as-synthesized powders were calcined in air for two hours at various temperatures between 700 to 1300 °C to investigate the evolution of crystalline phases. X-ray diffraction patterns for these heat treated LSC and SSC powders are shown in Figs. 3 and 4, respectively and the results are summarized in Table I. The as-prepared LSC powder shows weak crystallinity of the perovskite phase. SrCOj phase was also observed in the as-synthesized powder and after calcination at 700 °C. An unknown peak at 32° (probably Sr3Co2Û6 13, 83-375) appeared for the powder calcined at 800 and 900 °C. Formation of the perovskite phase, Lao.6Sro4Co03.x, is completed above 1000°C as observed by XRD results in Fig. 3. The asprepared SSC powder showed the presence of S1ÏI2O3, C03O4, and SrC03 phases. The desired Smo.jSro.5Co03.x perovskite phase emerged as the major phase after the powder was calcined at

26

• Novel Processing of Ceramics and Composites

Synthesis of Sm 0 5Sr0 5 Co0 3 _ x and La0 6Sr0 4Co03_x Nanopowders by Solution Combustion

Figure 3.—X-ray diffraction patterns of Lao.6Srrj 4C0O3-X powders made by solution-combustion synthesis after heat treatments at various temperatures for 2 h in air.

Figure 4.—X-ray diffraction patterns of Smo.5Srrj.5Co03-x powders made by solution-combustion synthesis after heat treatments at various temperatures for 2 h in air.

Novel Processing of Ceramics and Composites

•

27

Synthesis of Sm0.5Sr0.5CoO3_x and La0.6Sr0.4CoO3_x Nanopowders by Solution Combustion

700 °C. Secondary phases such as Sr3Co20613 remained even after the powder was heat treated at 1100 °C. Perovskite phase-pure Smo.sSro sCo03_x powder was obtained after heat treatment at 1200° C for 2 hours. Earlier investigation7 of SSC synthesis by solid-state reaction method indicated that the perovskite phase was formed after calcination at 1200°C for 6 hours. The products calcined at this temperature will have low porosity and non-ideal microstructure as cathode materials. Table. I. X-ray diffraction analysis of Sm0 5Sr0 5C0O3., and Lao éSr0 4Co03.x powders made by solution-combustion synthesis after heat treatments at various temperatures in air System

Lao 6 Sr 0 4 Co0 3 . x

m0 5 Sr 0 sCo0 3 . x

Heat treatment Time Temp.

CC)

As synthesized

700 800 900 1000 1100 1200 1300

(h)

Crystalline phases"

Average grain size (nm)b

Lao. 6 Sr 04 Co0 3 . x , SrCOj

12

2 2 2 2 2 2 2

Lao 6 Sr 0 4 Co0 3 . x , SrCCB Lao óSro 4 Co0 3 _ x , low intensity peak at 32" 28 Lao iSro 4 Co0 3 . x , low intensity peak at 32' 28

15 17 28 37 50

2 2 2 2

Smo 5 Sr 0 5 Co0 3 . x , SrC0 3 , Co 3 0 4 Sm 0 5 Sr 0 5 Co0 3 _ x , Sr 3 Co 2 0 6 , 3 , Co 3 0 4

As synthesized 700 800 900 1000

—

1100

2

Lao óSro 4 Co0 3 . x Lao6Sro4Co0 3 . x Lao6Sro4Co03_x Lao6Sr 0 4Co0 3 - x Sm 2 0 3 , Co 3 0 4 , SrC0 3

Sm0 sSr0 sCo0 3 . x , Sr3Co20613 Sm 0 sSr0 sCo0 3 . x , Sr3Co20,s l3 low intensity peak at 32" 28 Smo sSro sCo0 3 . x , Sr 3 Co 2 0 6 , 3 low intensity peak at 32° 26 Smo5Sr 05 Co0 3 . x Sm 0 5Sro5Co0 3 . x

— 15 15 25 38 41

2 2 "Phases in decreasing order of peak intensity b Calculated from Scherrer formula using FWHM of XRD peak in 47-48° range of 29. 1200 1300

The SEM micrographs of Lao óSro 4Co03.x and Smo 5Sr0 5Co03.x powders made by solution-combustion synthesis after heat treatments at different temperatures for 2 h in air are presented in Figures 5 and 6, respectively. The as prepared powders were highly porous and particles were linked together in agglomerates of different shapes and sizes. Substantial particle growth was observed upon calcination for two hours at 1000°C or higher temperatures. The particle size of samples calcined at 1000°C increased but the structure remained highly porous, which resembled the typical cathode structure for SOFC. Therefore, LSC and SSC powders

28

• Novel Processing of Ceramics and Composites

Synthesis of Smo.5Sro.5Co03_x and La06Sr0 ^CoO^,, Nanopowders by Solution Combustion

should be sintered around 1000"C for fabrication of cathode structures. After calcination at 1200°C. LSC became dense and lost porosity. SSC powder sintered into a dense pellet following heat treatment at 1200°C.

Figure 5.—SEM micrographs of Lao.6Sro.4Co03_x powders made by solution-combustion synthesis after heat treatments at different temperatures for 2 h in air.

3.3. Particle Size Analysis After each heat treatment of the as synthesized LSC and SSC powders, the average particle size was evaluated from X-ray line broadening analysis using the Scherrer equation": t = 0.9 X/(B cos 9B)

(3)

where t is the average particle size, X the wave length of Cu K<, radiation, B is the width (in radian) of the XRD diffraction peak at half its maximum intensity, and OB the Bragg diffraction angle of the line. Correction for the line broadening by the instrument was applied using a large particle size silicon standard and the relationship

Novel Processing of Ceramics and Composites

• 29

Synthesis of Sm0 5Sr0 5Co03_x and La0 6Sr0 4Co03_x Nanopowders by Solution Combustion

B2 = B2M - B2s

(4)

where BM and B s are the measured widths, at half maximum intensity, of the lines from the sample and the standard, respectively. Values of average grain sizes of the as synthesized SSC and LSC powders and of those after heat treatments at various temperatures are given in Table I.

Figure 6—SEM micrographs of Smn.sSro 5C0O3.X powders made by solution-combustion synthesis after heat treatments at different temperatures for 2 h in air,

The as synthesized powders had an average grain size of about 10-12 nm. A number of factors are responsible for the nanosize of the resulting powders. Before the reaction, all the reactants are uniformly mixed in solution at atomic or molecular level. So, during combustion, the nucleation process can occur through the rearrangement and short-distance diffusion of nearby atoms and molecules. Also, large volume of the gases evolved during the combustion reactions (1) and (2) limits the inter-particle contact. Moreover, the combustion process occurs at such a

30

• Novel Processing of Ceramics and Composites

Synthesis of Sm0 5Sr0 5Co03_x and La0 6Sr0 ^CoO^* Nanopowders by Solution Combustion

fast rate that sufficient energy and time are not available for long-distance diffusion or migration of the atoms or molecules which would result in growth of crystallites. Consequently, the initial nanosize of the powders is retained after the combustion reaction. The X-ray line broadening method can be used only for the size determination of small crystallites (< 100 nm). The values obtained are not the true particle size, but the average size of coherently diffracting domains; the latter being usually much smaller than the actual size of the particles. The crystallite size of the as-synthesized powder depends8'9 on the glycine to nitrate ratio used during the combustion synthesis. Powder made using a fuel-deficient system has the highest surface area. The powder surface area decreases as the glycine to nitrate ratio is increased. This has been attributed to an increase in the flame temperature during combustion which helps in the growth of crystal size. The average grain size of the SSC and LSC powders increased (Table I) with the increase in calcination temperature, as expected. 4. SUMMARY AND CONCLUSIONS Nanopowders of Smo.sSro 5C0O3.X (SSC) and Lao.öSro 4Co03.x (LSC) cathode materials for solid oxide fuel cells have been synthesized by the glycine-nitrate solution-combustion method. Formation of crystalline phases in both the powders started at relatively low temperatures. However, the as-synthesized powders had to be calcined at or above 1000 °C to yield phase pure perovskite products. The high temperature calcination caused significant reduction in surface area, coarsening of the powders, and sintering which is not favorable for forming the cathode structures for SOFC. The investigations of electrochemical activity of these materials and co-sintering with fuel cell electrolytes are being investigated and will be presented in the future. ACKNOWLEDGMENTS Thanks are due to Ralph Garlick for X-ray diffraction analysis. This work was supported by Low Emissions Alternative Power (LEAP) Project of the Vehicle Systems Program at NASA Glenn Research Center. REFERENCES 1. N. Q. Minh, Ceramic Fuel Cells, J. Am. Ceram. Soc., 76_[3], 563-588 (1993). 2. D. Stover, H.P. Buchkremer, S. Uhlenbruck, Processing and Properties of the Ceramic Conductive Multilayer Device SOFC, Ceram. Int., 30 [7], 1107-1113 (2004). 3. Y. Liu, S. Zha, M. Liu, Adv. Mater., 16 [3], 256-260 (2004). 4. S.-J. Kim, W. Lee, W.-J. Lee, S. D. Park, J. S. Song, and E. G. Lee, Preparation of Nanocrystalline Nickel Oxide-Yttria-Stabilized Zirconia Composite Powder by Solution Combustion with Ignition of Glycine Fuel, J. Mater. Res., 16 [12], 3621-3627 (2001). 5. L. A. Chick, L. R. Pederson, G. D. Maupin, J. L. Bates, L. E. Thomas, and G. J. Exarhos, Glycine-Nitrate Combustion Synthesis of Oxide Ceramic Powders, Mater. Lett., 10, 6-12 (1990). 6. M. Marinsek, K. Zupan, and J. Maeek, Ni-YSZ Cermet Anodes Prepared by Citrate/Nitrate Combustion Synthesis, J. Power Sources, 106, 178-188 (2002). 7. T. Ishihara, M. Honda, T. Shibayama, H. Minami, H. Nishiguchi, Y. Takita, Intermediate Temperature SOFCs Using a New LaGaÛ3 Based Oxide Ion Conductor. I. Doped SmCo03 as a New Cathode Material, J. Electrochem. Soc, 145 [9], 3177-3183(1998).

Novel Processing of Ceramics and Composites

• 31

Synthesis of Sm 0 5Sr0 5Co03_x and La0 6Sr0 4CoC>3_x Nanopowders by Solution Combustion

8. R. D. Purohit, S. Saha, and A. K. Tyagi, Nanocrystalline Thoria Powders via GlycineNitrate Combustion, J. Nuclear Mater., 288 [1], 7-10 (2001). 9. T. Ye, Z. Guiwen, Z. Weiping, and X. Shangda, Combustion Synthesis and Photoluminescence of Nanocrystalline Y2O3: Eu Phosphors, Mater. Res. Bull., 32, 501 (1997). 10. K. Prabhakaran, J. Joseph, N. M. Gokhale, S. C. Sharma, and R. Lai, Sucrose Combustion Synthesis of LaxSr(i.x)Mn03 (x < 0.2) Powders, Ceram. Int., 31 [2], 327-331 (2005). 11. B. D. Cullity, Elements ofX-Ray Diffraction, T* Edition, Addison-Wesley, Reading, MA, p. 284(1978).

32

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

CHEMISTRY PURIFICATION OF TITANIUM DIBORIDE POWDER SYNTHESISED BY COMBUSTION SYNTHESIS PROCESSES Wang Weimin* Fu Zhengyi Wang Hao State Key Lab of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology Wuhan 430070, P.R.China ABSTRACT: Combustion synthesis method with reductive process was used to synthesize TÍB2 powder from TÍO2 — B2O3 — Mg system. Main impurity phase can be cleaned out by acid washing treatment. The influences of acid washing conditions (acid concentration, treatment time, treatment temperature and so on) on the purity of TÍB2 powder were studied. Results showed that acid treatment temperature and amount of excessive acid were the main factors that effected purity of TiB2 powder. With increase of acid treatment temperature and amount of excessive acid, purity of TiB2 powder increased continuously. The analysis of thermodynamics and kinetics showed that the increase of acid treatment temperature could increase chemistry reaction speed constant (K), and accelerate acid washing processing. Diffusion mechanism controlled mainly acid treatment reaction. 1. INTRODUCTION TiB2 Ceramic has excellent physcc—chemical properties such as high melting point, high hardness, good corrosion resistance and high temperature mechanical properties, TiB2 ceramic has also an excellent electricity conductivity, which make TÍB2 materials have a wide application areas as advanced engineering ceramicsl1'2'. Traditional fabrication methods of TiB2 ceramic powder such as carbon -thermal reduce processing need a longer thermal treatment time and consume a great deal of energy, and however, this technology could not produce the TÍB2 powder with high purity, small particle size and lower price. Combustion Synthesis technology is a novel materials synthesis and processing technology which is receiving more and more interests in recent decades because of the following advantages: simple processes, less energy consume, high production efficiency and so on '3'. Combustion Synthesis method has been employed to synthesise TÍB2 ceramic powder from element Boron and Titanium '4|, but due to the high cost of element boron and titanium, the commercial production of TÍB2 ceramic is impossible by this method. Combustion Synthesis of TÍB2 ceramic powder from various oxides such as Boron Oxide and Titanium Oxide is a possible way to obtain TÍB2 ceramic with low costing and high purity'5'. In this processing, one of the key steps is the acid washing treatment to clean away impurity , which decide the final purity of TiB2 ceramic powder. In this paper, acid washing conditions effect the purity of TiB2 powder were studied, the thermodynamics and kinetics of acid washing processing were discussed.

Corresponding author: State Key Lab of Advanced Technology for Materials Synthesis and Processing Wuhan University of Technology, Wuhan 430070, P. R. China, Fax: 00862787879468, E-mail: [email protected]

33

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

2. EXPERIMENTAL The raw materials used in this paper were Mg powder, TÍO2 and B2O3 powder. Powder characteristics were shown in Table 1. TÍB2 powder can be synthesized according to following chemistry reaction: Ti0 2 + B2O3 +5 Mg = TiB2+ 5MgO

(1)

The main impurity in TÍB2 powder is MgO, which can be cleaned away by acid washing .The raw powders were mechanically mixed in dry condition after weighing .The mixture powder were axially compacted into cylinders, and then was putted into combustion synthesis chamber. Synthesis temperature and synthesis processing were monitored by means of an infrared thermometer and high—speed video camera, respectively. The ignition head is a tungsten coil, which can provide an ignition temperature in the range of 2000--3000 K by the changing the voltage. Specimens were synthesised in an argon atmosphere. After combustion synthesis, the synthesized production was ball milled for 1 hour. Ball milled fine powder was soaked in hydrochloric acid solution. After a certain time of soaking, the powder was cleaned using distilled water a few times, after that, filtering and vacuum-dry processing were followed. X-ray diffraction quantitative analysis method was used to measure the residual MgO content in TÍB2 powder. Scanning electronic microscopy was employed to reveal the microstructure of powder. 3. RESULTS AND DISCUSSION 3.1 Influence of combustion synthesis parameters on the mineral phase component of synthesized sample Literature[6) showed that combustion synthesis parameters such as density of sample, synthesis pressure, and diluent have great influences on processing characteristics of combustion synthesis and component of synthesized sample. Literature [7' showed that synthesized production consisting of only TÍB2 and MgO cannot be obtained according to the standard stoichiometric (Ti02: B203: Mg = 1:1:5), but the suitable component of reactant mixture have not been suggested. From analysis of combustion synthesis processes of this reaction system, the synthesis reaction consists of two steps. Firstly, metal Mg reduced B2O3 and TÍO2 to form element B and Ti, secondly, TÍB2 was synthesized by chemical combination of two monomer element. Obviously, in these reducing processes, the reduced degree of oxides has a direct influence on the mineral phase composition of synthesised product, metal Mg play an important role in controlling the reduced degree and phase composition. Study showed that when reducer Mg is not enough for oxides, 3MgO B2O3 appears in synthesised production according to following chemistry reaction: 3MgO+B203=3MgO B2O3

(2)

Fig.l. showed the influence of amount of Mg in reactant mixture on content of 3MgOB203 in synthesised sample. With increase of the Mg content, the content of 3MgO B2O3 decreased

34

• Novel Processing of Ceramics and Composites

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

remarkably, when the TÍO2: B2Û3: Mg is equal to 1:1:7, 3MgO B2O3 almost could not been detected. Figure 2 showed the X-ray diffraction result, the diffraction peak intensity of 3MgO B2O3 is very low. 3.2 Effects of acid washing conditions on the purity of TÍB2 powder Influences of acid concentration and amount of excessive acid on the residual MgO content in TÍB2 powder were shown in Fig.3. It can be found from Fig.3.a that under the same treatment temperature and time, the residual MgO content decreased with increase of acid concentration. This indicated that increased acid concentration was helpful to clean out MgO. When the acid concentration is 2M, the residual MgO content is less than 20wt%, but further increase of acid concentration seem to have no obvious influence on the residual MgO content. When acid concentration is fixed, the increase of amount of excessive acid resulted in the decrease of residual MgO content (Fig3.b). When the amount of excessive acid surpassed 50vol%, the residual MgO content in TÍB2 powder did not decrease obviously with continuous increase of amount of excessive acid. The results showed in Fig.3 indicated that it is impossible to get high pure TiB2 powder at room temperature. In order to obtain TÍB2 powder with high purity, it is necessary to increase acid washing temperature and time. Fig.4 showed the influence of acid washing temperature and time on residual MgO content in TÍB2 powder. From Fig.4.a, it can be seen that the residual MgO content decreased with increase of acid washing time. At room temperature (298K), when acid treatment time was more than 15hrs, the residual MgO content seems to no longer decrease, and the residual MgO content is about 15—18wt%, however, when acid washing temperature was raised to 333K, residual MgO content will decrease continuously and quickly with increase of acid washing time. MgO content in TiB2 powder can be respectively decreased to less than 8wt% and 2wt% after acid washing 2 and 5 hours. Increase of acid washing temperature can accelerate remarkably the acid washing processing, and shorten acid time, which have been proved also by experiment results shown in Fig.4b. Certainly, too high acid washing temperature is not better, which cause the increasing of residual MgO content due to volatilization of hydrochloric acid at high temperature. Scanning electronic microscopy was employed to observe the microstructure of TÍB2 powder. Fig .5 showed the morphology of pure TÍB2 powder, which is fine, uniform and agglomerate free. 3.3 DISCUSSION 3.31. The thermodynamics analysis In the combustion synthesis of TÍO2 ~ B2O3 — Mg system, the synthesized powder consists of MgO and TÍB2. During acid washing, the MgO was cleaned out and TiB2 did not take part in any chemistry reaction, the chemistry reaction of acid washing processing can be expressed by following reaction equation: MgO(s) + 2H+ =Mg 2 + + H 2 0

(3)

Novel Processing of Ceramics and Composites

• 35

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

The free energy change of acid washing reaction AG"2i)8 can be written as following: AG°298 = (AG° Ms 2++ AG°H2 o) - (AG°Mgo + AG°H2+)

(4)

According to equation (4), the free energy change of acid washing reaction at room temperature can be calculated, the calculation value is AG°29g = -122.58KJ, which is less than zero and indicates this chemistry reaction can initiate automatically at room temperature. When temperature of reaction system was increased, the system free energy changeAG"T will change. According to G.R. Kirchhoff principle, system free energy changes at different temperature can be calculated by equation (5) AG°T = A G ° 2 9 8 - TAS°2,8 +J"^ 8 ACpdT -T J ^

ACpd(lnT)

(5)

AS°298 is the change of standard entropy, ACp is average thermal capacity. When reaction temperature is 333K (60 °C), the system free energy changes AG" 333= -120.6kJ < 0, which is less than that of room temperature, showing the acid washing reaction is exothermal reaction. Chemistry reaction activation energy can be calculated according to the relationship between reaction free energy change and activation energy. At room temperature T = 298 K. the calculated reaction activation energy E = 83.68 kJ/mol. Chemistry reaction rate constant K is a function of temperature, increasing of temperature will cause change of reaction rate constant. According to Van't Hof theorem, chemistry reaction rate constant at different reaction temperature KT can be calculated: In (KT+iT /KT) = EAT/(R T)

(6)

Where, E is the activation energy, R is gas constant and T is temperature. If reaction temperature increase 10 K, the KT+IO /KT is about 2.7, which means thelO K increasing of reaction temperature will cause about 2.7 times increasing of reaction rate constant, chemistry reaction will also be accelerated significantly. The experimental results agreed well with this calculated result. 3.3.2. The kinetics analysis Acid washing treatment is a solid/liquid reaction process. During acid washing, mass transfer and chemistry reaction take place according to following steps: Transfer of reactant W to solid surface ( MgO) — Adsorption of reactant H+on solid surface ( MgO) — Chemistry reaction on solid surface (MgO) — Separation of reaction production from solid surface (MgO) — Transfer of reaction production to solution. In this multiphase reaction system, the speed of acid washing depends on the slowest steps in acid washing process. In HC1 — MgO reaction system, adsorption and separation is quicker, chemistry reaction MgO + H+ = Mg2+ + H 2 0 is also quicker, the diffusion step is the slowest. Diffusion processes determined the speed of acid washing processing.

36

• Novel Processing of Ceramics and Composites

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

When H ion diffuse to MgO solid surface, a liquid state diffusion layer with a concentration gradation was formed. H + ion diffuses to MgO solid surface through diffusion layer, and takes part in chemistry reaction continuously. The diffusion equation can be written as following: dn/dt - DS(dc/dx)

(7)

Where, D is diffusion coefficient, S is surface area of solid particle, dc/dx is concentration gradation. Concentration of hydrochloric acid on the MgO surface is almost zero due to quick solid/liquid chemistry reaction, and the concentration of outside of diffusion layer is about concentration of hydrochloric acid solution C. So that the diffusion equation can be changed to: dn/dt = DSC/A.

(8)

X is thickness of diffusion layer. Equation (8) showed chemistry reaction speed has a direct proportion with concentration of hydrochloric acid solution C, increasing of hydrochloric acid concentration will accelerate acid washing speed, which have been proved by our experiments. When we suppose MgO particles were uniform distribution in hydrochloric acid solution and the MgO solid particle is a ball with a radius ro? the relationship between the chemistry reaction degree of MgO solid particle and acid washing time can be expressed as following: r0/M Where, M and p are respectively molecular weight of MgO and density. Equation (9) showed that the longer the acid washing time , the higher the reaction degree a , and the lower the residual MgO content in TÍB2. This agrees with experiment results. However, in the combustion synthesis production, some T1B2 particle and MgO particle cohere each other, and even MgO particle was coated by TÍB2 particle. Hence, only prolongation of acid washing time could not clean out residual MgO completely. 4. CONCLUSION From above experiment study and analysis, following conclusion can be summarised: (1) In combustion synthesis processes, Mg plays an important role in controlling the mineral phase component of synthesised product. When the TÍO2 : B2O3 : Mg is proper, there is almost no impurity to be detected in synthesised powder. After acid washing treatment, pure TÍB2 powder can be obtained. (2) In acid washing processes, residual MgO content in TÍB2 powder depends on acid concentration, amount of excessive acid, treatment times, and treatment temperature. With increasing of acid concentration, amount of excessive acid, treatment time, residual MgO content decreased gradually, increase of acid washing temperature accelerated acid washing speed remarkably. (3) Thermodynamics analysis showed that acid washing processes can carry out automatically at room temperature but longer treatment time is needed. Increase of acid washing temperature can quicken this process. (4) Kinetics analysis showed that acid washing process is controlled by diffusion mechanism.

Novel Processing of Ceramics and Composites

• 37

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

ACKNOWLEDGEMENTS The authors are grateful to the project of National Hi-Tech Research and Develop of China for this work under Grant number: 2001AA333020 REFERENCE [1] Rieded Ralf(Ed.),Handbook of Ceramic Hard Materials, Weinheim: Weily-VCH,2000 [2] J.Matsushita, T.Suzuki, A.Sano, Wire Electronical Discharge Machining of TiB2 composites, J. Ceram. Soc. Jap., Vol 100(1992),219-222 [3] R. W.Rice, W.R.Grance and C.Conn, Reactant compact and product microstructure for TiC, TiB2 and TiC/ TiB2 from SPS processing, Ceram. Eng. Sei and Proce, Voll 1, pi 192, pi990-1203 [4] V.I.Yukhvid, Modifications of SHS Processes, Pure and Applied Chemistry, Vol64(7), p977988,1992 [5]Wang Weimin, Fu Zhengyi, Wang Hao, Chemistry Reaction Processes During Combustion Synthesis of B2O3—Ti02—Mg System, Journal of Materials Processing and Technology, Vol () 2002. [6] Pampuch R , Some Fundamental Versus Practical Aspects of Self-propagating Hightemperature Synthesis, Solid State Ionics, Vol 101(NOV),p 899-907,1997 [7] A.G.Merzhanove, Combustion processes that synthesize materials, J. of Mate. Proce. Tech, Vol 56 , p222, 1996

Raw materials

Mg Ti02 B203

38

Table 1 Characteristics of the raw materials powders Purity (%) 98 99 97

Main impurity Si Al Fe

Fe 2 0 3 , Sn0 2 S i 0 2 , Al 2 0 3

■ Novel Processing of Ceramics and Composites

Average particle size (p.m) 70 1 70

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

Fig. 1. Influence of amount of Mg in reactant mixture on the content of 3MgO B2O3 in synthesised sample

Fig.2 .X-ray diffraction pattern of synthesised product when Ti0 2 : B203; Mg (mole) is 1:1:7

Novel Processing of Ceramics and Composites

• 39

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

Fig.3. Influences of acid concentration (a) and amount of excessive acid (b) on the residual MgO content in T1B2 powder

40

• Novel Processing of Ceramics and Composites

Chemistry Purification of Titanium Diboride Powder Synthesised by Combustion Synthesis

Fig.4 .Influence of acid washing time (a) and acid washing temperature (b) on Residual MgO content in TÍB2 powder.

Novel Processing of Ceramics and Composites

• 41

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Reaction Forming

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

FABRICATION OF SILICON CARBIDE FROM BAMBOO CARBON TEMPLATES Yung-Jen Lin and Yi-Hsiang Chiu Dept. of Materials Engineering, Tatung University 40 Chungsan North Road, section 3 Taipei, Taiwan, 10451 ABSTRACT Porous ß- SiC with cellular structure was prepared using Moso bamboo as the bio-template. The bamboo was pyrolyzed into carbon and infiltrated with Si vapor at temperature ~ 1500°C to react to form SiC. The highly anisotropic structure of bamboo was faithfully retained after the reaction with negligible dimensional changes. The reaction depths were linearly proportional to the square roots of reaction time. After complete conversion into SiC, the bamboo-like SiC had bulk density of 1.00 g/cm3, open porosity of 66.7% and BET surface area of 18 m2/g. Due to the anisotropy of the structure, the reaction rate was anisotropic. Furthermore, the strengths of the bamboo-like SiC also showed difference in different directions. The compressive strengths were 114 MPa and 73 MPa and the bending ones were 12.8 MPa and 42.5 MPa in directions parallel and perpendicular to the growth direction, respectively. When compressed parallel to the growth direction, the samples showed progressive fracture with serrated stress-strain curves after maximum strength. In contrast, when the samples were compressed perpendicular to the growth direction, they showed abruptfractureat maximum strength. INTRODUCTION It has become an attractive field to fabricate ceramics by mimicking the structure of bio-organics.1" Porous ceramics and ceramic composites have been successfully synthesized using various bio-materials such as bones, shells, fish scales, rice husk and wood.4"6 The products are mostly carbides due to the fact that bio-materials consisted mainly of carbon. The bio-carbon is used as a reactant as well as a template in the reaction synthesis of carbide. During the synthesis, the structure of the bio-materials is retained. Nevertheless, oxide ceramics could also be fabricated using bio-materials as the templates via sol-gel processing.7 So far, the most widely used bio-materials for carbide synthesis are various kinds of wood. The appealing features of ceramics from wood lie in the uniqueness of the wood structures as well as its availability.8 The wood structure is cellular, anisotropic with fibrous structure. It is believed that the structures of wood should have reached optimal configuration after millions of years of natural evolution. Consequently, porous ceramics having the structure of wood should be also optimal in the strength. The cellular, porous ceramics acts as an ideal back-bone for

45

Fabrication of Silicon Carbide from Bamboo Carbon Templates

composites. When it is infiltrated with molten metals, metal-ceramic composites can be obtained. The reaction of wood charcoal with silicon or silicon oxide to form silicon carbide is well established.''3'6,9 Different kinds of wood, ebony, beech, oak, etc, would have different porosity, pore structure and thickness of cell walls, which would result in different mechanical properties of the ceramics.9" Therefore, the choice of starting bio-precursor is an important step in the processing. Bamboo is a group of herbaceous plants widely spread in sub-tropical Asian countries. It is hollow but strong and tough. It is so strong and tough that Chinese people do not regard it as grass. It could be used as a substitute of wood (timber) and find its applications in house, bridge and scaffolds.12 The superior mechanical properties of bamboo are partly contributed by its highly anisotropic structure. The cellular structure of bamboo is highly aligned along its growth direction between the nodes. Consequently, it would be very interesting to fabricate ceramics with highly aligned bamboo cellular structure. In this research, we used Moso bamboo, a common bamboo species in Taiwan, as the starting bio-precursor to form bio-templates for the fabrication of porous silicon carbide. The processing and properties of SiC with bamboo structure are reported. EXPERIMENTS Moso bamboo of ~ 4 years old was used as the bio-precursor. The samples were taken from lower part of the bamboo trees. It has a highly anisotropic cellular structure. The large pores channels (vessel), middle pores (tracheids) are perfectly aligned along the growth direction. Small pits in the cell walls interconnect these pores. Pieces of bamboo were first dried at 100CC for 24 h. Then, they were pyrolyzed inflowingargon to form charcoal preforms (bio-carbon templates). The pyrolysis was performed with l°C/min between room temperature and 500°C and with 5°C/min between 500°C to 1100°C. The bamboo pieces were soaked at 1100°C for 4 h. For reaction syntheses, the pyrolyzed charcoals were placed inside graphite crucibles with covers and were embedded in Si powder. To reduce the reaction of crucible with Si powder, the interior of crucible was coated with a layer of BN. Reaction was performed at 1400°C - 1550CC for various time ¡n flowing argon. During the synthesis, the charcoal reacted with infiltrated Si vapor to form SiC. After the reaction, the samples were heated at 600°C for 3 h in air to remove residual carbon. The weight changes before and after the heat treatment were regarded as the amount of residual carbon. Thermal analyses of the bamboo were done in DTA-TGA (model SDT-2960, TA Instruments Inc., New Castle, DE). X-ray diffractometer (Cu K«; D-5000, Siemens, Karlsruhe, Germany) was used to identify the phases of the samples after reaction. The samples were ground into powder for XRD analyses. The morphology of the samples was observed in a SEM (JSM-5600, JEOL, Tokyo, Japan) after the samples were coated with Au. Mechanical strengths were evaluated by

46

• Novel Processing of Ceramics and Composites

Fabrication of Silicon Carbide from Bamboo Carbon Templates

compression tests as well as bending tests. The samples for compression tests had the size of 4x4x4 mm3. The samples for bending tests were 3x4x36 mm3 unless specified otherwise. The crosshead speed was 0.3 mm/min. In both tests, the samples were placed in two perpendicular orientations to evaluate its anisotropy. The loading direction was either parallel or perpendicular to the growth direction. RESULTS AND DISCUSSIONS Figure 1 showed the DTA-TGA curves of the bamboo. There were three distinct stages of weight loss as the sample was heated. Below 130°C, the loss was due to evaporation of water. Between 230°C - 360°C drastic weight losses occurred. This stage of weight loss was due to the decomposition of organics (such as lignin, cellulose). Above 360°C, the weight loss was gradual. Therefore, the pyrolysis needed to be very slow between 200°C - 400°C to avoid cracking in the sample. The total weight loss after pyrolysis would reach 76%. After the samples were completed with pyrolysis at 1100°C for 4 h, the samples shrank significantly. The shrinkage along the growth direction (axial) was ~ 24% while the shrinkage perpendicular to the growth direction (radial and tangential) was ~ 30%. Despite the large weight loss and volume shrinkage, the charcoal retained the microstructure and morphology of the natural bamboo. Figures 2(a,b) revealed the cellular structure of bamboo charcoal. Large pore channels of ~ 80 |xm were surrounded by dense walls. The trachéal pores were of ~ 20 (im in diameter. Small pits of ~ 0.5 \xm existed in the cell walls (did not show in figures 2.) Figure 3 showed the XRD patterns of the charcoal and the samples after reaction synthesis at 1400°C to 1550°C for 4 h. The charcoal was amorphous. After reaction, it reacted with Si vapor to form ß-SiC. The shoulder to the left of ß-SiC peaks at ~ 36° indicated the presence of a-SiC. The amount of ß-SiC in the samples increased with the reaction temperatures. With sufficient supply of Si, samples with thickness of ~ 6 mm parallel to the growth direction could be completely transformed into SiC at 1500°C for 4 h. The completely transformed samples remained porous and the shapes of the charcoal were faithfully retained with volume changes less than 0.1%, see figure 4. The reaction depths parallel and perpendicular to the growth direction of the samples after synthesis at 1500°C were plotted against the square roots of reaction time in figure 5. In this figure, it showed linear relationships between the reaction depth and the square root of reaction time. This linear relationship indicated that the reaction is a diffusion control process.13 The reaction rate was twice as high in the growth direction as in the radial or tangential directions. The anisotropy in the reaction rates was caused by the anisotropy of the structure. The large pore channels along the growth direction provided fast transport routes for the Si vapor. Perpendicular to the growth direction, the vapor could only transport through small lateral pits in the cell walls.

