This Page Intentionally Left Blank
progress in Nanotechnology
This Page Intentionally Left Blank
Progress in Nanotechnology Processing
A Progress in Ceramic Technology series publication
@WILEY A John Wiley & Sons, Inc., Publication
Copyright 02010 by The American Ceramic 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) 750-4470, 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., I 1 1 River Street, Hoboken, NJ 07030, (201) 748-601 1, fax (201) 748-6008, or online at http://www.wiley.com/go/permission. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations 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 or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (31 7) 572-3993 or fax (3 17) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic format. For information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-in-Publication Data:
Progress in nanotechnology : processing. p. cm. - (Progress in ceramic technology) Includes index. ISBN 978-0-470-40839- 1 (cloth) I . Nanostructured materials. 2. Nanotechnology. TA4 18.9.N35P765 2010 620'.56c22 2009035822 Printed in the United States of America. 10 9 8 7 6 5 4 3 2 1
Contents
xi
Introduction Synthesis Methods for Powders
Freeze Casting as a Nanoparticle Material-Forming Method K. Lu and X. Zhu Int. J. of Appl. Ceram. Techno/., Vol. 5, Is. 3, p. 219-227
Preparation of a Nanoscale/SOFC-GradeYttria-Stabilized Zirconia Material: A Quasi-Optimization of the Hydrothermal Coprecipitation Process
3 13
Y-C Chang, M-C Lee. W-X Kao, and T-N Lin Int. J. of Appl. Ceram. Techno/., Vol. 5, Is. 6, p. 557-567
Synthesis of NanosizeTin Dioxide by a Novel Liquid-Phase Process
25
Y. Zhou, N. Dasgupta, and A. Virkar J. Am. Ceram. SOC., Vol. 91, No. 3, p. 1009-1012,2008
Fabrication of Nanocomposite Powders of Carbon Nanotubes and Montmorillonite
29
J. Feng and Q. Wang J. Am. Ceram. SOC.,Vol. 91, No. 3, p. 975-978,2008
Synthesis of Highly Dispersed Barium Titanate Nanoparticles by a Novel Solvothermal Method X. Wei, G.Xu, Z. Ren, Y. Wang, G. Shen, and G.Han
33
J. Am. Ceram. SOC., Vol. 91, No. 1, p. 315-318,2008
Continuous Production and Harvesting of Inorganic-Ceramic Nanoparticles S.A.E. Abdulla, P.A. Sermon, M. Worsley, and I.R. Collins CESP, Vol. 28, No. 6, p. 131-141,2008
Nanocrystalline Scandia Powders Via Oxalate Precipitation: The Effects of Solvent and Solution pH Z. Xiu, J-G. Li, X. Li, D. Huo, X. Sun, T. Ikegami, and T. lshigaki J. Am. Ceram. SOC.,Vol. 91, No. 2, p. 603-606,2008
A Pulse Combustion-Spray Pyrolysis Process for the Preparation of Nano- and Submicrometer-Sized Oxide Particles
37
49 53
W. Widiyastuti, Wei-Ning Wang, Agus Purwanto, 1. Wuled Lenggoro, and Kikuo Okuyama J. Am. Ceram. SOC., Vol. 90, No. 12, p. 3779-3785,2007
One-Step Synthesis of Luminescent Nanoparticles of Complex Oxide, Strontium Aluminate C. Li, Y. Imai, Y. Adachi, H. Yamada, K. Nishikubo, and C-N Xu
61
J. Am. Ceram. SOC.,Vol. 90, No. 7, p. 2273-2275,2007
Nan0 a-Al,O,
Powder Preparation by Calcining an Emulsion Precursor
Y-C Lee, S-B Wen, L. Wenglin, and C-P Lin J. Am. Ceram. SOC., Vol. 90, No. 6, p. 1723-1727,2007 Contents
65
V
Lanthanum Strontium Manganite Powders Synthesized by Gel-Casting for Solid Oxide Fuel Cell Cathode Materials
71
L. Zhang, Y. Zhang, Y. Zhen, and S. Jiang J. Am. Ceram. SOC.,Vol. 90, No. 5, p. 1406-1411,2007
Preparation of Matrix-Type Nickel Oxide/Samarium-Doped Ceria Composite Particles by Spray Pyrolysis
77
S. Suda, K. Kawahara, M. Kawano, H. Yoshida, and T. lnagaki J. Am. Ceram. SOC.,Vol. 90, No. 4, p. 1094-1100, 2007
Novel Low-Temperature Synthesis of Ferroelectric Neodymium-Doped Bismuth Titanate Nanoparticles
85
P. Prakash, A. Garg, M. Roy, and H. Verma J. Am. Ceram. SOC., Vol. 90, No. 4, p. 1295-1298, 2007
Hydrothermal Synthesis of CdMoO, Nano-Particles
89
X. Jiang, J. Ma, B. Lin, Y. Ren, J. Liu, X. Zhu, J. Tao, Y. Wang, and L. Xie J. Am. Ceram. SOC.,Vol. 90, No. 3, p. 977-979, 2007
Chromium-Doped Forsterite Nanoparticle Synthesis by Flame Spray Pyrolysis
93
T. Tani, S. Saeki, T. Susuki, and Y. Ohishi J. Am. Ceram. SOC.,Vol. 90, No. 3, p. 805-808,2007
Formation of AI,O,-Tic
Composite Nano-Particles Synthesized from Carbon-Coated Precursors
97
H. Kaga and R. Koc J. Am. Ceram. SOC.,Vol. 90, No. 2, p. 407-411,2007
Synthesis of Sm,~,Sr,,CoO,~,
and La,~,Sr,,4Co0,~,
Nanopowders by Solution Combustion Process
103
N. Bansal and Z. Zhong CT; Vol. 195, p. 23-32,2006
Colloidal Processing and Sintering of Nano-ZrO, Powders Using Polyethylenimine
113
Y. Hotta, C. Duran, K. Sato, and K. Watari Cx Vol. 190, p. 85-93,2006
Synthesis of High Purity p-SiAION Nanopowder from a Zeolite by Gas-Reduction-Nitridation
123
T. Yamakawaa, T. Wakihara, J. Tatami, K. Komeya, and T. Meguro CT Vol. 190. P.3-8.2006
A Novel Supercritical CO, Synthesis of Amorphous Hydrous Zirconia Nanoparticles, and Their Calcination to Zirconia
129
M-H Lee, H-Y Lin, and J. L. Thomas J. Am. Ceram. SOC.,Vol. 89, No. 12, p. 3624-3630,2006
Praseodymium-Doped Photo-Luminescent Strontium lndate Nanoparticles by Ultrasonic Spray Pyrolysis
137
S. E. Lin, K. Borgohain, and W. C. J. Wei J. Am. Ceram. SOC., Vol. 89, No. 10, p. 3266-3269, 2006
Nano-Blast Synthesis of Nano-size Ce0,-Gd,O,
Powders
141
Oleg Vasylkiv, Yoshio Sakka and Valeriy V. Skorokhod J. Am. Ceram. SOC., Vol. 89, No. 6, p. 1822-1826,2006
Sol-Gel Processing and Characterization of Phase-Pure Lead Zirconate Titanate Nano-Powders
147
Yasir Faheem and M. Shoaib J. Am. Ceram. SOC.,Vol. 89, No. 6, p. 2034-2037, 2006
Synthesis of AIN Nanopowder from -y-Al,O,
by Reduction-Nitridation in a Mixture of NH,-C,H,
151
Tomohiro Yamakawa, Junichi Tatami, Toru Wakihara, Katsutoshi Komeya, Takeshi Meguro, Kenneth J. D. MacKenzie, Shinichi Takagi, and Masahiro Yokouchi J. Am. Ceram. SOC.,Vol. 89, No. 1, p. 171-1 75, 2006
Membranes, Films, and Coatings
Microporous ZrO, Membrane Preparation by Liquid-Injection MOCVD
159
S. Mathur, E. Hemmer, S. Barth, J. Altmayer, N. Donia, 1. Kumakiri, N. Lecerf, and R. Bredesen CESP, Vol. 28, No. 6, p. 165-173, 2008 vi
Progress in Nanotechnology:Processing
Growth of Barium Hexaferrite Nanoparticle Coatings by Laser-Assisted Spray Pyrolysis
169
G. Dedigamuwa, P. Mukherjee, H. Srikanth, and S. Witanachchi CESP, Vol. 28, No. 6, p. 73-81, 2008
Two Phase MonaziteKenotime 3OLaPO4-7OYPO, Coating of Ceramic Fiber Tows
179
E. Boakye, R. Hay, P. Mogilevsky, and M. Cinibulk J. Am. Ceram. Soc., Vol. 91, No. 1, p. 17-25,2008
Template-Free Self-Assembly of a Nanoporous TiO, Thin Film
189
Y. Gao, M. Nagai, W-S Seo, and K. Koumoto J. Am. Ceram. SOC.,Vol. 90, No. 3, p. 831-837,2007
Nano-Sized Hydroxyapatite Coatings on Ti Substrate with TiO, Buffer Layer by E-beam Deposition
197
S-H Lee, H-E Kim, and H-W Kim J. Am. Ceram. SOC.,Vol. 90, No. 1, p. 50-56,2007
Sol-Gel Routes to Nanostructured Patterned FerroelectricThin Films with Novel Electronic and Optical Functions
205
M. Kuwabara, Y. J. Wu, J. Li, and T. Koga Ceramic Transactions, Vol. 196, p. 371-380, 2006
Preparation and Properties of Hydrothermally Stable y-Alumina-Based Composite Mesoporous Membranes
215
Md. Hasan Zahir, Koji Sato, Hiroshi Mori, Yuji Iwarnoto, Mikihiro Nornura, and Shin-ichi Nakao J. Am. Ceram. Soc., Vol. 89, No. 9, p. 2874-2880,2006
Synthesis and Tribological Behavior of Silicon Oxycarbonitride Thin Films Derived from Poly(Urea)Methyl Vinyl Silazane
223
T. Cross, R. Raj, T. Cross, S. Prasad, and D. Tallant Int. J. of Appl. Ceram. Techno/., Vol. 3 No. 2, p. 113-1 26, 2006
Synthesis and Tribology of Carbide-Derived Carbon Films
237
A. Erdemir, A. Kovalchenko,C. White, R. Zhu, A. Lee, M. J. McNallan, B. Carroll and Y. Gogotsi Int. J. of Appl. Ceram. Techno/.,Vol. 3, No. 3, p. 236-244,2006
Nanotubes, Nanorods, and Nanowires
Design, Fabrication and Electronic Structure of Oriented Metal Oxide Nanorod-Arrays
249
L. Vayssieres CESP, Vol. 28, No. 6, p. 187-193,2008
Electrospinningof Alumina Nanofibers
257
K. Lindqvist, E. Carlstrorn, A. Nelvig, and B. Hagstrorn CESP, Vol. 28, No. 6, p. 41-51,2008
ZnO Nanofiber and Nanoparticle Synthesized Through Electrospinningand Their Photocatalytic Activity Under Visible Light
269
H. Liu, J. Yang, J. Liang, Y. Huang, and C. Tang J. Am. Ceram. SOC.,Vol. 91, No. 4, p. 1287-1291,2008