Novel Processing of Ceramics and Composites

• 47

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Although the estimation of the critical pore diameter for effective Si vapor transport at 1400CC -1600°C in porous materials was ~ 1 urn,1314 the pits (~ 0.5 urn) in bamboo cell walls seemed to allow Si to transport through the walls. The formation of SiC was believed via solid-gas reaction: C(S> + Si(g> -> SiC(S). The SiC formation proceeded through the diffusion of Si vapor and through the diffusion of Si through SiC layer on the surface.13 This solid-gas reaction ensured that the structure of bio-carbon could be retained (as shown in Fig.4.) Smooth charcoal surface became rough as SiC grains formed on the surface of the cell walls. The pits in the cell walls were also retained (see inset in Fig. 4.) In order to evaluate the properties of the porous SiC, instead of partially reacted samples (SiC + residual carbon), the physical and strength measurements were performed with the samples reacted at 1500°C for 4 h or 8 h. Table 1 listed density and porosity of the porous SiC. The bulk density of the sample increased from 0.60 ± 0.05 g/cm3 of charcoal to 1.00 ± 0.05 g/cm3 of SiC. Since the dimension did not change, the weight increase after reaction was 67%. The solid density of the SiC was 3.07 g/cm3, which was close to the density of monolithic ß-SiC. The BET surface area was 18 m2/g and the pore size distribution measured by mercury porosimetry was shown in figure 6. In figure 6, there were two continuous ranges of pore size distribution: 0.01 p.m - 1 p.m and 5 |xm - 400 |im. The pores of 5 |im - 400 urn were the pore channels along the growth direction. However, the majority of pore volume was from pores of- 0.2 um in diameter. This should be regarded as the pore volume of the tracheids, which were connected with each other by the pits on the cell walls. Even if the tracheids had the cylindrical pores of- 20 u,m in diameter and ~ 100 um in length, they were connected with each other by small lateral pits on the cell walls. The pore diameter of- 0.2 um reflected the size of the neck of those pits in the cell walls. (The pits were estimated to be ~ 0.5 u,m in SEM in Fig. 4) Table II listed the strengths of the samples when loading direction was parallel or perpendicular to the growth direction. For charcoal, the maximum compressive strengths were about the same (~ 45 MPa) in both directions. However, the bending strength showed the anisotropic characteristics of the structure. After the samples were converted into porous SiC, the compressive and bending strength increased and the anisotropy of the compressive and bending strengths became apparent. The maximum compressive strengths in our experiments were 114 MPa and 73 MPa in directions parallel and perpendicular to the growth direction, respectively. The bending strengths were 12.8 MPa and 42.5 MPa in directions parallel and perpendicular to the growth direction, respectively. It was also noted that samples reacted at 1500°C for 8 h resulted in higher compressive strength than samples reacted for 4 h although both samples had been converted into SiC. Figures 7 and 8 showed the stress-strain curves during the compression perpendicular and parallel to the growth direction, respectively. In the compression tests, the charcoal exhibited

48

• Novel Processing of Ceramics and Composites

Fabrication of Silicon Carbide from Bamboo Carbon Templates

serrated stress-strain curves with extensive strain, typical stress-strain curves of laminated materials. The anisotropy of the porous SiC was also reflected in the compressive stress-strain curves. When compressed parallel to the growth direction, the samples behaved like laminated composites: serrated stress-strain curves with extensive strain. But when the samples were compressed perpendicular to the growth direction, they broke abruptly at maximum strength, much like brittle ceramics. The high anisotropy of the strengths resulted from the high anisotropy of the structure, which will need attentions in future applications. CONCLUSIONS Porous SiC with cellular structure was prepared using Moso bamboo as the bio-templates. The bamboo was pyrolyzed into carbon and infiltrated with Si vapor at temperature ~ 1500°C to react to form SiC. The highly anisotropic structure of bamboo was faithfully retained after the reaction with negligible dimensional changes. The reaction depths were linearly proportional to the square roots of reaction time. And the reaction rate was twice as high in the growth direction as in the radial or tangential directions. After complete conversion into SiC, the bamboo-like SiC had bulk density of 1.00 g/cm3, open porosity of 66.7%, BET surface area of 18 m2/g. The pore size distribution measurement indicated that the tracheid pores were mostly inter-connected by the lateral pits (~ 0.5 um) of the cell walls. Due to the anisotropy of the structure, the strengths of the bamboo-like SiC also showed difference in different directions. The compressive strengths were 114 MPa and 73 MPa and the bending ones were 12.8 MPa and 42.5 MPa in directions parallel and perpendicular to the growth direction, respectively. When compressed parallel to the growth direction, the samples behaved like laminated materials with serrated stress-strain curves after maximum strength. In contrast, when the samples were compressed perpendicular to the growth direction, they showed abrupt fracture at maximum strength. REFERENCES 1. J. Martinez-Fernandez, F.M. Valera-Feria and M. Singh, "High temperature compressive mechanical behavior of biomorphic silicon carbide ceramics," Scripta Materialia, 43 [9], 813-818(2000). 2. N.H. Thomson, B.L. Smith, G.D. Stucky, D.E. Morse and P.K. Hansma, "Methods for fabricating and characterizing a new generation of biomimetic materials," Mater. Sei. and Eng. C7, 37-43(1999). 3. J.M. Qian, J.P. Wang, GJ. Qiao and Z.H. Jin, "Preparation of porous SiC ceramic with a woodlike microstructure by sol-gel and carbothermal reduction processing," J. Eur. Ceram. Soc., 24, 3251-3259 (2004). 4. B. Chen, X. Peng, J.G Wang and S. Wu, "Laminated microstructure of Bivalva shell and

Novel Processing of Ceramics and Composites

• 49

Fabrication of Silicon Carbide from Bamboo Carbon Templates

research of biomimetic ceramic," Ceram. Int. 30, 2011-2014 (2004) 5. V. Rodriguez-Lugo, E. Rubio, 1. Gomez, L. Torres-Martinez and V.M. Castaño, "Synthesis of silicon carbide from rice husk," Int. J. Environ. Pollut. 18 [4], 378-387 (2002). 6. J.M. Qian, Z.H. Jin and X.W. Wang, "Porous SiC ceramics fabricated by reactive infiltration of gaseous silicon into charcoal," Ceram. Int., 30 [6], 947- 951 (2004). 7. H. Sieber, C. Rambo, J. Cao, E. Vogli and P. Greil, " Manufacturing of porous oxide ceramic by replication of plant morphologies," Key Eng. Mat. 206-213 [III], 2009-2012 (2001). 8. Y. Huang, Q. Zan, H. Guo and S. Cai, "Biomimetic structure design - a possible approach to change the brittleness of ceramics in nature," Mater. Sei. and Eng. C11, 9-12 (2000). 9. P. Greil, T. Lifka and A. Kaindl, "Biomorphic cellular silicon carbide ceramics from wood: I. Processing and microstructure," J. Eur. Ceram. Soc. 18 [14], 1961-1973 (1998). 10. M. Singh and J.A. Salem, "Mechanical properties and microstructure of biomorphic silicon carbide ceramics fabricated from wood precursors," J. Eur. Ceram. Soc. 22 [14-15], 2709 2717(2002). 11. F. M. Varela-Feria, M.J. Lopez-Robledo, J. Martinez-Fernandez, A.R. De Arellano-Lopez and M. Singh, " Precursor selection for property optimization in biomorphic SiC ceramics," Ceramic Eng. and Sei. Proceedings 23 [4], 681-687 (2002). 12. J. Fu, "Chinese Moso Bamboo: Its importance," Bamboo 22 [5], 5-7 (2001). 13. J.M. Qian, J.P. Wang and Z.H. Jin, "Preparation and properties of porous microcellular SiC ceramics by reactive infiltration of Si vapor into carbonized basswooe," Mater. Chem. Phy. 282, 648 - 653 (2003). 14. A. Kaindl, Cellular SiC ceramics from wood, Ph.D Thesis, University of Erlangen-Nuemberg, 2000 (in German).

50

• Novel Processing of Ceramics and Composites

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Table 1. Density and porosity of porous SiC measured with Archimedes* technique using water as the immersion medium. Porous SiC (1500°C/4h)

Bulk density (g/cm3)

Solid density (g/cm3)

Open porosity (%)

1.00

3.07

66.5

Table II. Mechanical strengths of the samples in different orientations. "^'-^--^I'roperty

Compressive strength \ . Loading Perpendicular 7~-~^direction Sample ^**""---^_^ Parallel to growth to growth direction direction Charcoal Porous SiC (1500°C/4h) Porous SiC (1500°C/8h)

Four point bending strength Parallel to growth direction

Perpendicular to growth direction 26.4 ±4.1

46.7

45.9

5.8 ±2.0

75.3

53.9

ll.Oil.2

114.0

73.4

12.8 ±1.7

42.5 ± 4.5*

#: due to the low reaction rate in the direction perpendicular to the growth direction, the bending bars of regular size (3x4x36 mm3) did not convert into SiC completely after l500°C/4 h. The bending strength of porous SiC were obtained in samples of a smaller size, 2x2x36 mm after 1500°C/8h, which had complete conversion into SiC.

Novel Processing of Ceramics and Composites

• 51

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Temperature (#J)

Figure I. DTA-TGA curves of the bamboo

Figure 2(a,b). SEM micrographs of bamboo charcoal, (a) Cross-section perpendicular to the growth direction, (b) Cross-section parallel to the growth direction.

52

• Novel Processing of Ceramics and Composites

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Figure 3. XRD patterns of the charcoal and the samples after reaction synthesis at various temperatures for 4 h.(a) charcoal, (b)1400°C (c) 1450°C (d) 1500°C (e) 1550°C. * indicates diffraction peaks from ß-SiC.

Figure 4. SEM micrograph of the sample completely transformed into SiC. The microstructure of bamboo is retained. Inset: magnified view of cell wall, showing SiC grains and pores (pits)

Novel Processing of Ceramics and Composites

• 53

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Figure 5. The reaction depths of the samples after synthesis at 1500°C versus the square roots of reaction time.

1000000

100000

10000

1000

100

Pore size Diameter (nm)

Figure 6. Mercury porosimetry of porous SiC. The sample was reacted at 1500°C for 4 h.

54

• Novel Processing of Ceramics and Composites

Fabrication of Silicon Carbide from Bamboo Carbon Templates

Figure 7. Stress-strain curves of compression perpendicular to the growth direction, (a) charcoal, (b) porous SiC after 1500°C/4h reaction, (c) porous SiC after 1500°C/8h.

Figure 8. Stress-strain curves of compression parallel to the growth direction, (a) charcoal, (b) porous SiC after 1500°C/4h reaction, (c) porous SiC after 1500°C/8h.

Novel Processing of Ceramics and Composites

• 55

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

EFFECTS OF PROCESS PARAMETERS ON POST REACTION SINTERING OF SILICON NITRIDE CERAMICS Toru Wakihara, Junichi Tatami*, Katsutoshi Komeya, Takeshi Meguro Graduate School of Environment and Information Sciences, Yokohama National University 79-7, Tokiwadai, Hodogaya-ku Yokohama, 240-8501, Japan ♦Corresponding author: [email protected], Tel: +81-45-339-3959, Fax: +81-45-339-3957 Hideki Kita, Naoki Kondo, Kiyoshi Hirao Advanced Industrial Science and Technology, Shimo-Shidami, Moriyama-ku, Nagoya 463-8560, Japan ABSTRACT To establish a method for the fabrication of silicon nitride ceramics at lower cost with keeping their high performances, a post sintering technique was focused on. At first, silicon powder compacts with oxide sintering aids (Y2O3 and AI2O3) were nitrided in nitrogen gas at 1450 °C. In the next step, silicon nitride compacts obtained are fired at 1850-1900 "C in nitrogen gas under 0.9 MPa. As a result, we could successfully fabricate silicon nitride ceramics with high strength (600 MPa in 3-point bending test) and high relative density (more than 93%). In the present study, the effects of processing parameters, e.g. raw materials, firing temperature, additive (Fe), were investigated. INTRODUCTION Silicon nitride (SÍ3N4) ceramics are widely utilized as structural materials with high strength, hardness, wear resistance and thermal stability. There is a limit, however, to the application for industrial uses because of its economic disadvantages. To establish a method for the fabrication of silicon nitride ceramics1"6 at low cost with keeping their high performances is increasingly important. We focused on, therefore, a post sintering technique which is known as a lower cost process; silicon powder (several dollars/kg at cheapest) compacts with oxide sintering aids (typically Y2O3 and AI2O37 ) are nitrided and densified in nitrogen gas. However, there are drawbacks that it takes time for nitridation of silicon and is difficult to fabricate high strength sintered body. It has been reported that addition of Fe promotes nitridation of silicon powder compacts2 although detailed nitridation and densification behavior have not been well clarified. In the present study, the effects of processing parameters, e.g. raw materials, firing temperature, additive (Fe), were investigated. EXPERIMENTAL PROCEDURE Si powder was used as a raw material. Y2O3 (RU, Shinetsu Chemical Co.) and AI2O3 (AKP-30, Sumitomo Chemical Co., Japan) as sintering aids and Fe2Û3 (Japan Pure Chemical Co. Ltd.) as an additive were added to Si powder. The raw-powder batch composition was (89-x)Si : xFe : 8Y2O3 : 3AI2O3, where x=0.1, 0.7, 1.2, 2.2 in weight. These powders were milled and mixed in ethanol with a dispersant by ball-milling using sialon balls for 2, 4 and 7 days. After mixing, the ethanol in the slurry was evaporated using a mantle heater. After paraffin as a binder and dioctyl phthalate as a lubricant were added, the mixed powders were sieved (using a #50

57

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

sieve made of nylon) to obtain granules. The granules were molded into 40x40><5 mm by uniaxial pressing at 35 MPa, followed by cold isostatic pressing at 200 MPa. The organic binder was eliminated at 500 °C for 6 h in 41/min N2 flow. After dewaxing, the green bodies were set on carbon boat and fired at 1450 °C for 2 hours in 0.1 MPa N2 (TP-150, Tokai Konetsu Kogyo Co. Ltd., Japan). During this process, nearly all Si changes into SÍ3N4. In the next step, the bodies were set into silicon nitride crucibles and fired at 1850 or 1900 °C for 2 hours in 0.9 MPa N2 for densification using a gas-pressure-sintering furnace (Himulti 5000, Fujidenpa Kogyo Co., Japan). Table 1 shows the compositions and heating conditions of the samples. Density was measured by the Archimedes method, the phase present was identified by X-ray diffraction (XRD: Multiflex, Rigaku Co., Japan), and the microstructure was observed by a scanningelectron microscope (SEM: JSM-5200, JEOL, Japan). The sintered specimens were machined into 3x4x40 mm and their mechanical properties were evaluated. Bending strength was measured by 3-point bending test with a lower span of 30 mm and a crosshcad speed of 0.5 mm/min. Table 1 Compositions and firing conditions of the samples

Sample

Mixing (days)

Fe (wt%)

Firing temperature (X.)

1 2

3

;1 '\ 1

0.7 0.7 0.7

-

4 5 6 7

i 1 1 r

0.1 0.7 1.2 2.2

1850 1850 1850 1850

8 9 10

r 1 r

0.7 1.2 2.2

1900 1900 1900

RESULTS AND DISSCUSSION Figure la shows a raw Si powder. It was found that Si powder is composed of ca. 10-50 urn particles. Figures lb, lc and Id show the raw powders after milling for 2, 4 and 7 days; indicating that the mean particle size decreased as milling period increased. Figure 2 shows particle size analysis for the mixed powder. Average diametes (¿50) were 4.6, 1.8 and 1.6 um after 2, 4 and 7 days ball milling, respectively. Green bodies prepared from these powders (Sample 1, 2 and 3, milled for 2,4 and 7 days, respectively: 0.7 wt% of Fe was added) were fired at 1450 °C as mentioned in experimental procedure. However, nitridation did not proceed enough except Sample 3 and Si was oozed from the sintered body using a raw powder milled for

58

• Novel Processing of Ceramics and Composites

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

2 and 4 days. It is speculated that nitndation did not proceed enough since Si particles were too large, and therefore, unreacted Si (melting point: 1420 °C) was oozed from the sintered body. Therefore, the raw powder milled for 7 days was used for the fabrication of SÍ3N4 ceramics hereafter.

S S

I' |

z0

S

h

1

1

1 1

1 1rj

1

1

r ¿4|M»T

ft § t

fS

60

M?

$

Jr

-

~0*f

0.1

g té

J • "

'

' ' ' ' 1

ê i

'

ff

; t

20

m

*

fí fí f: ff

§

amount (%)

Figure 1 SEM images of the powders

. . . 2 days

'

"

4 days

0 2

4

6

8

2

4 6 8

1 10 Particle diameter (urn)

2

4

6

8

100

Figure 2 Particle size analysis of the mixed powder (2, 4, and 7 days)

Novel Processing of Ceramics and Composites

• 59

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

In the next step, effect of the amount of Fe on nitridation behavior was investigated. Figure 3 shows XRD patterns of Samples 5, 6 and 7, which were fired at 1450 °C for 2 hours. Note that XRD pattern of Sample 4, in which 0.1 wt% of Fe was added, was eliminated since nitridation did not proceed enough and Si was oozed from sintered body (the same result was obtained as Sample 1 and 2). It was found that addition of Fe is inevitable for the promotion of nitridation in the present synthesis condition. As shown in Figure 3, main phase was a mixture of a- and p-SÍ3N4, and there is no clear difference between three samples. Nitridation was completed for the most part in all samples and almost no residual unreacted Si was confirmed. Formation of FeS¡2, melilite (SÍ3Y2O3N4) and crystobalite (SÍO2)8 were also confirmed, and it was found that peak intensities of them increased with increasing the amount of Fe. Figure 4 shows XRD patterns of Samples 5, 6 and 7 fired at 1850 °C for 2 hours. Main phase was ßSÍ3N4; indicating that phase transformation from a- to P-SÍ3N4 was completed during firing. Note that the same result was obtained in Samples 8, 9 and 10 (fired at 1900 °C for 2 hours). It was found that peak intensity of FeS¡2 increases with increasing the amount of Fe. This result well agrees with XRD patterns shown in Figure 3; indicating that FeS¡2 did not change into other phases during densification process. Figure 5 shows SEM images of fractured surfaces of Samples 5, 6 and 7. It was found that densification proceeded in all samples and decrease of grain size and elongated grain of P-SÍ3N4 were confirmed with increasing the amount of Fe.

Figure 3 XRD patterns of Samples 5, 6 and 7 fired at 1450°C for2hours

60

• Novel Processing of Ceramics and Composites

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

• ß-Si3N, ■ AI,Y409(YAM) * FeSi,

Sample 7 Sample 6 Sample 5 20

25

30

35

40

45

50

55

60

2 0 /degrees

Figure 4 XRD patterns of Samples 5, 6 and 7 firedatl850°Cfor2hours Bending strength and relative density of SÍ3N4 ceramics obtained (Sample 5-10) are shown in Table 2. Densification was performed at 1850 °C (Sample 4-6) and 1900 X. (Sample 7-9), respectively. Bending strength was 400-500MPa in the samples fired at 1850 °C (Sample 4-6), and a lowering of bending strength was confirmed with increasing the amount of Fe. It is thought that decrease of elongated grain is the cause of a lowering the bending strength. Bending strengths of the samples fired at 1900 °C are higher (600 MPa) than those fired at 1850 °C; indicating that densification was promoted at higher temperature. As a result, for the fabrication of SÍ3N4 ceramics by a post reaction sintering method, important process parameters are summarized as follows; • Nitridation of Si is largely affected by the particle size of raw Si powder • Addition of Fe plays an important role for the promotion of nitridation, but too much amount of Fe in sintered body lowers the bending strength CONCLUSIONS A post sintering technique was applied for the fabrication of silicon nitride ceramics. As a result, we could successfully fabricate silicon nitride ceramics with high strength (600 MPa in 3-point bending test) and high relative density (more than 93%). Especially, the effects of processing parameters, e.g. raw materials, firing temperature, oxide sintering aids, were investigated. Mixing process of raw powder and addition of Fe in raw powder play an important role for the fabrication of SÍ3N4 ceramics.

Novel Processing of Ceramics and Composites

■ 61

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

Low magnification

High magnification

Sample 5

Sample 6

Sample 7

Figure 5 SEM images of sintered bodies (Samples 5, 6 and 7) Table 2 Bending strength and relative density of SÍ3N4 ceramics (Average of six specimens for each sample)

Sample

Bending strength (MPa)

Iterative density (%)

S 6 7

500 430 440

93.4 92.8 92.6

8 9 10

600 590 600

93.7 93.7 93.4

REFERENCES 'K. Komeya and H. Inoue, Japanese paient No. 734272 (1973). 2 J.A. Mangels and G..J. Tennenhouse, "Densification of Reaction-Bonded Silicon Nitride," Am. Ceram. Soc. Bull., 59, 1216-18 (1980).

62

• Novel Processing of Ceramics and Composites

Effects of Process Parameters on Post Reaction Sintering of Silicon Nitride Ceramics

3

Y. Kaga, M.I. Jones, K. Hirao and S. Kanzaki, "Fabrication of Elongated a-SiA10N via a Reaction-Bonding Process," J. Am. Ceram. Soc, 87, 956-59 (2004). 4 B.R. Zhang, F. Marino, V.M. Sglavo, "Post-hot-pressing and high-temperature bending strength of reaction-bonded silicon nitride-molybdenum disilicide and silicon nitride-tungsten suicide composites," J./lw. Ceram. Soc., 81, 1344-48 (1998). 5 S. Y. Lee, "Fabrication of SÍ3N4/SÍC composite by reaction-bonding and gas-pressure sintering " J. Am. Ceram. Soc., 81, 1262-68 (1998). D.S. Park, H.D. Kim, K.T. Lim, K.S. Bang and C. Park, "Microstructure of reactionbonded silicon nitride fabricated under static nitrogen pressure" Mater. Sei. Eng. A, 405, 158-62 (2005). 7 K. Komeya and F. Noda, "Aluminum Nitride and Silicon Nitride for High Temperature Gas Turbine Engines," Society ofAutomotive Engineers Paper No. 740237 (1974). 8 S.M. Boycr and A.J. Moulson, "Mechanism for Nitridation of Fe-Contaminated Silicon,' J. Mater. Sei., 13, 1637-46 (1978).

Novel Processing of Ceramics and Composites

■ 63

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Polymer Processing

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

SYNTHESIS OF CARBON/Fe-Ni-Cu ALLOY COMPOSITE BY CARBONIZATION OF ORGANOMETALLIC POLYMERS AND THEIR MAGNETIC PROPERTIES Yukiko Uchida, Makoto Nakanishi, Tatsuo Fujii, Jun Takada, Akinori Muto, Yusaku Sakata Department of Material Chemistry, Graduate School of Natural Science and Technology, Okayama University Tsushima-naka 3-1-1 Okayama 700-8530, Japan ABSTRACT Carbon/Fe-Ni-Cu alloy composites were synthesized by carbonization of organometallic polymers obtained from ion exchanged chelate resins. The carbonization was performed at 900 °C in a N2 gasflow.The composition of the Fe-Ni-Cu alloy in a carbon matrix can be controlled by the adjusting solution compositions in the ion exchange process. The carbon obtained from the organometallic polymer was crystallized to a turbostratic structure. X-ray diffraction patterns revealed that the Cu, Fe and Ni elements in the ion exchanged chelate resins were crystallized to a Fe-Ni-Cu alloy after carbonization at 900°C. This result indicates that the formation of the ternary alloy in the carbon matrix is coincident with the crystallization of the well-ordered turbostratic structure of the carbon matrix. The alloy particles with an average size of about 200-700 run were widely dispersed in the carbon matrix. The saturation magnetization and coercivity of the composites depended on the alloy composition. The saturation magnetization and coercivity of the carbon/Feo.52Nio.42Cuo.o6 alloy composite were 7.9 emu/g and 167 Oe, respectively. INTRODUCTION Nanometer-scale metallic compounds dispersed in carbon matrices are expected to have unique properties making the best of both the carbon and metallic compounds. There have been many studies on preparation and characterization of metallic nanoparticles hybridized with a carbon matrix as a heterogeneous catalyst'"4, an electric double layer capacitor5 and so on. Especially, the carbothermal reduction of a metal ion-exchanged resin (MIER-CTR) was widely used for the preparation of the carbon composite material containing the highly dispersed nanometer scale metallic compounds''5. After carbonization, metal elements adsorbed into the resin were transformed into various crystals such as metals, carbides, nitrides, sulfides or phosphates, which were highly dispersed in the carbon matrix. Nanometer scale metals and metallic compounds played important roles in enhancing the catalytic, electric and other properties of carbon composites.

67

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

Recently various methods were reported to prepare the Ni, Co, Fe-Co and Fe-Ni alloy composites with a carbon matrix applied for magnetic materials6"10. Hirano et al. demonstrated the synthesis of Fe-Co alloy particles dispersed in carbon by pressure pyrolysis of organoiron-organocobalt copolymers under 125 MPa6. We have already reported that the carbon/Fe-Co alloy composites were prepared by the polymerized complex method at ambient pressure9. In the previous study, it has been revealed that the composite carbonized at 600°C had very small coercivity of 14 Oe and exhibited relatively large permeability spectra in the GHz frequency range9. As far as our knowledge there are no reports on the preparation of carbon/alloy composites by using the MIER-CTR method. In addition, the carbon composites with multinary alloys or intermetallic compounds including more than three kinds of metal elements were relatively unfamiliar. In this study, we tried to prepare the magnetic ternary alloy composites with the carbon matrix by using the MIER-CTR method. We chose the composition of Fe-45Ni-5Cu, which is known as the Permalloy B or the Radiometal. The Fe-Ni binary alloys are well known as soft magnetic alloys with high permeability. Low Ni alloys containing approximately 45% to 50% Ni have a quite high saturation magnetization and comparatively high initial and maximum permeabilities". In order to improve the magnetic properties small additives such as Mo, Cr and Cu are known to be effective. We also discuss the magnetic carbon/Fe-45Ni-5Cu alloy composite materials prepared by using the MIER-CTR method. EXPERIMENTS Carbon/Fe-Ni-Cu alloy composites were prepared by using the following procedure. The ion exchanged resin of chelate type IRC748 (ÓRGANO Co.) having an iminodiacetic acid group (-C-N-(COONa)2) was used as the raw material. Before the ion exchange reaction, the resin was fully exchanged with H* by the conventional procedure. Then, the resin was dipped in the mixed aqueous solutions of Fe(N03)3 , Ni(N03)2 and Cu(N03)2, to adsorb the metal ions Fe3+, Ni2+and Cu2+ at room temperature for 2 hours. The ¡on exchage conditions are shown in Table I. The metallic ion concentrations in the solution during the ion exchange reaction were analyzed by inductively coupled plasma spectrometry (ICP) (PERKIN ELMER Optima 2000 DV). The obtained metal ion exchanged resin was used as a precursor for the preparation of the carbon/Fe-Ni-Cu alloy composite. The resin wasfiltrated,and naturally dried at room temperature for 1 week. Then, the resin was thermally dried at 110°C for 5 hours and heated at 900°C for 3 hours in a N2 gas flow (0.3 dm3/min). The prepared carbon/Fe-Ni-Cu alloy composites were characterized by an X-ray diffraction (XRD) technique using Cu Ka radiation (RIGAKU RINT-2500). The lattice parameters and crystalline size of the Fe-Ni-Cu alloy were estimated in comparison with diffraction lines of Si powder as an internal standard. The amount of each metal

68

• Novel Processing of Ceramics and Composites

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

contained in the composite was determined by X-ray fluorescence analysis (XRF) (RIGAKU ZSXIOOs). The morphology and atomic distribution of the obtained composites were analyzed by using a field emission scanning electron microscope (FESEM) with an energy dispersive X-ray spectrometer (EDS) (HITACHI S-4300). The magnetic properties of the samples were examined by a vibrating sample magnetometer (VSM) (TOEI VSM-5-15) Table I Ion exchange conditions Sample name Sample 1 Sample 2 Sample 3 Sample 4

Metal composition ratio Ni Cu Fe 1 1 1 1

0 2 2.5 5

0 0.1 0.1 0.1

Total metal concentration (mol/ldm3-resin) 0.07 0.22 0.26 0.44

RESULTS AND DISCUSSION The XRD patterns of the carbon/Fe, Fe3C and carbon/Fe-Ni-Cu composites carbonized at 900 °C are shown in Fig. 1. The 002 line assigned to turbostratic carbon appeared at about 20=26° in all samples. All carbon matrices were crystallized to turbostratic carbon. Moreover, the adsorbed Fe ions into the resin were crystallized to Fe3C and or-Fe (bcc structure) for sample 1, while the samples containing ternary Fe, Ni and Cu (Samples 2, 3 and 4) ions showed two intense diffraction lines at about 20=44° and 51°, which were assigned to the /// and 200 lines of the fee structure according to the extinction rule of X-ray diffraction of the fee structure. The lattice constant of the obtainedjfcc structure was about 0.389 nm, as summarized in Table II. These values are not consistent with those of the bulk Ni, Cu or j¿Fe having a.fee structure, as shown in Table II. The Cu, Fe and Ni elements in the ion exchanged chelate resins were crystallized to a monophasic Fe-Ni-Cu alloy in the carbon matrix with a well-ordered turbostratic structure after carbonization at 900°C. As no Fe3C phase appeared in Samples 2,3 and 4, the Ni and Cu loading was effective to suppress the formation of FejC. The same phenomenon was observed in the carbon/Fe-Co composite system prepared from a different precursor6,9. The formation of Fe3C in carbon matrix was strongly suppressed when the second metal elements like Ni or Co coexisted in the carbon precursors. The alloy composition and total metal content in the composites determined by the XRF measurements are summarized in Table III. The Ni content was increased with increasing the Ni concentration in the aqueous solution for the ion exchange process. However the Fe and Cu

Novel Processing of Ceramics and Composites

• 69

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

contents were decreased with increasing the Ni composition though the both Fe and Cu concentrations in aqueous solution were unchanged. The total metal content of ternary alloy samples also depended on the Ni content up to the 11.6 mass% for the sample 4. The metal composition ratio for Samples 2 and 3 were nearly consistent with those for the expected values of Fe-45Ni-5Cu. We conclude the composition of Fe-Ni-Cu alloy in carbon matrix can be controlled by the solution compositions used in the ion exchange process.

Ocarbon ▼Fe-Ni-Cu alloy Va-Fe xFe,C

8 o

Sample 4 Sample 3 Sample 2 Sample 1

20

30 40 50 20(deg.)Cu-Ka

Figure I XRD patterns of obtained composites carbonized at 900°C. Sample 1 is a carbon/Fe, Fe3C composite. Samples 2, 3 and 4 are carbon/Fe-Ni-Cu composites having different metal compositions.

Table II Lattice constant for the Fe-Ni-Cu alloy and constituent metals ' Sample name

er-axis length (nm)

Sample 2 Sample 3 Sample 4 Ni Cu

0.3594 0.3595 0.3588 0.35238 0.36147 0.36468

j'Fe

70

• Novel Processing of Ceramics and Composites

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

Table 111 Metal composition and content Sample name Sample 2 Sample 3 Sample 4

Metal compositions Fe (at%) Ni (at%) Cu (at%) 51.8 46.9 34.7

41.9 47.4 61.3

6.3 5.7 3.9

Total metal content (mass%) 9.0 9.4 11.6

Figure II (a) and (b) show typical FESEM images for the carbon/Fe. FejC composite (Sample 1) and carbon/Fe-Ni-Cu alloy composite (Sample 2), respectively. The carbonized samples had the spherical shape as same as that of the chelate resin used for the starting material. The metallic crystals were hardly observed in the cross section image of the composites. The surface SEM images clearly suggested the metallic crystals preferably grew on the surface of the composites. The EDS ¡mages of Sample 2 (not shown) confirmed that the small particles observed in the carbon matrix corresponded to the Fe-Ni-Cu alloy. The metal composition of individual particles confirmed by the EDS point analysis was nearly unchanged from each other. The same surface morphology was also observed in Samples 3 and 4. The grain size of metallic particles was very different depending on the obtained phases of metallic crystals. The grain size of Fe or Fe3C for sample 1 was about 100 ran, while that of the Fe-Ni-Cu alloy for sample 2 was 200-700 ran.