Synthesis of Carbon Nanotubes and Silicon Carbide Nanofibers as Composite Reinforcing Materials
275
H. Li, A. Kothari, and B. W. Sheldon CESP, Vol. 27, No. 8, p. 41-48,2007
Polymer Fiber Assisted Processing of Ceramic Oxide Nan0 and Submicron Fibers
283
S. Shukla, E. Brinley, H. J. Cho, and S. Seal, CESP, Vol. 27, NO. 8, p. 57-68,2007
Growth of Quasi-Aligned AIN Nanofibers by Nitriding Combustion Synthesis
295
M. Radwan and Y. Miyamoto J. Am. Ceram. SOC.,Vol. 90, No. 8, p. 2347-2351, .2007
Contents
vii
Synthesis and Optical Properties of Mullite Nanowires
301
H-K Seong, U. Kim, M-H Kim, H-J Choi, Y. Lee, and W-S Seo J. Am. Ceram. SOC.,Vol. 90, No. 6, p. 1937-1939, 2007
(Nao,,Ko,2)o,,Bi,,5Ti03 Nanowires: Low-Temperature Sol-Gel-Hydrothermal Synthesis and Densification Y-D Hou, L. Hou, T-T Zhang, M-K Zhu, H. Wang, and H. Yan
305
Synthesis and Characterization of Ce,Gd,O,_,
311
J. Am. Ceram. SOC.,Vol. 90, No. 6, p. 1738-1743,2007
Nanorods
J. S. Lee and S. Kim
J. Am. Ceram. SOC.,Vol. 90, No. 2, p. 661-663,2007
Synthesis and Characterization of Cubic Silicon Carbide (p-Sic) and Trigonal Silicon Nitride (a-Si,N,) Nanowires
315
K. Saulig-Wenger, M. Bechelany, D. Cornu, S. Bernard, F. Chassagneux, I? Miele, and T. Epiciers CESP, VOl. 27, NO. 8, p. 81-88,2007
Synthesis of Boron Nitride Nanotubes for Engineering Applications
323
J. Hurst, D. Hull, and D. Gorican CESP, Vol. 27, NO. 8, p. 95-102,2007
Novel Process of Submicron-Scale Ceramic Rod Array Formation on Metallic Substrate
331
K. Okamoto, S. Hayakawa, K. Tsuru, and A. Osaka CT; VOl. 195, p. 133-138,2006
Tin Oxide Nanoparticle-Functionalized Multi-Walled Carbon Nanotubes by the Vapor Phase Method
337
W. Fan, L. Gao, and J. Sun J. Am. Ceram. SOC.,Vol. 89, No. 8. p. 2671-2673, 2006
Electrospinning: A Simple and Versatile Technique for Producing Ceramic Nanofibers and Nanotubes
341
D. Li, J. McCann, Y. Xia, and M. Marquez J. Am. Ceram. SOC.,Vol. 89, No. 6, p. 1861-1 869, 2006
Nanocomposites and Nanostructures
Chemical Precipitation Synthesis and Optical Properties of ZnO/SiO, Nanocomposites
353
H. Yang, Y. Xiao, K. Liu, and Q. Feng J. Am. Ceram. SOC.,Vol. 91, No. 5, p. 1591-1596, 2008
Low-Temperature Processing of Dense Hydroxyapatite-Zirconia Composites
359
Y. Nayak, R. Rana, S. Pratihar, and S. Bhattacharyya Int. J. Appl. Ceram. Techno/.,Vol. 5, No. 1, p. 29-36, 2008
Synthesis and Characterization of Chalcogenide Nanocomposites
367
J. Martin and G. Nolas ESP, VOl. 28, NO.8, p. 221-226, 2008
Self Assembled Functional Nanostructures and Devices
373
C. S. Ozkan CESP, VOl. 28, NO.6, p. 91-94, 2008
Carbon Nanotube (CNT) and Carbon Fiber Reinforced High Toughness Reaction Bonded Composites
377
I?Karandikar, G. Evans, and M. Aghajanian CESP, Vol. 28, No. 6, 2008, p. 53-63, 2007
Syntheis, Characterization and Measurements of Electrical Properties of Alumina-Titania Nanocomposites
389
V. Somani and S. Kalita ESP, Vol. 27, No. 8, p. 11-22, 2007
A New Ternary Nanolaminate Carbide: Ti3SnC,
401
S. Dubois, T. Cabioc'h, P. Chartier, V. Gauthier, and M. Jaouen J. Am. Ceram. SOC.,Vol. 90, No. 8, p. 2642-2644, 2007
viii
Progress in Nanotechnology: Processing
Fabrication of a Nano-Si,N,/Nano-C Composite by High-Energy Ball Milling and Spark Plasma Sintering
405
X. Xu, T. Nishirnura, N. Hirosaki, R-J Xie, and H. Tanaka J. Am. Ceram. SOC., Vol. 90, No. 4, p. 1058-1062, 2007
Conversion of Bamboo to Biomorphic Composites Containing Silica and Silicon Carbide Nanowires
41 1
T. L. Y. Cheung and D. H. L. Ng J. Am. Ceram. SOC., Vol. 90, No. 2, p. 559464,2007
Novel Processing to Produce Polymer/Ceramic Nanocomposites by Atomic Layer Deposition
41 7
X. Liang, L. Hakirn, G-D Zhan, J. McCorrnick, S. George, A. Weirner, J. Spencer II, K. Buechler, J. Blackson, C. Wood,
and J. Dorgan J. Am. Ceram. SOC.,Vol. 90, No. 1, p. 57-63,2007
Intra-Type Nanocompositesfor Strengthened and Toughened Ceramic Materials
425
S. Choi, S. Honda, S. Hashirnoto, and H. Awaji C7; VOl. 190, p. 173-180,2006
Prepation and Properties of Mullite-Based Iron Multi-Functional Nanocomposites
433
H. Wang, W. Wang, Z. Fu, T. Sekino, and K. Niihara C7; Vol. 190, p. 203-211,2006
Electrospinning of Ceramic Nanofibers and Nanofiber Composites
443
J. Yuh, H. Park, and W. Sigmund Ceramic Transactions, Vol. 190, p. 9-19, 2006
Microstructure and Properties of Spark Plasma-Sintered Zr0,-ZrB,
Nanoceramic Composites
455
B. Basu, T. Venkateswaran, and D-Y Kim J. Am. Ceram. SOC., Vol. 89, No. 8, p. 2405-2412,2006
Homogeneous Zr0,-Al,O,
Composite Prepared by Nano-ZrO, Particle Multilayer-CoatedAI,O,
Particles
463
Y. Jia, Y. Hotta, K. Sato, and K. Watari J. Am. Ceram. SOC.,Vol. 89, No. 3, p. 1103-1106, 2006
Preparation of a Highly Conductive AI,O,/TiN Growth
lnterlayer Nanocomposite through Selective Matrix Grain
467
X. Jin and L. Gao J. Am. Ceram. SOC., Vol. 89, No. 3, p. 1129-1132, 2006
Preparation and Microstructure of Multi-Wall Carbon Nanotubes-ToughenedAI,O,
Composite
471
J. Fan, D. Zhao, M. Wu, Z. Xu, and J. Song J. Am. Ceram. SOC., Vol. 89, No. 2, p. 750-753,2006
Three-DimensionalAssemblies of Zirconia Nanocrystals Via Shape-Preserving Reactive Conversion of Diatom Microshells
475
S. Shian, Y. Cai, M. Weatherspoon, S. Allan, and K. Sandhage J. Am. Ceram. SOC., Vol. 89, No. 2, p. 694-698,2006
Contents
ix
This Page Intentionally Left Blank
Introduction
Although nanotechnology is still an emerging industry, it represents a huge potential in a variety of markets that include biomedical, electronics, and energy totaling billions of dollars. However, before these markets are realized, processing methods must be developed that can produce quality nanomaterials and structures. Whether the material is a powder, thin film, wire, or composite, an optimal processing method is needed. Powders of various compositions can be made by a wide range of methods, including freeze casting, chemical, hydrothermal synthesis, and solution combustion, among others. Each method has its limitations and advantages. The methods to make thin films and coatings include chemical vapor deposition, spray pyrolysis, and sol gel. Wires can be made by electrospinning or hydrothermal synthesis. Other methods are under development for making composites and other structures. This edition of Progress in Ceramic Technology series contains a select compilation of articles on the topic of nanomaterials processing of powders; thin films, wires and tubes; and composites that were previously published in The American Ceramic Society Bulletin, Journal of the American Ceramic Society, International Journal ofApplied Ceramic Technology, Ceramic Engineering and Science Proceedings (CESP) and Ceramic Transactions (CT). The American Ceramic Society contributes to the progress of nanotechnology by providing forums for information exchange during its various meetings and by publishing articles in its various journals and proceedings. For other books on nanotechnology, including Progress in Nanotechnology: Applications, visit the ACerS bookstore at www.ceramics.org or the ACerS-Wiley webpage at www.wi1ey.comJgolceramic.s.
Introduction
xi
This Page Intentionally Left Blank
Synthesis Methods for Powders
This Page Intentionally Left Blank
Freeze Casting as a Nanoparticle Material-Forming
Method
Kathy Lu*’+and Xiaojing Zhu* Materials Science and Engineering Department, Virginia Polytechnic Institute and State Universiy, Blacksburg, Virginia 24061
Nanoparticle material forming is challenging because of loose packing and agglomeration issues intrinsic to nanoparticles. Liquid processing shows great potential to overcome such hurdles. This study is focused on nanoparticle colloidal processing and freeze-casting forming. A l 2 O 3 nanoparticle suspensions are examined, and microstructure evolution of A203nanoparticle suspension during freeze casting is discussed. The “Fines” effect influences nanoparticle packing on freeze-cast sample surfaces. Trapped air bubbles in the suspension lead to a porous bulk microstructure. Prerest is necessary for dense and homogeneous green microstructure formation. The green strength, fracture mode, and ability to form fine features by freeze casting are also evaluated.