Figure II SEM images of surfaces of (a) Sample 1 and (b) Sample 2.

Novel Processing of Ceramics and Composites

• 71

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

Figure III shows magnetic hysteresis curves for the carbon/Fe-Ni-Cu alloy composites measured at room temperature. The saturation magnetization Ms per composite of the Sample 4 was about 14 emu/g, which was the largest value among other samples. It is clearly observed that Ms per composite increased with increasing metal content of the carbon composites, as shown in Table III. Moreover, Sample 2 has the smallest coercivity H0 about 167 Oe, only a quarter of Hz value of Sample 1. The formation of the ternary alloy particles seemed to enhance the soft magnetic performance. The values of Ms and // c for samples with different metal compositions are summarized in Table IV. The Ms per alloy was determined from the total alloy content in the composites shown in Table III. The relationship between the magnetic properties and the crystalline size of the Fe-Ni-Cu alloy in the carbon matrix is shown in Figure IV. Its crystalline size was estimated by using the Sherrer formula'3 from the XRD patterns shown in Figure I. The decrease of Ms per alloy correlates with the Cu content of the Fe-Ni-Cu alloy. Though the crystalline size remained to be almost constant in these samples the coercivity did clearly change. It is revealed that the coercivity also depended on the composition of the Fe-Ni-Cu alloy. The low Hc of both Samples 2 and 3 mean their soft magnetic property. Since the Samples 2 and 3 are comparatively consistent with the metal composition ratio of Fe-45Ni-5Cu, their soft magnetic performance is superior to that of Sample 4. 15 „ O) 3

fc
10 5

C

o

0

4) C D)

-5

CO N

CO

2

=P: Sample 4 Sample 3

Sample 2

Sample 1

10 .1«;

-10

0

10

Magnetic field (kOe) Figure HI Magnetic hysteresis curves for samples.

72

• Novel Processing of Ceramics and Composites

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

Table IV Magnetic properties of samples.

Figure IV Relationship between magnetic properties and crystalline sizes of the Fe-Ni-Cu alloy in composite. CONCLUSION We successfully obtained the carbon composite material in which the nanometer scale Fe-Ni-Cu ternary alloy was highly dispersed by using the MIER-CTR method. The metal composition could be controlled to compose the Fe-45Ni-5Cu alloy by adjusting solution compositions used in the ion exchange process. The saturation magnetization and coercivity of the composite depended on the alloy compositions. The sharp hysteresis suggested high permeability of the carbon/Fe-Ni-Cu alloy in the high frequency. AKNOWLEDGEMENT A part of this work was financially supported by the grant-in-aid from The Circle for the Promotion of Science and Engineering, Japan. The authors are grateful to Dr. Ishikawa of ÓRGANO Co. for providing chelate resin. We also thank Dr. Fujii of Industrial Technology Center

Novel Processing of Ceramics and Composites

• 73

Synthesis of Carbon/Fe-Ni-Cu Alloy Composite by Carbonization of Organometallic Polymers

of Okayama Prefecture for XRF measurements and Prof. Fukuhara of Okayama University of Science for ICP measurements. REFERENCES 'A. Muto, K. Ida, T. Bhaskar, Md.A. Uddin, S. Takashima, T. Hirai, and Y. Sakata, "Preparation of novel TÍP2O7 carbon composite using ion-exchanged resin (C467) and evaluation for photocatalytic decomposition of 2-propanol", Appl. Catal. A, 260, 163-168 (2004). 2 T. Matsui, T. Okita, Y. Fujii, T. Hakata, T. Imai, T. Bhaskar, and Y. Sakata, "Preparation, characterization and reactivity of calcium-carbon, iron-calcium-carbon composites for dechlorination", Appl. Catal. A, 261, 135-141 (2004). 3 B. H. Lipshutz, S. Tasler, W. Chrisman, B. Spliethoff, and B. Tesche, "On the Nature of the 'Heterogeneous' Catalyst: Nickel-on-Charcoal", J. Org. Chem.,6%, 1177-1189(2003). 4 B. H. Lipshutz, and B. Frieman, "Microwave accelerated, Ni/C-catalyzed cross-coupling of in situ-derived zirconocenes", Tetrahedron, 60, 1309-1316 (2004). 5 Y. Sakata, A. Muto, Md. Aznar Uddin, N. Yamada, C. Marumo, S. Ibaraki and K. Kojima, "Preparation of a Carbon Electrode for Electric Double -Layer Capacitors by Carbonization of Metal-Ion-Exchanged Resin", Electorchem. Solid State Lett., 3, 1-3 (2000). 6 S. Hirano, T. Yogo, K. Kikuta, and S. Naka, "Synthesis and properties of carbons dispersed with Fe-Co alloy by pressure pyrolysis of organoiron-organocobalt copolymer", J. Mater. Sei., 21, 1951-1955 (1986). 7 Y. Kaburagi, H. Hatori, A. Yoshida, Y. Hishiyama, and M. Inagaki, "Carbon films containing transition metal particles of nano and submicron size", Synth. Met., 125, 171-182 (2002). 8 M. Kodama, and H. Honda, "Preparation of ultrafine alloy dispersed carbons using anisotropic coal-tar pitch", J. Mater. Sei. Lett., 19, 1197-1199 (2000). 9 Y Uchida, K. Oishi, M. Nakanishi, T. Fujii, J. Takada, Y, Kusano, and T. Kikuchi, "Preparation and Magnetic Properties of Carbon/Fe-Co alloy Composite by Polymerized Complex Method" J. Jpn. Soc. Powder Powder Metallurgy, 52, 640-645 (2005). ,0 F. Goutfer-Wurmser, H. Konno, Y. Kaburagi, K. Oshida and M. Inagaki, "Formation of nickel dispersed carbon spheres from che late resin and their magnetic properties", Synth. Met. ,118, 33-38(2001). "The ASM International Handbook Committee, Metals handbook, vol. 2: Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, ASM International, 771-773 (1990). 12 J. Emsley, THE ELEMENTS, 3rd ed., Oxford University Press, 53 (1998). 13 B. E. Warren, X-Ray Diffraction, Dover Publications, 251-254 (1990).

74

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Electrochemical Deposition

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

FABRICATION OF YSZ THIN FILMS IN AN AQUEOUS SOLUTION BY ELECTROCHEMICAL DEPOSITION Atsushi Saiki, Hiroki Uno, Satoka Ui, Takashi Hashizume, Kiyoshi Terayama Department of Material Systems Engineering and Life Science, Faculty of Engineering, Toyama University, 3190 Gofuku, Toyama 930-8555 JAPAN ABSTRACT In our present study, YSZ thin films were fabricated on glass substrates in an aqueous complex solution by electro-chemical deposition. Zirconia is very important for their many characters, and we aim to make their thin films for large area and homogeneous thickness by using environment friendly materials and low cost method. Composition and concentration of the precursor aqueous solution is 94mol%Zr02-6mol%YOi 5 and 0.073 mol/dm3 (YSZ solution). ZrO(N03)2-2H20 and Y(N03)3-6H20 were used Zr and Y sources. Ammonia solution in appropriate quantities (0-0.5 vol%) was added to adjust the acidity and to occur a radical exchange reaction between ZrO(N03)2 and NH4OH to produce an intermediate form of ZrO(OH)2. After the mixture was stirred for 1 h in an ice bath, a homogeneous colorless and transparent solution was obtained. YSZ thin film was deposited on the glass substrate which was placed on the carbon electrode by applying the voltage 2.0-4.5V for 300s. As-deposited films were flat and transparent but amorphous and so heat-treated at temperature 573-773K for 1 h afterwards. From XRD patterns of the films, the peaks of YSZ were observed at samples annealed at 673, 773 K. After this heat treatment, transparent, pure, relatively hard thin films which thicknesses were about 0.05 to 0.3um, were get on glass surface. Composition of the films were almost same with that of aqueous solution. It was confirmed that the crystallization of the films were depended on the applied voltage and the amount of added Nfyaq.. INTRODUCTION Zirconia is very important for their thermal, mechanical, and chemical stability,1"5 and their thin films have attracted much attention for applications such as, buffer layers for growing electric devices,6"7 thermal-shield or corrosion-resistant coatings,8"9 ionic conductors,10"" oxygen sensors'2 and optical coatings.13 One of the fabrication method for zirconia thin films by using relatively environment friendly materials is to use aqueous solutions as raw materials. And mainly three kinds of techniques have been investigated for direct deposition of Zr0 2 thin films using aqueous solutions. The first method is the liquid-phase deposition (LPD) method.14"15 However, cause of the hexafluorozirconate salt (M2ZrF6: M = Na, K. NH4, etc.) used as the zirconium source, it is difficult to eliminate the residual fluorine in the as-prepared thin film even by annealing, which restricts application of the films. And the second is self-assembled monolayer (SAM) technique.16"18 But in them a large amount of chlorine or sulfur were also tend to present in the as-deposited thin films because of uncompleted hydrolysis reaction of zirconium ethoxide, as zirconium sulfate (Zr(S04)2-4H2Û) and HC1 are used as raw materials. And the last is the decomposition reaction of aqueous peroxozirconium complex solution using zirconium oxynitrate dihydrate (ZrO(N03)2-2H20) as Zr source.19"20 In this method pure amorphous Zr0 2 thin film at room temperature. In the present study, we aim to make zirconia thin films by using environment friendly materials, at low temperature, and use low cost method for large area and homogeneous

77

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

thickness. By applying electro-chemical deposition method to aqueous solution using simple materials such as ZrO(NOj)2-2H20 and NH3-H2O, a transparent, highly pure, amorphous Z1O2precursor thin film was synthesized on a glass substrate surface at room temperature. A crystalline phase can be obtained after annealing at 673-773 K for 1 h in air. EXPERIMENTAL PROCEDURE Preparation of Thin Films and Characterization Composition and concentration of the precursor aqueous solution is 94mol%ZrC<26mol%YOi 5 and 0.073 mol/dm3 (YSZ solution). Zirconium oxynitrate dihydrate (ZrO(NC<3)22H20, 99.99% ) and Yttrium nitrate hexahydrate (Y(N03)3-6H20n99.99%.), were selected as the Zr and Y sources. Ammonia solution (NH3-H2O, 17%,) in appropriate quantities (0-0.5 vol%) was added to adjust the acidity and to occur a radical exchange reaction between ZrO(NC<3)2 and NH4OH to produce an intermediate form of ZrO(OH)2. After the mixture was stirred for 1 h in an ice bath, a homogeneous colorless and transparent solution was obtained. Glass substrates (9.0 x 9.0 mm2) were cleaned ultrasonically in ethanol. and after the substrate was dried, they were further irradiated by UV light for 20 min for improving wettability. YSZ thin film was deposited on the glass substrate which was placed on the carbon electrode by applying the voltage 2.0-4.5V for 300s. Figure 1 is schematic diagram of the cell for thin film deposition. As-deposited films were flat and transparent but amorphous and so heat-treated at temperature of 573-773K for 1 h afterwards. Characterization of Thin Films The crystallization of the films was investigated using X-ray difjfractometry (XRD, 40 kV, 30 mA, CuKa, RintlOOO, R1GAKU). The chemical composition of the films were analyzed by X-ray fluorescence spectrometer (EDS, Eagle-uProbe, ED AX). Morphology of the films were observed by scanning electron microscope (SEM, S-3500, HITACHI), and optical microscope (BHSM-313MB, OLIMPUS).

Fig. 1. Schematic diagram of the cell for thin film deposition

78

• Novel Processing of Ceramics and Composites

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

RESULTS AND DISCUSSION During the film deposition constant voltage of 2.0 to 4.5 V was applied between electrodes. In electro-chemical deposition the radical exchange and precipitation reaction occurred by electric field assist and so were influenced by the applied voltage. Figure 2 shows the current change during deposition time. Film was mainly deposited on the glass substrate surface of negative carbon electrode side. Current become almost stable after 200 s. In this case, amount of added Nfhaq was 0.50 vol% to the YSZ solution. When the amount of added NF^aq was less than this case, the current change in early stage became moderate than that shown in figure 2. When the amount of added NF^aq exceeds 0.75 vol% or higher voltage was applied, a precipitation like gel tended to arise easily and became hard to deposit thin film. Stability of the ZrO(NÛ3)2, ZrO(OH)2 and deposition of the film by the assist of applied voltage was thought to be affected by the subtle acidity change.

300

Time /s Fig. 2. Relation between current and the reaction time As-deposited films were flat and transparent but amorphous and no obvious diffraction peaks were detected. Therefore samples were heat-treated at temperature 573-773K for 1 h afterwards. After heat treatment, transparent, pure, relatively hard thin films on the glass substrates. Figure 3 showed the SEM micrographs of the YSZ thin films after heat treatment at 773 K. Film thicknesses were about 0.05 to 0.3 um. When amount of added NF^aq was larger, the thickness of the films tended to be thicker for the same deposition time, and so cracks tended to occur during heat treatment due to shrinkage. Thinner films shrank mainly along the direction perpendicular to the substrate and film areas without cracks became large. Figure 4 showed XRD profiles for the YSZ thin films after heat treatment. From XRD patterns of the films, the peaks of YSZ (tetragonal phase) were observed at samples annealed at 673, 773 K. Definite crystallization can't be detected at samples annealed 573 K for 1 h. Composition of the films was almost same with that of aqueous solution. By the mist deposition method using same solution in our study monoclinic phase was easily generated and composition tends to deviate due to film growth condition.20 The electro-chemical deposition method thought

Novel Processing of Ceramics and Composites

• 79

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

to be easy, environment friendly, low temperature, and l o w cost method to fabricate Y S Z films compared other thin film preparation methods.

Figure 3. S E M micrographs o f the Y S Z thin films after heat treatment at 773 K. for 1 h. Applied voltage w a s 2.5 V during deposition and the amount o f added N H i a q were (a) 0.25vol% and (b) 0.50vol%.

30

40 50 2 6 /degree

60

Fig. 4. X R D profiles for the Y S Z thin films after heat treatment at 773 K for 1 h with applied voltage at the deposition. Figure 5,6 and 7 showed the relation between peak intensities, F W H M , lattice constant and applied voltage at the deposition. Course o f the higher values o f peak intensity from the films deposited by using Y S Z solution with added NH 3 aq amount o f 0.5 v o l % w a s mainly the thickness o f the films mentioned above. In addition F W H M of them are small. This indicated that Y S Z thin film crystallization process from aqueous zirconium oxynitrate solution is progressed by adding a small amount o f N f y a q . Well crystallized films w e r e formed when the applied voltage condition at the deposition w a s 2.5 V . Lattice constant of the annealed films

80

• Novel Processing of Ceramics and Composites

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

indicates higher values at 2.5 V than other applied voltages. When higher voltage than 2.5 V was applied deposition rate became large and so defect in the precipitates or films thought to increase and remained after heat treatment. It was confirmed that the degree of crystallization of the films were depended on the applied voltage and the amount of added NH3aq. Added NH 3 aq

3.0 3.5 Voltage / V

Fig. 5. Relation between (111) peak intensities and applied voltage at the deposition. (Heat treatment: 773 K for 1 h.)

3.0

3.5

4.0

4.5

Voltage / V

Fig. 6. Relation between FWHM of (111) peaks and applied voltage at the deposition. (Heat treatment: 773 K for 1 h.)

Fig. 7. Relation between lattice constants and applied voltage at the deposition. (Heat treatment: 773 K for 1 h.)

Novel Processing of Ceramics and Composites

• 81

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

CONCLUSION In our present study, YSZ thin films were fabricated on glass substrates in a aqueous complex solution by electro-chemical deposition. Zr0(N03)2-2Ff.20 and Y(N03)3-6HÎO were used Zr and Y sources and composition of the precursor aqueous solution is 94mol%Zr026mol%YOi 5. Ammonia solution in appropriate quantities was added to adjust the acidity and to occur a radical exchange reaction between ZrO(NÛ3)2 and NH4OH to produce an intermediate form of ZrO(OH)2. YSZ thin film was deposited on the glass substrate which was placed on the carbon electrode by applying the voltage 2.0-4.5V for 300s. As-deposited films were flat and transparent but amorphous. After heat-treated at temperature 573-773K for 1 h, the XRD peaks of YSZ were observed at samples annealed at 673, 773 K. After this heat treatment, transparent, pure, relatively hard thin films were formed on glass surface. Composition of the films was almost same with that of aqueous solution. It was confirmed that the degree of crystallization of the films were depended on the applied voltage and the amount of added NF^aq. REFERENCES 1 J. Eichler, U. Eisele, J. Rodel, "Mechanical Properties of Monoclinic Zirconia", J. Am. Ceram. Soc, 87, 1401-1403 (2004) 2 F. Bondioli, C. Leonelli, T. Manfredini, A. M. Ferrari, M.C. Caracoche, P. C. Rivas, A. M. Rodriguez, "Microwave-Hydrothermal Synthesis and Hyperfine Characterization of Praseodymium-Doped Nanometric Zirconia Powders", J. Am. Ceram. Soc, 88, 633-638(2005) 3 O. Vasylkiv, Y. Sakka, Y. Maeda, V. V. Skorokhod, "Sonochemical Preparation and Properties of Pt-2Y-TZP Nano-Composites", J. Am. Ceram. Soc, 88, 639-644 (2005) 4 K. Mukae, N. Mizutani, A. Saiki, X. Li, J. Nowotny, Z. Zhang, T. Bak, C.C.Sorrell, " Effect of surface preparation of zirconia on its reactivity with oxygen", J. Aust. Ceramic Soc, 34,76-79(1998) 5 T. Kiguchi, A.Saiki, K.Shinozaki, K. Terayama, N.Mizutani, "Effect of axial ratio on critical stress of ferroelastic domain switching in ceria-partially-stabilized zirconia", J. Ceram. Soc. Japan, 105, 871G875 (1997). 6 N. Wakiya, K. Shinozaki, N. Mizutani, "Improvement of magnetic properties of ( 111 )epitaxial nickel-zinc-ferrite thin films deposited on Si platform", Key Engineering Materials, 269, 245-248 (2004) 7 C-H. Chen, N. Wakiya, A. Saiki, K. Shinozaki, N. Mizutani, "Thickness and Roughness Analysis on YSZ/Si(001) Epitaxial Films with Ultra Thin Si02 Interface by X-Ray Reflectivity", Key Engineering Materials, 181-182, 121-124 (2000) G. R. Dickinson, C. Petorak, K. Bowman, R. W. Trice, "Stress Relaxation of Compression Loaded Plasma-Sprayed 7 Wt% Y203-ZrÛ2 Stand-Alone Coatings", J. Am. Ceram. Soc, 88, 2202-2208 (2005) 9 N. Wu, Z. Chen, S. X. Mao,"Hot Corrosion Mechanism of Composition Alumina/Yttria-Stabilized Zirconia Coating in Molten Sulfate-Vanadate Salt", J. Am. Ceram. Soc, 88,675-682 (2005) 10 Y. Ohya, M. Murayama, Y. Takahashi, "Electrical Properties of Zr0 2 Thin Films Doped With ln 2 0 3 by Sol-Gel Method", Key Engineering Materials, 169-170, 176-178 (1999) " K. Sasaki, L. J. Gauckler, "Microstructure-Property Relations of SOFC Electrodes: Importance of Microstructural Optimization of La(Sr)Mn03 Cathodes on Zr02(Y2Û3) Electrolytes", Key Engineering Materials, 169-170, 201-204 (1999)

82

• Novel Processing of Ceramics and Composites

Fabrication of YSZ Thin Films in an Aqueous Solution by Electrochemical Deposition

12

(1995)

A. Bastianini, G. A. Battiston, R. Gerbasi, M. Porchia, S. Daolio, J. Phys D 5, 525-532.

13

H. Wendel, H. Holzschuh, H. Suhr, G. Erker. S. Dehnicke, M. Mena, Mod. Phys. Lett. 54,1215-1217(1990) 14 T. Yao, T. Irui, A. Ariyoshi, "Novel Method for Zirconium Oxide Synthesis from Aqueous Solution", J. Am. Ceram. Soc, 79, 3329 (1996). 15 N. Ozawa, T. Yao, Trans. Mater. Res. Soc. Jpn., 18, 321 (2003) 16 M. Agarwal, M. R. De Guier, A. H. Heuer, "Synthesis of Zr0 2 and Y203-Doped Zr0 2 Thin Films Using Self-Assembled Monolayers", J. Am. Ceram. Soc, 80. 2967 (1997) 17 T. P. Niesen, M. R. De Guire, J. Bill, F. Aldinger, M. Ruhle, A. Fischer, F. Jentoft, R. Schlogl, J. Mater. Res., 14,2464-2475 (1999) 18 Y. Gao, Y. Masuda, T. Yonezawa, K. Koumoto, "Site-Selective Deposition and Micropatteming of Zirconia Thin Films on Templates of Self-Assembled Monolayers", J. Ceram. Soc. Jpn., 110, 379,(2002) 19 Y. Gao, Y. Masuda, H. Ohta, and K. Koumoto, "Room-Temperature Preparation of Zr02 Precursor Thin film in an Aqueous Peroxozirconium-Complex Solution", Chem. Mater. 16, 2615-2622(2004) 20 A. Saiki, Y. Fujisawa, T. Hashizume, K. Terayama, "Yttria Stabilized Zirconia Thin films formation from an aqueous solution by mist deposition", Proceedings ofPACRIMó, PACRIM-S10-14-2005 (2005)

Novel Processing of Ceramics and Composites

• 83

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Plasma Synthesis

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

PREPARATION AND CHARACTERIZATION OF EPITAXIAL Fej-.T^Oj SOLID SOLUTION FILMS Tatsuo Fujii, Hideki Hashimoto, Yusuke Takada, Makoto Nakanislii, and Jun Takada Department of Applied Chemistry, Faculty of Engineering, Okayama University Tsushima-naka 3-1-1 Okayama 700-8530, Japan ABSTRACT Thin films of hematite-ilmenite (a-Fe20j-FeTi03) solid solution series are proposed to be one of the new high-temperature magnetic semiconductors. We have successfully prepared wellcrystallized epitaxial Fe2-.TTi,03filmswith various Ti content, .r, between 0.6 and 1.0. The films were epitaxially formed on cx-AkOj^OOl) single-crystalline substrates by reactive helicon plasma sputtering technique. The films prepared at lower substrate temperature below 700 °C had the R3c structure, in which the Fe and Ti ions randomly occupied octahedral interstices. While thefilmsprepared at higher substrate temperature above 700 °C had the R3 structure, in which the octahedral interstices occupied by the Fe and Ti ions were arranged in an ordered way along the e-axis. Moreover the films prepared at the higher substrate temperature consisted of nearly stoichiometric oxygen concentration. However, with decreasing the substrate temperature the films seemed to have large cation vacancies and Fe2+ ions were gradually oxidized to Fe3+. With decreasing the Ti contentfromx= 1.0 to 0.6, the electric resistivity of the stoichiometric films was sharply decreased from 10 to 0.1 fícm because of the formation of mixed valence states between Fe2+ and Fe3+. The stoichiometric films with *=0.6 were ferrimagnetic semiconductors at room temperature when the films had the R3 structure. INTRODUCTION Solid solution series between hematite (a-Fe^Oi) and ilmenite (FeTiOj) are one of the candidates for noble electronics and spintronics materials. Both of oc-Fe203 and FeTi0.i are antiferromagnetic insulators. However, the intermediate compositions between them are ferrimagnetic and also semiconducting.12 Recent theoretical calculation predicted that the Tisubstituted tx-Fe20j had possibility to lead a magnetic semiconductor with strongly spinpolarized transport properties and with very high Curie temperature (Tc) of about 1000 K..3 The carrier type of Fe2-,Tit03 solid solutions can be controlled form n-type to p-type by carefully selecting the Ti content, x. The Fe2.1Ti.tO3 with ,r<0.73 are n-type semiconductor while those with *>0.73 are p-type one.2 Both a-Fe2Û3 and FeTi03 are characterized by the crystal structure derived from the corundum structure as a-Fe203. The structure consists of hexagonal closepacking oxygen ions with the two thirds of octahedral interstices filled by metal cations. If all cation sites are crystallographically equivalent as a-Fe2Û3, the structure has R3c symmetry. While FeTi03 has R3 symmetry, in which alternate Fe sites along the c-axis are replaced by Ti. The equilibrium phase diagram of a-Fe203-FeTi03 clearly indicated the boundary between the disordered (R3c) and ordered (R3) phases at about x=0.5. Only the solid solutions with R3 symmetry exhibit large ferrimagnetic moments.1 We had previously reported the preparation of well-crystallized epitaxial Fe2-.rTii03 films by various thin-film preparation methods such as reactive vapor deposition and sputtering techniques.4"7 The epitaxial films of the disordered phase with the R3c symmetry easily formed on the a-AW^OOl) substrate. However the films of

87

Preparation and Characterization of Epitaxial Fe2_xTix03 Solid Solution Films

the ordered phase with the R3 symmetry required much strict preparation conditions such as the higher substrate temperature and severely optimized oxygen pressure. Some films contained smaller Fe2+/Fe3+ ratio than that expected from the stoichiometric Ve2-xTixO). Fe2+ ions in Fe23+ xT\xOj seemed to easily oxidize to Fe introducing the oxygen nonstoichiometry. In the present study we report the structural, magnetic and electric properties of epitaxial Fe2-xTix03 films with various Ti content, x, between 0.6 and 1.0 as a function of the substrate temperature during the deposition. All films were prepared by reactive sputtering technique with ultra-high vacuum system to control the oxygen pressure precisely. EXPERIMENTS Fe2-^TiJ:03filmswere prepared on a-Al2O3(001) single-crystalline substrate by using helicon plasma sputtering system with base pressure of 10"7 Pa. Schematic illustration of the helicon plasma sputtering system we used is shown in Fig. 1. A Fe-Ti alloy target with three different compositions of Fei-yTiy (y=0A, 0.5, 0.6) was placed on the cathode. During the sputtering deposition, applied rf powers to the cathode and the helicon coil were fixed to 120 W and 50 W, respectively. 10.0 ccm of pure Ar gas and 0.6 ccm of pure O2 gas were introduced into the system. Residual gas species during the sputtering deposition were monitored in situ by a mass analyzer unit in order to control the oxygen partial pressure precisely. Substrate temperature (Ts) was fixed to certain values ranging from 400 to 750°C. Thickness of all sample films was fixed to 100 nm. The crystal structure of the deposited films was identified by using a conventional x-ray diffraction (XRD) technique with monochromated Cu Ka radiation and 6/28-scaning from 20=15° to RHEED Gun

-Heater

U)

k 1.0

1

1— ■

■ -

1 '

•

S 0.9

-

•

£0.8

Q-MASS

-

_

.c TMP

10.7

RP

-

c

£0.6 0

¡-

•

1

0.4

1

0.5

'

-

0.6

Ti content, y, in Fei-/Tïj> targets

Fig.l. Schematic illustration of the helicon plasma sputtering system we used.

Fig.2. Ti content, x, of the prepared Fe2.jTix03 films as a function of the target composition.

90°. However the scannings between 29=41.0 and 42.5° were skipped to protect the x-ray counter form the strong reflection intensity of a-Al2O3(006) for the single-crystalline substrate. The Fe/Ti content ratio of the deposited films was analyzed by energy dispersive x-ray

88

• Novel Processing of Ceramics and Composites

Preparation and Characterization of Epitaxial Fe2-xTix03 Solid Solution Films

spectroscopy (EDS). The electric and magnetic properties were characterized by conversion electron Mössbauer spectroscopy (CEMS), vibrating sample magnetometer (VSM), and dc fourprobe resistivity measurements. RESULTS AND DISCUSSION Fig. 2 shows the analyzed Ti content, x, of the prepared Fe2-iTij,03 films as a function of the target composition, v, of the Fe^Tiy alloy targets. The prepared ¥e2-xTixO}filmshad much smaller Ti content than that of the target alloys. The difference in the Fe/Ti ratio between the films and the targets was probably due to the difference in the sputtering yields between Fe and Ti atoms. The sputtering yield of Ti atoms by Ar+ bombardments was much smaller than that of Fe atoms.8 The films prepared by the Ti-rich target, y=Q.6, gave the nearly stoichiometric FeTiCh. While the films prepared by the Fe-rich target, y=0A, gave the Fe1.4Tio.6O3 solid solution composition, which should have the femmagnetism with the high Tc above room temperature in the bulk system.' Shown in Fig. 3 are typical XRD patterns of Fei 02TÍ0.98O3filmsprepared by the Feo.-iTioó target at various substrate temperatures between 400 and 700°C. The XRD patterns clearly indicated that the (OOl)-oriented FeTi03 were epitaxially formed on the substrate over the all Ts's. The films having the corundum structure with the R3c symmetry showed only two diffractions indexed as (006) and (0012). While the ilmenite structure with the R3 symmetry possessed additional two diffractions indexed as (003) and (009). Only the film deposited at higher TS(=700°C) had the R3 structure.

2 e (deg.) Cu-Ka

Velocity (mm/s)

Fig.3. XRD patterns of Fe1.02Tio.9sO3filmsas a function of the substrate temperature (Ts).

Fig.4. Room temperature CEMS spectra of Fei 02TÍ098O3films.The spectra were fitted to assume the two asymmetric doublets.

Novel Processing of Ceramics and Composites

• 89

Preparation and Characterization of Epitaxial Fe2_xTix03 Solid Solution Films

Table I. Fitted parameters for the room temperature CEMS spectra shown in Fig. 4. Substrate Quadrupole .„ Isomer shifts ,.„. Intensity Area splitting temperature ratio (%) (mm/s) (mm/s) (°C) 2+ Fe 1.08 75 0.75 1 :3 700 25 Fe3+ 0.27 1 :3 0.72 66 Fe2+ 0.91 1.35 1:2.5 600 0.27 34 Fe3+ 1 :2.5 0.73 64 Fe2+ 1.00 1.27 1:1.4 500 0.68 36 Fe3+ 1 : 1.4 0.38 Fe2+ 1.01 1 : 1 48 1.49 400 52 Fe3+ 0.68 1 :1 0.42

Other films deposited at lower Ts(<700°C) had the R3c structure. The order and disorder cation arrangements between the R3 and R3c structures were seriously influenced by the Ts- In order to produce the ordered array of Fe and Ti ions along the c-axis, the higher substrate temperature should be required to supply the thermal activation energy to the Fe and Ti ions moving from the randomly occupied octahedral interstices of the R3c structure. The Ts also had large influence on the oxygen nonstoichiometry of the deposited films. Fig. 4 shows the room temperature CEMS spectra of the Fe1.02Tio.98O3filmsprepared at various Ts's. All sample films were paramagnetic at room temperature. The CEMS spectra consisted of two asymmetric doublets, though the stoichiometric FeTi03 should have one doublet peaks assigned to the octahedral Fe2+ ions.9 Moreover the optical arraignment of the incident y-ray direction parallel to the principal electric-field gradient axis (c-axis) of epitaxial FeTiÛ3 should produce the asymmetric doublet pattern with an intensity ratio of 1:3. The fitted parameters for the CEMS spectra were listed in Table I. The one with larger isomer shifts was assigned to the Fe2+ component, while the other with smaller isomer shifts was due to the Fe3+ component. The spectrum for the Fe1.02Tio.9sO3filmprepared at Ts=700°C was decomposed mainly to the Fe2+ component. Only a small amount of Fe3+ ions were contaminated in the film. However, with decreasing the Ts, the films contained an increasing amount of Fe3+ ions. Moreover the anisotropic doublet intensity ratio of 1:3 for the perfect crystal was gradually changed to the isotropic ratio of 1:1. This suggested the local symmetry around the Fe ions in FeTiÛ3 films was broken to introduce the structural vacancies. Fe2+ ions in FeTiÛ3 could be oxidized to Fe3+ ions to form the cation vacancies for the charge compensation. The higher Ts should be required to produce the stoichiometric films. With decreasing the Ti content,^ of Fei-yTi^ targets, the required substrate temperature to form the stoichiometric ilmenite with the R3 symmetry seemed to increase gradually. Fig. 5 shows the typical XRD patterns of the Fei 4TÍ0.6O3filmsprepared by the Feo.ôTio.4 target. The film prepared at Ts=700°C had the corundum structure with the R3c symmetry. Only the film deposited at much higher substrate temperature of Ts=750°C showed the week diffraction line

90

• Novel Processing of Ceramics and Composites

Preparation and Characterization of Epitaxial Fe2_xTix03 Solid Solution Films

indexed as (009) assigned to the ilmenite structure with the R3 symmetry. The relative intensities of the (003) and (009) diffractions, characteristic to the formation of the R3 structure, were considerably small than

+5

• Fe M T bB°3 Osubstrate

~i

13.96

1

1

r

Ü

v 13.94 a» E

a O

TsfC) • O 750

2. 13.92 fi

13.90 0.6 0.7 0.8 0.9 1.0 Ti content, x, in Fe2xTii03 films

700 40 60 26>(deg.)Cu-Ka

80

Fig.5. XRD patterns of Fei 4TÍ06O3 films as a function of the substrate temperature (Ts).

Fig.6. Out-of-plane lattice parameter, c, for the deposited films as a function of the Ti content.

the other films prepared from the Ti-rich targets as Feo 5TÍ0.5 and Feo.4Tio.6- The solid solution films with the higher Fe concentration made it difficult to form the R3 structure, probably because the difference in the formation energies between the R3 and R3c structures was reduced. Fig. 6 shows the out-of-plane lattice parameter, c, of the deposited films at TS=700°C as a function of the Ti content. The c-axis parameter was lineally decreased with decreasing the Ti 400

•1-

O

3

§ 200

078K V RT

c 0

TO N V

gi 100 5

i>-200 n 5 -300

-400

c 0

,S

0 7

2

-10 0 10 Magnetic Field (kOe) fr'ig.7. Room and low temperature magnetization curves of Fei igTio.8203 film prepared at TS=700°C.

-

7

e

u

0.6 0.8 1.0 Ti content, x, in Fe2xTii03 films Fig.8. Room and low temperature saturation magnetization of Fe2-iTij;03 films with the R3 structure as a function of the Ti content.

Novel Processing of Ceramics and Composites

• 91

Preparation and Characterization of Epitaxial Fe 2 . x Ti x 0 3 Solid Solution Films

1

1 ' ' ' ' i ' ' ' ' i ' ' ' ' i'

Fel02Tl0 98O3 1803

I

Fei 18TI082O3

Fei 4Tio 603

1

■

1

-

-10

■

■

.

1

.

.

.

.

1

.

■

.

■

1

-5 0 5 Velocity (mm/s)

.

■

■

.

1

0.6 0.8 1.0 Ti content, x, in Fe2-xTn03 films

■

10

Fig.9. Room temperature CEMS spectra of various Fe2-iTixO}filmswith the R3 structure.

Fig. 10. Electric resistivity measured at room temperature for the nearly stoichiometric Fe2-iTi,03 films as a function of the Ti content.

content of the deposited films. The linear relationship between the lattice parameter and the composition seemed to be governed by the Vegard's law for the solid solution system. However the c-axis length of FeTiCb bulk crystals reported in literature was c=14.083 Â,10 which was much larger than that for our prepared FeTiÛ3 films. One of the possibilities for this deviation should be due to the introduced metal vacancies into the thin film structures as discussed above. A small amount of Fe2+ ions in the deposited FeTiÛ3 films were oxidized to Fe3+ ions to introduce the vacancies. The solid solution films of the R3 structure were ferrimagnetic below the Tc. Typical magnetization curves for an Fei isTio 82O3filmdeposited at Ts=700°C are shown in Fig. 7. The Tc of the Fei i8Tio8203filmwas measured to be about 200 K. Therefore the magnetization curve measured at 78 K indicated the large hysteresis loop with the large saturation magnetization. While the one measured at room temperature was paramagnetic. The saturation magnetization values at room and low temperatures are plotted in Fig. 8 as a function of the Ti content of the Fe2-jrTix03 films with the R3 symmetry. The film of FeTiÛ3 had no saturation magnetization even at low temperature because it should be antiferromagnetic. On the other hand the composition of Fei 4TÍ06O3 was expected to have large ferromagnetic moments at room temperature, because it should have higher Tc of about 400 K.1 However the observed saturation magnetization of the Fei.4Tio603 film was very small even at 78 K. The disappointing small magnetization of the prepared Fei 4TÍ0.6O3filmwas due to the inferior cation ordering between Fe and Ti ions to form the R3 structure. The inferior cation ordering in the Fei 4TÍ06O3filmwas fully consistent with the XRD results for the film as discussed above. Fig. 9 shows the room temperature CEMS spectra of various Fe2.xTiI03 films with different Ti contents. All films had the R3 structure prepared at higher T s 's. The films of Ti-rich compositions, Fei 02TÍ098O3 and Fe1.i8Tio.s2O3, had the paramagnetic doublet patterns because of their lower Tc or Néel temperatures below the room temperature. However, the spectrum for the

92

• Novel Processing of Ceramics and Composites

Preparation and Characterization of Epitaxial Fe2-xTi)<03 Solid Solution Films

Fe1.4Tio.6O3 film clearly showed the magnetically split sextet pattern. This was a clear proof that the Fei 4TÍ06O3filmhad the higher Tc above the room temperature, in spite of ¡ts small saturation magnetization. The Tc of Fe2-jrTr,03 films was linearly increased with decreasing the Ti concentration. The electric properties of Fe2-,Tij03 films were also influenced by the Ti content of the films. Fig. 10 shows the electric resistivity of nearly stoichiometric Fe2-J[Ti;r03 films measured at room temperature as a function of the Ti concentration. The temperature dependence of the resistivities suggested their semiconducting natures for all films. With decreasing the Ti content fromx=1.0 to 0.6, the resistivity of Fe2-,Ti;r03 was sharply dropped in two digits from 10 to 0.1 Qcm. The mixed valence states between Fe2+ and Fe3+ ions could produce the good conductivity. The films of the Ti-rich solid solution compounds with the R3 structure showed large possibilities to be a magnetic semiconductor at room temperature. CONCLUSION Well-crystallized epitaxial Fe2.ITiJt03 films with various Ti content, x, between 0.6 and 1.0 were successfully prepared on the a-Al2O3(0001) single-crystalline substrate. The films prepared at lower Ts below 700 °C had the R3c structure. With decreasing the Ts, the prepared films seemed to contain larger amount of cation vacancies due to the partly oxidization of Fe2+ ions to Fe3+. Only the films prepared at higher Ts above 700°C forx=0.98 and 0.82, and 750°C for*=0.6 had the nearly stoichiometric R3 structure. With decreasing the Ti content for the solid solution films, the formation of the R3 structure became more difficult. The obtained Fei 4TÍ0.6O3 film showing the R3 symmetry had fairly small ferrimagnetic moments both at room and low temperatures due to the inferior cation ordering between the Fe and Ti sites. By the way, with decreasing the Ti content from x=\ .0 to 0.6, the electric resistivity of the stoichiometric Fe2. jTijÛ3 was sharply dropped from 10 to 0.1 Qcm because of the formation of mixed valence states between Fe + and Fe +. The Fe1.4Tio.6O3filmswith the R3 symmetry were ferrimagnetic semiconductors at room temperature. REFERENCES 'Y. Ishikawa, "Magnetic properties of ilmenite-hematite system at low temperature", J. Phys. Soc. Jpn., 17, 1835-1844 (1962). 2 Y. Ishikawa, "Electrical Properties of FeTi03-Fe2Û3 Solid Solution Series", J. Phys. Soc. Jpn., 13,37-42(1958). 3 W.H. Butler, A. Bandyopadhyay, and R. Srinivasan, "Electronic and magnetic structure of a 1000 K magnetic semiconductor: a-hematite (Ti)", J. Appl. Phys., 93, 7882-7884 (2003). 4 T. Fujii, K. Ayama, M. Nakanishi and J. Takada, "Effect of Ti doping on crystallographic and magnetic properties of epitaxial hematite films", J. Magn. Soc. Jpn., 22, Suppl. SI, 206-209 (1998). 5 T. Fujii, K. Ayama, M. Nakanishi, M. Sohma, K. Kawaguchi and J. Takada, "Electric and magnetic properties of epitaxial Fe2.JTiI03+g films", Mater. Res. Soc. Symp. Proc, 623, 191196, (2000). 6 T. Fujii, M. Sadai, M. Kayano, M. Nakanishi and J. Takada, "Nonstoichiometry of epitaxial FeTi03+5 films", Mater. Res. Soc. Symp. Proc, 746, Q6.10.1-6, (2003). 7 T. Fujii, M. Kayano, Y. Takada, M. Nakanishi, and J. Takada, "Ilmenite-hematite solid solution films for novel electronic devices ", J. Solid State Ionics, 172, 289-292, (2004).

Novel Processing of Ceramics and Composites

• 93

Preparation and Characterization of Epitaxial Fe 2 . x Ti x 0 3 Solid Solution Films

8

N. Laegreid and G. K. Wehner, J. Appl. Phys., "Sputtering Yields of Metals for Ar+ and Ne Ions with Energies from 50 to 600 eV", 32, 365-369 (1961). 9 G. Shirane, D. E. Cox, W. J. Takei and S. L. Ruby, "A Study of the Magnetic Properties of the FeTi03-aFe203 System by Neutron Diffraction and the Mössbauer Effect", J. Phys. Soc. Jpn., 17, 1598-1611(1962). 10 B. A. Wechsler and C. T. Prewitt, "Crystal structure of ilmenite (FeTi03) at high temperature and at high pressure", Am. Mineral., 69, 176-185 (1984). +

94

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Solid Freeform Fabrication

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society MICROTOMOGRAPHY OF SOLID FREEFORM FABRICATION Jay C. Hanan, James E. Smay Oklahoma State University 218 Engineering North Stillwater, OK 74078 Francesco DeCarlo, Yong Chu Argonne National Lab 9700 S. Cass Ave Argonne, IL, 60439 ABSTRACT Solid freeform fabricated ceramics promise unprecedented control over shape and dimension of final ceramic forms. The advanced processing technique lends itself to intricate shapes of 3-D structures. For some applications, precise control over final dimensions is critical. An advanced tomographic analysis technique was used to observe potential inhomogeneities in the fabrication process. The analysis technique can be used before and after sintering to trace sources of dimensional error leading to refinement of the processing method. In addition, a new instrument is under construction which, in concert with the tomographic method, will allow residual and applied strain measurements in as processed ceramic components. Examples of the tomographic method applied to solid freeform fabrication are presented along with the application of stress measurements. INTRODUCTION Through recent developments in ceramic rapid prototype processing, specifically robocasting, control of material type and part shape has become possible.1' 2 Using ceramics with similar sintering temperatures, but different coefficients of thermal expansion (C.T.E), residual stress distributions may be engineered. A favorable residual stress could then lead to improved fracture behavior. C.T.E. mismatch causing compressive stress on tensile surfaces will decrease the effective tensile load at the specimen surface, limiting damage initiation. A potential near-term application of these ceramic composites is dental crowns.3 Other applications are also under consideration; however, the process parameters and mechanical characteristics must be understood. For this study, the objective is to nondestructively examine robocast ceramics. Primarily an examination of porosity with an analysis of potential to measure interface characteristics and residual strain is desired. Thus a method to make specimens amenable for X-ray micro-tomography (uCT) of these layered composites was developed. Tomography is a method of 3-D imaging reconstructed from multiple views about a rotation axis. X-ray micro-tomography is a form of tomography with high spatial resolution and the capability to reconstruct a complete volume rather than a surface as in conventional microscopic techniques. It provides a powerful tool for the examination of microstructure. In addition, since the method is non-destructive, |xCT enables in-situ

97

Microtomography of Solid Freeform Fabrication

measurements of the microstructural response to changing environments such as applied stress, temperature, or chemical potentials. Micro tomography has seen a decade of development but has only recently become efficient and capable of high resolution over a significant field of view. Recent advances include voxel resolutions down to 150 nm and full resolution scans on the order of milliseconds 5' Many of the developments in (iCT require use of synchrotron radiation due to the high flux and wide range of available energies. For this study, synchrotron X-ray |aCT was employed to observe the microstructural evolution of freeform fabricated monolithic and composite ceramics during sintering and to identify regions of interest for analysis of prescribed residual strains. Microtomography provided understanding of void formation and shrinkage in 3-D space in monolithic and composite specimens. METHOD Freeform fabrication: The robocasting freeform fabrication process is based on the direct writing of colloidal gels to assemble three-dimensional objects by printing of stacked 2-D layers. The colloidal inks are housed in syringes and dispensed through nozzles at a controlled volumetric flow rate while the nozzle is moved in the x-y plane (Figure 1). The process consists of first designing the part to be built and formulating the appropriate ink. Next, the ink is printed into the desired shape and then dried and sintered. Each step of the process requires careful attention to detail. For example, when printing with multiple nozzles, one must control relative tip positions, on-off switching of ink flow, compatible Theological properties of the ink, and compatible sintering behavior. Here, a two nozzle printing scheme was used where the inks were aqueous gels of alumina (A1203 with 0.5 w% MgO) and zirconia (Zr0 2 with 5.4 mol% Y). Sintering was for 2 hours at 1450°C.

Figure 1 Schematic of robocasting process. Inset photograph taken during the composite sample fabrication shows construction of 80% zirconia phase.

98

• Novel Processing of Ceramics and Composites

Microtomography of Solid Freeform Fabrication

The solids loading in each of the inks was held fixed at 47% by volume. Mixed compositions with 4:1 and 1:1 volumetric ratios were prepared by mixing aliquots of the pure inks. Four sets of samples were prepared: (i) pure alumina, (ii) pure zirconia, (iii) pure 4:1 and 1:1 mixtures, (iv) layered composites with 4:1 on pure alumina and 1:1 on pure alumina, and (v) shell composites with pure alumina shells on 4:1 or 1:1 cores. The sample dimensions were kept small to facilitate X-ray transmission and high-resolution tomography. A printing sequence consisting of stacked circles was used to form cylinders for layered composites and shell composites. The extruded filament is 200 urn in diameter and the deposition speed was lOmm/s.

layered composite

shell composite

diameter - 1.8 mm, height - 3 mm

diameter ~ 1.8 mm, height - 3 mm

Figure 2 Left: Photograph of layered composite after tomographic analysis (coloration due to radiation exposure). Schematic of layered (Center) and shell composites (Right). Tomography The yiCT experiments were performed at the bending magnet beamline number 2 (2BM) of the Advanced Photon Source (APS) at the Argonne National Laboratory. This bending magnet beamline is under development for simultaneous tomography and diffraction measurements allowing imaging and simultaneous characterization of stress. The advantage to combining the techniques is the ability to unambiguously locate cracks, composite interfaces, or other areas of interest. The spatial resolution, while dependent on sample characteristics and the optics used, was 2.5 urn per pixel for the tomographic analysis.

Novel Processing of Ceramics and Composites

• 99

Microtomography of Solid Freeform Fabrication

Tomography in general is a method of 3-dimentional measurement of X-ray attenuation, /: I/Io = exp(-/ux)

(1)

where /« is the intensity of the unattenuated beam and x is the thickness of the homogeneous material and ju the linear attenuation coefficient. Radiographs contain information on the attenuation and phase in 2-dimetions from the entire thickness of the sample they expose. At short detector distances the phase component may be neglected. When the sample is rotated in front of an imaging detector, many radiographs associated with each rotation angle, <> | , are observed. The observed attenuation from each radiograph as a function of (j> is dependent on the sample attenuation in both directions of an intersecting plane. Thus, the contributions to the attenuation in 3-D can be decoupled from the many 2-D radiographs through a procedure called reconstruction.8 The resolution is experimentally determined by the optical magnification of an X-ray scintillator plate placed behind the sample (Figure 3). For tomography, an optical magnification of 2.5x with a camera pixel count of 1024x1024 in the x and y directions was used. After reconstruction the 3-dimentional dimension of a voxel is 2.5x2.5x2.5 urn3.

Figure 3 Photograph with the Diffraction and Transmission beam path labeled. Each beam emanates from the sample at the center of the diffractometer. The tomography samples were fabricated cylindrically symmetric so that absorption could be tuned to the imager's greatest sensitivity for all directions. At 2BM using 17.85 keV radiation the integration time for tomography was 7 seconds. At least 724

100

• Novel Processing of Ceramics and Composites

Microtomography of Solid Freeform Fabrication

radiographs were combined at equal steps about 180° of the (j> rotational axis including bright field frames before and after the § rotation and a dark field frame completing the measurement. The frames are normalized to the bright field and combined using a reconstruction algorithm into 1024 slices along the axis of the tomographed sample. RESULTS Micro-Tomography Microtomography of the alumina cylinder before and after sintering provided the density change. An example of a volume rendered tomograph showing the surface of the sample is shown in Figure 4. 9 The power of tomography is more evident through the surface rendered image (Figure 5) which exposes the 3-dimetional distribution of voids within the sample.

Figure 4 Volume rendered image of the sintered alumina cylinder.

Quantitative analysis was preformed on 2-D slices of the tomographs. The indention (visible in Figure 4) on the top surface of the sample was used as a reference to align the data sets. From the registered slices a comparison of void size distribution and shrinkage was performed. The largest void was observed to shrink to 35% of its original area. From measurements of void area on an example tomographic slice (Figure 6). The total area of voids greater than 26 (im2 decreased to 70% of the initial area after sintering. The distribution of these voids is given in Figure 7. An expected decrease in size and number was observed.

Novel Processing of Ceramics and Composites

• 101

Microtomography of Solid Freeform Fabrication

Figure S Surface rendered image of a 0.25 mm thick section of the sintered alumina cylinder. The section is tilted upward with a perspective of looking through the underside of the section. The outer surface is clearly visible. The dispersion of voids is in 3-dimentions.

Figure 6 Measured void areas (red) are labeled in green. The center black area is a slice of the indention used to register the rotation of the cylinder between measurements (not a void).

102

• Novel Processing of Ceramics and Composites

Microtomography of Solid Freeform Fabrication

Figure 7 Distribution of void areas as seen in (Figure 6) for before and after sintering.

Similar measurements were performed in the composite samples (for example Figure 8). The analysis is underway.

Figure 8 Volume rendered tomograph of a layered composite.

Novel Processing of Ceramics and Composites

• 103

Microtomography of Solid Freeform Fabrication

CONCLUSIONS A procedure to robocast ceramic monolithic cylinders and layered composites suitable for X-ray microtomography study was completed. A new instrument is under construction developed for in-situ tomography and diffraction microscopy optimized for deformation and fracture mechanics of metals and oxides. From tomography, real-space measurements are performed, from diffraction residual and applied strain measurements may be performed. Tomographic analysis observed shrinking of voids from sintering of the alumina ceramic freeform precursor. The technique can observe density to trace sources of dimensional error leading to refinement of the processing method. Zirconium/Alumina composites were also characterized using combined microtomographic imaging and micro-X-ray diffraction. The analysis is underway, and future publications will show cracking was observed with u-CT and strain measurements using X-ray diffraction reveal the residual stress from the interface. ACKNOWLEDGMENTS Sponsorship by NIH-NIDCR. Use of the Advanced Photon Source was supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. W-31-109-Eng-38. REFERENCES 1

S. Nadkarni,, J. E. Smay, "Concentrated Barium Titanate and Barium Zirconate Colloidal Gels for Direct Writing of Periodic Dielectric Structures," J. Am. Ceram. Soc., (2005 in press) 2 J. E. Smay, B. A. Tuttle, J. Cesarano III, J. A. Lewis, "Directed Colloidal Assembly of Linear and Annular PZT Arrays", J. Am. Ceram. Soc, 87(2), 293-295, (2004). 3 J. E. Smay, "Solid Freeform Fabrication of Graded Composition Dental Crowns." 6' Pacific Rim Conference on Ceramic and Glass Technology, American Ceramic Society (2005). 4 J. C. Hanan, C. Veazey, M. D. Demetriou, F. DeCarlo, J. S. Thompson, "Microtomography of Amorphous Metal During Thermo-Plastic Foaming." Adv. XRayAnal, 49, in press, (2005). 5 A. Sasov, "Laboratory system for X-ray nanotomography." Adv. X-Ray Anal., 49, in press, (2005). 6 L. Salvo, P. Cloetens, E. Maire, S. Zabler, J.J. Blandin, J.Y. Buffiere, W. Ludwig, E. Boiler, D. Bellet, C. Josserond, "X-ray micro-tomography an attractive characterisation technique in materials science." Nuclear Instruments and Methods in Physics Research B 200, 273-286, (2003). 7 J. E Smay, J. Cesarano III, J. A. Lewis,"Colloidal Inks for Directed Assembly of 3-D Periodic Structures", Langmuir, 18(14), 5429-37, (2002). 8 A. C. Kak, M. Slaney, Principles of Computerized Tomographic Imaging, IEEE Press (1988). 9 3-D data analysis software, http://www.amiravis.com/

104

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Floe Casting

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

FABRICATION AND EVALUATION OF TRANSPARENT AMORPHOUS Si0 2 SINTERED BODY THROUGH FLOC-CASTING D. Hiratsuka, J. Tatami, T. Wakihara, K. Komeya, T. Meguro, Graduate School of Environment and Information Sciences, Yokohama National University 79-7 Tokiwadai, Hodogaya-Ku, Yokohama 240-8501, Japan M. Ibukiyama, DENKI KAGAKU KOGYO K. K Material Research Center, 3-5-1, Asahimachi, Machida, Tokyo 194-8560, Japan ABSTRACT We prepared dense green compacts of high purity spherical fused SÍO2 powder by the floc-casting technique and fired them in atmosphere to obtain transparent SÍO2 sintered bodies. Highly concentrated and dispersed slurries were prepared. The slurries were solidified in humidity-controlled oven. Homogeneous and dense green compacts were fabricated through vacuuming, centrifuging and floc-casting technique1. The compacts were fired at 1400 °C for 30 min in atmosphere. No crystallization was observed in the sintered bodies. The SÍO2 sintered bodies with a thickness of 2 mm had high transmissivity in the range of visible light. INTRODUCTION Transparent SÍO2 glass is an important material with many desirable properties, such as low thermal expansion coefficient, high electrical insulation, high chemical resistance and high UV transparency, etc. It is frequently used for high efficiency lamps, crucibles for melting high-purity silicon, IC photo-mask substrate, and lens material for excimer stepper equipment. However, because SÍO2 glass has high softening point (~1600 C), fabrication of transparent SÍO2 glass by a melting method requires high temperatures of 2000-2300 °C. So many researches have been performed for the production of transparent SÍO2 glass in low energy cost using sol-gel and sintering method.2"11 The sol-gel method has, however, several drawbacks; for example, it takes a long time to fabricate and raw materials are expensive. In the previous study on sintering SÍO2 glass,5'12 high-purity powder prepared by sol-gel technique was used as a raw material. Since the powder has many hydroxyl groups, atmosphere control is needed to eliminate them and to control devitrification. Although fused SÍO2 powder has few hydroxyl groups, there are no reports on their sintering to fabricate transparent SÍO2 sintered body. In order to obtain high transparency, because residual pores must be eliminated during sintering process, high density and homogeneity is essentially required to the green body. Uematsu et al. prepared dense and homogeneous compacts with a floc-casting technique in AI2O3. In this method, a green body is made through the control of dispersion and flocculation of highly concentrated slurry by pH value or temperature. The purpose of this study is to fabricate transparent SÍO2 sintered body through floc-casting technique. 107

Fabrication and Evaluation of Transparent Amorphous Si0 2 Sintered Body

EXPERIMENTAL PROCEDURE Spherical fused Si0 2 powder (purity >99.99 %, diameter: 0.54 urn, DENKI KAGAKU KOGYO Co. Ltd., Japan.) was used as a starting material, which has little amount of chemical impurities. SÍO2 powder was dispersed in ion-exchanged water with a dispersant (SERUNA D-735, Chukyoyushi, Co. Ltd., Japan). In order to prepare a highly concentrated and dispersed slurry, wetting and flowing points of the powder and dispersant system were evaluated. In this study, wetting and flowing points were defined as quantities of water when the powder becomes a lump and the lump becomes liquid, respectively. The most suitable condition for the highly concentrated and dispersed slurry was decided from the smallest value of wetting and flowing points. Based on the result, SÍO2 slurry was prepared by ball milling for 24 h. After vacuuming, highly concentrated slurry was cast into a silicone resin mold, followed by centrifuging the slurry. The slurries were centrifuged. The solidification was took place in humidity-controlled oven. Green compacts were demolded and dried at 110 °C for 12 h. The green density was estimated from the weight and bulk volume. The internal structures of green compacts were observed by liquid immersion technique13 using water. Green compacts were fired at 1400 °C for 30 min in air. Density of the sintered bodies was measured by the Archimedes method. Devitrification due to crystallization was characterized by X-ray diffractometer (XRD: MultiFlex, Rigaku. Co. Ltd.,). The internal structure of sintered body was characterized by a polarization microscope (ECLIPSE E600 POL. Nicon. Co. LTD.,). Microstructure was observed by scanning electron microscope (SEM: JSM-5200, JEOL. Co. Ltd.,). Light transmission was measured for the sample having 2 mm thick in the range of 200 to 1100 nm by spectrophotometer (UV-VIS: UV mini-1240, SHIMAZU. Co. Ltd.,). RESULTS AND DISCUSSION Figure 1 shows the result of wetting and flowing points measurement. In the figure, an arrow indicates optimum condition for preparation of highly concentrated and dispersed slurry. As a result, SÍO2 slurry with 60.9 vol.% (=78.3 wt.%) were obtained by 1.2 wt.% dispersant addition. This is higher values compared with general concentration for slip-casting. This high concentration seems to result from the round shape and the wide size distribution of the powder. Relative density of green compacts prepared by floe-casting was 64.7 %. The internal structures of green compacts are shown in Figures 2. It was found that a floc-cast green body had homogeneous structure with few defects although compacts compared with uniaxial pressing. Relative density of SÍO2 sintered bodies was reached to 99.7 %. They were transparent inspite of 7mm thickness (Figure 3 (a) and (b)). On the other hand, the sintered body prepared by uniaxial pressing in dry process was opaque (Figure 3 (c)). Figure 4 shows X-ray profiles of starting powder and sintered body. No crystallization was detected. A typical internal structure of the SÍO2 sintered body was shown in Figures 5. Few residual pores were observed and no crystallization was found in sintered body (b). SEM

108

• Novel Processing of Ceramics and Composites

Fabrication and Evaluation of Transparent Amorphous Si0 2 Sintered Body

,

> o 4 'S <3 ( a. o

II

-

3

•«. optimal value

S--/. •

» flowing point - O- - wetting point

13 0

0.5 1.0 1.5 Amount of dispersant / wt % Fig. 1 Wetting and flowing points measurement

Fig. 2 Internal structure of green compact (a) floc-casting, (b) uniaxial pressing in dry process

Fig. 3 Snapshots of sintered bodies (a), (b) floc-castine and (c) uniaxial pressine in dry process

Novel Processing of Ceramics and Composites

■ 109

Fabrication and Evaluation of Transparent Amorphous Si0 2 Sintered Body

G

Ü

10

20

30 40 261 degree

50

60

Fig. 4 XRD profiles of (a) sintered body and (b) starting powder

Fig. 5 Internal structure of sintered body (a) open nicol, (b) cross nicol

Fig. 6 A SEM photograph of sintering body

110

• Novel Processing of Ceramics and Composites

Fabrication and Evaluation of Transparent Amorphous S¡0 2 Sintered Body

photograph of the sintered body is shown in Fig. 6. The shape of residual pore was spherical, and pore size was less than 10 u.m. Figure 7 shows transmission of 2mm thick of SÍO2 sintered body in the range of 200 to 1100 nm by UV-VIS spectrometer. The sintered body had a high transmissivity in the range of visible light and its maximal value was 88.7 % (X=\ 100 nm). Consequently, it was shown that high transparency resulted from high density, few pores and no crystallization due to homogeneous green body by floc-casting.

100r

^

s> 8Ü60 -' CO

co

E 40 to

c

F 20-

1-

■

oL

200

400

600

800

¡Wnm

1000

Fig. 7 Transmissivity of the SÍO2 sintered body (UV-VIS) CONCLUSIONS Green compacts by floc-casting technique had high density and homogeneous structure. Although they were fired at 1400 C in air, the sintering bodies had no devitrification. The 2 mm thick sample had 88.7 % of light transmission at X= 1100 nm. Consequently, it was shown that high transparency resulted from high density, few pores and no crystallization due to homogeneous green body by floc-casting. REFERENCES 'L-C. Guo, Y. Zhang, N. Uchida and K. Uematsu, "Influence of Temperature on Stability of Aqueous Alumina Slurry Containing Polyelectrolyte Dispersant" J. Eur. Ceram. Soc, 17,345-50(1997) 2 M. Nogami and Y. Moriyama, "GLASS FORMATION THROUGH HYDROLYSIS OF Si(OC2H5)4 WITH NH4OH AND HC1 SOLUTION." J. Non-Cryst. Solids, 37, 191-200(1980) 3 E. M. Rabinovich, D. W. Johnson, Jr., J. B. MacChesney and E. M. Vogel, "Preparation of transparent high-silica glass articles from colloidal gels" J. Non-Cryst. Solids., 47,435-39(1982) 4 E. M. Rabinovich, D. W. Johnson, Jr., J. B. MacChesney and E. M. Vogel, "SOL-GEL PREPARATION OF TRANSPARENT SILICA GLASS." J. Non-Cryst. Solids., 63, 155-61 (1983)

Novel Processing of Ceramics and Composites

• 111

Fabrication and Evaluation of Transparent Amorphous Si0 2 Sintered Body

5

E. M. Rabinovich, "PREPARATION OF GLASS BY SINTERING." J. Mater. Sei., 20,4259-97(1985) 6 M. Toki, S. Miyashita, T.Takeuchi, S. Kanbe and A. Kochi, "LARGE-SIZE SILICA GLASS PRODUCED BY A NEW SOL-GEL PROCESS." J. Non-Cryst. Solids., 100,479-82(1987) 7 L. Siqiang and T. Kairong, "LOW TEMPERATURE SYNTHESIS OF MONOLITHIC SILICA GLASS FROM THE SYSTEM Si(OC2H5)4-H20-HCl-HOCH2CH2OH BY THE SOL-GEL METHOD" J. Non-Cryst. Solids., 100, 254-62 (1987) 8 M. Toki, T. Takeuchi, S. Miyashita and S. Kanbe, "Fabrication of high-purity silica glass through the WSPA-sol-gel process" J. Mater. Sei., 27, 2857-62 (1992) E. M. Rabinovich, D. W. Johnson, Jr., J. B. MacChesney and E. M. Vogel, "PREPARATION OF HIGH-SILICA GLASSES FROM COLLOIDAL GELS" J. Am. Ceram. Soc., 66, 683-99 (1983) 10 M. D. Sacks and T. Y. Tseng, "PREPARATION OF Si0 2 GLASS FROM MODEL POWDER COMPACTS" J. Am. Ceram. Soc., 67, 526-37 (1984) n R. Ciasen, "PREPARATION AND SINTERING OF HIGH-DENSITY GREEN BODIES TO HIGH-PURITY SILICA GLASSES." J. Non-Cryst. Solids., 89, 335-44 (1987) 12 J. P. Williams, Y. - S . Su, W. R. Strzegowski, B. L. Butler, H. Hoover and V. O. Altemose, "Direct Determination of Water in Glass" Am. Ceram. Soc. Bull., 55, 524-27 (1976) 13 K. Uematsu, J.-Y. Kim, M. Miyashita, N. Uchida and K. Saito, "Direct observation of internal structure in spray-dried alumina granules" J. Am. Ceram. Soc, 73, 2555-57 (1990)

112

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Solution Deposition

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

YTTRIA STABILIZED ZIRCONIA THIN FILMS FORMATION FROM AN AQUEOUS SOLUTION BY MIST DEPOSITION Atsushi Saiki, Yukimine Fujisawa, Takashi Hashizume, Kiyoshi Terayama Department of Material Systems Engineering and Life Science, Faculty of Engineering, Toyama University 3190 Gofuku, Toyama 930-8555 JAPAN ABSTRACT In our present study, YSZ thin films were fabricated from aqueous solution by mist deposition. Composition and concentration of the solution is 94mol%Zr02-6mol%YOi.5 and 0.02-0.15 mol/dm3. ZrO(N03)2-2H20 and Y(N03)3-6H20 were used as Zr and Y sources. The solution was misted by using ultrasonic nebulizer and deposited on the heated glass substrates, which were heated from 373 to 623 K. By using this method transparent, pure, amorphous thin films were grown up on glass surface when substrate temperature was between 423 to 523 K, and which thickness were about 0.2 to 0.4um. Crystalline YSZ films could be obtained after annealing at 773 K for 30 min in air. XRD profiles from the thin films deposited at 473 K showed tetragonal phase. Diffraction peaks indicate that films were mainly preferentially oriented to (111) direction to the surface of the glass substrate. Milky and rough YSZ films were formed when deposition temperature were over 523 K. Monoclinic phase was detected when deposition temperature was higher than 523 K. or lower than 423 K after heat treatment. INTRODUCTION Zirconia are very important for their thermal, mechanical, and chemical stability,1"5 and their thin films have attracted much attention for applications such as, buffer layers for growing electric devices, " ionic conductors,9"" thermal-shield or corrosion-resistant coatings, " and oxygen sensors.14 Especially the thickness, flatness and quality of YSZ buffer layer thought to affect the structure of upper layers and electrical properties of devices. One of the fabrication method for zirconia thin films by using relatively environment friendly materials is to use aqueous solutions as raw materials. And some kinds of techniques have been investigated for direct deposition of Zr0 2 thin films using aqueous solutions. At the liquid-phase deposition the hexafluorozirconate salt (M2ZrF6: M = Na, K. NFU, etc.) was used as the method,1 zirconium source, therefore it is difficult to eliminate the residual fluorine in the as-prepared thin film even by heat treatment, which restricts application of the film. And at the self-assembled monolayer technique, 18"20 zirconium sulfate (Zr(SO4)2-4H20) and HC1 were used as raw materials. However in that films a large amount of chlorine or sulfur was also tend to present in the as-deposited films because of uncompleted hydrolysis reaction of zirconium ethoxide. On the other hand at the decomposition reaction method of aqueous peroxozirconium complex solution using zirconium oxynitrate dihydrate (ZrO(N03)2-2H20) as Zr source,21 pure amorphous Zr0 2 thin films were fabricated at room temperature. In the present study, we aim to make zirconia thin films by using environment friendly materials, at low temperature, in atmospheric pressure and using low cost method for large area and homogeneous thickness. By applying mist deposition method using aqueous solution with

115

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

simple materials such as ZrO(NC>3)2-2H20 and NH3-H2O, a transparent, highly pure, amorphous ZrC>2 thin film was synthesized on a glass substrate surface at 473 K. A crystalline phase can be obtained after annealing at 773 K for 30 min in air. EXPERIMENTAL PROCEDURE Preparation of Thin Films Composition and concentration of the precursor aqueous solution is 94mol%ZrC»26mol%YOi5 and 0.02-0.15 mol/dm3 ( 0.02-0.15 M YSZ solution). Zirconium oxynitrate dihydrate (ZrO(N03)2-2H20, 99.99% ) and Yttrium nitrate hexahydrate (Y(N03)3-6H20, 99.99%.), were selected as the Zr and Y sources. Ammonia solution (NH3-H2O, 17%.) in appropriate quantities (0-0.5 vol%) was added to adjust the acidity and to occur a radical exchange reaction between ZrO(N03)2 and NH4OH to produce an intermediate form of ZrO(OH)7. After the mixture was stirred for 1 h in an ice bath, a homogeneous colorless and transparent solution was obtained. Glass substrates (10 x 10 mm2) were cleaned ultrasonically in ethanol, and after the substrate was dried, they were further irradiated by UV light for 30 min for improving wettability. The solution was misted by using ultrasonic nebulizer and deposited for 30 min on the heated glass substrates, which was heated from 373 to 623 K. Figure 1 showed the schematic diagram of the chamber for thin film deposition. As-deposition films were transparent but amorphous and so heat treated at temperature of 773 K for 30 min in air afterwards. T.C. Chamber Partition.