Introduction The huge surface area intrinsic to nanoparticles serves as one of the most striking advantages as well as disadvantages for nanoparticle material forming. Because of the natural tendency of nanoparticles in forming agglomerates, wet forming utilizing a colloidal suspension has become the most active research area in nanoceramics. For almost all the nanoparticle wet forming processes, the first step is to produce a stable colloidal suspension. A green body with a uniform
Supported by Petroleum Research Fund, administered through American Chemical Society. Presented at the 31st International Conference on Advanced Ceramics and Composites, Dayrona Beach, FL, January 21-26, 2007. ‘Member, The American Ceramic Society. ‘Wu&t.edu 02008 The American Ceramic Society
Synthesis Methods for Powders
microstructure can then be produced from the h l l y stabilized colloidal suspension of nanoparticles. Freeze casting is a process that pours the suspension into a nonporous mold, freezes the suspension, demolds the frozen sample, and then dries the sample under vacuum. Liquid-state to solid-state conversion is realized through phase transformation of the dispersing medium such as water. The process has the potential to form near net-shape complex geometry parts with low pressure and often environmentally benign advantages. l T 2 The key requirements are that the suspension is stable and ~nagglomerated.~ When the freeze-casting condition is properly controlled, water separates from solid phases through sublimation and no capillary force exists to cause hard agglomerates or cracks.* Freeze casting has been practiced for sometime. 1,2,5,6 However, the studies were mainly focused on micrometer-sized particles and large-sized samples. For example, A 1 2 0 3 particles of 0.4 pm were freeze cast
3
and > 98% sintered density was achieved.’ Enclosed shells of Al2O3 bodies encapsulating steel parts were fabricated.8 Porous and layered-hybrid materials were freeze cast.’ However, application of freeze casting to nanoparticle systems and understanding the freezecasting behaviors of nanoparticle suspensions have not been explored. One challenge is that nanoparticle research in forming bulk components is still evolving; considerable effort is still needed in achieving high solids loading suspensions. The other challenge is the drastic capillary pressure increase with decreasing particle size; it takes a prolonged duration of time and much improved control to remove water completely by sublimation. Our past work has addressed the first challenge to a certain extent. 10-12 To address the second challenge successfully, it is essential to understand the nanoparticle microstructure evolution from the colloidal state to the freeze-dried state. High particle packing efficiency is preferred for postfreeze casting, handling, and densification purposes. Freeze-cast sample strength and failure mode can be used to understand and evaluate this aspect. With the decrease of particle size to nanoscale, the feature sizes of nanoparticle samples can also be correspondingly decreased. The potential in forming fine features by freeze casting should be explored. This study is focused on A 1 2 0 3 nanoparticle suspension viscosity evaluation and the corresponding freeze casting process for solid sample formation. Microstructural evolution of nanoparticle suspensions from a colloidal state to a freeze-dried state is examined. The surface and bulk microstructures under different prerest conditions are compared. Special nanoparticle phenomena during the suspension to solid-state transformation are explained. Equibiaxial strength and fracture mode of the freeze-cast samples are examined. Fine features produced by the freeze-casting process are presented.
Experimental Procedure A1203 dry nanoparticles with a specific surface area of 45 m2/g were used in this study (Nanophase Technologies, Romeoville, IL). The particles can be redispersed in water by ball milling as reported before by transmission electron microscopy (TEM) analysis. l o In this study, the particle size distribution was further measured by a dynamic light-scattering measurement
4
Fig. I . A1203 nanoparticle size distribution porn dynamic light scattering analysis: (a) weight basis, (b) number basis. as shown in Fig. 1 (Zetasizer Nan0 ZS, Malvern Instruments Inc., Southborough, MA). Agglomerates are absent. However, A203 nanoparticles are polydispersed, consistent with the TEM analysis. Even though on the weight basis there seems to be 10 wt% of > 100 nm particles, on the number basis, large particles are negligible ( 1.0 would be necessary to chelate all cations in the solution. But SP allows a non-equilibrium rapid thermal process and will lead to some different results from the synthesis under mild conditions. In this study, the ratio of EG/cation was adjusted to be 0.25, 0.50, 1.0, and 5.0 in order to obtain some information on the relationship between the degree of chelation and the morphology of NiO/SDC composite particles. As for heating for the chelation with the solution containing EG, we investigated two processes. One process aimed at the chelation during SP. The solution without heating was diluted with deionized water until the NiOjSDC concentration was 0.1 mol/ dm3 and the EG-mixed solution was spray pyrolyzed directly. The other aimed at sufficient chelating before the pyrolysis by heating the solution. The solution containing EG was heated to 80°C for 5 h in air to allow sufficient chelation. Several minutes after the solution was heated to 8 0 T , it generated a heat by the
Fig. 1. Morphology of Ni/SDC composite particles capsule type versus matrix type. SDC, samarium-doped ceria; Ni, nickel.
vide a large contact area between Ni and SDC. As compared with capsule-type composites, a matrix-type composite particle is expected to have a large contact area because the matrix-type composite is comprised of nanometer-sized Ni and SDC particles (Fig. I). SOFC anodes derived from the matrix-type composite particles can have improved anodic polarization properties relative to the capsule-type ones. Matrix-type NiOjSDC composite particles were thus synthesized by SP using a chelating reagent. The use of an aqueous solution composed of Ni, Sm, and Ce salts resulted in capsuletype composites. The formation mechanism of capsule-type composites is still unclear, but the difference in mass transfer rates between NiO and SDC during firing could possibly lead to separate inner NiO and outer SDC. Therefore, the addition of a binding agent among Ni, Sm, and Ce can result in a matrix-type composite. In this study, ethylene glycol (EG) was used as a binding agent, and the effects of EG on the morphology of NiO/ SDC composite particles were examined by SP. The synthesis conditions and electrochemical performance were also investi-
H
H
I
1
-
HO - C - C - O H
I
I
H
H
A
+
M+2 = > I HNO3
..*...* M .* .*
0
No
\\ /c
HO
- \ OH
+
H
H
I
1
HO-C-C-OH
I
I
H
H
c-c + H
nH20
H
n Fig. 2. Chelation scheme by the Pechini process.
Fig. 3. SEM images of NiOjSDC composite particles. The particles were synthesized by SP using EG-containing solutions at an EG ratio to the cations: (a) 0.25, (b) 0.50, and (c) 1.00. SDC, samarium-doped ceria; SEM, scanning electron microscope; EG, ethylene glycol; SP, Spray pyrolysis.
78
Progress in Nanotechnology: Processing
Fig.4. SEM image of Nio/sDC composite Particles. The Particles were synthesized by spray pyrolyzing diluted chelated solutions at an EG/cation ratio of 5.0. SDC, samarium,oped ceria; SEM, scanning electron microscope; EG, ethylene glycol.
exothermaic chelating reaction, and the temperature of the solution temporarily reached 120°C. After cooling to room temperature, the chelated solution was then diluted with deionized water and the diluted solution was spray pyrolyzed. In the SP instrument, an aqueous solution was atomized with an ultrasonic mist generator with an oscillation frequency of 1.7 MHz. The solution was placed in an atomizing vessel and the atomizer transformed the solution into a few micrometer-sized mists. Ultrasonic atomizers increased the temperature of the solutions. Cold water was then circulated around the atomizing vessel because atomization at a relatively high temperature led to inhomogeneous mists.23 The mists were transferred into a firing zone using dry air as a carrier gas. Four furnaces were arranged in series, and the temperatures of the furnaces were set at 200", 400", SOO", and 1000°C. The mists were dried, decomposed, reacted, and sintered during passage through the various furnaces. The composite particles were thus obtained continuously. The particles were collected using a membrane filter with 0.2 pm pores, and calcined at 1000°C for 24 h in air. The morphologies of the particles and anode microstructures obtained with the particles were observed with a scanning electron microscope (SEM, Hitachi S4500, Tokyo, Japan) and a transmission electron microscope (TEM, Topcon 002EB, Tokyo, Japan), The compositions of the composite particles were investigated with an energy-dispersive X-ray spectrometer equipped with TEM (EDS/TEM).
Fig.5. TEM image and EDSFEM analysis of a NiOjSDC composite particle. The particle was synthesized with chelated solutions at an EG/cation ratio of 5.0. TEM, transmission electron microscope; SDC, samarium-doped ceria; EG, ethylene glycol; EDS, energy dispersive X-ray spectrometer.
Synthesis Methods for Powders
79
Fig. 6. TEM image and EDS/TEM analysisof NiOjSDC flakes. The NiOjSDC flake was synthesized by spray pyrolyzing an EG-containingsolution at an EG/cation ratio of 1.OO.TEM, transmission electron microscope; SDC, samarium-doped ceria; EG, ethylene glycol; EDS, energy dispersive X-ray
spectrometer.
SOFC single cells were prepared by attaching the NiOjSDC composite particles on an electrolyte and a cathode on the other side of the electrolyte, and SOFC performances were examined for the single cells. Lao.sSro.,Gao.8Mgo.z03-~ (LSGM) disks with a thickness of 0.2 or 0.1 mm were prepared as a solid electrolyte.
Fig. 7. Power densities and cell voltages of SOFC single cells composed of the anodes synthesized using EG. The thickness of the LSGM electrolyte was adjusted at 200 pm. The measurement temperature was 750°C. NiOjSDC composite particles were synthesized with either diluted-chelated solutions at an EG/cation ratio of 5 . 0 ( 0 , 0 ) or EGmixed solutions at an EG/cation ratio of 1.0 (0,+), 0.5( O, ), and 0.25 (A, A). SOFC, solid oxide fuel cells; LSGM, LaoSSrO ~ M g o@ - s ; SDC, samariumdoped ceria; EG, ethylene
glycol.
The anode paste was composed of polyethylene glycol (PEG, average molecular weight: 400, Wako Pure Chemical Ind. Ltd.) and the NiOjSDC composite particles were screen printed on the LSGM disk and sintered at 1250°Cfor 2 h in air. The anode paste was prepared by wet blending with the particles, PEG, and ethanol for 15 min using an alumina mortar. SEM observations of the composite particles before and after the wet blending indicated that this wet blending did not break up the composite morphology. After the anode paste was attached on the LSGM disk, the cathode paste composed of PEG and L%.zSro.8C003 (LSC, Seimi Chemical Co. Ltd., Kanagawa, Japan) was also screen printed on the other side of the LSGM disk and sintered at 1000°Cfor 4 h in air.*'When SOFC performance was investigated using LSGM disks with a thickness of 0.2 mm, a reference electrode was attached on the side of the disk with a Pt wire and glass flit-free Pt paste (Tanaka Precious Metals). Hydrogen humidified at 23°C (3%) was used as a fuel gas, whereas dry air was used an oxidation gas. After the complete reduction of NiO to Ni metal with 3% humidified hydrogen at 800°C for 30 min, SOFC performances and anode polarizations were measured at 650"-750"C. The anode polarizations were measured using the reference electrodes and the anodes by the current interruption method. Therefore, the ohmic loss estimated by the anode polarization measurement contains almost half the resistance of the LSGM electrolyte.