-Heater IIIIIFIINU—4-Substrate

-Mist

Water Buth Ultrasonic Nebulizer Figure 1. Schematic diagram of the chamber for thin film deposition. Characterization of the films The crystallization of the films were investigated using X-ray diffractometry (XRD, 40 kV, 30 mA, CuKa, RintlOOO, RIGAKU). The chemical composition of the films was analyzed by X-ray fluorescence spectrometer (XRF, PW2404, PHILIPS). Morphology of the films were observed by atomic force microscopy (AFM, DEGITAL INSTRUMENT) and scanning electron microscope' (SEM, S-3500, HITACHI).

116

• Novel Processing of Ceramics and Composites

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

RESULTS AND DISCUSSION The YSZ precursor aqueous solution was misted by using ultrasonic nebulizer and deposited for 30 min on the heated glass substrates, which was heated from the backside of the substrate and the temperature range of 373 to 623 K. By using this method flat, transparent, pure, amorphous thin films were grown up on glass surface when the substrate temperature was between 423 to523 K. If the ammonia solution was added to the YSZ solution, the solution became unstable, and composition deviation became to occur easily. Although small amount of ammonia addition to the same YSZ solution at another film fabrication method of the electrochemical deposition22 was effective for film growth, in this mist deposition method ammonia solution did not be used. Thickness of the films were about 0.2 to 0.4um. When the substrate temperature was higher than 523 K, not the film but the powders were deposited and they could be removed easily. Mists thought to be dried independently by the radiant heat from the substrate before adhering to the glass surface. On the other hand when the substrate temperature was lower then 423 K, mist reached to the substrate surface before previously adhered mist dry out. Therefore only condensation of the YSZ solution thought to take place on the substrate surface instead of an appropriate film growth. As-deposition films were transparent but amorphous and so heat treated at temperature of 773 K for 30 min in air afterwards. In case of the heat-treatment temperature of 573 K it took 2 h for enough crystallization. Figure 2 showed the XRD profiles for the YSZ thin films after heat treatment at 773 K for 30 min deposited from mist of 0.10 M YSZ solution with different heated substrate temperature. XRD profiles from the thin films deposited at different temperatures showed monoclinic phase except the film grown at 473 K which showed tetragonal phase. Diffraction peaks indicate that films were mainly preferentially oriented to (111) direction perpendicular to the surface of the glass substrate. Milky and rough YSZ films were formed when deposition temperature were over 523 K. Monoclinic phase was detected when deposition temperature was higher than 523 K or lower than 423 K after annealing. Aqueous solution in the deposition chamber in the present state was also heated slightly by the radiant heat from the substrate. Therefore solubility ratio change in the solution of the yttrium nitrate compared to the zirconium oxynitrate might cause the composition deviation in the mist and precipitation of monoclinic phase. Figure 3 showed the XRD profiles for the YSZ thin films using different concentration of YSZ solution. From 0.15 M to 0.07 M, almost same films with (111) oriented tetragonal phase were grown, but at thinner concentration of 0.02 M, (111) oriented monoclinic phase arose. It turned out to be the phase difference due to the solution concentration and the temperature at the deposition period even if same heat treatment was carried out afterwards.

Novel Processing of Ceramics and Composites

• 117

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

^■—~~V^-ivv

/v~

Fig. 2. XRD profiles for the YSZ thin films after heat treatment at 773 K for 30 min deposited from mist of 0.10 M YSZ solution with different heated substrate temperature, (a) 423 K, (b) 473 K, (c) 523 K, (d) 573 K.

30

32

26/deg

Fig. 3. XRD profiles for the YSZ thin films after heat treatment at 773 K for 30 min deposited from mist at 473 K with different YSZ solution concentration, (a) 0.02 M, (b) 0.07 M, (c) 0.10 M,(d) 0.15 M. Figure 4 showed SEM micrographs of the surface of YSZ thin films after heat treatment with different YSZ solution concentration. Films with solution concentration of (b) 0.07 M and (c) 0.10 M had large flat, transparent, pure oriented tetragonal YSZ area were fabricated. Cracks

118

• Novel Processing of Ceramics and Composites

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

observed in the micrographs developed during heat treatment afterward due to the shrinkage by dehydration from the film. As deposition times were constant and concentration of the film (d) was high, the thickness of the films was largest and the many cracks developed and became milky and rough surface. For this sample more moderate heat treatment condition needed to get flat and transparent film. For the sample (b) 0.07 M and (c) 0.10 M, AFM images were also observed (Figure 5). They had very flat surfaces which mean roughness was less than 1 nanometer for measured areas. Film deposited from 0.07 M liad more smooth surface than that from 0.10 M due to film growth rate of which the former is small than the later. It was confirmed that condition of fabricate flat, transparent, pure, having oriented tetragonal phase YSZ thin film were limited and depended on solution concentration, substrate temperature, heat treatment condition, film thickness and others.

Fig. 4. SEM micrographs of the surface of YSZ thin films after heat treatment at 773 K for 30 min deposited from mist at 473 K with different YSZ solution concentration, (a) 0.02 M, (b) 0.07 M, (c) 0.10 M, (d) 0.15 M.

Fig. 5. AFM images of YSZ films deposited at 473 K, (a) 0.07 M, (b) 0.10 M

Novel Processing of Ceramics and Composites

• 119

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

CONCLUSION In our present study, YSZ thin films were fabricated from aqueous complex solution by mist deposition. ZrO(N03)2-2H20 and Y(N03)3-6H20 were used Zr and Y sources. Composition is 94mol%Zr02-6mol%YOi 5 and concentration of the solution for fabricating flat, transparent, pure, oriented tetragonal phase of YSZ was 0.07-0.10 mol/dm3. Ammonia solution was not used for instability of the aqueous solution. The solution was misted by using ultrasonic nebulizer and deposited on the heated glass substrates, which were heated from 373 to 623 K. By using this method transparent, pure, amorphous thin films were grown up on glass surface when substrate temperature was between 423 to 523 K. Crystalline YSZ films could be obtained after annealing at 773 K for 30 min in air. XRD profiles from the thin films deposited at 473 K only showed tetragonal phase and showed monoclinic phase at other deposition temperature condition. Diffraction peaks indicate that films were mainly preferentially oriented to (111) direction to the surface of the glass substrate. REFERENCES 1

J. Eichler, U. Eisele, J. Rodel, "Mechanical Properties of Monoclinic Zirconia", J. Am. Ceram. Soc, 87, 1401-1403 (2004) 2 F. Bondioli, C. Leonelli, T. Manfredini, A. M. Ferrari, M.C. Caracoche, P. C. Rivas, A. M. Rodriguez, "Microwave-Hydrothermal Synthesis and Hyperfine Characterization of Praseodymium-Doped Nanometric Zirconia Powders", J. Am. Ceram. Soc, 88, 633-638(2005) O. Vasylkiv, Y. Sakka, Y. Maeda, V. V. Skorokhod, "Sonochemical Preparation and Properties of Pt-2Y-TZP Nano-Composites", J. Am. Ceram. Soc, 88, 639-644 (2005) 4 K. Mukae, N. Mizutani, A. Saiki, X. Li, J. Nowotny, Z. Zhang, T. Bak, C.C.Sorrell, " Effect of Surface Preparation of Zirconia on Its Reactivity with Oxygen", J. Aust. Ceramic Soc, 34,76-79(1998) 5 T. Kiguchi, A.Saiki, K.Shinozaki, K. Terayama, N.Mizutani, "Effect of Axial Ratio on Critical Stress of Ferroelastic Domain Switching in Ceria-Partially-Stabilized Zirconia", J. Ceram. Soc. Japan, 105, 871D875 (1997). 6 N. Wakiya, K. Shinozaki, N. Mizutani, "Improvement of Magnetic Properties of (111)Epitaxial Nickel-Zinc-Ferrite Thin Films Deposited on Si Platform", Key Engineering Materials, 269, 245-248 (2004) 7 C-H. Chen, N. Wakiya, A. Saiki, K. Shinozaki, N. Mizutani, "Thickness and Roughness Analysis on YSZ/Si(001) Epitaxial Films with Ultra Thin SÍO2 Interface by X-Ray Reflectivity", Key Engineering Materials, 181-182, 121-124 (2000) ' Y. Komatsu, T. Sato. S. Ito, K. Akashi, Thin Solid Films, 341, 132-135 (1999). 9 Y. Ohya, M. Murayama, Y. Takahashi, "Electrical Properities of ZrÛ2 Thin Films Doped With ln 2 0 3 by Sol-Gel Method", Key Engineering Materials, 169-170, 176-178 (1999) 10 K. Sasaki, L. J. Gauckler, "Microstructure-Property Relations of SOFC Electrodes: Importance of Microstructural Optimaization of La(Sr)Mn03 Cathodes on Zr02(Y2Û3) Electrolytes", Key Engineering Materials, 169-170, 201-204 (1999) " G-Z Cao, W. Brinkman II. K. J .Dc Vries, A. Burggraaf., J. Am. Ceram. Soc, 76, 2201-2208(1993).

120

• Novel Processing of Ceramics and Composites

Yttria Stabilized Zirconia Thin Films Formation from an Aqueous Solution by Mist Deposition

G. R. Dickinson, C. Petorak, K. Bowman, R. W. Trice, "Stress Relaxtion of Compression Loaded Plasma-Sprayed 7 Wt% Y203-Zr02 Stand-Alone Coatings", J. Am. Seram. Soc, 88, 2202-2208 (2005) 13 N. Wu, Z. Chen, S. X. Mao, "Hot Corrosion Mechanism of Composition Alumina/Yttria-Stabilized Zirconia Coating in Molten Sulfate-Vanadate Salt", J. Am. Ceram. Soc, 88,675-682 (2005) 14 A. Bastianini, G. A. Battiston, R. Gerbasi, M. Porchia, S. Daolio, J. Phys 0 5, 525-532. (1995) 15 T. Yao, J. Mater. Res. 13, 1091 (1998) 16 T. Yao, T. Irai, A. Ariyoshi, "Novel Method for Zirconium Oxide Synthesis from Aqueous Solution", J. Am. Ceram. Soc., 79, 3329-3330 (1996). 17 N. Ozawa, T. Yao, Trans. Mater. Res. Soc. Jpn., 18, 321 (2003) 18 Y. Gao, Y. Masuda, T. Yonezawa, K. Koumoto, "Site-Selective Deposition and Micropatterning of Zirconia Thin Films on Templates of Self-Assembled Monolayers", J. Ceram. Soc. Jpn., 110, 379-385, (2002) 19 M. Agarwal, M. R. De Guier, A. H. Heuer, "Synthesis of ZrC>2 and Y2C>3-Doped Zr0 2 Thin Films Using Self-Assembled Monolayers", J. Am. Ceram. Soc., 80, 2967-2981 (1997) 20 T. P. Niesen, M. R. De Guire, J. Bill, F. Aldinger, M. Ruhle, A. Fischer, F. Jentoft, R. Schlogl, J. Mater. Res., 14, 2464-2475 (1999) 21 Y. Gao, Y. Masuda, H. Ohta, and K. Koumoto, "Room-Temperature Preparation of ZrC"2 Precursor Thin film in an Aqueous Peroxozirconium-Complex Solution", Chem. Mater. 16, 2615-2622(2004) 22 A. Saiki, H. Uno, S. Ui, T. Hashizume, K. Terayama, "Fabrication of YSZ Thin Films in an Aqueous Solution by Electro-Chemical Deposition", Proceedings ofPACRIMó, PACRIM-S10-16-2005 (2005)

Novel Processing of Ceramics and Composites

• 121

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Nanopowders and Nanorods

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

SYNTHESIS AND STRUCTURAL CHARACTERIZATION OF NANOAPATITE CERAMICS POWDERS FOR BIOMEDICAL APPLICATIONS Kanae Ando1, Mizuki Ohkubo1, Satoshi Hayakawa1, Kanji Tsuru1-2, Akiyoshi Osaka1,2, Eiji Fujii3, Koji Kawabata3, Christian Bonhomme4, and Florence Babonneau4 1

Faculty of Engineering, Okay ama University Tsushima, Okayama, 700-8530, Japan 2 Research Center for Biomédical Engineering, Okayama University 3 Industrial Technology Center of Okayama Prefecture Haga, Okayama, 701-1296, Japan 4 Laboratoire de Chimie de la Matire Condense de Paris, Universite Pierre et Marie Curie, CNRS Jussieu, 75252 Paris Cedex 05, France Corresponding author: S. Hayakawa; [email protected] ABSTRACT Zn/HAp particles were prepared by soaking hydroxyapatite (HAp), derived through a wet chemical method, into aqueous solutions containing various amounts of zinc nitrate. Obtained samples were characterized by inductively coupled plasma emission spectroscopy, X-ray diffraction (XRD), Brunauer-Emmett-Teller surface area analysis (SA), and transmission electron microscopy. Their zeta potential was also measured. The analyzed Zn content was almost proportional to the content in the aqueous solutions. The Zn2+ ions were partially replaced the Ca2+ ions in the apatite, and hence, little change was observed in crystallinity, particle size and SA. Adsorption of bovine serum albumin (BSA) and ß2-microglobulin (ß2-MG) in the solutions containing both BSA and B2-MG was examined. As the Zn2+ ion content in the apatites increased, the adsorbed amount of BSA was almost constant, whereas that of B2-MG increased. Thus, the Zn incorporation increased the selectivity, and it is suggested that the Zn-substituted site should be the active ones for ß2-MG adsorption. INTRODUCTION Hydroxyapatite (HAp) is a mineral and the major inorganic component of human bone and tooth. Solid or porous HAp ceramics have been used as hard tissue repairing materials in clinic. Zittle et al. reported protein adsorption onto calcium phosphates in 1951'. Since then, hydroxyapatite has also been used as a column packing material for chromatography to separate proteins or enzymes2'3'4. Recently, much attention has been focused on it as an adsorbent for removing pathogenic proteins from blood in blood purification therapy.

125

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

After Gejyo el al. ß2-microglobulin (ß2-MG) is the precursor protein, amyloid, that causes dialysis-related amyloidosis, and it is accumulated in the patient blood due to long-term dialysis treatment5. Current dialysis-membranes separate B2-MG from the blood plasma by a molecular size effect. If one attempts to remove JJ2-MG effectively by the dialysis treatment, the essential proteins are also removed since B2-MG is as large as those. Recently, a few studies have been conducted on preparing hydroxyapatite for selective adsorption of protein to overcome the problem in terms of removing pathogenic substances without removing the essential proteins6"9. Fujii et al. reported that Zn-containing hydroxyapatite (ZnHAp) had high selectivity on &2-MG in physiological saline solution containing bovine serum albumin (BSA) and ÍS2-MG10. As the zinc content of ZnHAp increased, the primary particle size of ZnHAp decreased. They concluded that the selective protein adsorption of ZnHAp was improved by optimum distribution of the surface charge with the decrease of the primary particle size. In this study, another method of incorporation of zinc ions into HAp lattice is proposed, and the effects of zinc-incorporation on the protein adsorption property are examined in detail. We introduced zinc ions on the hydroxyapatite surface by soaking the HAp particles derived form a wet chemical method nitrate (the products are denoted as Zn/HAp: note the difference from the previous code name ZnHAp after Fujii et al.10). The adsorption of BSA and B2-MG on Zn/HAp was evaluated in terms of the selectivity for B2-MG as Zn/HAp's were contacted with physiological saline containing both proteins. Their composition, crystallite size, and surface area were correlated to the protein adsorption property. EXPERIMENTAL Hydroxyapatite was synthesized from reagent-grade calcium nitrate (Ca(N03)2*4H20) and diammonium hydrogen phosphate ((NH^HPO,!) by a wet chemical method. (NH^HPC^ aqueous solution (0.3 mol/L) whose pH was adjusted to 10 by addition of a 28 mass% NH4OH aqueous solution was added under rigorous stirring to 0.5 mol/mL Ca(NC>3)2 aqueous solution at a feeding rate of 3 mL/min under an N2 atmosphere at 60CC. After completion of the addition, the precipitates were aged for 24 h, washed with distilled water, and dried at 105°C for 48 h. The derived cakes were milled and sieved to obtain particles of <150 um in size. The HAp particles were suspended in 0-11.5 mM Zn(NÛ3)2 aqueous solutions. They were then aged at 80°C for 24 h, washed with distilled water and dried at 105°C for 48 h. We denoted those samples as Zn/HAp. Zn, Ca, P ion contents of the samples was analyzed by inductively coupled plasma emission spectroscopy (ICP, ICPS-7500, SHIMADZU). The crystalline phases were identified by an X-ray diffractometer (XRD, RINT2000, RIGAKU; 40 kV-200 mA), and the lattice parameters were also derived from their XRD peaks. The surface area (SA) were measured with

126

• Novel Processing of Ceramics and Composites

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

the BET N2 adsorption method (Micromentics GEMINI2370, SHIMADZU). Morphology of those apatite particles was observed by transmission electron microscopy (TEM, JEM-2010D JEOL). Zeta potential was measured with ZETASIZER (ZETASIZER 3000HSA, MALVERN), used physiological saline (0.142 mol/L NaCl aqueous solution, pH7.4) as solvent. Both BSA and ß2-MG were dissolved in saline (pH7.6) to prepare a mixed-protein solution so as to attain the concentration of BSA to be 70 mg/mL and that of ß2-MG to be 30 ug/mL. The solution containing 30 \ig/mL of ß^MG should simulate a blood plasma of patients suffering from amiloidosis. The amounts of BSA and IVMG adsorbed on the apatite particles were quantitatively analyzed by colorimetry, that is, the enzyme immunoassay method: each samples (0.1 g) was soaked in 1.0 mL of saline (pH7.4) in a glass test tube. One mL of the mixed-protein solution was added to the test tube and left in contact for 6 h with the sample powder. Then, the sample powders werefilteredwith the cellulose membrane, and optical absorptions of the filtrates were measured at ca. 420 nm for ß2-MG and at ca. 630 nm for BSA. RESULTS Figure 1 shows the Zn2+ content of Zn/HAps as a function of the concentration of the Zn(NÛ3)2 aqueous solution, indicating that the analyzed content was almost proportional to the Zn(N03)2 content in the solution.

0 4 8 12 Concentration of Zn(NO 3)2 (mmol/L) Figure 1 The analyzed Zn2+ ion content in the apatite particles as a function of the Zn(N03)2 concentration in the starting solutions.

Novel Processing of Ceramics and Composites

• 127

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

Figure 2 shows the XRD patterns of the HAp and HAps after soaked into the Zn(NOj)2 aqueous solution. All peaks were assigned to hydroxyapatite (JCPDS: 09-0432), and the peaks of .

Î 1 .£ ?

X

Î.

2.3%

-~_JL»/Ui V L _ A ^ W W X ^ i m 0.9% _~_AJ/LJV. 20

A_^JUA

30 40 20 / degree

50

Fig 2 XRD patterns of HAp and Zn/HAp. The analyzed Zn contents in the apatite particles. • : apatite. byproducts weren't detected such as Parascholzite ([CaZ^iPO^^fyO], JCPDS: 35-0495), Zinc phosphate hydrate ([Zn2P207n5H20], JCPDS: 07-0087), Zinc oxide (ZnO, JCPDS: 21 -1486), and Zincite (ZnO, JCPDS: 36-1451). The crystallinity changed little after soaked in the Zn(NOj)2 solution. In Figure 3 the lattice parameters, a0 and c¡¡, were plotted as a function of the molar ratio Zn/(Ca+Zn) derived from the ICP analysis. As the Zn/(Ca+Zn) ratio increased, the an decreased from 9.414Â to 9.378Â, c0 decreased from 6.878Â to 6.819Â. Figure 4 shows the transmission electron micrographs of the Zn/HAp samples. Their average particle size was ca. 120nm. In contrast to the lattice parameters, the particle size and morphology changed little, and SA remained almost constant with the increase in the Zn * content. Similarly, zeta potential remained at ca. 0 mV, regardless of the Zn2+ content.

0.5 1.0 1.5 Zn/(Ca+Zn) Figure 3 Hexagonal lattice parameters of HAp and Zn/HAp particles as a function of their molar ratio Zn/(Ca+Zn).

128

• Novel Processing of Ceramics and Composites

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

Figure 4 TEM micrographs of HAp and Zn/HAp. The analyzed contents in the apatite particles are presented. Bar: 100 nm. Figure 5 shows the amount of BSA and B2-MG adsorbed per unit area as a function of the Zn2+ content of Zn/HAp. As the Zn2+ content increased, the amount of ß2-MG adsorbed per unit area increased from 1.6 ug/ml to 2.5 ug/ml, while that of BSA changed little. Therefore, this indicated that the fraction of selective adsorption for ßi-MG increased with the Zn2+ content of Zn/HAp. St

^b

3

>

3

3 3¡

o 2

"S S

S ai' î 'S ÇsJ c g 0 00

1

o

-*Î

ß2-MG

*~T

u

1

1.0

2 03

> 03 11

BSA O

O 1

2.0

3

3a.

B

3

Zn ion content (mass%)

of

o ZÍ-

Figure 5 Amount of BSA and ß^-MG adsorbed on Zn/HAp as a function of the analyzed Zn2+ ion content in the apatite particles.

Novel Processing of Ceramics and Composites

• 129

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

DISCUSSION Miyaji et al. synthesized Zn-substituted Ca hydroxyapatites by a similar wet chemical method, and reported that the lattice constants of the apatites decreased with the ratio Zn/(Ca+Zn)". The present results shown in Figure 2 well agreed with them. Thus, as Miyaji et al. mentioned, the decrease in the lattice constants resulted from the partial substitution of smaller Zn2+ ions for the Ca2+ ions in the HAp. Moreover, Miyaji et al. reported that the crystallinity of their Zn-substituted Ca hydroxyapatites significantly decreased with the increase in the Zn fraction. In contrast, the present study indicated that Zn2+ ions were introduced into HAp lattice without changing crystallinity and particle size of the apatites when the soaking method was employed. Fujii et al. synthesized nano-crystalline zinc-containing hydroxyapatite, denoted as ZnHAp, by a wet chemical method and found that the meso-pores were produced by the agglomeration of the primary particles of ZnHAp10. When the Zn2+ ion content of ZnHAp increased, the amount of BSA adsorbed per unit area decreased drastically, while that of ß2-MG changed little. This result was explained in terms of the relative size of BSA and pores: since BSA (4x4x16 nm) is larger than the meso-pores of ZnHAp, the ZnHAp particles could not involve BSA molecules in the meso-pores and thus the efficient surface area for BSA adsorption must have been decreased by the formation of the meso-pores. That is, the key factor was the change in the particle size in the case of wet-chemically derived ZnHAp. In the present Zn/HAp's, on the contrary to Fujii et al., little changes were observed for the particle size, SA, or crystallinity. Yet, the amount of ß2-MG adsorbed or the selectivity increased with the Zn2+ ion content. This result strongly suggests, therefore, that the substituted zinc ions in the HAp lattice of the surface should serve the active sites for fc-MG adsorption. Indeed, the fraction of BÎ-MG adsorbed on ZnHAp was 94.9 % based on the report by Fujii et al., while that of Zn/HAp was 95.6 %. CONCLUSION We synthesized Zn/HAp by the impregnation of synthesized HAp into solution containing zinc ions. As the concentration of Zn(N03)2 aqueous solution increased, Zn2+ ion content of Zn/HAp increased, and it is considered that Ca2+ ions in HAp lattice were replaced by Zn2+ ions. We introduced zinc ions into the HAp lattice with little change of crystallinity, particle size, or SA by the impregnation method. Protein adsorption properties for Zn/HAp were examined using a mixed protein solution containing BSA and ß2-MG Although Zn2+ ion content of Zn/HAp increased, the adsorbed amount of BSA was constant, whereas that of ß2-MG increased. It was suggested that the substituted zinc ions in HAp lattice should be the active sites for ß2-MG adsorption. ACKNOWLEDGEMENTS This study was supported by a Grant-in-Aid for Scientific Research from the Japan

130

• Novel Processing of Ceramics and Composites

Synthesis and Structural Characterization of Nanoapatite Ceramics Powders

Society for the Promotion of Science (No. 16360330, Z1416005) and The Hosokawa Powder Technology Foundation. REFERENCES 'C.A. Zittle, "Adsorption studies of enzymes, and other proteins," Advances in Enzymol, 14,319-74(1951). 2 A. Tiselius, S. Hjerten, and O. Levin, "Protein chromatography on calcium phosphate columns," Arch. Biochem. And Biophys., 65, 132-55 (1956). 3 T. Kawasaki, K. Ikeda, S. Takahashi, and Y. Kuboki, "Further study of hydroxyapatite high-performance liquid chromatography using both proteins and nucleic acids, and a new technique to increase Chromatographie efficiency," Eur. J. Biochem., 155, 249-57 (1986). 4 T. Kawasaki, W. Kobayashi, K. Ikeda, S. Takahashi, and H. Monma, "High-performance liquid chromatography using spherical aggregates of hydroxyapatite micro-crystals as adsorbent," Eur. J. Biochem., 157, 291-95 (1986). 5 F. Gejyo, T. Yamada, S. Odani, Y. Nakagawa, M. Ayakawa, T. Kunitomo, H. Kataoka, M. Suzuki, Y. Hirasawa, T. Shirahama, A. Cohen, and K. Schmid, "A new form of amyloid protein associated with chronic hemodialysis was identifyied as ß2-microglobulin," Biochem. Biophys. Res. Commum., 129, 701-06 (1985). 6 S. Takashima, S. Hayakawa, C. Ohtsuki, and A. Osaka, "Adsorption of proteins by calcium phosphate with varied Ca to P ratios," Bioceramics, 9, 217-20 (1996). 7 S. Takashima, Y. Kusudo, S. Takemoto, K. Tsuru, S. Hayakawa, and A. Osaka, "Synthesis of carbonate-hydroxy apatite and selective adsorption activity against specific pathogenic substances," Bioceramics, 14, 175-78 (2002). 8 S. Hayakawa, Y. Kusudo, S. Takemoto, K. Tsuru, and A. Osaka, "Hydroxy-carbonate apatite, blood compatibility and adsorption of specific pathogenic proteins," Bioceramics: Materials and Applications IV, 147, 111-19 (2003). S. Takemoto, Y Kusudo, K.Tsuru, S. Hayakawa, A. Osaka, and S. Takashima, "Selective protein adsorption and blood compatibility of hydroxy-carbonate apatites," J. Biomed. Mater. Res., 69A, 544-51 (2004). I0 E. Fujii, M. Ohkubo, S. Hayakawa, K. Tsuru, A. Osaka, and K. Kawabata, "Selective protein adsorption property and characterization of nano-crystalline zinc-containing hydroxyapatite," submitted to Acta Biomaterialia. "F. Miyaji, Y. Kono, and Y. Suyama, "Formation and structure of zinc-substituted calcium hydroxyapatite," Mater. Res. Bull, 40, 209-20 (2005).

Novel Processing of Ceramics and Composites

• 131

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

NOVEL PROCESS OF SUBMICRON-SCALE CERAMIC ROD ARRAY FORMATION ON METALLIC SUBSTRATE Kazuya Okamoto*, Satoshi Hayakawa*, Kanji Tsuru*, **, Akiyoshi Osaka*, ** *Biomaterials Laboratory, Graduate School of Natural Science and Technology, Okayama University, Tsushima, Okayama, 700-8530, Japan **Research Center for Biomédical Engineering, Okayama University, Tsushima, Okayama, 700-8530, Japan ABSTRACT Nano-scale rods of titania and tungsten oxide were fabricated on metallic titanium and tungsten substrates by heat-treatment in vacuo. The surface structure and morphology were characterized by Thin-film X-ray diffraction analysis (TF-XRD) and scanning electron microscopy. The TF-XRD analysis indicated that the thermal treatment in vacuo (6.7 xlO2 Pa) gave Ti02 (rutile: PDF#21-1276) and WO, (PDF#20-1324) nano-scale rod arrays. The mechanism of the nano-scale rod array formation was explained by the structure matching due to topotaxy and the crystal growth habit. INTRODUCTION One-dimensional nanostructures of polymers, metals and semiconductors in the forms of rods, tubes and others have attracted much attention for a broad range of potential applications, including catalysts, electronics, sensors, photonics, micromechanical devices, and biomédical devices'". Possible factors to optimize various chemical, biological and physical properties of the rod arrays are the chemical states of the metals, like valence and oxygen coordination, and the shape and size of the rods. Actually, Liu et al. pointed out that the rod shape was dependent on the growth mechanism5*. They prepared submicrometer-scale rod arrays of titania on titanium (Ti) substrates with a glass-coating technique that involved selective dissolution of chemically unstable ingredients of the glass into hot water". Recently, Gu et al. prepared tungsten oxide nanowires with heating metallic tungsten (W) in an argon flow7, where the oxide nanowires growth resulted from the oxygen leakage of their system. Therefore, it is likely that the oxygen partial pressure in the heating atmosphere due to the oxygen leakage is one of the key factors in the rod formation process of these metallic oxides, or the partial pressure of oxygen during the heat-treatment controls the oxidation and the growth of rods. In this study, then, we heat-treated Ti and W substrates in an electric furnace in vacuo, and created new nanometer- to submicrometer scale rod array structures on the surfaces of metallic substrates.

133

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

EXPERIMENTAL PROCEDURE A sheet of commercially available pure Ti (ce-Ti) and W (Nilaco, Japan) was cut to give pieces of Ti and W substrates with 10x10x0.1 mm in size, which were subsequently cleaned in acetone. These substrates denoted as Ti_NT and W_NT. The Ti_NT samples were heated up to 700°C at a rate of 10°C/min in an electric furnace (MILA-3000, ULVAC-RIKO, Japan) in vacuo with vacuum pump ((6.7 xlO2 Pa); GLD-051, ULVAC KIKO, Japan), and kept there for 6 h, while the W_NT samples were heated up to 500°C at the same heating rate and kept there for 3 h. Then, the furnace was turned off, and the samples were let cooled down to room temperature in the furnace, before these samples were cleaned in acetone gently. These samples were denoted as Ti_VHT and W_VHT. For comparison, some of the Ti_NT and W_NT samples were heated in air under the same heating and cooling schedule. They were denoted as Ti_HT and W_HT. After they were coated with gold of 30 nm in thickness, their surface microstructure was observed under a scanning electron microscope (JSM-6300, JEOL, Japan) which was operated under 20 kV acceleration and 300 mA emission current. The crystalline phases of the samples were identified using thin-film X-ray diffraction (TF-XRD: CuKa) patterns taken by an X-ray diffract meter (RINT2500, Rigaku, Japan) operated at 40 kV-200 mA and at a scanning step of 0.027s. RESULTS Fig. 1 shows SEM photographs of Ti_HT and Ti_VHT. The titanium oxide covered the whole surface of Ti_HT and Ti_VHT. In Ti_VHT the random aggregation of submicron-scale rods with 1 u.m in length was observed on the whole surface. TF-XRD patterns in Fig. 2 show a sharp and strong diffraction peak at 27.48° assigned to the (110) diffraction of rutile (PDF#21 -1276). The peaks at 38° and 40° were assigned to a-Ti. Although the heat atmosphere was different, Ti_HT and Ti_VHT showed almost the same oxide

Fig. 1 SEM photographs of surface of the titanium substrates heated in air (Ti_HT) and in vacuo (Ti VHT).