111.
Results and Discussion
EG was added in the nitric acid solution of Ni, Sm, and Ce salts at the EG/cation molar ratio of 0.25, 0.50, and 1.00, and the EG-mixed solutions were directly spray pyrolyzed to allow the partial or full chelation and the pyrolysis to occur almost simProgress in Nanotechnology: Processing
Fig.8. Ohmic loss and overpotential properties of SOFC anodes. The anodes were obtained NiOjSDC composite particles prepared with a dilutedchelated solution at an EG/cation ratio of 5.0 (@), and EG-mixed solutions at an EG/cation ratio of 1.0 (+),0.5 ( ), and 0.25 (A).SOFC, solid oxide fuel cells; SDC, samarium-doped ceria; EG, ethylene glycol.
ultaneously by firing into the series of furnaces. The NiOjSDC composite particles obtained were observed with SEM. Figure 3 shows the particles synthesized with EG containing mixed solutions at the various ratios. The mixed solution at the ratio of 0.25 resulted in round composite particles and the particle sizes were 0.3-3 pm. This particle size distribution is large as compared with capsule-type NiOjSDC composite particles synthesized with EG-free solutions. The composite particle size slightly increased with an increase in the EG ratio. The particles obtained with the mixed solutions at the EG ratio of 0.5 and 1.0 contained flakes and their fragments. The addition of E C would bind the cations by chelating, but the addition at the EG/cation ratio larger than 0.5 resulted in the formation of flakes. The EG/ cation ratio larger than 1.0 would lead to full chelation but formed a large amount of flakes. The amount of flakes increased with an increase in the EG/cation ratio, but SEM and TEM observations did not clarify the relationship between the EG/ cation ratio and the dispersion of NiO or SDC in the composite particles. The flakes had a large volume, which prevented the composite particles from forming a homogeneous anode paste for screen printing. Therefore, simultaneous chelation during SP without pre-heating would not be suitable for synthesis of matrix-type composite particles. The EG solutions were then heated to 80°C and the chelated viscous solutions were diluted with deionized water. The solution was then spray pyrolyzed as soon as possible to avoid hydrolysis of the solution with added water. Figure 4 shows an SEM image of NiOjSDC composite particles using the chelated solution at an EG ratio of 5.0. The composite particles had many dimples and some particles were broken into pieces. Small voids were observed in the particles, but solid particles were obtained by SP with the solution. The particle size distribution was relatively small and these particles would be applicable as anode precursor particles. In order to clarify whether the composite particles had matrix-type morphology or not, TEM observation was carried out. The composite particles obtained with the fully chelated solution at an EG/cation ratio of 5.0 were calcined at 1000°C for 24 h in air. The calcined particles were observed with TEM, and detailed compositions were investigated with EDS/TEM (Fig. 5). This TEM and EDS/ TEM images reveal a composite particle comprised of 50-100 nanometer-sized particles, and both Ni and Ce were dispersed together in the composite particle. The images in Fig. 5 are relatively unclear because the calcined particles were too large or thick to observe detailed composite morphologies with TEM or EDS/TEM. The solid particles obtained with the chelated solution are supposed to have a composite morphology similar to the flakes synthesized with the solution at an EG/cation ratio larger than 1.0. The flakes were then observed with TEM and Synthesis Methods for Powders
detailed compositions in the flakes were analyzed with EDS/ TEM because the flakes were thin enough to investigate with TEM. Figure 6 shows a TEM image and the results of EDS/ TEM for the flakes. These flakes were obtained by spray pyrolyzing the EG-mixed solution at an EG/cation ratio of 1.0. The flake also comprised 50-100 nanometer-sized particles. EDS/ TEM analysis showed that the small particles were either NiO or SDC, and that NiO and SDC particles were dispersed together in the flake. Apparently, this flake and the composite particle in Fig. 5 were matrix-type composite particles and flake-free matrix-type composite particles were obtained with the chelated solution. Therefore, adequate addition of EG in the solution for SP converted capsule-type to matrix-type composition particles. SOFC performance was then measured using the matrix-type NiO/SDC composite particles. Four kinds of matrix-type particles were prepared by the SP process with or without pre-heating. Solid particles were obtained by pre-heating the chelated solution at an EG/cation ratio of 5.0, whereas the composite particles containing the flakes were obtained by spray pyrolyzing directly the EG-mixed solutions at an EG/cation ratios of 0.25,0.50, and 1.0. SOFC single cells were fabricated using these NiOjSDC composite particles as anode precursor particles. Power densities and cell voltages for the single cells prepared with LSGM electrolytes with a thickness of 200 pm were measured at 750°C. Figure 7 shows the electrochemical performance of the single cells with anodes derived from various matrix-type composite precursors. The NiOjSDC composite particles obtained using the chelated solution at an EG/cation ratio of 5.0 showed the highest power densities as compared with other matrix-type composite particles. The relationship between SOFC performances and the morphology of the composite particles is still vague. For example, whether the flakes decreased SOFC cell performance or not, and whether pre-heating changed micro- or nano-dispersion in the matrix-type composite particles are under consideration. The flakes, which were not formed using the chelated solutions, had a large volume. The large volume has some advantages such as an easy diffusion of a fuel gas in the anode, but it may have a disadvantage related to the difficult formation of Ni or SDC skeleton network during reduction from NiO to Ni. If the volume of the flakes is too much, the flakes would prevent Ni or SDC from forming a skeleton network. Electrode polarizations for the anode derived from matrix-type composite particles were also measured at 750°C as shown in Fig. 8. The use of the chelated solution at an EG/cation ratio of 5.0 lowered ohmic loss and overpotential voltage as compared with other matrix-type composite anodes. The solid matrix-type composite particles with a relatively small particle size distribution appear to produce low anode polariza81
Fig. 9. Power densities and cell voltages of an SOFC single cell composed of anodes prepared with matrix type and capsule-type NiOjSDC composite particles. The thickness of the LSGM electrolyte was adjusted at 100 pm. The measurement temperatures were 650°C (A, A), 700°C (0, m), and 750°C (0,0 ) . SOFC, solid oxide fuel cells; SDC, samarium-doped ceria; LSGM, Lao.sSro.,Gao.sMgo.zO~_s.
tions. The NiOjSDC composite particles prepared with the EGmixed solution at an EG/cation ratio of 0.25 did not contain flakes, but the particles showed higher ohmic losses as compared with the particles prepared with the mixed solution at an EG/ cation ratio of 0.50 and 1.O. Backscattered SEM images of the
anode derived from various matrix-type particles did not show any apparent difference in the anode microstructure depending on the EG/cation ratios and volume of the flakes. The flakes may have some advantages as the formation of an Ni or SDC network. But the flakes also resulted in inhomogeneous anode
(a) SOFC' mode dcrived from matrix-t> pe paTtic1es.
Fig. 10. Backscattered SEM images of SOFC anodes derived from matrix-type and capsule-type NiOjSDC composite particles. SOFC, solid oxide fuel cells; SDC, samarium-doped ceria; SEM, scanning electron microscope.
82
Progress in Nanotechnology: Processing
pastes and the inhomogeneous pastes would affect anode performance. Rheological investigation of anode pastes for screen printing would be additionally required to make good use of the flakes and to clarify the relationship between EG/cation ratios of the matrix-type particles prepared with the EG-mixed solutions and anode performance. It has been noted that NiO reacts with the LSGM electrolyte to form an insulating layer of LaNiO, at the interface between LSGM electrolytes and Ni-containing anodes.“ However, severe degradation owing to this reaction was not observed using the NiOjSDC composite particles obtained by SP. Micro- or nano-composition may suppress this reaction. Finally, the electrochemical performances of the SOFC single cells composed of the anodes derived from matrix-type composite particles and capsule-type particles were measured using an LSGM electrolyte with a thickness of 100 pm at 650”, 700”, and 750°C. Figure 9 shows SOFC power densities and cell voltages for both the single cells. Flake-free matrix-type composite particles were synthesized by SP with the chelated solutions at EG/ cation ratios of 5.0, whereas capsule-type composite particles were synthesized by SP without EG. Both the composite particles had the same NiOjSDC compositions. Matrix-type composite particles resulted in high-performance SOFC anodes as compared with capsule-type composite particles under these measurement conditions, and the maximum power densities of 1.69 W/cm2 at 750°C, 1.05 W/cm2 at 700”C, and 0.53W/cm2 at 650°C were found. Figure 10 shows backscattered SEM images of the anodes prepared with matrix-type and capsule-type composite particles. Partly collapsed composite particles in which some SDC small particles surrounded inner spherical Ni particles remained in the anode derived from the capsule-type composites, whereas nanometer-sized SDC or Ni particles were connected together in the anode derived from matrix-type composites. The matrix-type composite-derived anode showed poor porosity but preferable contact between SDC and Ni as compared with the capsule-type composite-derived anode. The capsule-type. composite anode would suppress Ni sintering as compared with the matrix-type one. However, the Ni sintering would be reduced by decreasing SOFC working temperatures. Matrix-type composite particles were then effective as precursors of high-performance IT-SOFC anodes.
IV. Conclusions Matrix-type NiOjSDC composite particles were synthesized by SP. An adequate addition of EG in the solution for SP converted capsule-type to matrix-type composition particles. The direct SP of the mixed solution containing EG resulted in NiO/SDC composite particles and flakes, whereas the pyrolysis of the solution chelated with EG by pre-heating resulted in solid particles with many dimples on the surface and relatively small particle size distributions. The solid matrix-type composite particles also showed a higher anode performance than the capsule-type composite particles. The matrix-type composite-derived anode showed poor porosity but preferable contact between SDC and Ni as compared with the capsule-type composite-derived anode. The matrix-type solid composite particles were effective as precursors of high-performance IT-SOFC anodes. References ‘V. V. Kharton, A. V. Kovalevsky, A. P. Viskup, A. L. Shaula. F. M. Figueiredo, E. N. Naumovich, and F. M. B. Marques, “Oxygen Transport in
Synthesis Methods for Powders
C Q , ~ G ~202-s-Based ,, Composite Membranes,” Solid Slate tonics. 160. 247-8 (2003). 2J. C. C. Abrantes, D. Perez-Coll, P. Nunez, and J. R. Frade, “Electronic y-6 Ceramics Under Reducing Conditions,” ElecfroTransport in Ce,,.8Sm,, 201 chem. Acfa, 48, 27614 (2003). ’F. Y. Wang. S. Chen, and S. Cheng, “Gd’+ and Sm3+Co-Doped Ceria Based Electrolytes for Intermediate Temperature Solid Oxide Fuel Cells,” Elecfrochem. Commun., 6. 7 4 3 4 (2004). 4V. Agdrwal and M. Liu, “Colloidal Processing of BaCeO,-Based Electrolyte Films,” J. Elecfrochem. Soc., 143, 323944 (1996). 5Z. Wu and M. Liu, “Stability of BaCe,,8Gdn203in a H20-Containing Atmosphere at Intermediate Temperatures,” J. Electrochem. Soc., 144, 217C-5 (1997). 6N. Sammes, G. Tompsett, Y. Zhang, A. Cartner, and R.Torrens, “The Structural and Mechanical Properties of (Ce02)l_,(GdO, .5)x Electrolytes.” Denki Kagaku, 64,674-80 (1996). 7Y. lkuma, “Nonstoichiometry and Diffusion in Ceria-Based Solid Solutions,” J. Soc. Inorg. Mafer. Jpn., 12, 213-20 (2005). ‘K. Higashi, K. Sonoda, H. Ono. S. Sameshima, and Y. Hirata. “Synthesis and Sintering of Rare-Earth-Doped Ceria Powder by the Oxalate Coprecipitation Method,” J. Maier. Res.. 14, 957-6 (1999). ’C. Milliken, S. Guruswamy, and A. Khandkar, “Evaluation of Ceria Electrolytes in Solid Oxide Fuel Cells Electric Power Generation,” J. Elecfrochem. Soc.. 146, 872-2 (1999). ‘OK. Yashira, S. Onuma, A. Kaimai, Y. Nigara, T. Kawdda, J. Mizusaki, K. Kawamura, T. Horita, and H. Yokokawa. ”Mass Transport Properties of Ce,,&do at the Surface and in the Bulk,” Solid Sfare tonics, 152-153, 469-76 (2002). ‘IT. lshihara, H. Minami, H. Matsuda, H. Nishiguchi, and Y. Takita, “Decreased Operating Temperature of Solid Oxide Fuel Cells (SOFCs) by the Application of LaGaO,-Based Oxide as Electrolyte,” Chem. Commun., 929-30 (1996). ”P. R. Slater, J. T. S. Irvine, T. Ishihara. and Y. Takita, “The Structure of the Oxide Ion Conductor L+,ySro lG+ 8M& 202.8s by Powder Neutron Diffraction.” Solid Stare tonics, 107, 319-23 (1998). ”I. Taniguchi, R. C. van Landshoot, and J. Schoonman, “Electrostatic Spray Deposition of Gdo.iC~yO1.g5and LaoySrg.lGa08Mgo202.87 Thin Films,” Solid Stare lonics, 160, 271-9 (2003). I4A. C. Tas, P. J. Majewski, and F. Aldinger, “Chemical Preparation of Pure and Strontium- and/or Magnesiumdoped Lanthanum Gallate Powder,” J . Am. (2000). Ceram. Sac., 83. 295”M. Feng, J. 9. Goodenough, K. Huang, and C. Milliken, “Fuel Cells With Do d Lanthanum Gallate Electrolyte,” J . Power Sources, 63,47-51 (19%). Huang, R. Tichy. J. 9. Goodenough, and C. Milliken, “Superior Perovskite Oxide-Ion Conductor; Strontium-and Magnesium-Doped LaGaO,: 111. Performance Tests of Single Ceramic Fuel Cells,” J. Am. Ceram. Soc., 81, 2581-5 (1998). ”Y. Matsuzaki, Y. Baba, and T. Sakurai, “High Electric Conversion Efficiency and Electrochemical Properties of Anode-supported SOFCs,” Solid Sfafetonics, 174, 81-6 (2004). “Y. J. Leng, S. H. Chan, K. A. Khor, and S. P. Jiang, “Performance Evaluation of Anode-supported Solid Oxide Fuel Cells with Thin Film YSZ Electrolyte.” Inf. J. Hydrogen-Energ., 29, 102S33 (2004). ”S. de Souza, S. J. Visco, and L. C. De Jonghe, “Thin-Film Solid Oxide Fuel Cell With High Performance at Low Temperature,” Solid State lonics, 98, 57-71 (1997). , Matsuzaki. Y. Baba. and T. Sakurai, “Diffusion Characteristics in AnodeSubstrates for AnodaSupported SOFCs,” Elecfrochemisfry,73, 484-8 (2005). ”S. Suda, M. Itagaki, E. Node, S. Takabashi, M. Kawano, H. Yoshida, and T. Inagaki, “Preparation of SOFC Anode Composites by Spray Pyrolysis,” J . Euro. Ceram. Soc., 26, 593-7 (2006). 22 M. Kawano, K. Hashino, H. Yoshida, H. Ijichi, S. Takahashi. S. Suda, and T. Inagaki, “Synthesis and Characterizations of Composite Particles for Solid Oxide Fuel Cell Anodes by Spray Pyrolysis and Intermediate Temperature Cell Performance,” J. Power Sources, 152, 1969 (2005). 23S.Suda, S. Takahashi, M. Kawano, H. Yoshida, and T. Inagaki, “Effects of Atomization Conditions on Morphology and SOFC Anode Performance of Spray Pyrolyzed N i 0 - S ~ , 2 C e o , 8 0 Composite 1,y Particles,” So/idSrafetonics, 177, 121% 25 (2006). 24R.C. West, M. J. Astle, and W. H. Beyer (eds.) C R C Hondhook of Chemisfry and Physics, pp. E-124, CRC Press Inc. Florida. USA. 1989. 25 H. Yahiro, Y. Eguchi, K. Eguchi, and H. Arai, “Oxygen Ion Conductivity of the Ceria-Samarium Oxide System with Fluorite Structure,” J . Appl. Elecfrmhem.. 18, 527-31 (1988). ’%. Zhao and R. J. Gorte. “A Comparison of Ceria and Sm-doped Ceria for Hydrocarbon Oxidation Reactions,” Appl. Cafal.A: General, 277, 129-36 (2004). 27R.Maric, S. Ohara, T. Fukui, H. Yoshida, M. Nishimura, T. Inagaki, and K. Miura, “Solid Oxide Fuel Cells with Doped Lanthanum Gallate Electrolyte and LaSrCoO, Cathode, and Ni-Samaria-Doped Ceria Cermet Anode.” J. Elecfro0 chem. Soc., 146. 200610 (1999).
‘9°K.
\ -
83
This Page Intentionally Left Blank
Novel Low-Temperature Synthesis of Ferroelectric Neodymium-Doped Bismuth Titanate Nanoparticles Prem Prakash and Ashish Gargt Department of Materials and Metallurgical Engineering, Indian Institute of Technology, Kanpur 208016, India
Mukesh Kumar Roy and Harish Chandra Verma Department of Physics, Indian Institute of Technology, Kanpur 208016, India
In this study, we report on the synthesis of nanopowders offerroelectric Bi3.5N&.5Ti3012 ceramic at temperatures below 500°C via a simple chemical method using citric acid as a solvent. The calcined powders were characterized using X-ray diffraction (XRD), differential scanning calorimetry (DSC), and transmission electron microscopy (TEM). Heating the as-dried powders in air first leads to crystallization of the BizTi207 phase at -310°C,followed by crystallization of the perovskite Nddoped BhTi3Ol2 phase at -490°C as suggested by the peaks in the DSC analysis and confirmed by the evolution of phases in XRD patterns of the powders calcined at various temperatures. TEM of particles calcined at 550°C for 1 h in air showed an average particle size of 50-60 nm. The temperature dependence of capacitance of nanopowders calcined at 700°C for 1 h in air showed a Curie temperature of 615°C evincing a ferroelectric transition.
-
I.
Introduction
L
ANTHANIDE-doped ferroelectric bismuth titanate (Bi4Ti3OL2) has been extensively studied from both the technological and fundamental points of view because of its potential applications in the non-volatile ferroelectric random access memory devices as well as sensors and actuators. Bi4Ti3012is one of the well-studied compounds of the Aurivillius family' and is represented by the general formula (Bi202)2+ [Am-lB,03m+ with m = 3 consisting of three perovskite-structured (Bi2Ti3010)2units sandwiched between (Bi202)2+ layers along the c axis of the Bi4Ti30L2unit cell, i.e., one layer is present on each side of the stack of three perovskite layers along the c axis. Upon thermal cycling, it undergoes reversible ferroelectric- araelectric phase transition at a Curie temperature of 675"C, a transition from a paraelectric tetragonal Z4/mmm structure, as proposed by Aurivillus,' to an orthorhombic structure.24 Doping with a lanthanide element such as La or Nd creates a significant amount of increment in the fatigue resistance of Bi4Ti3OI2,especially in the thin-film form.5 The current emphasis on nanotechnology has led to the drive toward the synthesis of ferroelectric materials in nanosized form. Nanosized ferroelectrics are further important to enhance the understanding of size effects in ferroelectrics. Conventionally, Bi4Ti3OI2ceramics have been prepared by solid-state reaction in a stoichiometric oxide mixture of Bi203 and Ti02, which
-
9,
Z. Bdrbe--contributing editor
Manuscript No. 22374. Received October 20, 2006; approved November 27, 2006. Supported by Department of Science and Technology, New Delhi (Grant no. SR/FTP/ ETA-032/2002). 'Author to whom correspondence should be addressed. e-mail:
[email protected] Synthesis Methods for Powders
are calcined at temperatures above 800°C to yield a fully crystallized perovskite Bi4Ti3OI2phase.637However, this leads to unwanted coarsening and agglomeration of the powders, which is undesirable in many applications. Efforts have been made to decrease the calcination temperatures, and chemical processes have proved to be successful to some extent. The powders of Bi4Ti301 have been synthesized by various chemical rocesses such as co-pre~ipitation,~.~ sol-gel,'03'1 hydr~thermal,'.'~ molten salt route,14 metalor anic decomposition,15 and mechanically assisted synthesis," but reports are limited in nature. Although some of these processes offer crystallization of the perovskite phase at temperatures below 800"C, nano-crystallinity is not achieved. On the other hand, hydrothermal synthesis leads to c stallized nanosized Bi4Ti3012at temperatures as low as 24OoC,' but the processing route remains complex, making the process unviable for large-scale production demanding high throughput at lower costs. Thus, low-temperature synthesis of Bi4Tt3OI2at a nanoscale using a simple process is important and remains a challenge. In this study, we report on the low-temperature synthesis of nanosized powders of well-crystallized ferroelectric Bi3,5Nd0.5Ti3012r using a chemical route. The powders were characterized using X-ray diffraction (XRD), differential scanning calorimetry and thermo-gravimetric analysis (DSCTGA), and transmission electron microscopy (TEM). Dielectric measurements were made to ascertain the ferroelectric nature of the particles.