134

• Novel Processing of Ceramics and Composites

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

XRD patterns.

However, the diffraction peak width at 27.48° is narrow in Ti_VHT compared

with that in Ti_HT, indicating that the crystallite size of rutile is larger in Ti_VHT compared with that in Ti HT.

{1 ÏV)

▼ : rutile

_JL

■ : a-Ti

I

r

...

.

_

7

j

r"T

, . Ti_HT

7 T

, . TLVHT

(0 2

i \.on

20

i

i

25

30

_. AT i

35

-i

40

Ti_NT ■

i '

45

50

26/dcgrcc Fig. 2 TF-XRD patterns of the titanium substrate. See text for the sample codes.

Fig. 3 shows SEM photographs of W_HT and WJVHT, where both had nm-size crystallites but in different morphology. in size were formed in W_HT.

Aggregates of block-like crystallite of around 100 nm

In contrast, nm-size rod arrays were formed in W_VHT, where

the size of each rod was tens of nanometers in diameter, and 100-300 nm in length. shows TF-XRD patterns for W_NT, W_HT, and W_VHT. for the heat-treated samples.

Fig. 4

The presence of WO, was confirmed

Note that W_HT gave a peak at 23.25°, assigned to (001) of WO,

(PDF#20-1324) whereas WJVHT gave one at 23.53° assigned to (020) of WO,.

Fig. 3 SEM photographs of surface of the tungsten substrates heated in air (W_HT) and in vacuo (WJVHT).

Novel Processing of Ceramics and Composites

• 135

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

(001)

0:W • : WO,

•

(110) (021)

Î Ü

Î

D

(211) D

1

(200)

(020)

w VHTA

•

n D

W_NT

~—r

10

20

-r-430 40

_•. • S

50 26/dcgrcc

D A

D

D

60

70

80

Fig. 4 TF-XRD patterns of the tungsten substrates. See text for the sample codes. DISCUSSION The present Ti substrates gave stronger (002) diffraction than (101). This means that more rx-Ti grains exposed their (002) plane to the heating atmosphere, and were oxidized. For Ti_VHT with the rutile nano-rod arrays in Fig. 2 the rutile (110) diffraction was by far strong accompanying very weak diffractions. We can interpret the result by considering that the topotaxy between the oc-Ti (002) plane and rutile (110) plane. That is, a good correlation has been already pointed out between the Ti-Ti atomic distance of the a-Ti (002) plane (2.9506 Â; 1 nm = 10Â) and that of the rutile (110) plane (2.9587 A), and it favors the nucleation and subsequent growth of rutile layers with the (110) plane developed preferably810. Moreover, similar titania nano-rod arrays were formed on Ti due to heating at 700°C under the nitrogen atmosphere (not shown here). Therefore, those results strongly suggest that the oxygen leakage of the vacuum system during the heat-treatment controls the oxidation and the growth of rods by the vapor-solid growth process as reported by Funahashi et al." and that the decrease in the oxygen partial pressure in the heating atmosphere is favorable for obtaining the nano- and micro-scale rod array structure due to the topotaxial nucleation and growth of the rutile rods on the titanium substrate. Then, let us examine if a similar topotaxial nucleation and growth mechanism is applicable to the WO, rod formation on the tungsten substrate. The strength of the XRD peaks of W (211) and WO, (020) in Fig. 4 suggests the topotaxial correlation between those planes. Fig. 5 compares the atomic arrangements on the W (211) plane and WO, (020) plane. Contrary

136

• Novel Processing of Ceramics and Composites

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

to the expectation, the topotaxial structure matching between those planes are not so good as that between the a-Ti (002) plane and rutile (110) plane. Thus, we can conclude that this inferior structure matching caused the random arrangements of the WO, nano-rod array. Indeed, Gu et al. reported that the growth direction of the nanowire of tungsten trioxide is along <010>, and this crystal growth habit is in good agreement with Fig. 4 where the (020) diffraction was the strongest.

Fig. 5 Comparison of the atomic arrangements based on the crystal structures of W (211) plane and WO, (020) plane. CONCLUSION We fabricated nano- and submicron-scale rods of titania and tungsten oxide on metallic titanium and metallic tungsten surface by heat-treatment in vacuo. TF-XRD analysis indicated that thermal treatment in vacuo gave Ti02 (rutile: PDF#21-1276) and WO, (PDF#20-1324) nanoand submicron-scale rod array. The mechanism of nano- and submicron-scale rod array formation was explained by the structure matching due to topotaxy and the crystal growth habit.

Novel Processing of Ceramics and Composites

• 137

Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate

ACKNOWLEDGMENT This study was supported by The Kazuchika Okura Memorial Foundation. REFERENCES 'M. Gopal, W.J. Moberly Chan, L.C.De Jonghe, "Room temperature synthesis of crystalline metal oxides," J. Mater. Sei., 32[22], 6001-6008. (1997). ! U. Diebold, "The surface science of titanium dioxide," Surf. Sei. Rept. 48, 53-229 (2003). ! Z.W. Pan, Z.R. Dai and Z.L. Wang, "Nanobelts of semiconducting oxides," Science, 291 [5510], 1947-1949.(2001). 4 P. Si, X. Bian, H. Li and Y. Liu, "Synthesis of ZnO nanowhiskers by a simple method," Mater. Lett., 57[24-25], 4079-4082. (2003). *Y. Liu, K. Tsuru, S. Hayakawa, A. Osaka, "Titanate Nano-rods Grown on Titanium Substrates," The 20th International Japan-Korea Seminar on Ceramics, 299-302, 2003.11.20-22. "Y. Liu, K. Tsuru, S. Hayakawa, A. Osaka, "In Vitro Bioactive Nano-Crystalline TiO, Layers Grown at Glass-Coating/Titanium Interfase," J. Ceram. Soc. Japan, 112[8], 452-457 (2004). 7 G. Gu, B. Zheng, W. Q. Han, S. Roth, J. Liu, "Tungsten Oxide Nanowires on Tungsten Substrates," Nano. Lett., 8[2] 849-851 (2002). k Y. Liu, K. Okamoto, S. Hayakawa, K. Tsuru, A. Osaka, "Preparation of Titanate Nano-Rod Array on Titanium Substrates by Novel Microflux Method," in Ceramic Nanomaterials and Nanotechnology III, (Ceramic Transactions, Volume 159), American Ceramic Society, Westerville, Ohio, USA, pp. 193-200, (2004). *S. Hayakawa, Y. Liu, K. Okamoto, K. Tsuru, and A. Osaka, "Formation of Titania Submicron-Scale Rod Arrays on Titanium Substrate and In Vitro Biocompatibility," in Nanoscale Materials Science in Biology and Medicine, edited by Cato T. Laurencin and Edward A. Botchwey (Mater. Res. Soc. Symp. Proc. 845, Warrendale, PA , 2005), AA6.9. "Y. Liu, K. Tsuru, S. Hayakawa and A. Osaka, "Topotaxial nucleation and growth of Ti02 submicron-scale rod array on titanium substrates via sodium tetraborate glass coating," J. Ceram, Soc. Japan, 112[10], 567-571, (2004). "R. Funahashi, I. Matsubara and M. Shikano, "Growth of Co-based Oxide Whiskers", Chem. Mater., 13,4473, (2001).

138

■ Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Coatings and Films

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

NOVEL PROCESS FOR SURFACE TREATMENT OF AIN - CHARACTERIZATION AND APPLICATION Takehiko Yoneda and Motonobu Teramoto Advanced Materials Business Division, Tokuyama Corporation 40, Wadai, Tsukuba, Japan, 300-4247 Kazuya Takada Corporate Development Division, Tokuyama Corporation 3-3-1, Shibuya-ku, Tokyo, Japan, 150-8383 Hiroyuki Fukuyama Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University 2-1-1 Katahira, Aoba-ku, Sendai, Japan, 980-8577

ABSTRACT Aluminum nitride (AIN), with its high thermal conductivity, high electrical resistivity and close thermal expansion match to silicon, has attained wide use. Some representative examples are electronic substrates, heat sinks, cutting tools and refractory materials. One of its drawbacks, however, is the vulnerability to chemical attacks. The theoretically excellent properties of AIN are significantly suppressed by reactions with water or water vapor and chemical corrosions with an alkaline solution. These shortcomings have been remedied by the present invention: a novel surface treatment process to form a chemically stable and mechanically strong alumina layer on AIN. This innovative process has been developed on the basis of the high-temperature oxidation behavior of AIN, and expands the potential applications of AIN. A Cu post-fired metallization process has been developed associated with the surface-treated substrate. The Cu post-fired metallization substrate has excellent adhesion strength and sufficient reliability. INTRODUCTION An AIN substrate is used to various electronic circuits or semiconductor process equipments as the material having the characteristics, which are high thermal conductivity, high electric resistivity, and resistance to halogen gas plasma1"2. However, AIN has drawbacks of corrosion by water, water vapor or alkaline solution. On the other hand, for metallization processes of AIN substrate, there are few choices except of thin film processes. However, thin film processes are expensive. Developments of low-cost technologies such as thick film processes (post-fired) are required. We have developed a novel surface treatment process of AIN substrate based on the fundamentals of high-temperature oxidation of AIN3. The AIN substrate treated by the newly developed process has resistance to water and alkaline solution without degrading physical

141

Novel Process for Surface Treatment of AIN—Characterization and Application

properties of the original one. Simultaneously, we have developed a Cu post-fired metallization process customized for this substrate. This combination has significantly improved adhesion strength and reliability, and has been expanded to various applications such as LED elements, Peltier elements and electro power devices. EXPERIMENTAL A1N substrates (Grade SH-30, Tokuyama, Tokyo, Japan) were set into the high temperature atmosphere furnace (Model 202-4608, Tokai Konetsu Kogyo, Tokyo, Japan). After the reaction chamber in the furnace was evacuated to 0.01 Pa by a vacuum pump, the chamber was refilled vvith Ni gas (purity: 99.9995 %, moisture content: 2.5 ppm). After that, temperature was raised to 1200 °C at a rate of 3.3 °C /min while N2 gas was flowed at a velocity of 7 mm/sec. The flow of N2 gas was stopped when temperature reached 1200 CC. 0 2 gas (purity: 99.999%, moisture content: 2.5ppm) was alternatively flowed at the same velocity, and that was held for one hour. After the A1N substrate was cooled down to a room temperature at a rate of 3.3 °C /min, we obtained a surface-treated A1N substrates (it is called N-substrate). We performed similar experiments with different holding time and temperature. For comparison, we made specimens by the conventional process (it is called C-substrate), which was performed in the air (moisture content: 2.3 %) through the all processes. CHARACTERIZATION The surface-treated A1N substrates obtained by the experiments were characterized by using XRD (Model RINT1200, Rigaku, Tokyo, Japan), FE-SEM (Model JSM-6400, JEOL, Tokyo, Japan) and TEM (Model TECNAI F20, FEI, Oregon, USA). The resistance of the substrates to an alkaline solution was evaluated as follows3"4. After the dry substrate was weighed, the substrates were soaked in an aqueous carbonate solution at 60 °C for one hour, and then the dry weight of the sample was measured again. This procedure was repeated up to 100 hours. We have also developed a Cu post-fired metallization process for A1N substrate and evaluated the characteristics of metallization'"9. A Cu-conductor paste (Grade CUX-R106, Mitsuboshi Belting, Kobe, Japan) was printed on the N-substrates with the thickness of 40 (im and the area of 2 mm x 2 mm, then dried at 170 °C and sintered at 900 °C. After that, adhesion strength of the metallized area of the Cu post-fired substrate was measured by a strength test machine. RESULTS AND DISCUSSION New Oxidation Process of A1N Figure 1 shows XRD patterns of the substrates, which reveals that the oxide layer of N-substrate consisted of a-alumina. Figure 2 shows cross section and surface of the substrates observed by FE-SEM. There are many protuberant striate patterns on the surface but they are different from cracks. The thickness of the oxide layer of the N-substrate is not as thick as that in the C-substrate and cracks are hardly observed on the surface of the N-substrate. Figure 3 shows cross section of the substrates observed by TEM. There are numerous elliptical voids in the oxide layer. In the vicinity of the boundary of the oxide layer and A1N, the voids are unclear for the N-substrate as shown in Fig. 3 (a).

142

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—Characterization and Application

Figure 1 X-ray diffraction patterns of (a) the original AIN substrate and (b) surface-treated AIN substrate by newly developed process.

Figure 2 SEM photographs of surface-treated substrate by (a), (b) newly developed process and (c), (d) conventional process, (a) and (c): cross sectional view, (b) and (d): surface view.

Novel Processing of Ceramics and Composites

■ 143

Novel Process for Surface Treatment of AIN—Characterization and Application

Figure 3 TEM photographs of surface-treated substrate by (a) newly developed process and (b) conventional process. Figure 4 shows oxidation kinetics of the A1N substrate as a parameter of temperatures. There were linear relations between the square of thickness of the oxide layer and holding time for the N-substrate but there were not such the relations for the C-substrate. The parabolic rate constant was calculated from the slope of these straight lines, and the activation energy was calculated to be 361 kJ/mol by Arrhenius plots.

Figure 4 Oxidation kinetics of A1N substrate.

144

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—Characterization and Application

Table 1 shows physical properties of the N-substrate. Although bending strength falls, sufficient strength has been maintained in practical use, and other physical properties are almost equal with the original substrate. Table 1 Comparison of physical properties of the A1N substrate treated by newly developed process with the original A1N substrate.

Bending strength Thermal conductivity Density Dielectric constant Dielectric loss

unit

AIN substrate

Surface-treated AIN substrate by newly developed process

MPa

490

W/(m K)

188

,*m3

3.34 9.2 4.5E-04

9.1 3.60E-04

1 1

350 187 3.34

Volume resistivity

Í2 Em

1.20E+14

2.90E+13

Dielectric strength

kV/mm

36

30

Resistance to Alkaline Solution As shown in Figure 5, weight loss of the N-substrate after 100 hours was below 0.5g/m2, and it was 1/250 of the original substrate and 1/5 of the C-substrate. From the results, it was confirmed that the resistance to the alkaline solution has significantly been improved by the newly developed process.

Figure 5 Weight loss of the substrates after soaking in a carbonate aqueous solution (pH=10,60 D).

Cu Post-Fired Metallization Process

Novel Processing of Ceramics and Composites

• 145

Novel Process for Surface Treatment of AIN—Characterization and Application

As shown in Figure 6, the adhesion strength of the N-substrate has 86 MPa, which is about 1.5 times larger than the value of the original substrate or C-substrate. Figure 7 shows the adhesion strength of the Cu post-fired substrate coated with Ni/P layer manufactured by plating after heat cycle tests were performed (1 cycle: -50°C - +125°C - -50°C, disclosure time: 10min, cycle number: 500, 1000, 3000 cycles). From the results, it was confirmed the Cu post-fired metallization on the N-substrate has excellent adhesion strength and sufficient reliability even under severe environments.

Figure 6 Adhesion strength of Cu post-fired substrates. (a) original, (b) conventional process, (c) newly developed process.

Figure 7 Adhesion strength of Cu post-fired substrate after heat cycle test.

146

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—Characterization and Application

CONCLUSION We have developed the novel process of surface treatment of AIN substrate based on the oxidation behavior of AIN. The oxide layer formed on surface-treated AIN substrate consisted of a-alumina and had many protuberant striate patterns at the surface. The physical properties are almost equal to those of the original substrate except for bending strength. The treated substrate has an excellent resistance to the alkaline solution. The oxide layer grows according to the parabolic rate law, and the activation energy of this process has been determined to be 361 kJ/mol. Furthermore, we have also developed the Cu post-fired metallization process associated with the surface-treated substrate. It has been confirmed that the Cu post-fired metallization substrate has excellent adhesion strength and sufficient reliability even under severe environments. REFERENCES 'H. Itoh et al., "Oxidation Resistance of AIN Coated Graphite Prepared by Plasma Enhanced CVD",J. Ceram. Soc, Japan, 94 [1], 136-150, (1986). 2 H. Fujimori et al., "Corrosion Resistance of Aluminum Nitride as Semiconductor Equipment Parts Studied by Micro-Raman Spectroscopy", J. Ceram. Soc, Japan, 111 [12], 935-938, (2003). 3 H. Fukuyama, T. Tanoue, and K. Nagata, "Novel Process for Surface Treatment of AIN High-Temperature Oxidation Behavior of AIN -," Proceedings of Pac Rim-6, Maui, Hawaii, Sep., (2005). 4 Y. Kurihara et al. "The Influence of Moisture on Surface Properties and Insulation Characteristics of AIN Substrates", IEEE TRANS. COMP. MAMUF. TECHNOL., 12 [3], 330-334(1989). 5 D. Suryanarayana et al. "Behavior of Aluminum Nitride Ceramic Surfaces Under Hydrothermal Oxidation Treatments", IEEE TRANS. COMP. MAMUF. TECHNOL., 12 [4], 566-570,(1989). 6 P. K-Weiss and J. Gobrecht, "Directly Bonded Copper Metallization of AIN Substrates for Power Hybrids", Materials Research Society, Symp. Proc, 40, 399-404, (1985). 7 Y. Yoshino, "Role of Oxygen in Bonding Copper to Alumina", J. Am. Ceram. Soc, 72, [8], 1322-1327,(1989). S. Mellul et al., "Optimization of Nitrogen Atmosphere for the Manufacturing Copper Thick-Films over Aluminum Nitride Substrates", Proc. of 6th I.M.C., 279-286, (1990). 9 Y. Yoshino et al., "Structure and Bond Strength of Copper-Alumina Interface", J. Am. Ceram. Soc, 75, [10], 2756-2760, (1992).

Novel Processing of Ceramics and Composites

• 147

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

NOVEL PROCESS FOR SURFACE TREATMENT OF AIN — HIGH-TEMPERATURE OXIDATION BEHAVIOR OF AIN Hiroyuki Fukuyama Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University 2-1-1 Katahira, Aoba-ku, Sendai, Japan, 980-8577 Tetsuharu Tanoue and Kazuhiro Nagata Department of Chemistry and Materials Science, Tokyo Institute of Technology 2-12-1 Ookayama, Meguro-ku Tokyo, Japan, 152-8552 ABSTRACT In order to understand fundamental mechanism of oxidation of aluminum nitride (AIN) and develop a new surface treatment forming a protective oxide scale on AIN, the oxidation behavior of AIN powder has been studied in an oxygen or oxygen/nitrogen atmosphere by thermogravimetry-differential thermal analysis and in-situ x-ray diffractometry. In the initial stage of heating AIN, oxygen mostly dissolves into AIN. When the temperature reaches a threshold temperature, an abrupt oxidation occurred. The mechanism of the abrupt oxidation is discussed in association with the dissolved oxygen. On the basis of the results and discussion, a new surface treatment is proposed to form a crack-free protective oxide scale on AIN. INTRODUCTION Aluminum nitride (AIN) is one of the most promising non-oxide ceramics due to its characteristic properties such as high thermal conductivity1, excellent electric insulation performance, thermal expansion property close to silicon and resistance to halogen gas plasma. Therefore, many applications are currently proposed as a heat sink for LSI and laser diode (LD), insulator and substrates for power transistors, and some parts for semiconductor manufacturing equipments. However, AIN has some shortcomings, i.e., reactivity with moisture forming NH3 gas and a poor resistance to an alkaline solution. To overcome the problems associated with surface reactivity, the surface treatments forming a protective layer have been required. An AI2O3 scale formed on AIN surface obtained by high-temperature oxidation is expected to be a possible protective layer. In the past few decades, many fundamental studies have been conducted on high-temperature oxidation behavior of AIN2"7. Water vapor significantly had an influence on oxidation behavior of AIN as Kuromitsu et al. presented4. However, many investigators conducted oxidation experiments just in air without paying much attention to residual water vapor content in the gas phase2'5"7. Thus, the oxidation kinetics disagrees among the investigators. In the present paper, the oxidation behavior of AIN in a dry oxygen or oxygen/nitrogen atmosphere has been intensively studied by thermogravimetry-differential thermal analysis (TGDTA) and in-situ x-ray diffractometry for AIN powder. On the basis of the results, a new process for oxidation of AIN is proposed to obtain a crack-free oxide scale.

149

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

EXPERIMENTAL Thermogravimetry-Differential Thermal Analysis (TG-DTA) A1N powder (R-15, TOYO ALUMINIUM K. K., Japan) was used for the TG-DTA. It should be noted that this powder was produced by nitriding metallic aluminum. Average grain size of the powder was 13 (im and its purity is 99.9 % in mass. Impurity content in the A1N powder is given in Table I. The oxidation kinetics of the A1N powder was studied by TG-DTA under flowing oxygen gas or oxygen/nitrogen gas mixture at a rate of 50 NmL/min. High purity oxygen (99.999 %) and nitrogen (99.9999 %) gases were used and dried by passing it through columns containing silica gel and P2O5 before introducing the gas into the TG-DTA apparatus. The sample powder was set in an alumina crucible (inner diameter: 5 mm) and heated by infrared radiation at a rate of 75 K/min. Table I Impurity content in the A1N powder (R-15, TOYO ALUMINIUM K. K., Japan) 0 4100

C 10

Concentration of impurity (mass ppm) Fe Si Cu Mn Mg Zn 441 230 3 6 6 5

Ni 15

Cr 11

Ti 21

In-Situ High-Temperature X-Ray Diffractometry The x-ray diffractometer incorporating a platinum-wire resistance furnace was used for in-situ studying oxidation behavior of the A1N powder. The A1N powder was placed on a platinum holder and the holder was set in the platinum-wire resistance furnace. The furnace was surrounded by insulator and x-ray was allowed to pass through windows made of a nickel film and a polyimide film (Kapton, DuPont). The high purity oxygen gas dried in the manner described above was intorduced into the furnace at a rate of 100 - 200 NmL/min. The sample powder was heated at a rate of 30 K/min from room temperature to 973 K. After that, the sample was heated at a rate of 25 K/min from 973 to 1573 K. Cu-Ka was used as an x-ray source. The xray apparatus was operated with accelerating voltage and current of 40 kV and 200 mA, respectively. During the course of heating sample, in-situ x-ray diffraction was conducted in a range of 28 from 30 to 40 degree to cover all characteristic peaks of possible aluminum oxides such as OC-AI2O3, 9-AI2O3, Y-AI2O3 and S-AI2O38"10. The x-ray scanning rate was 5 degree/min. Dissolution of Oxygen into A1N Powder To elucidate phenomena of dissolution of oxygen into the A1N powder, the A1N powder was kept at 1173 K, at which no obvious oxidation was occurred as presented from the results of the TG-DTA (see RESULTS section). The sample powder placed in an alumina crucible was set in a one-end closed alumina tube. The alumina tube was first evacuated and subsequently the dried oxygen gas was blown down to the sample held in the crucible at a rate of 100 - 200 NmL/min for several time periods from 1 to 48 h. After holding the sample for the desired time periods, the sample was quenched and subject to the x-ray diffractometry for evaluating shift of lattice constant of A1N caused by dissolution of oxygen. The dissolved oxygen content was quantitatively determined by the following mass balances, (total increase in mass of sample) - (decrease in mass of nitrogen) = (total mass of oxygen) (total amount of oxygen) - (amount of oxygen as (X-AI2O3) = (amount of dissolved oxygen)

150

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

The total increase in mass of sample was determined by thermogravimetry (TG). The nitrogen content in the sample was determined by an inert gas fusion method (LECO, EF-400, TC-436). To complete the extraction of nitrogen, the sample powder was encapsulated in a nickel container and the container was heated in a graphite crucible for the LECO nitrogen analyzer. The amount of 01-AI2O3 was determined by the external standard method of x-ray diffraction using Cu-K 0 as an x-ray source. The peak of (012) plane of a-Al 2 0 3 was used. The peak profile was fitted by a Gaussian function. The amount of C1-AI2O3 was evaluated by the relative integral intensity of the peak with respect to that of pure OC-AI2O3. Thus, amount of dissolved oxygen was evaluated by subtracting oxygen existed as (X-AI2O3 from total oxygen content. RESULTS TG-DTA Figure 1 shows the typical (a) TG and (b) DTA profile of the AIN powder on heating in an oxygen gas stream. The ordinate is defined as follows, (Relative oxidation ratio, %) =

}MAMi.

^,4/203 -

2M

AIN

x

— x 100 W

(1)

i

where Mc is relative molar mass of c, W¡ the initial weight and AW an increase in weight. According to the above definition, 100 % means a complete oxidation of the AIN powder. In the early stage of heating, a slight increase in mass was only observed. Subsequently, an abrupt increase in mass was occurred around 1373 K associated with a large exothermic peak. Thus, the oxidation of the AIN powder presents a characteristic behavior. Figure 2 shows the TG profile with changing an oxygen/nitrogen-mixing ratio. With decreasing the ratio, the abrupt increase in mass became more moderate. In-Situ High-Temperature X-Ray Diffractometry Figure 3 shows the results of the in-situ x-ray diffraction of the AIN powder on heating in an oxygen gas stream. The peaks for CI-AI2O3 were emerged at 1473 K. No polymorphic AI2O3 such as O-AI2O3, Y-AI2O3 and 8-AI2O3 were appeared. It is concluded that the abrupt increase in mass associated with the large exothermic peak presented in Fig. 1 is caused by the following oxidation reaction, AlN(s) + 3/20 2 (g) = 1 /2a-Al 2 0 3 (s) +1 /2N2(g)

(2).

Below the threshold temperature, 1373 K, oxygen mostly dissolves into the AIN powder.

Novel Processing of Ceramics and Composites

• 151

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

Figure 1 TG-DTA profile of the AIN powder heated in an oxygen gas stream.

Figure 2 TG profile of the AIN powder with changing a oxygen/nitrogen-mixing ratio.

152

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

Figure 3 In-situ x-ray diffraction for oxidation of the A1N powder in an oxygen atmosphere. Dissolution of Oxygen into A1N Powder Figure 4 shows the variation of the lattice constants of a- and c-axes of a hexagonal A1N crystal caused by the dissolution of oxygen during heating the A1N powder in an oxygen gas stream at 1173 K. Both lattice constants decreased with time. This may be interpreted by the oxygen dissolution into nitrogen sites. Nitrogen atoms having a larger ionic radius (1.71 angstrom)1 were replaced by the dissolved oxygen atoms having a smaller ionic radius (1.40 angstrom)". Figure 5 shows the variations with time of contents of nitrogen and dissolved oxygen in the A1N powder heated for several hours in an oxygen gas stream at 1173 K. As the oxygen content increased, the equimolar amount of nitrogen decreased. DISCUSSION Mechanism of Dissolution of Oxygen into A1N The oxidation behavior of the A1N powder is characterized as the initial stage of dissolution of oxygen followed by the abrupt oxidation as shown in Fig. 1. In the initial stage, the reaction of dissolution of oxygen may be expressed as

Novel Processing of Ceramics and Composites

• 153

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

10

20 30 Time , ( / h

40

50

Figure 4 Variations with time of lattice constants of a- and c-axes of AIN caused by the dissolution of oxygen in the AIN powder heated for several hours in an oxygen gas stream at 1173 K.

Figure 5 Variations with time of concentrations of nitrogen and dissolved oxygen in the AIN powder heated for several hours in an oxygen gas stream at 1173 K. NN+^02(g) = 0 ^ + e - + | N 2

154

• Novel Processing of Ceramics and Composites

(3)

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

using Kröger-Vink notation, because the equimolar nitrogen was replaced by the dissolved oxygen as shown in Fig. 5. On the basis of the law of mass action, concentration of dissolved oxygen is given by [0*N]~ (PoJns.f1

(4)

where [O^ ] is the concentration of the dissolved oxygen and/?¡ is activity of i with respect to 1 bar of ideal gas. Mechanism of Abrupt Oxidation of AIN The abrupt oxidation became more moderate with decreasing ratio poi/pta as shown in Fig. 2, where decreasing ratio poilpm results in decreasing concentration of dissolved oxygen as Eq. (4) tells. On the basis of the results, the mechanism of the abrupt oxidation is proposed as follows, (1) In the initial stage of heating, oxygen just dissolves in the AIN powder as illustrated in Fig. 6(a). (2) When the temperature reaches the threshold temperature, 1373 K, the dissolved oxygen is triggered to be transformed into (X-AI2O3 in the AIN. (3) The formed 01-AI2O3 gives rise to microscopic cracks in the vicinity of surface of the AIN, which creates more surface area for oxidation, leading to the abrupt increase in mass as illustrated in Fig. 6(b).

Figure 6 Illustration of (a) dissolution of oxygen into AIN and (b) abrupt oxidation caused by the dissolved oxygen. New Process for Surface Treatment of AIN Based on the nature of oxidation of AIN as described above, a new surface treatment of AIN is proposed to form a crack-free protective oxide scale on AIN. The point is how to oxidize

Novel Processing of Ceramics and Composites

• 155

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

AIN without dissolution of oxygen before the oxidation begins at the threshold temperature. The following procedure is given as an example. (1) Dissolution of oxygen is completely prevented by heating AIN in a nitrogen atmosphere until the threshold temperature. (2) At a given temperature above the threshold, the nitrogen gas is subsequently switched to oxygen gas to allowed AIN to be oxidized. Some advantageous effects of this new process are presented in detail by Yoneda from the practical point of view in the Pac Rim-6 conference. CONCLUSION The oxidation behavior of the AIN powder, which is manufactured by nitriding metallic aluminum, has been studied in the oxygen and oxygen/nitrogen atmosphere. In the initial stage of heating AIN, oxygen mostly dissolves into AIN by the following reaction, NÏj+^0 2 (g) = 0 ^ + e - + i N 2 . When the temperature reaches the threshold temperature, 1373 K, the abrupt oxidation occurred by the following reaction, AlN(s) + 3/202(g) = l/2a-Al203(s) +l/2N2(g). The abrupt oxidation may be caused by creating more surface area associated with microscopic cracks generated by the transformation of the dissolved oxygen into (X-AI2O3 in the vicinity of surface of the AIN powder. On the basis of the results, the new surface treatment has been proposed to form a crack-free protective oxide scale on AIN. REFERENCES 'G. A. Slack, R. A. Tanzilli, R. O. Pohl, and J. W. Vandersande, "The Intrinsic Thermal Conductivity of AIN," J. Phys. Chem. Solids, 48, (7), 641-47(1987). 2 A. D. Katnani, and K. I. Papathomas, "Kinetics and Initial Stages of Oxidation of Aluminum Nitride: Thermogravimetric Analysis and X-Ray Photoelectron Spectroscopy Study," J. Vac. Techno!. A, 5, (4), 1335-40(1987). 3 T. Sato, K. Haryu, T. Endo, and M. Shimada, "High Temperature Oxidation of HotPressed Aluminium Nitride by Water Vapor," J. Mater. Sei., 22, 2277-80(1987). 4 Y. Kuromitsu, H. Yoshida, S. Ohno, H. Masuda, H. Takebe, and K. Morinaga, "Oxidation of Sintered Aluminum Nitride by Oxygen and Water Vapor," J. Ceram. Soc. Jap., 100,(1), 70-4(1992). 5 A. Bellosi, E. Landi, and A. Tampieri, "Oxidation Behavior of Aluminum Nitride," J. Mater. Res., 8, (3), 565-72(1993). 6 P. S. Wang, S. G. Malghan, S. M. Hsu, and T. Wittberg, "The Oxidation of an Aluminum Nitride Powder Studied by Bremsstrahlung-Excited Auger Electron Spectroscopy and X-Ray Photoelectron Spectroscopy," J. Mater. Res., 10, (2), 302-5(1995). 7 E. W. Osborne, and M. G. Norton, "Oxidation of aluminum nitride," J. Mater. Sei., 33, 3859-65(1998).

156

• Novel Processing of Ceramics and Composites

Novel Process for Surface Treatment of AIN—High-Temperature Oxidation Behavior of AIN

G. C. Ryubicki, and J. L. Smialek, "Effect of the G-OC-AI2O3 Transformation on the Oxidation Behavior of ß-NiAl+Zr," Oxidation of Metals, 31, 275-304(1989). 9 A. Andoh, S. Taniguchi, and T. Shibata: High-Temperature Corrosion and Protection 2000, 297-303(2000). 10 W. Wintruff: Krist.Teck, 9, (4), 391-403(1974). n R. D. Shannon, "Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides," Acta Cryst., A32, 751-67(1976). 12 T. Yoneda, N. Teramoto, K. Takada, and H. Fukuyama, "Novel Process for Surface Treatment of AIN - Characterization and Application -," Proceedings of Pac Rim-6, Maui, Hawaii, Sep., (2005).