r
11. Experimental Procedure (1) Synthesis Neodymium-doped BbTi3012 nanopowders (Bi3.5Nd0.5Ti3012 or BNdT) were synthesized by the wet chemistry route. Initially, a solution was made by dissolving the Bi and Nd precursors (hydrated bismuth and neodymium nitrates) in a citric acid solvent. Ethylene amine was used to stabilize the solution. Titanium isopropoxide was used as a titanium precursor and was stabilized using acetyl acetone before adding to the solvent containing Bi and Nd precursors. Extra 10% Bismuth was added to overcome the loss of Bi during heat treatment. A measured amount of sol (200 mL) was dried at 110°C while vigorously stirring using a hot place cum magnetic stirrer until it became viscous enough so that the sol constituents could not move freely. It was dried further in an ultrasonic bath at the same temperature. The dried powder was calcined at various temperatures, from 200" to 700°C for 1 h in air.
-
(2) Characterization Analysis of the crystallinity, phase purity, and particle size of the calcined powders was carried out using a Seifert X-ray diffractometer using CuKu radiation (h = 1.540598 A). DSC and TGA were conducted using a METTLER TOLEDO DSC (822e;
85
Greifensee, Switzerland) equipment to investigate the crystallization of the Bi4Ti3OI2phase in as-synthesized powders at a heating rate of 10"C/min from room temperature to 700°C in air. The particle size and phase purity of the synthesized powders were further confirmed by using JEOL-TEM (Tokyo, Japan), operated at 200 kV. To prepare the samples for TEM investigation, calcined BNdT samples were ultrasonicated in isopropanol for 10 min. A few drops of the suspension were placed on the Formvar-coated 3 mm diameter and 200 mesh copper grid. The concentration of BNdT nanoparticles in the suspension was optimized to avoid agglomeration of the particles. The average particle size was also calculated from the bright-field (BF) images, and a selected area diffraction pattern (SADP) was taken to affirm the phase purity of the particles. The stoichiometry of the crystallized powders was examined using EDAX (energydispersive X-ray analysis) in a scanning electron microscope (FEI Quanta 200, Eindhoven, the Netherlands). Crystallized powders were cold pressed into pellets of 10 mm diameter. The faces of the green pellets were graphite coated to perform capacitance versus temperature measurements to determine ferroelectric transition using a Hewlett Packard 4192A (Melrose, MA) impedance analyzer at a frequency of 500 kHz. The sample was placed inside a horizontal tube furnace and was connected with silver wires on both the surfaces. A chromel/alumel thermocouple was placed beside the pellet to measure the sample temperature accurately. The temperature was controlled using a PID controller and a heating rate of l"C/min was maintained to attain the desired temperature uniformity.
111. Results and Discussion Figure 1 represents the XRD pattern of the BNdT samples calcined for 1 h in air at various temperatures: 300", 350", S O " , 600", and 700°C. The pattern shows that the samples calcined up to 300°C do not exhibit any significant phase formation. Crystallization is observed to begin at 350°C with the appearance of a non-ferroelectric Bi2Ti207phase7.17-19(JCPDS file number: 3118). However, upon heat treating the samples further up to 7 W C , Bi2Ti207was observed to transform into perovskite Nddoped B&Ti3012(JCPDS file number: 35-795a) as suggested by the gradual disappearance of Bi2Ti2O7peaks and enhancement of the peaks of the Nd-doped Bi4Ti3OI2phase. EDAX analysis of powders calcined at 550" and 700°C with the fully crystallized Nd-doped Bi4Ti3OI2phase showed a powder composition of Bi~.~+o.~Ndo.5+o.04Ti~O12, suggesting that the stoichiometry was maintained in all samples. Crystallite sizes of the heat-treated powders were calculated utilizing XRD peak-broadening data and using Scherrer's equation
Fig. 1. &20 patterns of Bi3sN& 5Ti3012nanopowders calcined at various temperatures.
heating and was evident by the broad peak between 100" and 250°C and significant weight loss as suggested by the TGA measurements (not shown here). In this figure, a strong exothermic peak (A) is observed at -31O"C, corresponding to the crystallization of the Bi2Ti207phase as suggested by the XRD results. Upon heating the sample further, another peak (B) is observed at -490°C, which can be ascribed to the formation of perovskite Bi4Ti3012phase, substantiated by the XRD results of powders heat treated above 500°C (Fig. I). Another peak (C) at 625°C could be indicative of a reversible ferroelectric-paraelectric phase transition and could be suggestive of the ferroelectric nature of the synthesized powders. This was further substantiated by the dielectric measurements of the powders as explained later. Previously, studies have been carried out to investigate the crystallization behavior of amorphous B&Ti3012.7*19 These studies also predicted the formation of perovskite Bi4Ti3012via a two-step process where Bi2Ti207was the first phase to form. It was reported that the formation of a metastable Bi2Ti207phase occurs at 580"C19and -608"C,7 followed by crystallization of a perovskite Bi4Ti3OI2phase at -85O"Cl9 and -831"C.7 Our results show the crystallization of perovskite Nd-doped Bi4. Ti3012 below 500"C, more than 300°C lower than the above values. To the best of our knowledge, this is also the lowest crystallization temperature reported for Bi4Ti3OI2 powders made using a simple chemical solution route. It is envisaged that this occurs due to the combined effect of vigorous stirring
-
-
10
7
8-
where d is the crystallite size, k is the shape factor (-0.9), h is the wavelength of the radiation used, is the full-width at halfmaximum of the diffraction intensity of selected peak, and 8 is the Bragg angle. The instrumental broadening was calculated for standard samples and was subtracted from the measured broadening. Measurements were made using broadening of the (117) peak of Bi4Ti3OI2occurring at 28 = 30.157". The calculated particle sizes for samples calcined for I h at S O " , 600", and 700°C were approximately 18, 35, and 40 nm, respectively. An increase in the particle size upon heating up to 700°C is indicative of particle coarsening due to enhanced diffusion and a reduction in the overall surface energy. In order to confirm and analyze the phase evolution observed from XRD patterns, thermal analysis was performed on uncalcined samples. Figure 2 shows the DSC scan of as-synthesized powders upon heating up to 700°C in air. The removal of the volatile chemical constituents takes place in the earlier stages of 86
z Y
Q
I3
64:
2-
1 100
200
300 400 500 Temperature "C
600
700
Fig. 2. Differential scanning calorimetry curve for as-prepared Bi,,,N~.sTi301zpowders (air atmosphere).
Progress in Nanotechnology:Processing
nanopowders calcined for 1 h in air at (a) 550"C, (b) W C , and Fig. 3. Bright-field transmission electron microscopy micrographs of Bi3.5N&.STi3012 (c) selected area diffraction pattern of the powders calcined at 600°C.
as well as ultrasonication of the solution during the drying stage, which helps to mix the different constituents of the mixture effectively and results in lower crystallization temperatures of various phases. In addition, vigorous stirring and ultrasonication during drying avoids the gel formation, which could be the key to the lowering of crystallization temperatures, eventually leading to nanosized particles. Formation of gel can lead to larger separation between the constituents, leading to higher crystallization temperatures. Figures 3(aHc) show the BF TEM micrographs of Bi4Ti3OI2 powders calcined at 550" and 600°C for 1 h in air and a selected area diffraction pattern of powders calcined at 600°C. It is evident from the micrograph (Figs. 3(a) and (b)) that the average particle size of the powders is between 50 and 60 nm, with the size ranging between -40 and 90 nm, which is reasonably uniform. A comparison of size measurements with those made using XRD peak widths suggests that some particles consist of more than one crystallite. The rings in the selected area diffraction obtained from powders calcined at 6OO0C, as shown in Fig. 3(c), were indexed and correspond to (012), (1 13), and (002) planes of Nd-doped Bi4Ti3OI2. Figure 4 shows the capacitance versus temperature plot for powders heat treated at 700°C for 1 h in air. It should be noted that we chose the nanopowders calcined at 700°C for the measurements to avoid any ambiguities or effects related to size growth that can occur due to heating the low-temperature-calSynthesis Methods for Powders
cined particles up to 800°C during electrical measurements. It is observed that the capacitance increases slowly until about 500"-550"C, and then starts to increase monotonically until the
0.4
G C
v
a, 0
a C
c
'U 0.2 m a
s
100
200
3M)
400
500
600
700
800
Sample Temperature ("C) Fig. 4. Capacitancetemperature plot for neodymium-doped bismuth titanate nanopowders calcined at 700°C (1 h, air) and measured at a frequency of 500 kHz.
87
-
transition from the ferroelectric to paraelectric phase at 615”C, followed by a sharp decline at the transition temperature. This temperature is comparable with the endothermic peak observed in DSC scan (Fig. 2), approximately at the same temperature. The transition process was found to be repeatable and reversible. This plot clearly shows that the nanopowders fabricated in the present study are ferroelectric in nature. Further studies are being carried out to thoroughly investigate the ferroelectric properties of the nanopowders, especially those calcined at lower temperatures.
IV. Conclusions Bi3.5Nd0.5Ti3012 nanopowders have been successfully synthesized at temperatures as low as -500°C via a simple chemical method using citric acid as a solvent. The crystallite sizes measured from XRD patterns were approximately 18,35, and 40 nm, for powders calcined for 1 h in air at 550”, 600”, and 700°C, respectively. The combination of XRD observations with DSC scans suggests a two-step transformation: formation of a nonferroelectric, non-perovskite Bi2Ti207 phase at 310”C, followed by formation of perovskite ferroelectric Nd-doped Bi4Ti3OI2 at -490°C. The sizes of well-crystallized powders calcined at 550°C (1 h in air) were -40-90 nm as shown by TEM, two to three times larger than those computed from Scherrer’s formula. The presence of a ferroelectric-paraelectric phase transition at -615°C is confirmed from the capacitance versus temperature measurements.
-
Acknowledgments The assistance of Mr. S.C. Barthwal (IIT Kanpur) with TEM, and Mr P. Padaikathan, and Dr. U Ramamurthy (both from Department of Metallurgy, Indian Institute of Science, Bangalore) with the DSC measurements is gratefully acknowledged.
References ’B. Aurivillus, “Mixed Bismuth Oxides with Layer Lattices I 1 Structure of Bi4TiZOI2.”Ark. Kemi.,1 [58] 499 (1949).
88
’J. F. Dorian, R. E. Newnham, and D. K. Smith, “Crystal Structure of Bi4Ti3OI2,”Ferroelectrics,3, 17 (1971). ’R. E. Newnham, R. W. Wolfe, and J. F. Dorian, “Structural Basis of Ferroelectricity in the Bismuth Titanate Family,” M a w . Res. Bull., 6, 1029 (1971). 4A. D. Rae, J. G. Thompson, R. L. Withers, and A. C. Willis, “Structure Refinement of Commensurately Modulated Bismuth Titanate, Bi4Ti3012.”Acra. Cr s f , 846,474 (1990). ‘B.’H. Park, B. S. Kang, S. D. Bu, T. W. Noh, J. Lee, and W. Jo. “LanthanumSubstituted Bismuth Titanate for Use in Non-Volatile Memories,” Nature, 401. 682 (1999). %. C. Subbarao, “Ferroelectricity in Bi4Ti3OI2and its Solid Solutions.” Phys. Rev., I22 131 804 (1961). 7S. Kojima, A. Hushur. F. Jiang. S. Hamazaki, T. Takashige, M. S. Jang, and S . Shimada, “Crystallization of Amorphous Bismuth Titanate.” J. Non-Crysr. Solids, 293-2295, 250 (2001). *Y. H. Wang, C. P. Huang, and Y. Y. Zhu, “Variable-Temperature Raman Scattering and X-Ray Diffraction Studies of BiZ2sNdo75Ti3012Ceramics,” Sol. State Comm., 139,229 (2006). ’M. Villegas, C. Moure, J. F. Fernandez, and P. Duran, “Low-Temperature Sintering of Submicronic Randomly Oriented Bi4Ti3012Materials,” Ceram. Int.. 22, 15 (1996). ’OX. Q. Chen, H. Y. Qi, Y. J. Qi, and C. J. Lu, “Ferroelectric and Dielectric Properties of Bismuth Neodymium Titanate Ceramics Prepared Using Sol-Gel Derived Fine Powders,” Phys. L p f f .A, 346 [I-31 204 (2005). “A. V. Prasada Rao, A. 1. Robin, and S. Komarneni, “Bismuth Titanate From Nanocomposite and SoCGel Processes,” M a w . Left.. 28, 469 (1996). ”D. Chen and X. Jiao, “Hydrothermal Synthesis and Characterization of Bi4Ti3012Powders From Different Precursors,” Mater. Re$. Bull., 36, 355 (2001). ”Y. Shi, C. Cao, and S. Feng, “Hydrothermal Synthesis and Characterization of Bi.,Ti3012,” Mater. L p f f . , 46, 270 (2000). I4Y. Kan, X. Jin, P. Wang, Y. Lia, Y.-8. Cheng, and D. Yan, “Anisotropic Grain Growth of Bi4Ti3OI2in Molten Salt Fluxes,” Mater. Res. Bull., 38, 567 (2003). ”W. L. Liu, H. R. Xia, H. Han, and X. Q. Wang, “Structural and Dielectrical Properties of Bismuth Titanate Nanoparticles Prepared by Metalorganic Decomposition Method.” J. Crysf. Growth, 269 [ 2 4 ] 499 (2004). I6B. D. Stojanovic, C. 0.Paiva-Santos, C. Jovalekic, A. 2.Simoes, F. M. Filho, Z. Lazarevic, and J. A. Varela. “Mechanically Activating Formation of Layered Structured Bismuth Titanate,” Mater. Chem. Phys., 96. 471 (2006). I7S. Shimada, K. Kodaria, and T. Matsushita, “Crystal Growth of Bismuth Titanates and Titanium Oxide From Melts in the System Bi203-V20rTi02.” J. Crysr. Growrh, 41, 317 (1977). ‘*W.-F. Su and Y.-T. Lu, “Synthesis, Phase Transformation and Dielectric Properties of Sol-gel Derived Bi2Ti207Ceramics,” Mater. Ckem. Phys., 80 [3] 632 (2003). I9Y. Yoneda, J. Mizuki, S. Kohara, S. Hamazaki, and M. Takashige, “Crystallization Process of Ferroelectric Bi4Ti3OI2From Amorphous State,” J. App. Pkys., 99,074108 (2006).
Progress in Nanotechnology:Processing
Hydrothermal Synthesis of CdMo04 Nano-Particles Xiaohui Jiang, Junfeng Ma,+,$Botao Lin, Yang Ren, Jun Liu, Xiaoyi Zhu, and Jiantao Tao Institute of Materials Science and Engineering, Ocean University of China, Qingdao 266003, China
Yonggang Wang and Lijin Xie College of Chemistry and Chemical Engineering, Ocean University of China, Qingdao 266003, China
CdMo04 nano-particles were successfully synthesized by a hydrothermal process at a low temperature of 9WC, and the powders were characterized in detail by X-ray diffraction, transmission electron microscopy, scanning electron microscopy, and photoluminescent spectra techniques (PL), respectively. CdMo04 particles could be obtained under the hydrothermal condition from micrometer to nanometer sizes by varying their precursors. The PL spectra results showed that the optical properties of CdMo04 crystallites obviously relied on their particle sizes.
1. Introduction
R
ECENTLY,
considerable attention has been paid to molybdates due to their electrochemistry properties,' special luminescence, and unique structure^.^-^ Metal molybdates gf relatively large bivalent cations (MMo04, ionic radius >0.99 A, M = Ca, Ba, Sr, Pb) are of a so-called scheelite structure in which the molybdenum atom adopts tetrahedral coordination,6 where the emission spectrum of the metal molybdates is mainly attributed to the charge-transfer transitions within the [Moog-] ~ o m p l e x . ~Calcium - ~ ~ molybdate (CaMo04) as an important functional material has been used in various fields, such as in photoluminescence (PL)" and microwave applications,12 and lead molybdate (PbMo04) single crystals are extensively used in acousto-optical and high-voltage measurement devices.l 3 Ternary oxides with the general formula MMo04 (M = Cu, Zn, Ni, Fe) are also characterized as cathode materials for rechargeable batteries,' the idea being to take advantage of the charge couple Mo6+/Mo4+where in the metal-redox oxidation state change by two units is very attractive for the development of batteries with a high capacity and also with high-energy density.I4-l8Cobalt molybdate (CoMo04) as another attractive compound has exhibited a wide range of applications in industrial catalysis for properties related to its s t r u ~ t u r e . ' ~ * ~ ~ However, there have been few reports on CdMo04, whether in its synthesis process or properties, to the best our knowled e Even though there possibly exist environmental drawbacks2 22 of cadmium compounds such as other heavy metal compounds as Lead Zirconate titanate (PZT), it is necessary to study CdMo04 as a potentially functional material with relatively low toxic Cd(II).23 Therefore, it is of significance both in fundamental and applied fields to synthesize and characterize
F.
G . Pdtzke--contrihuting editor
Manuscript No. 21820. Received May 24, 2006; approved November 3, 2006. 'Author to whom correspondenceshould he addressed. e-mail:
[email protected] 'COkge of Chemistry and Chemical Engineering,Ocean University of China, Qingdao, China.
Synthesis Methods for Powders
CdMo04 powders with different particle sizes, especially of nanometer scale. It is well known that molybdate powders can be prepared by many methods, e.g. the pulsed laser ablation process for synthesizi;g C ~ M O O the ~ , citrate ~ ~ complex method for BaMoo4, and the spray pyrolysis route to N ~ M o O ~ . ~ ~ Nevertheless, among these routes, there are still such limitations as needing a higher calcined temperature and a longer holding time to eliminate organic material or requiring complicated instrumentations and manipulations. The hydrothermal method has been considered to be a good synthesis process for some inorganic powders27 because of (1) available synthesis of crystallized products at a low reaction temperature, (2) flexibility in the design of reaction conditions, (3) uniformity of production composition, phase, and microstructure, and (4) simplicity of equipment and processing. Here, we report on the hydrothermal synthesis of CdMo04 crystallites. The particle size of CdMoO? powders was about 3-5 pm by a direct hydrothermal route, while the CdMo04 crystallites could be reduced to 30-50 nm in particle size under the same hydrothermal condition by introducing a microemulsion route to the precursor of CdMo04. The PL properties of CdMo04 crystallites with different particle sizes were also studied for the first time. 11. Experimental Procedure
(1) Synthesis of CdMoOl Precursors Sodium molybdate (Na2Mo04 2H20) and cadmium nitrate (Cd(N03)2 4H20) were used as the starting materials, and both of them were of analytical grade without any further purification. In our work, two methods were used to prepare the precursor of CdMo04 for the subsequent hydrothermal treatment. ( A ) Normal Precursor Preparation: Appropriate amounts of Na2Mo04 2H20 and Cd(N03)2 4H20 were dissolved in distilled water of 250 mL to form 1M aqueous solutions, respectively. Then, both solutions were mixed together with vigorous magnetic stirring at room temperature to form a suspension, which was called the normal precursor. ( B ) Microemulsion Precursor Preparation: Cyclohexane was used as the oil phase, OP was used as the surfactant, and nbutanol as the co-surfactant. A typical process was as follows: two kinds of microemulsions (M I and M 11) with different aqueous phases were obtained (see Table I). The aqueous phase in the M I was a solution of Cd(N03)*,while the aqueous phase in the M I1 contained Na2Mo04. The concentration of both of the aqueous phases was 0.1M . These two microemulsions were mixed together and stirred for 30 min at room temperature. The appearance of a white suspension indicated the formation of microemulsion precursor of CdMo04. The precipitate could be separated by a centrifuge, and washed with acetone and absolute
-
-
89
Table 1.
Aqueous phase Surfactant Co-surfactant Oil ohase
Ingredients of the Microemulsions
0.1 M Cd(N0& OP n-butanol Cvclohexane
0.1 M NazMoO4 OP n-butanol Cvclohexane
8 15 10
I0
ethanol several times to eliminate excess OP. The final product was called the microemulsion precursor.