Novel Processing of Ceramics and Composites

■ 157

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

MYCRONID1M BASED LONG-LASTING BN-HARDCOATING AS RELEASE AGENT AND PROTECTION AGAINST CORROSION FOR ALUMINUM FOUNDRY APPLICATIONS Jochen Greim, Martin Engler, Krishna Uibel, Christoph Lesniak ESK Ceramics GmbH & Co. KG Max-Schaidhauf-Straße 25 Kempten, D-87437, Germany ABSTRACT Boron Nitride powder containing suspensions (MYCRONID™) were developed for applications in aluminum processing (e.g. as coatings for foundry consumables in non-pressure applications). MYCRONID™ suspensions consist of dispersed hexagonal BN-particles (1-10 um) in a reactive TEOS derived sol-gel-binder. These suspensions are applied by conventional coating technologies. After drying and firing (500°C) dense, wear resistant glass ceramic-BN composite layers in the range of 3-20 um were obtained. Coated substrates (hot-work steel or conventional refractory ceramics) were exposed to Al-melts at temperatures of 750°C and showed excellent corrosion resistance, oxidation protection and thermal stress stability. It was observed, that only with thoroughly cleaned and pretreated substrates, the excellent results were achieved. INTRODUCTION Aluminum is the most important light metal; the world consumption is around 25 million tons per year. Primary aluminum industries and foundries handle aluminum melts (usually T > 700°C) which are chemically highly aggressive. The Al-melts attack melting crucibles, molds, runners, tubes, spoons and stirrers and limit the life time of such parts. Due to that one observes another problem: the contamination of the aluminum with the corrosion products. Therefore the protection of surfaces in contact with Al-melts is an important issue in the Al-processing industry. Up to now only coatings with a limited life time ("single use coatings") were available, acting either as corrosion protection or release agent. Examples are powder coatings based on bone ash, silicates, marble dust and Boron Nitride. The limited life time of these coatings requires a frequent recoating: usually it is necessary to remove the former coating completely and to clean the surface of the substrate carefully. The resulting downtimes have a negative impact on the productivity of foundries. In order to achieve higher productivities and to improve the product quality, protective release coatings with a long life time are essential. Such coatings must withstand the chemical and corrosive attack of the molten metal at high temperatures (T > 700°C) and huge thermal shocks during e.g. casting processes. This paper describes new Boron Nitride based coating systems for hot-work steel and refractory ceramics fulfilling the respective requirements of the Al-processing industries. BORON NITRIDE AND ITS USE IN ALUMINUM PROCESSING INDUSTRY One of the few materials stable under the extreme conditions, occurring during the processing of molten aluminum is hexagonal Boron Nitride (hBN). hBN is isoelectronic to graphite and is often referred to as the „white graphite". Some general properties of hBN are

159

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

listed in Table 1. It is important to note here, that hBN crystallizes in platelets with hexagonal shape (Fig.l ), and that hBN particles show only a very low surface chemical activity, due to the low density of functional groups. Table 1: Physical properties of Boron Nitride Chcm. formula Molecular weigh! (g/mol) Crystal structure Spec, weight (g/cm3) Melting point (°C)

BN 24.82 hexagonal 2.25 2700-3(100i">

i lurches:* (Moris') Friction coefficient El. resistivity (£2 cm) Therm, conductivity (W/m K)

1-2«) 0 . 2 - 0 . 7 « !1 > I2!:*) 60 II"» 120 l">

Linear coefficient of thermal expansion al20-IOOO°C{IO-6K-!)

7.51 0.71

IP ±">

*) These values are taken from hot-pressed BN (ihe symbols li + J. indicate values lakeu parallel and perpendicular to the pressing direct ion) **) decomposition

Figure 1 : SEM picture of hexagonal Boron Nitride powder (hBN); average particle size ~ 8 um One of the most interesting properties of Boron Nitride for the aluminum industry is its poor wettability by metal melts, like aluminum and magnesium and the alloys thereof. Fig. 2 shows the contact angles of liquid aluminum on Boron Nitride, Silicon Nitride and Alumina in the temperature range up to 1100°C. High contact angles mean poor wetting, low contact angles spreading of the liquid. Up to a temperature of 900°C Boron Nitride shows wetting angles of about 160°, i.e. aluminum melts do not wet BN. Between 900°C-1000°C the wetting angle of

160

• Novel Processing of Ceramics and Composites

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

contact gradually decreases.

E

Î

?

180 160 140 120 100

1

80 60 40

20 0 650

750

850

950

1050

1150

temperature [°C]

Figure 2: Wetting behavior of aluminum on Boron Nitride (BN), Silicon Nitride (SÍ3N4) and Alumina (A1203) ' Taking into account the low chemical reactivity and the described wetting behavior, hBN combines all the properties that make it the material of choice for the use in Al-metallurgy2. hBN is used since many years in other metallurgical applications, being it sintered parts like break rings or powder coatings (made from suspensions or electrostatically coated). CONVENTIONAL BORON NITRIDE COATINGS There are a few disadvantages with the classic BN-coatings: they adhere poorly on the substrates and are removed with the slightest mechanical force - the material crumbles away, even when containing refractory binders (aluminum-silicates or -phosphates). The coatings survive only a short time under work load and have to be repaired or renewed frequently. The reason for the limited life time is the poor adhesion of the hBN-particles due to the low chemical reactivity: conventional refractory binders cannot bind chemically to the hBN-particles and so the adhesion is mainly caused by weak van-der-Waals forces. NEW, LONG-LASTING BORON NITRIDE HARDCOAT1NG The request for long-lasting BN-coatings in the aluminum-processing industry requires a different approach. Very reactive binders are necessary for the almost inert hBN-particles; the binder must allow a chemical bonding to the few surface functional groups of hBN-particles and to a broad variety of substrate types. On the other hand, the binder must withstand the stresses occurring in the target applications (T up to 750°C, thermo-shock, corrosion, and oxidation). These requirements can only be matched by an inorganic high temperature stable binder. Inorganic sol-gel-coatings are already well known in literature, also for applications at high temperatures3. Based on these concepts a new binder concept for hBN was developed. By dispersing fine BN-powders (1-10 um) in an ethanol based binder-sol, the new MYCRONID ' suspensions were obtained. A variation of the binder to hBN ratio is possible and gives a broad range offlexibilityregarding applications4.

Novel Processing of Ceramics and Composites

• 161

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

Conventional coating technologies like painting, dip-coating or spraying can be used to apply MYCRONID^TM suspensions. Spraying results generally in the smoothest surfaces and the thinnest layers. The hardening of the applied sol-gel coatings and the transformation into a protective layer is achieved by an appropriate thermal treatment: first, a gelation step at elevated temperatures (> room temperature) gives the coating a certain mechanical stability. In a second temperature step (around 500°C), the binder matrix is densified and the binder reacts chemically with the hBNparticles and the substrate. The results are nearly pore free, glass-ceramic BN composite layers, showing a much higher mechanical stability (wear resistance) than conventional BN powder coatings.

Figure 3: Model describing BN-hardcoatings made of MYCRONIDIM suspensions; sol-gelbinders form a dense glass-ceramic matrix for BN-particles (aligned parallel to the substrate; see SEM picture) After the application of the coating and prior to the drying and densification, the Boron Nitride platelets go through a self-orientation process and align parallel to the coated surface. Fig. 3 shows a model of the described new coating systems derived from MYCRONIDIM. Due to the very low thickness (< 50 um), the layers exhibit excellent thermo shock resistance. For many applications in the nonferrous metallurgy such Si02-based sol-gel-binders are suitable, as it will be shown below. NEW BN-HARDCOATINGS AND THEIR PERFORMANCE IN LIQUID ALUMINUM MYCRONID 'M suspensions were applied on hot-work steel and silicate based refractory ceramics. The samples were ultrasonic cleaned in ethanol prior to coating. Samples were either

162

• Novel Processing of Ceramics and Composites

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

coated by spray-coating (airbrush) or dipping (automated). MYCRONID™ suspensions had a viscosity of 20 mPas. Coated samples were dried in a drying furnace at temperatures up to 120°C for 30 minutes. The dried samples then were heated up to 500°C for 30 minutes to form the glass-ceramic BN-hardcoating layer. Crack free coatings showed a thickness between 3 and 20 Urn. To determine the corrosion protection and the release properties of the BN-hardcoatings, coated and uncoated samples were exposed to liquid aluminum under different conditions: • Static tests were carried out, rotating the samples in Al-melt at 750°C up to 120 h, • In dynamic tests the samples were dipped into Al-melt at 750°C up to 3000 cycles. Already after the first melt contact, uncoated hot-work steel samples are covered with aluminum. A complete dissolution of the uncoated steel substrate with the melt was observed after 3 h.

Figure 4: Hot-work steel samples in contact with liquid aluminum at 750°C a) Uncoated sample, 1 h of Al-melt contact b) BN-hardcoated sample, 120 h of Al-melt contact Coated samples showed the same non wetting behavior as sintered BN, though the hBN is embedded in a binder matrix. The samples were stable during the whole testing period, being it 120 h rotation in melt or 3000 cycles of dipping respectively (Fig. 4b). Occasionally observed adhesions of Al on the BN coatings wereflakingoff without any external force. Samples tested under the above described conditions were carefully analyzed by SEM at the contact interface melt/substrate and melt/BN-hardcoating respectively.

Novel Processing of Ceramics and Composites

• 163

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

Fig. 5a shows the interface of an uncoated substrate after only 2 min of melt contact. Three different regions can be distinguished: on the left side of the picture only pure aluminum was detected by EDX. At the contact face of melt and substrate aluminum, iron and its alloying elements were detected. This is a clear indication for heavy corrosion in the melt contact zone This means that dissolution of the steel in the aluminum melt starts immediately after melt contact. Fig. 5b shows a SEM cross section of a BN-hardcoated substrate (hot-work steel) after 120 hours of melt contact. No corrosion or oxidation can be found, the BN-hardcoating layer is intact, thus providing a corrosion protection to the substrate. On the left side of the picture pure aluminum was detected by EDX, but the Al does not stick to the coating, proving the non wetting properties of the new coating system. The observed non-wetting behavior of the coatings is attributed to the embedded BN-particles, and the corrosion and oxidation protection is explained by the parallel orientation of the BNplatelets in the matrix (see Fig. 3 above).

Figure 5: SEM cross section of hot-work tool steel samples in contact with molten aluminum at 750°C a) Uncoated sample after 2 min. of Al-melt contact b) BN-hardcoated sample, 120 h of Al-melt contact, in the solidified melt The same positive results regarding corrosion protection and non-wetting were obtained with BN-hard coatings on silicate based refractory ceramics. CONCLUSION AND OUTLOOK A new class of BN-suspensions, based on a sol-gel-binder system was developed for applications in the aluminum processing industry (MYCRONID™). These new suspensions form long-lasting BN-hard coatings showing very interesting properties in laboratory tests: in contact with aluminum melts, they act not only as a release agent but provide also protection against oxidation and corrosion. The system was successfully applied to hot-work steel and refractory ceramics.

164

• Novel Processing of Ceramics and Composites

MYCRONID™ Based Long-Lasting BN-Hardcoating as Release Agent

MYCRONID suspension derived wet coatings are transformed into dense, wear resistant glass-ceramic-BN composite layers by a thermal densification step. So far MYCRONID™ suspensions have to be considered as "expert systems". The application of such formulations is more demanding than for standard, commercial available BNsuspensions. To achieve the described performance a thorough cleaning of the surfaces to be coated is a must. On different materials (ceramics, metals, glass) the solid content and the BN to binder ratio have to be adapted. For the demands of different applications, as wear, oxidation, corrosion, thermal stress and their combinations, the best performing layer thickness has to be adjusted, i.e. the coating application itself (spray-coating, dip-coating, painting) has to be modified. The BN-hardcoatings have to be absolutely defect free (no pores, no cracks), as failure would begin immediately under those challenging ambient conditions. The results presented were achieved using an ethanol solvent based suspension. Flammable liquids might not be acceptable in every location or application. Therefore water based systems, would provide an interesting extension of the presented concept. Boron Nitride hardcoatings will not replace technical solutions made of bulk ceramic material, but could ease the work in many applications, making them more reliable and economic. Also other application fields beside the aluminum industry that require high temperature stability and corrosion resistance in combination with release properties could make use of the presented coating systems (e.g. glass melting and forming, hot gas systems). REFERENCES 1

Gießerei 8 (1993), pp. 256-259 Rudolph, S.; Klein, F.: Gießerei-Praxis (1992) Nr. 6, S. 81 - 84. 3 C. J. Brinker and G. W. Scherer: Sol-Gel Science, Acad. Press, San Diego, 1990. 4 DE 103 26 769 B3 2

Novel Processing of Ceramics and Composites

• 165

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

A STUDY INTO EPOXY COMPOSITES FOR HIGH-VOLATGE DEVICE ENCAPSULATION Dr. Ammer K. Jadoon A.W.E Aldermaston Berkshire RG7 4PR United Kingdom Prof. John C. Fothergill and Mr. Andy Wilby University of Leicester Leicester LEI 7RH United Kingdom ABSTRACT An initial study was carried out to identify and characterise an encapsulant material for a high-voltage device. The first phase material specifications were to have high dielectric breakdown strength, good mechanical and thermal shock properties; a wide operating temperature range and to be produced by a simple and commercially viable route. Two epoxy-alumina composite based systems were looked at, where both systems had a DGEBA resin and over 50 wt% paniculate alumina but different hardeners (curing agents). System 1 had a DDM hardener while system 2 had a DTD, both belonging to the same aromatic amine family. It was hoped to produce a composite material with tailored properties. A simple processing route was used to produce the epoxy-composite and the addition of a large amount of alumina was found not to effect the curing kinetics and formation of the epoxy based matrix. This is unusual due to the large amount of alumina filler incorporated, but highly desirable. The tensile and thermo-mechanical (thermal shock) properties were found to improve by the addition of the alumina filler. The glass-transition temperature (Tg) was unchanged, indicating that the filler had no adverse effects on the formation of the epoxy matrix. A uniform and homogenous distribution of the alumina was found (no sedimentation) and was unaffected by the casting mould type (top/side pour). The dielectric breakdown results showed that the addition of alumina filler lowered the breakdown strength. 1 INTRODUCTION Epoxy resins are amongst the most important and widely used thermosetting polymeric materialsfl]. Their desirable properties include heat, moisture and chemical resistance, good dielectric and mechanical properties, dimensional and thermal stability, good creep resistance, low shrinkage during cure and adhesion to many substrates[2]. These properties have made them ideal candidate matrices for various important applications including adhesives, electronic encapsulation and matrix resins for high performance fibre reinforced composites in the aerospace and electronic industries. However, brittleness and poor crack propagation resistance due to high cross-linking densities are inherent problems that many researchers have tried to address in different ways. The inclusion of second phase rubber particles (elastomeric material) to improve the mechanical properties was a technique used for many years. The major disadvantage of this was the lowering of the elastic modulus, yield strength and creep resistance[2].

169

Study into Epoxy Composites for High-Voltage Device Encapsulation

In later years this was followed by the addition of dispersed rigid thermoplastics such as polyethersulphone (PES) to the epoxy matrix [2]. The major disadvantages were found to be the lowering of the Tgand reduction in the modulus and tensile strength. The use of alumina (AI2O3) powder offers a cost effective technique of producing an epoxy composite with enhanced mechanical and dielectric properties. The physical, electrical, ultimate mechanical properties and performance of epoxy resins is dependant upon the degree of cure and filler content. The basic parameter governing the state of the material is the chemical conversion. The cross linking polymerisation reactions of a thermoset polymer generally involve the transformation of a fluid into a rubber (gelation) and then into a solid glass (vitrification) as a result of further reactions. The final properties of the cross linked epoxy resin depend on the reaction kinetics of the curing reaction. Thus, knowledge of the reaction kinetics during cure and how the rate varies with cure temperature, time and filler content is important to ensure the optimum processing conditions are used to produce the desired final properties in the epoxy resin. Fillers are generally added to thermosetting polymers with the following in mind; a. Improve dimensional stability - heat distortion temperature, shrinkage and thermal expansion. b. Mechanical properties - stiffness, compressive and tensile strength c. Electrical/optical properties Generally, fillers tend to be stiffer than the matrix and the most common fillers for epoxy resins are silicates, calcium carbonate, glass spheres and occasionally alumina and kaolin. However, they tend to be used in small quantities with 20wt% being a maximum. As the curing reactions are exothermic, DSC is an excellent tool to measure the heat change with time and this can be related to the curing process. However, these parameters are often not sensitive enough to measure small changes in chemical conversion, especially at high conversion and in diffusion controlled regimes. A parameter that shows considerable increase accompanying the changes in chemical conversion during cure of a thermosetting material is the Tg. The fact that Tg increases non-linearly with conversion in cross-linking systems makes it more sensitive in the later stages of cure. Once cured Tg is a useful quality assurance (QA) parameter as it indicates extent of cure. This paper presents the results of an initial study to identify and characterise an encapsulant material for a high-voltage device. The first phase material specifications for the material were; a. High dielectric breakdown strength b. Able to withstand vigorous thermal cycling in the -45 to 70°C range c. A high Tg (>90°C) d. Adequate mechanical properties (able to withstand robust handling and retain properties with time) e. Simple processing route Two epoxy-alumina composite based systems were looked at, where both systems had a DGEBA resin and over 50wt% paniculate alumina but different hardeners (system 1 had a DDM hardener while system 2 had a DTD). The mechanical, thermal shock, dielectric and thermomechanical (Tg) properties were looked at in addition to the curing kinetics of the epoxy resin matrix and various effects of alumina filler.

170

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

2 EXPERIMENTAL 2.1 Materials & Sample Preparation An unmodified DGEBA resin was used with room temperature viscosity of 1200-1400 mPas. Two different hardeners were used, the first was a 3. 3'-diethyl-4-4'diaminodiphenylmethane (DDM) formulated aromatic amine and the second was a diethyltoluenediamme (DTD) based formulated aromatic amine. All were supplied by Robnor Resins. UK. High grade alumina powder with an average particle size of 5[im was used as filler with over 50 wt% used in the resin composite. This was left in a furnace at 110°C for a few days to 'dry' the alumina. The resin, hardener, filler and casting moulds were left overnight in an air furnace at 50°C. The mixing process for the epoxy-composite initially involved mixing the pre-heated resin and filler in a laboratory mixer for at least 5 minutes at 5000 rpm. This mixture was then placed in a heated furnace (40°C) for 15 minutes to assist with the wetting process. Hardener was then added and mixed for 5 minutes at 5000 rpm. The next step involved placing the mixture on a hot plate (70°C) vacuum chamber and degassed to remove air bubbles. All pouring into moulds was under vacuum. The unfilled resin (no alumina) process was similar, involving direct mixing of the resin and hardener followed by vacuum degassing. Curing was carried out in an air furnace using the determined schedule (55°C for 15 hours, 80°C for 24 hours). Figure 1 is a schematic of the process.

Figure 1. Schematic of processing route to produce epoxy composite.

Novel Processing of Ceramics and Composites

• 171

Study into Epoxy Composites for High-Voltage Device Encapsulation

2.2 Characterisation and Testing 2.2.1 Differential Scanning Calorimetry (DSC) As the curing of epoxy resins is exothermic, the DSC is an excellent tool for obtaining reaction rates/kinetics and degree of cure by a combination of dynamic and isothermal experiments. This can be used to assess what effects a large amount of alumina can have on the reaction kinetics and curing process. Thermal analysis was performed using a DuPont 910 DSC. Kinetic parameters such as activation energy (Ea) provide us with useful data regarding material processability. Activation energy was calculated according to ASTM method E698-79, otherwise known as the Ozawa-Kissinger method. 2.2.2 Thermogravimetric Analysis (TGA) This was used to determine the amount (wt%) of remnant filler to ensure uniform distribution in cast samples. 2.2.3 Tensile testing The mechanical properties of the material were quantified by tensile testing. To insure greater accuracy six samples were tested per material. A computer controlled Instron machine was used with a 2mm/minute cross-head speed. The samples were pulled to failure and the stress and strain at failure and Young's modulus were recorded. 2.3.4 Dynamic Mechanical Analysis (DMA) Technique in which the elastic & viscous response of a sample under oscillating load is monitored against temperature/time/frequency. Three-point bend mode was used with epoxy composite bars (20mm x 4mm x 3mm). A number of 're-run' samples were carried out, where after cooling to room temperature from a previous DMA run, the sample would be put through the same cycle and the effects upon Tg looked at. A Perkin-Elmer 7e DMA was used. 2.3.5 Thermal Cycling This was carried out on an encapsulated 'nut and bolt'. A nut and bolt was chosen due to the complex shape and possibility of high stress concentration sites in the encapsulant material as a result of this. A steel hex-headed bolt (BS1768) with a steel nut (BS1768) was used and mounted on the end where the thread was fully engaged and the face of the nut was flush with the end of the bolt. This was grit blasted and then cleaned ultrasonically in acetone prior to use. The assembly was suspended by a wire into the resin material and held by an over hanging bridge. The unit was then oven cured. A vigorous thermal schedule was chosen to over test the material, as given below in Figure 2.. The temperature range was -45 to 70°C and this was repeated 100 times. Six samples were tested per system type. After completion the samples were examined by digital radiography using a microfocal x-ray system looking for cracks and delamination (samples were also examined prior to thermal cycling to exclude the possibility of defects being present prior to thermal cycling).

172

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

-70T

-45T

d\

K

Figure 2. Thermal cycling schedule Dielectric testing This was carried out at the University of Leicester. A pin-plate set up was used with a hypodermic needle electrode (55um diameter). The material tested was in disk form (100mm diameter and lmm thick). Each disk was cleaned with IPA (Iso-Propyl Alcohol or propan-2-ol). The disk was submerged in silicone oil and the pin electrode applied under its own weight and that of the connecting copper strip. This did not leave an obvious indent or mark on the surface of the specimen but did hold the pin firmly in place. Experiments were carried out under ambient laboratory conditions. The temperature was 18±2°C. 2.4

2.4.1 Pulsed Voltages Three pulses were applied at each voltage level. The first voltage level was nominally 30 kV on the DC power supply, corresponding to a pulse with a peak voltage of 20 kV. The power supply voltage was increased in steps of nominally 10 kV, i.e. pulse peaks of 6.7 kV. The experiment was stopped when either breakdown or flashover occurred as evidenced by the specimen current waveform. The power supply was a PK Glassman 0-400kV pulse generator, further details given in [3]. 2.4.2 DC Voltage The sample was positioned on a brass electrode of approximately 25 mm diameter. This was smaller than the electrode used for the pulse breakdown tests in order to try to reduce flashover effects. The same needle set-up was used to apply a positive DC voltage to the top surface. The whole set-up was contained in silicone oil at room temperature. The voltage was increased in steps every 20 seconds according to BS2782: Part 2: Method 201, i.e. using the following step voltages (kV) 30, 32, 34, 36, 38, 40,42,44,46, 48 50, 55, 60, 65, 70, 75, 80, 85, 90, 95 100, 110, 120, 130. 140, 150, 160. 170, 180, 190 The reported breakdown voltage is the voltage at which breakdown actually took place. At high voltages (>~70kV) the oil was observed to be in vigorous motion, probably due to electro-hydrodynamic effects.

Novel Processing of Ceramics and Composites

• 173

Study into Epoxy Composites for High-Voltage Device Encapsulation

3 RESULTS & DISCUSSION 3.1 Effects of large amount of particulate filler on curing kinetics Conventional use of particulate filler is usually quite low (less than 10wt%) and the use of over 50wt% filler could possibly have an effect on the curing kinetics. The final properties of the cured epoxy composite depend upon the curing reaction kinetics and are therefore important. As the curing of most thermosetting resins is exothermic, the reaction kinetics can be followed using a DSC. This method assumes that the sample is fully homogenous; the curing reaction occurs in sequential elementary steps; there is no temperature or compositional gradients; the reaction process is activated and no geometric or interfacial effects are present. The curing of an epoxy resin can be autocatalytic or nth-order. Numerous isothermal DSC scans showed the curing reaction to be nth-order, where exothermic reactions occur at time 0, as shown in Figure 3. Therefore, a single rate constant model was used based on the OzawaKissinger approach detailed in ASTM standard E-698. It is important to note that the curing process is thermally activated and hence temperature dependant. As the curing reaction is kinetically controlled up to 75-90% of conversion, the temperature dependence of the rate constant can be given by an Arrhenius relationship. Figure 4a-c shows DSC plots of varied heating rate for system 1 unfilled prepreg. By viewing all of the plots, it can be seen that the curing process is exothermic and effected by heating rate. With low heating rates of 2 and 5°C per minute, two exothermal peaks are present in all cases. Higher heating rates of 10 and 20°C per minute have a single exothermic peak. Same phenomena seen with/without filler for both systems. The first peak of the lower heating rate curve corresponds to the nucleophylic attack of the primary amine function with the epoxy group leading to a secondary amine. This is believed to result in the formation of some secondary amine structures and some linear or branched mers [4,5], The second peak is due to the continuation of the polymerisation reaction leading to the formation of the three-dimensional network structure by the addition of the secondary amine to the epoxy monomer. The curing process common to all liquid thermosets includes gelation, which is the initial formation of an infinite 3D network and vitrification, which is the end process of network formation occurring. During vitrification the Tg rises to the temperature of the cure reaction, Teure- Cure reactions tend to proceed beyond vitrification causing the Tg to be higher than the Teure-

Heating rate plays a major role in this process. With a low heating rate the Teure (cure temperature) is above the initial Tg of the prepreg, chemical reactions occur such that the prepreg Tg rises at a faster rate than the Teure- Vitrification always occurs when Tg equals Tcure and the reaction rate slows down as diffusion control becomes more pronounced and reactant concentration level. After vitrification the Tg increases at the same rate at the Tcure and eventually the Teure rises above the Tg [5]. With a high heating rate the prepreg Tg does not reach TCure- So the reaction can proceed to completion entirely in the rubbery state without encountering vitrification and also explain the difference in the DSC plots with varied heating rates. Most epoxy resin prepregs left at room temperature will not cure, although limited reactions will occur. The curing process is thermally activated and temperature dependant and so

174

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

temperatures of 50°C or more are required for a specified period of time for complete curing. So an increase in heating rate not only follows with increased Tp, Ti and TF (peak maximum, initiation and final temperature of exotherm) but also a decrease in cure time. This was experimentally verified by Rosu et al [6] on a neat epoxy resin system where the Tp, Ti, TF and AH were found to increase with heating rate at a given temperature, while the cure time decreased. The influence and effect of heating rate on the curing process is shown and optimally a moderately high heating rate would be beneficial, especially when curing large volumes of resin The exotherm peak parameter values from the various DSC plots are given in Table 1. The overall trend is that the filler lowers the AH; whilst uneffecting the Tp, Ti and TF. Very importantly this shows that the addition of filler in both systems does not effect the curing dynamics and reaction. The significant reduction in AH can be attributed to the reduced volume of resin-hardener. For both systems the addition of filler does not have a detrimental effect upon the activation energy (Ea) given in Table II. This is encouraging proving that the cross-linking polymerisation process is not effected in both systems. The activation energy values determined seem to agree with those reported in open literature of 50-60 kJ/mol for epoxy resins [7]. An effect on the resin-curing agent stoichiometry caused by the addition of filler is unlikely, as this was comprehensively shown in work by Gupta et al [8]. They studied the effects of varied resin and hardener content in a neat epoxy resin. Increased resin content was shown to increase the AH, Ea, Ti, TF and Tp significantly due to increased cross-linking reactions. As the curing process involves the formation of a network and the cross-link density increases with increasing resin content, they put forth that diffusion becomes more and more difficult with increased resin content and so increased Ea and Tg to drive the reaction and formation of a good cross-linked network. They related the lowering of these parameters with the formation of a poorly cross-linked network and incomplete curing. Our work shows that the addition of filler only lowers the AH and does not effect Ea, Ti, TF and Tp and hence curing kinetics. There have been few studies on the effect of filler on curing kinetics. Dutta et al. [9] carried out a study into the kinetic effects caused by the presence of filler during the cure of an epoxy-amine system. Carbon and silica were individually evaluated up to 10wt%. In both cases the activation energy was unaffected by the increase in filler content, however, the carbon was found to effect the reaction rate K and cause an increase in AH, while the silica had no effect. The carbon filler was believed to have a catalytic effect on the curing process, while the silica was relatively inert in behaviour. This work highlights the importance of investigating the effects of filler on the curing kinetics, properties and shows that they can have effect. While the dynamic DSC scans and kinetic parameters for both systems show little variance due to the addition of filler, the exact reason for this is unclear. Fillers that have known catalytic effects are attributed to the presence of surface chemical complexes such as phenolics, carboxylics, quinines, hydroquinones and lactones [10]. The catalytic effects of these groups on epoxy curing have been established. The alumina filler used in both systems appears to have no significant effect on the curing process and kinetics. This can be attributed to either a lower specific surface area or relatively complex free surface; however, these postulations have to be further investigated.

Novel Processing of Ceramics and Composites

• 175

Study into Epoxy Composites for High-Voltage Device Encapsulation

Figure 3. DSC isothermal scan of epoxy composite prepreg

Figure 4a. DSC dynamic scan at heating rate of 2°C/minute

Figure 4b. DSC dynamic scan at heating rate of 5°C/minute

Figure 4c. DSC dynamic scan at heating rate of 20°C/minute

176

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

Table I. DSC plot parameters for various heating rates (system 1 unfilled/filled). Heating rate (K/min) System 1 unfilled 2 5 10 20 System 1 filled 2 5 10 20

T,rc)

TP(°C)

T F (°C)

Heat of reaction (J/g)

82.18 97.26 109.99 131.47

134.62 156.34 178.46 196.58

213.23 238.47 254.86 291.69

385 373 362 342

88.94 101.91 120.10 132.71

133.24 156.67 175.80 196.46

218.18 237.18 241.20 278.36

107.2 100.2 102.4 95.6

Sample

Activation energy (J/mol) System 1 unfilled 52 222 System 1 filled System 2 unfilled System 2 Filled

50 560

Table II. Activation energy for both systems with and without filler

54 672 56 241

3.2 Paniculate Filler Sedimentation With high filler loading it is essential to have a homogenous and uniform distribution of particulate filler in the matrix. An uneven distribution caused by sedimentation would result in varied mechanical, thermo-mechanical and dielectric properties across the cured resin. These property variations would lead to localised stress variations and possibly cracking/delamination upon cure or in service. Two types of casting mould can be used, the first being a top pour type and the other a side pour. As shown in Figure 4, a number of samples from various positions (top, middle and bottom) of cast were analysed using TGA and the remnant filler content determined. Table III shows the results and the variations from the desired loading level. The results are encouraging and show that for both systems sedimentation is not occurring irrespective of pour type. A difference of ±5 wt% would be of concern and warrant an investigation into the effects on final properties; however, the variation seen is less than 3 wt%. These results indicate that for both systems, pour

Novel Processing of Ceramics and Composites

•

177

Study into Epoxy Composites for High-Voltage Device Encapsulation

type does not have a significant effect on filler sedimentation and that uniform filler distribution occurs.

Figure 4. Top pour & side pour casts showing positions where samples were taken from (numbered 2,6 and 8) Table III. TGA results showing average deviation from desired loading Top pour mould Side pour mould System 1

2.6 wt%

1.8 wt%

System 2

1.7 wt%

1.4wt%

3.3 Mechanial Properties 3.3.1 Tensile Table IV presents the results of the tensile testing, outlining the Young's modulus and failure stress (conventionally termed as strength). What these results show is that the addition of filler greatly increases the Young's modulus and failure stress, producing a material with enhanced tensile mechanical properties. The effects of annealing/post-curing show a decrease in the tensile failure strength, while modulus remains unchanged. This is due to an increase in free volume as opposed to increased cross-link density. This was also observed by Gupta el al [8], when studying an epoxy system. In tension, most particulate filled cross-linked resins fail in a brittle manner before reaching the yield point. Brittle fracture is usually triggered by the appearance and/or activation of a critical defect. Understanding the mechanisms of failure with respect to process variation is important. The resin matrix bears the load with small strain deformation. Increasing strain gives rise to localised damage at the site of the largest stress concentration. There are three basic damage mechanisms [6]; i. Particle-matrix debonding ii. Void formation Hi. Plastic deformation

178

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

These mechanisms and the stress at which they occur are influenced by particle size and distribution, interfacial adhesion and matrix deformability. The use of small, well-dispersed and weakly adhered particles initiates a large number of small and stable voids, which permit further deformation of the matrix ligaments between particles. Voids created from large particles in a brittle matrix tend to give rise to a critical crack and instantaneous brittle fracture. This is why it is essential to control the particle size distribution, especially at the top end. Similar problems are found when using fillers in large amounts due to formation of matrix-free interfacial regions. Larger strains are required for interfacial void formation when good adhesion/bonding exists between particle and matrix. Factors such as particle size and ligament properties come into play, where large particle are susceptible to defects giving rise to brittle fracture at a higher stress than if poor particle-matrix bonding existed. Over the years work by various authors has looked at parameters that effect the mechanical properties of cast epoxy resins. Grillet et al. [11] found that the correct choice of hardener was paramount and effected the final mechanical properties. Pálmese et al. [12] found that epoxy-amino stoichiometery and cross-link density had an effect, while Ellis et al. [13] showed that cure schedule and test temperature effected the final properties of the cured system. The tensile strength and modulus has been reported in the range of 40-130 MPa and 2.0-4.1 GPa [14]. Our results fall within these values. Table IV. Tensile test results Sample System 1 filled 55°C @ 15hrs / 80°C @ 24hrs

Stress at Strain at Average Young's Modulus rail (MPa) failure (%) (MPa) 57.29 13 110 0.4423 (±3.86) (±0.0636)

System 2 filled 55°C@15hrs/80°C@24hrs

53.13 (±9.13)

0.4953 (±0.1023)

13 290

System 2 unfilled 55°C @ 15hrs / 80°C @ 24hrs

32.70 (±0.23)

1.10 (±0.09)

3 755

System 2 filled (anneal 1 ) 55°C @ 15hrs / 80°C @ 24hrs 40°C @ 6 hrs

46.25 (±11.52)

0.3825 (±0.0898)

13 960

3.3.2

Dynamic Mechanical Properties Tan 5 represents the overall mechanical properties of a material i.e. relative amounts of energy stored & lost. Often referred to as loss factor (ratio of damping to elasticity), it is an indicator of visco-elastic behaviour and Tg Figure 5a-b shows the Tan 8 as a function of temperature plots for system 1, where the Tg is approximately 120°C and is independent of pour type, sample position and dwell/start temperature. The close fit of the plots indicates good homogeneity in the cast. These results are very encouraging. The re-run samples show an increase in Tg, as would be expected. On the initial run after crossing the Tg, two processes occur that lead to an increase in Tg. The first is the loss of water,

Novel Processing of Ceramics and Composites

• 179

Study into Epoxy Composites for High-Voltage Device Encapsulation

as water can be thought of acting as a 'plasticizer'. The second is reaction curing, where the curing reaction of previously un-reacted groups occur in the rubbery state. With system 2 (Figure 5c-d), the close fit of the plots shows good homogeneity and dwell temperature does not effect the Tg of the material. Kaeble el al. [15] have extensively studied the dynamic mechanical response/properties of numerous epoxy resin systems. The peaks seen in Tan S plots are known a-transitions and are synonymous with the Tg (peak seen at or slightly above Tg). It is associated with co-operative rotational motion in the cured resin of the segments between cross-links normally involving 2050 atoms along the main chain. The Tg is highly dependant up on the rigidity of the molecule, where higher temperatures are required for the molecular motions associated with he softening/deformation at Tg for rigid/highly constrained molecular network. Hale et al. [16] showed that there was a one-one relationship between Tg and degree of cure and resin/hardener stoichiometry. Stoichiometric variations of the resin/hardener tend to cause unreacted epoxy or amino groups resulting in lowering of the Tg drastically. Therefore, Tg is a useful QA parameter.

Figure 5a-d. Tan 5 as a function of temperature plots investigating pour type effects on Tg(2, 6 & 8 indicates position where sample taken from cast) Figure 6a-d show Tan 8 as a function of temperature DMA plots investigating the effects of filler addition on System 2. It can be seen that the addition of filler has not effect on the Tg. For the unfilled system there are a number of transition like peaks which are due to noise from

180

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

deformation (sagging) of the samples close to the Tg (an equipment phenomena). Similar results were seen for system 1.

Figure 6a-d. DMA results investigating the effects of filler addition on Tg (System 2). 3.3.3 Thermal Cycling An extremely harsh thermal schedule was chosen to test the encapsulant material. Table V outlines the results, which are encouraging. System 1 filled had no failures due to cracking/delamination (6/6 pass), while system 2 filled had one failure due to delamination (5/6 pass) and system 2 unfilled had complete failure due to cracking/delamination (0/6 pass). These results show that the addition of filler does improve mechanical properties & increases resistance to crack propagation. This also indicates good bonding/adhesion between particulate & matrix in filled systems. Figure 6 shows cracking in an unfilled sample.

Novel Processing of Ceramics and Composites

• 181

Study into Epoxy Composites for High-Voltage Device Encapsulation

Figure 6. Radiograph showing cracks in an unfilled sample (system 2) after thermal cycling 4 Dielectric Breakdown It is now generally accepted that addition of filler to a polymeric material does not enhance the dielectric strength, rather mechanical properties [17]. However, work using small amounts (up to 10wt%) nano-sized particles have been shown to improve dielectric strength compared to micron sized particles, due to a very high specific surface area [18]. The results of the dielectric breakdown study carried out at the University of Leicester are given in Figure 7. With the pulsed voltage, it was found that the filled samples broke down at about 80kV, while the unfilled samples experienced flashover above 80kV. This is believed to be due to the experimental set-up, where a large electrode base plate was used. With DC testing, all the samples broke down (i.e. no flashover) and this was due to use of a smaller electrode base plate. What the results show is that the unfilled samples break down at a considerably higher voltage compared to the filled samples. These results indicate that addition of filler lowers the breakdown strength. SEM and EDX analysis of the puncture and exit hole for all filled samples was not useful in determining a possible breakdown mechanism. There was no variation in chemical composition in the puncture/exit region compared to the bulk and non-tested samples. Figure 8a shows a micrograph of the puncture hole and Figure 8b shows the exit hole for system 2 filled.

Figure 7. Bar chart showing mean pulse and DC breakdown strength (F=flashover, rest are breakdown)

182

• Novel Processing of Ceramics and Composites

Study into Epoxy Composites for High-Voltage Device Encapsulation

Figure 8 a-b. SEM micrographs of puncture & exit hole in system 2 filled (pulsed) 4. CONCLUSIONS This work has shown that the addition of a large amount (>50wt %) alumina filler to two epoxy based matrices significantly improves the mechanical and thermal shock properties, whilst uneffecting the Tg and curing kinetics/reaction. The use of different hardeners (from same aromatic amine family) was found to have little difference on the overall properties. Filler distribution was uniform and homogenous in two mould cast types (top/side pour) and eliminated the possibility of varied properties due to even filler distribution. The dielectric results showed that the addition of filler lowered the dielectric breakdown strength. A simple processing route has been put forth to produce an epoxy-alumina composite with enhanced mechanical and thermo-mechanical properties with a wide operating temperature range. The dielectric properties are good but it is felt that they can be further improved.

Novel Processing of Ceramics and Composites

• 183

Study into Epoxy Composites for High-Voltage Device Encapsulation

5. REFERENCES 1 J. Lange, N. Altmann, CT. Kelly & P. J. Halley, Polymer 41 p.5449 (2000) 2 J. D. Ferry, "Viscoelastic properties of polymers" Wiley (1980) 3 E. Kuffel, W.S. Zaengl & J. Kuffel, "High Voltage Engineering: Fundamentals" , 2nd edition Elsevier p.65 (2001) 4 K. Horie, H. Hinra, M. Sawada, I. Milda & H. Kambe, J. Poly. Sei. (part Al) 8 p.1357 (1970) 5 Polymer Matrix Composites, edited by R. Talreja & J. A. E. Manson, Pergamon (2003) 6 D. Rosu, A. Mititelu & C. N. Cascaval, Polymer Testing 23 p.209 (2004) 7 V. L. Zvetkov Polymer 42 p.6687 (2001) 8 A, Gupta, R. Singhal & A. K. Nagapal, J. Appl. Poly. Sel Vol 92, p.687 (2004) 9 A. Dutta & M. E. Ryan, J. Appl. Poly. Sei. Vol 24, p.635 (1979) 10 J. B. Enns & J. K. Gillham, J. Appl. Poly. Sei. Vol 28 p.2567 (1983) 11 A. C. Grillet, J. Galy, J. F. Geread & J. P. Pascalt, Polymer 32 p. 1885 (1991) 12 G. R. Pálmese, O. A. Anderson & V. M. Karbhari, Composites part A 304 p.l 1 (1999) 13 B. Ellis, M.S. Found & J. R. Bell, J. Appl. Poly. Sei. 9 p. 1493 (1996) 14 V. B. Gupta, L. T. Drazl & R. Omlor, J. Mat. Sei 20 p.3429 ( 1985) 15 D. H. Kaeble, J. Moacannin & A. Gupta in "Epoxy resins chemistry & technology" edited by C. A. May & Y. Tanaka (1988) 16 A. Hale, C. W. Macosko & H. E. Blair, Macromolecules 24(9) p.2610 (1991) l7 J. K. Nelson & J. C. Fothergill; Nanotechnology 15 p.1-10 (2004) 18 H. Z. Ding & B. R. Varlow; IEEE 2004 Annual conference report on electrical insulation & dielectric phenomena © British Crown Copyright 2005/MoD

184

• Novel Processing of Ceramics and Composites

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Geopolymers

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

ADVANCES IN UNDERSTANDING THE SYNTHESIS MECHANISMS OF NEW GEOPOLYMERIC MATERIALS Kenneth J.D. MacKenzie, Dan Brew, Ross Fletcher, Catherine Nicholson and Raymond Vagana The MacDiarmid Institute for Advanced Materials and Nanotechnology Victoria University of Wellington and Industrial Research Ltd., P.O. Box 600, Wellington New Zealand Martin Schmücker Institute of Materials, German Aerospace Centre D-5000 Cologne, Germany ABSTRACT Conventionally, geopolymeric materials are prepared from aluminosilicate clay minerals and sodium silicate, and were thought to exist in a restricted range of Si/Al compositions. Recent work presented here indicates that neither of these conditions is strictly true. Viable silica-rich geopolymers have been prepared from metakaolinite and added silica with Si/Al ratios up to 300, with the higher-silica members showing unexpected plastic deformation properties. Materials with the physical and structural characteristics of aluminosilicate geopolymers have also been produced using a range of alternative sources of Al and Si of mineral and non-mineral origin. The incorporation of tetrahedral borate and phosphate units into the framework structure has also been accomplished, as evidenced by XRD, multinuclear MAS NMR and electron microscopy. These results shed fresh light on fundamental questions concerning the mechanism of the geopolymerisation reaction and have implications for the definition of geopolymers themselves. INTRODUCTION Conventional understanding of geopolymers based on the original patent literature suggests that they are typically aluminosilicates formed by the condensation at ambient temperature of silica and alumina units (typically from dehydroxylated 1:1 layer lattice silicates such as metakaolinite with additional sodium silicate) under conditions of high pH and controlled water content. The most common tetrahedral polymerising units have been described' as sialate (silicomaluminium = 1), silalate siloxo (silicon:aluminium = 2) and sialate disiloxo (silicon:aluminium = 3); other possible silicomaluminium ratios are not generally considered. Charge balance in the polymer units is normally achieved by the presence of monovalent alkali cations. The unique and characteristic physical and chemical attributes of the resulting products are their achievement of strength at ambient temperatures, their X-ray amorphous characteristics, their solely tetrahedral Al and Si coordination state as revealed by solid state MAS NMR2 and their thermal stability up to high temperatures (>1000°C), with the retention of their Al and Si coordination and their essentially X-ray amorphous character at these high temperatures3. These considerations have led to the proposal of a "structure" consisting of tetrahedral Al and Si units randomly distributed in a 3D structure with the Si in predominantly SiQ4(3Al) sites and the charge-balancing hydrated cations located in the spaced within this random network4. These

187

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

conclusions are consistent with SEM/EDX and TEM/EDX studies of both unheated and heated polysialate geopolymers5. However, this self-consistent picture of the formation and character of geopolymers leaves a number of unanswered questions, including: * How much wider can the composition range be in aluminosilicate geopolymers without loss of the defining geopolymer characteristics? * Could other aluminosilicate clays be used as starting materials? * Do the clays have to be thermally activated, or are other activation methods applicable? * Are clay mineral reactants essential or can other alumina and silica sources be used? * Can we substitute other tetrahedral elements into the aluminosilicate geopolymer structure? This work addresses these questions and in doing so sheds new light on the synthesis mechanisms operating in these systems. EXPERIMENTAL Geopolymer compositions were prepared from a wide range of starting materials, dictated by the questions posed above. To test the composition limits of aluminosilicate geopolymerisation, a series of samples were prepared using metakaolinite (dehydroxylated at 800°C) with the composition adjusted to produce silica:alumina (S:A) ratios from 1 to 300 by the addition of fine silica (Aerosil), and for S:A compositions <1 by the addition of amorphous palumina. The possibility of viable geopolymerisation reactions occurring in 2:1 layer lattice alummosilicates was investigated in a series of experiments using natural pyrophyllite, both dehydroxylated at 800°C, undehydroxylated and ground in a planetary ball mill for 20-60 hours. Other experiments to determine the effect of partial dehydroxylation on the geopolymerisation reactions were carried out with the 1:1 layer lattice aluminosilicate halloysite dehydroxylated to various degrees by heating for 2 hours at temperatures ranging from 200 to 1000°C, the degree of dehydroxylation being monitored in each case by weight loss measurements. To examine the possible use of soluble forms of alumina and silica in place of solid alummosilicates, a number of mixtures were made of sodium alumínate and sodium silicate formed in situ by dissolving silica fume in sodium hydroxide solution. The possibility of incorporating tetrahedral phosphate or borate units into aluminosilicate geopolymers was investigated by making a number of metakaolinite compositions containing aluminium phosphate or sodium tetraborate. In all these experiments, the compositions of the mixtures were designed on the basis of the calculated molar ratios of the components. After mixing, the samples were placed in polythene moulds, sealed with plastic film and cured at 40-65°C. The plastic film was then removed and the samples dried at the curing temperature. The attainment of geopolymerisation was judged on the basis of several previously-established criteria2 (attainment of strength at ambient temperature, X-ray amorphous structure, tetrahedral coordination of the network-forming elements deduced by 27A1, 29Si, 23Na, "B and 31P solid-state MAS NMR). Details of the MAS NMR spectroscopy of these nuclides can be found elsewhere6. RESULTS AND DISCUSSION Composition range of viable aluminosilicate geopolymers. Compositions with S:A <1 set and harden, but contain mixtures of crystalline phases including gibbsite (Al(OH)3, a zeolite-type sodium aluminosilicate hydrate (PDF no. 31-1271)

188

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

and Na2CÛ3.H20, the latter probably formed by atmospheric carbonation of excess NaOH7. This formation of crystalline phases appears to result from the violation of Lowenstein's Rule. Silica-rich compositions with S:A >1 cure to viable geopolymers up to very highly siliceous compositions (S:A = 300). These materials all show the characteristic amorphous X-ray diffraction trace (Figure 1), and up to S:A values of about 100, show a single exclusively tetrahedral Al peak in the 7A1 MAS NMR spectra (Figure 2A). At S:A ratios between 100 and 300, a broad octahedral Al resonance begins to reappear, possibly reflecting the additional water required to formulate these highly siliceous compositions which might also encourage the formation of predominantly octahedral hydrated Al species7. This additional water content is also implicated in the physical characteristics of the geopolymers with S:A>24, which exhibit increasingly plastic properties, and on heating at = 120°C, evolve the additional water to form a stable foam (Figure 3).

Figure 1. X-ray diffractograms of a selection of higher-silica metakaolinite-based geopolymers showing the typical amorphous characteristics over the entire composition range.

Novel Processing of Ceramics and Composites

• 189

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

" A I shift (ppm)

»Si shift (ppm)

Figure 2. A selection of typical 11.7T MAS NMR spectra of high-silica geopolymers derived from metakaoiinite. A. 27A1 spectra, B. 29Si spectra.

Figure 3. Geopolymer with S/A = 300 heated to various stages of foaming. From reference 7. The 29Si MAS NMR spectra of these samples (Figure 2B) all display the typical broad geopolymer tetrahedral resonance centred at about -93 ppm, indicating that, apart from their

190

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

mechanical plasticity, all these materials display the typical physico-chemical characteristics of true geopolymers. Geopolymers from 2:1 layer lattice aluminosilicates (pyrophyllite) All attempts to produce viable geopolymers according to the above criteria from this mineral were unsuccessful; although many of the compositions set to reasonable strengths, none were Xray amorphous and all contained a variety of crystalline phases. This behavior was possibly due to the protection of the Al-containing layer by the surrounding silica sheets which prevent access of the alkaline solution. Attempts to thermally activate this mineral by heating in stages of 200°C up to 800°C also failed to promote the formation of an X-ray amorphous product, possibly due to the fact that on heating, pyrophyllite develops a much more crystalline dehydroxylate phase than its 1:1 counterpart (metakaolinite). Unsuccessful attempts were made to overcome this apparent resistance to alkaline attack by mechanochemical activation (high-energy grinding) to break down the crystalline structure and render the mineral more reactive. Compensation for the lack of reactive silica species by the use of additional sodium silicate solution was also unsuccessful. The effect of dehydroxylation of geopolymer formation in halloysite. Halloysite is a 1:1 aluminosilicate of similar chemical composition to kaolinite but with characteristic tubular rather than platy morphology. Geopolymers produced from undehydroxylated halloysite set to a relatively weak and crumbly material which, although effectively X-ray amorphous (Figure 4B) and displays a typical geopolymer 29Si MAS NMR spectrum (Figure 5), retains a significant proportion of its original octahedral Al (Figure 6), and has therefore not undergone complete geopolymerisation. Progressive dehydroxylation at 200, 400 600, 800 and 1000°C results in similar behaviour up to 600°C, by which temperature the bulk of the water has been lost and conversion of the solely octahedral Al of the parent mineral to a mixture of 6, 5 and 4-fold coordination has occurred. The cured product from halloysite activated at this temperature was hard, X-ray amorphous, and showed the NMR characteristics of a conventional geopolymer. The effect on the physical properties of geopolymers prepared from halloysite dehydroxylated at 600°C of increasing the soluble silicate content was investigated by adjusting the formulations so as to increase the content of sodium silicate while maintaining the molar ratios at conventional values (Si02:Al203 = 3.4, Na20:Si02 = 0.4, H20:Na20 = 11.4, Na20:Al2Û3 = 1.4). All the samples set, but at higher sodium silicate contents (>16g sodium silicate to 15g dehydroxylated halloysite), the strengths deteriorated and the products became increasingly friable, even though the X-ray and NMR results indicated that geopolymerisation had occurred. A parallel series of experiments in which the NaOH content was progressively increased while maintaining the molar ratios at the above values indicated that the tendency to flash-setting increased with increasing NaOH contents (>10g NaOH to 15h dehydroxylated halloysite). Thus, although any desired series of molar ratios of reagents can be achieved in claybased systems by varying the relative proportions of the components, in complex systems such as these there is considerable scope for any particular target formulation to be optimized in terms of the sodium silicate and sodium hydroxide contents to produce a smooth, strong product with a useful setting time.

Novel Processing of Ceramics and Composites

• 191

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

A. Undehydroxylated halloysite

Figure 4. XRD diffractograms of unheated halloysite and its geopolymerised product. A. Undehydroxylated halloysite

"AI shift (ppm) B. Geopolymer product

--3;

¿

■ ¿ ■ ¿

j

"AI shift (ppm)

3

Figure 5. 11.7T 27A1 MAS NMR spectra of unheated halloysite and its geopolymer Other possible methods for activating 1:1 clays for geopolymerisation. On the basis that high-energy grinding of a 1:1 aluminosilicate clay mineral can produce a mixture of 4, 5 and 6-fold coordinated Al sites similar to the effect of thermal dehydroxylation, mechanochemical activation of undehydroxylated halloysite was carried out in a planetary ball mill for 20 hr. This treatment failed to produce a viable geopolymer, suggesting that such

192

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

mechanochemical treatment does not sufficiently disrupt the clay mineral structure to render the aluminium labile and promote geopolymerisation. Chemical methods for increasing the lability of the alumina and silica components of the clay mineral were also investigated. These included treatment of undehydroxylated halloysite with 0.1 M NaOH for 1 to 24 hours, which resulted in an increasing proportion of the octahedral aluminium becoming tetrahedral. This coordination change becomes noticeable after about 3 hours alkali treatment, and after 24 hours of soaking under these conditions, all the aluminium is in tetrahedral coordination, but with no sign of 5-fold Al coordination. These results suggest the formation of an incipient zeolite-type structure in which the Al may prove to be sufficiently labile to undergo geopolymerisation, provided the activation reaction has not been subjected to a sufficient temperature and sufficiently long time to achieve significant crystallinity. By contrast, treatment of the undehydroxylated clay with 0.1M HC1 for periods of time varying from 1 to 24 hr produced no sign of change in the aluminium environment, as judged by the 27A1 MAS NMR spectra of the treated clay, in which the octahedral coordination of the starting material was fully retained. Geopolymers from soluble alumina and silica sources. A variety of approaches were made to the production of geopolymeric materials by this route, and only a few of the more promising results will be presented here. Viable materials with all the characteristics of geopolymers can be made by mixing alkaline sodium alumínate and sodium silicate solutions (Fig. 6,7). This material had a reproducible crushing strength, measured in triplicate on 50mm cubes, of 26.2 MPa. The reactions are sensitive to the order in which the ingredients are mixed; if the order of mixing the reagents encourages the formation of aluminosilicate gels, these tend to be resistant to further reaction to form geopolymers and their formation effectively removes labile alumínate and silicate from the system.

°20 CoKo

Figure 6. XRD of geopolymer from sodium alumínate, fumed silica and NaOH, S/A = 8.3, cured at 40°C.

Novel Processing of Ceramics and Composites

• 193

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

A

B

27

Al shift (ppm)

"SI shift (ppm)

Figure 7. 11.7T MAS NMR spectra of geopolymer from sodium alumínate, fumed silica and NaOH, S/A = 8.3, cured at 40°C for 48 hr. From this it follows that the formation of geopolymers from pre-formed aluminosilicate gels is difficult; even though their Al speciation as judged by 27A1 MAS NMR is not dissimilar to that of dehydrated clay, and contains a mixture of 6, 5 and 4-fold coordinated species, their insolubility renders them impervious to further network-forming reactions. Further, the reactions between alumínate and silicate solutions in alkaline conditions can be extremely exothermic, leading to runaway conditions unless the reaction mixture is cooled. Nevertheless, this approach to geopolymerisation has considerable potential for the facile incorporation of other ions including transition metals and radioactive species into the geopolymer matrix. Aluminosilicate geopolymers containing other tetrahedral groups. Viable aluminosilicate geopolymers containing tetrahedral phosphate or borate groups have been produced by reaction of aluminium phosphate or sodium tetraborate with standard metakaolinite-based formulations. A number of compositions in which a varying proportion of the silica was replaced by phosphate were found to produce materials which hardened at 65°C, were X-ray amorphous and showed tetrahedral 27A1 and 29Si MAS NMR spectra characteristic of a geopolymer (Figs.8, 9). The 3 'P MAS NMR spectra of the starting materials indicate the presence of the phosphorus in essentially one tetrahedral site, whereas after geopolymerisation reaction, the spectrum showed that the major tetrahedral phosphorus had shifted, indicating a significant change in the chemical environment of the phosphorus (Figure 10). In addition, two new, much less-populated tetrahedral PO4 sites can be distinguished in the cured material. Attempts to produce totally silica-free phosphate geopolymers by this route have so far proved unsuccessful, as has the use of various sodium phosphates which tend to form separate crystalline hydrated phosphate phases.

194

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

" S I shift (ppm)

Figure 8. A. XRD diffractogram and B. 29Si MAS NMR spectrum of phosphate geopolymer from AIPO4, metakaolinite, sodium silicate and NaOH, cured at 65°C. S/A = 3.17, P/S = 0.02, N/S = 0.19. A . Reactant

" A I shift ( p p m ) B. G a o p o l y m e r

" A I shift ( p p m )

Figure 9. 11.7T 27A1 MAS NMR spectra of: A. the AIPO4 reactant and B. the resulting phosphorus geopolymer described in Figure 8.

Novel Processing of Ceramics and Composites

• 195

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

Figure 10. 11.7T "P MAS NMR spectra of: A. the AIPO4 reactant and B. the resulting phosphorus geopolymer described in Figure 8.

Figure 11. SEM and elemental maps of the phosphorus geopolymer described in Figure 8. Although the microstructure of the phosphate geopolymer as seen by FEG-SEM (Figure 11) suggests the presence of discrete grains, possibly relicts of the reactant metakaolinite and AIPO4, elemental maps indicate a much more homogeneous distribution of all the constitutent elements throughout the sample.

196

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

Replacement of a proportion of the silicate groups of the standard metakaolinite-based formulations with tetrahedral borate groups has proved possible by the use of sodium tetraborate. In these cases, the geopolymer cures at 65°C to a hard material which is essentially X-ray amorphous and shows the normal 27A1 and 29Si MAS NMR spectra (Figure 12).

Figure 12. A. X-ray diffractogram, B. 27A1 MAS NMR spectrum and C. 1 l.&T 29Si MAS NMR spectrum of a geopolymer from sodium tetraborate, metakaolinite and NaOH, cured at 65°C, S/A= 2.02, B/S = 0.25, N/S = 0.56. The reactant tetraborate has an "B MAS NMR spectrum (Figure 13) showing the presence of a single resonance corresponding to tetrahedral B0 4 groups and a quadrupolar lineshape typical of trigonal BO3 groups. After geopolymerisation, all the boron has assumed tetrahedral sites at the expense of the trigonal BO3 groups, consistent with the incorporation of the boron into the geopolymer structure. As with the phosphate-containing geopoiymers, attempts to produce silicate-free borate geopoiymers have so far proved unsuccessful despite the use of a number of alternative starting materials including amorphous boron alumínate gels of differing starting compositions.

Novel Processing of Ceramics and Composites

• 197

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

"B shift (ppm)

B. Geopolymer

,1

B shift (ppm)

Figure 13. 11.7T "B MAS NMR spectra of: A. the reactant and B. the resulting borongeopolymer described in Figure 12. Implications of these results for the geopolymerisation mechanism. The observation that geopolymerisation does not proceed readily in unactivated 1:1 aluminosilicate clays such as undehydroxylated halloysite, or in 2:1 layer lattice clays where the alumínate layer is protected by the enveloping silicate sheets, or where insoluble aluminosilicate gels are formed suggests that the lability and solubility of the Al source plays a more important role in geopolymerisation than has previously been recognized. Where the aluminium source is sufficiently accessible, the alumínate reactions proceed more rapidly than those of the silicate, and are thus more difficult to monitor. The chemical form and concentration of the silica source in geopolymerisation also exerts an effect on the physical properties of the product, since at least a proportion of the silicate must be soluble. This soluble component may be self-generated (by the action of the alkaline environment on the solid silica source present) or it may be supplied in the form of additional sodium silicate, or both. In the former case, the solubility of the silica source in the alkaline hydroxide is influenced by the physical state of the solid, suggesting that finely divided materials are preferable. However, the more finely divided the silica source, the greater the water demand, which can, however be turned to advantage as in the generation of self-foaming highly siliceous products. Although the presence of additional sodium silicate is unnecessary for geopolymerisation to proceed, up to certain concentration limits it assists the strength and smoothness of the product, but if this optimum concentration is exceeded, the strength of the product is degraded. Irrespective of whether it is provided in solid or dissolved form, sufficient silica must be present in the formulation to satisfy Lowenstein's rule, ruling out the possibility of homogeneous geopolymers with S:A ratios <1. Such materials, if formed are likely to be geopolymer composites, with the additional alumina present as a filler, or as assemblages of several crystalline phases.

198

• Novel Processing of Ceramics and Composites

Advances in Understanding the Synthesis Mechanisms of New Geopolymeric Materials

CONCLUSIONS S:A compositions <1 give mixtures of crystalline aluminosilicates, probably reflecting the violation of Lowenstein's Rule. Geopolymers with S:A >1 are viable up to very highly siliceous compositions (S/A = 300), but above S/A = 24 the product is plastic and evolves water on heating at « 120°C to form a stable foam. The 2:1 layer lattice aluminosilicate mineral pyrophyllite does not produce a viable geopolymer, possibly due to the protection of the Al-containing layer by the silica sheets and the development of a crystalline dehydroxylated phase on thermal activation Undehydroxylated halloysite sets to a weak material which retains a proportion of its original octahedral Al. Thermal dehydroxylation at = 500°C is necessary to produce a viable geopolymer. Mechanochemical activation of the halloysite is not sufficiently disruptive of the clay mineral structure to promote geopolymerisation. Viable materials with all the characteristics of geopolymers can be made by mixing alkaline sodium alumínate and sodium silicate solutions but the reactions are sensitive to the order in which the ingredients are mixed. Viable aluminosilicate geopolymers containing tetrahedral phosphate or borate groups can be produced. These observations suggest that the lability and solubility of the Al source is equally important as that of the Si source, but the alumínate reactions proceed faster than those of the silicate and are thus more difficult to monitor. ACKNOWLEDGEMENTS This work was supported by funding from the MacDiarmid Institute for Advanced Materials and Nanotechnology and by funding from the New Zealand Foundation for Research, Science and Technology under contract CO8X0302. REFERENCES ' j . Davidovits, "Geopolymers: Inorganic Polymeric New Materials", J. Thermal Anal, 37,1633-56(1991). 2 K.J.D. MacKenzie, "What are These Things Called Geopolymers? A Physico-Chemical Perspective", Ceramic Trans., 153, 175-86 (2003). 3 V.F.F. Barbosa and K.J.D. MacKenzie, "Thermal Behaviour of Inorganic Geopolymers and Composites Derived from Sodium Polysialate". Mater. Res. Bull, 38, 319-31, (2003). 4 V.F.F. Barbosa, K.J.D. MacKenzie, and C. Thaumaturgo, "Synthesis and Characterisation of Materials Based on Inorganic Polymers of Alumina and Silica: Sodium Polysialate Polymers". Int. J. Inorg. Mater., 2, 309-17 (2000). 5 M.Schmiicker and K.J.D. MacKenzie, "Microstructure of Sodium Polysialate Geopolymer". Ceram. Internat., 31,433-37 (2005). 6 K.J.D. MacKenzie and M.E. Smith, "Multinuclear Solid-State NMR of Inorganic Materials", Pergamon Materials Series Volume 6, Pergamon/Elsevier, Oxford, (2002). 7 R.A. Fletcher, K.J.D. MacKenzie, C.L. Nicholson and S. Shimada, "The Composition Range of Aluminosilicate Geopolymers"./. Eur. Ceram Soc, 25, 1471-77 (2005).

Novel Processing of Ceramics and Composites

• 199

Novel Processing of Ceramics and Composites Edited by Narottam P. Bansal, J. P. Singh, James E. Smay and Tatsuki Ohji Copyright © 2006 The American Ceramics Society

Author Index

Ando, K., 125 Babonneau, F., 125 Bansal, N. P., 23 Bonhomme, C , 125 Brew, D., 187 Chiu, Y.-H., 45 Chu, Y., 97 DeCarlo, F., 97 Engler, M., 159 Fletcher, R., 187 Fothergill, J. C , 169 Fujii, E., 125 Fujii, T., 67, 87 Fujisawa, Y., 115 Fukuyama, H., 141, 149 Goto, T., 3, 13 Greim,J., 159 Hanan, J. C , 97 Hao, W., 33 Hashimoto, H., 87 Hashizume, T., 77, 115 Hayakawa, S., 125, 133 Hirao, K., 57

Hiratsuka, D., 107 Ibukiyama, M., 107 Jadoon, A. K.,169 Kawabata, K., 125 Kimura, T., 3, 13 Kita, H., 57 Komeya, K., 57, 107 Kondo, N., 57 Lesniak, C , 159 Lin, Y.-J., 45 MacKenzie, K. J. D., 187 Meguro, T., 107 Muto, A., 67 Nagata, K., 149 Nakanishi, M., 67, 87 Nicholson, C , 187 Ohkubo, M., 125 Okamoto, K., 133 Osaka, A., 125, 133 Saiki,A.,77, 115 Sakata, Y., 67 Schmücker, M.

Novel Processing of Ceramics and Composites

• 201

Smay,J. E.,97 Takada, J., 67, 87 Takada, K., 141 Takada, Y., 87 Takeshi, M., 57 Tatami, J., 57, 107 Tanoue, T., 149 Teramoto, M., 141 Terayama, K., 77, 115 Tsuru, K., 125, 133 Uchida, Y., 67 Ui, S., 77

202

Uibel, K., 159 Uno, H., 77 Vagana, R., 187 Wakihara, T., 57, 107 Weimin, W., 33 Wilby, A., 169 Yoneda, T., 141 Zhengyi, F., 33 Zhong, Z., 23

• Novel Processing of Ceramics and Composites