( 2 ) Synthesis of CdMoO, Crystallites The normal precursor of 500 mL was poured into a 1000 mL stainless-steel autoclave, and hydrothermally treated at 90°C for 20 h. Similarly, the above-obtained microemulsion precursor was also placed in the same autoclave with distilled water, filling up to 500 mL of capacity, and then hydrothermally treated at 90°C for 20 h. The two kinds of CdMo04 powders could be obtained by filtrating, washing, and drying after the hydrothermal process. (3) Characterization XRD analysis was carried out using an X-ray powder diffractometer (XRD, D/max, Rigaku, Tokyo, Japan) with CUKEradiation. The morphology and particle size of the as-prepared powders were observed by using a transmission electron microscope (TEM, JEM-l200EX, JEOL, Tokyo, Japan), and scanning electron microscope (SEM, JSM-840, JEOL). The roomtemperature luminescent spectra were recorded on a spectrofluorometer (PL, Fluorolog-3, Jobin Yvon, Edison, FL). 111. Results and Discussion
( I ) XRD Analysis Figure 1 shows the XRD patterns of the samples prepared at 90°C for 20 h by the hydrothermal method with the normal precursor, and microemulsion precursor, respectively. It indicates that the good crystallization of tetragonal phase CdMo04 can easily be achieved under the hydrothermal condition with the different precursors. Both the XRD patterns are well consistent with the reported data (JCPDS: 88-0182), and no other impurities can be found. On the other hand, the XRD pattern obtained when using the normal precursor is much stronger in intensity than of using the microemulsion precursor. This means that the crystallization and development of the CdMo04 phase
Fig. 2. Transmission electron microscope (TEM) and scanning electron microscope (SEM) images of the CdMo04 crystallites prepared at 90°C for 20 h by the hydrothermal process: (a) SEM, the product from the normal precursor, and (b) TEM, the product from the microemulsion precursor.
I 350
400
450
500
550
600
wavelength (nrn)
Fig. 1. X-ray diffraction patterns of the samples obtained by the hydrothermal process at 90°C for 20 h: (a) with the normal precursor, and (b) with the microemulsion precursor, respectively.
90
Fig. 3. Photoluminescence spectra of the as-prepared CdMo04 crystallites by the hydrothermal process with (a) the normal precursor, and (b) the microemulsion precursor, respectively.
Progress in Nanotechnology: Processing
seem to be sensitive to the precursor used in the hydrothermal process.
(2) Particle Size and Morphology of CdMoOd Crystallites SEM and TEM images of the CdMo04 powders obtained from the different precursors are shown in Fig. 2. The normal precursor resulted in a perfectly and homogeneously spherical crystallization of CdMo04 phase about 3-5 pm in particle size (Fig. 2(a)), while the nanometer scale of CdMo04 powders, which is much smaller than the former, could be prepared by using the microemulsion precursor, and the resultant particles were estimated at about 3&50 nm in size and of nearly homogeneous and polygonal morphology (Fig. 2(b)). Obviously, the particle size and morphology of CdMo04 crystallites strongly rely on the precursor species even though both the products have the same crystalline phase of CdMo04. Here, nano-size CdMo04 crystallites, prepared with the microemulsion precursor, should grow in a controlled manner in much smaller regions, determined by the microemulsion process, whereas the normal precursor, obtained from direct precipitation, would lead to an extensive growth of CdMo04 crystallites to larger particle sizes (Fig. 2(a)) in an uncontrolled manner. It is reasonable that the particle size and morphology of the tetragonal CdMo04 crystallites can be tailored under the hydrothermal condition by varying the precursor used.
( 3 ) PL Analysis Figure 3 presents the PL spectra of the CdMo04 samples corresponding to Fig. 2. Both the samples exhibit the same emission peak position at 390 nm in the PL spectra. Nevertheless, it is noteworthy that the emission peak of 390 nm could be obtained by a longer excitation wavelength (330 nm) in the case of nanoparticle CdMo04 than that used in the micrometer-sized ones (320 nm). Furthermore, as shown in Fig. 3, the nano-particle CdMo04 has a higher PL intensity than that of the micrometersized powder. This shows that the CdMo04 nano-particles should have better PL properties than the micrometer-sized ones. IV.
Conclusions
CdMo04 nano-particles with a good crystallization could be synthesized at 90°C for 20 h by a hydrothermal method, where a microemulsion precursor was used. But under the same hydrothermal condition, only the normal precursor afford led to micrometer-sized CdMo04 crystallites. The particle size and morphology of CdMo04 crystallites can be tailored by varying the precursor species. The improved PL properties of CdMo04 crystallites can be obtained with by decreasing the particle size to a nanometer scale.
References IN. N. Leyzerovich. K. G. Bramnik, T. Buhrmester, H. Ehrenberg, and H. Fuess, “Electrochemical Intercalation of Lithium in Ternary Metal Molybdates MMo04 (M: Cu, Zn. Ni and Fe),” J. Power Sources, 127, 76-84 (2004). ’V. B. Mikhailika, H. Krausa, D. Wahla, H. Ehrenbergb, and M. S. Mykhaylyk, “Optical and Luminescence Studies of ZnMo04 Using Vacuum Ultraviolet Synchrotron Radiation,” Nuclear Instrum. Methods Phys. Res. A , 562, 5 1 3 4 (2006).
Synthesis Methods for Powders
’V. B. Mikhailik, H. Kraus, D. Wahl, and M. S. Mykhaylyk. “Studies of Electronic Excitations in MgMo04, CaMo04 and CdMo04 Crystals Using VUV Synchrotron Radiation,” Phys. Star. Sol. ( h i , 242, R17-9 (ZOOS). 4D. Spassky, S. Ivanov, 1. Kitaeva, V. Kolobanov, V. Mikhailin, L. Ivleva, and I. Voronina, “Optical and Luminescent Properties of a Series of Molybddte Single Crystals of Scheelite Crystal Structure,” Phys. Srar. Sol. ( c ) . 2. 65-9 (ZOOS). ’V. B. Mikhailik, H. Kraus, M. Itoh, D. Iri, and M. Uchida, “Radiative Decay of Self-Trapped Excitons in CaMo04 and MgMo04 Crystals.” J. Phys. Condens. Matrer., 17, 7209-18 (2005). 6A. Karipides and D. A. Hailer, “The Crystal Structure of Tetraphenylgermanium,” Acfa CrystaNogr. B, 28, 288S92 (1972). ’R. Grasser, E. Pitt. A. Scharmann, and G. Zimmerer, “Optical Properties of CaW04 and CaMo04 Crystals in the 4 to 25 eV Region,” Phys. Srar. Sol. ( b ) , 69, 35948 (1975). ‘S. B. Mikhrin, A. N. Mishin, A. S. Potapov, P. A. Rodnyi. and A. S. Voloshinovskii, Nucl. Insrrum. Merh. A.. 486,295-7 (2002). 9D. A. Spassky, S. N. Ivanov, V. N. Kolobanov, V. V. Mikhailin. V. N. Zemskov, B. 1. Zadneprovski, and L. 1. Potkin, “Optical and Luminescent Properties of the Lead and Barium Molybdates,” Rad. Meas., 38, 607-10 (2004). “B. K. Chandrasekhar and W. B. White, “Luminescence of Single Crystal CaMo04,” Marer. Res. Bull., 25, 1 5 1 H (1990). “R. Grasser, E. Pitt, A. Scharmann, and G . Zimmerer, “Optical Properties of CaW04 and CaMo04 Crystals in the 4 to 25 eV Region,” Phys. Star. Sol. ( h ) , 69, 35758 (1975). L. F. Johnson, G. D. Boyd, K. Nassau, and R. R. Soden, “Continuous O p eration of a Solid-state Optical Maser,” Phys. Rev., 126, 14069 (1962). ”A. N. Belsky, V. V. Mikhailin, A. N. Vasil’ev, I. Dafinei. P. Lecoq, C. Pedrini, P. Chevallier, P. Dhez, and P. Martin, “Fast Luminescence of Undoped PbW04 Cr stal,” Chem. Phys. Lett., 243, 552-8 (1995). ‘P. A. Christian, J. N. Carides, F. J. DiSalvo, and J . V. Waszczak. J. Elecfrochem. Soc.. 127 [Ill2315 (1980). I5J. 0. Besenhard, J. Heydecke, E. Wudy, H. P. Fritz, and W. Foag, “Characteristics of Molybdenum Oxide and Chromium Oxide Cathodes in Primary and Secondary Organic Electrolyte Lithium Batteries. Part II. Transport Properties,” Solid Srare Ionics, 8. 61-71 (1983). ‘q.Tsumura and M. Inagaki, “Lithium Insertion/Extraction Reaction on Cr stalline MOO,,” Solid State Ionics, 104, 183-9 (1997). y7A.Yu. N. Kumagaj, Z. Liu, and I. Lee, “Preparation of Sodium Molybdenum Oxides by a Solution Technique and Their Electrochemical Performance in Lithium Intercalation,” Solid Srare lonics, 106, 11-18 (1998). “R. H. Sanchez, L. Trevino, A. F. Fuentes, A. Martinez-de la Cruz, and L. M. Torres-Martinez, “Electrochemical Lithium Insertion in Two Polymorphs of a Reduced Molybdenum Oxide,” J . Solid Srare Elecfrochem., 4, 21CL-5 (2000). I9P. K. Pandey, N. S. Bhave, and R. B. Kharat, “Preparation and Characterization of Spray Deposited NiMoO, Thin Films for Photovoltaic Electrochemical Studies,” Mater. Res. Bull., 41, 1160-9 (2006). 2oJ. L. Brito and A. L. Barhosa, “Effect of Phase Composition of the Oxidic Precursor on the HDS Activity of the Sulfided Molybdates of Fe(ll). Co(11). and Ni(ll),” J. Catal., 171, 467-75 (1997). ’IN. Serpone, P. Marathamuthu, P. Pichat, E. Pelizzetti. and H. Hidaka, “Exploiting the Interparticle Electron Transfer Process in the Photocatalysed Oxidation of Phenol, 2-Chlorophenol and Pentachlorophenol: Chemical Evidence for Electron and Hole Transfer Between Coupled Semiconductors,” J. Phororhem. Phorobiol., 85, 247-55 (1995). 22 H. Fujii. M. Ohtaki, and K. Eguchi, “Preparation and Photocatalytic Activities of a Semiconductor Composite of CdS Embedded in a TiO’ Gel as a Stable Oxide Semiconducting Matrix,” J. Mu/. Catal. A , , 129, 61-68 (1998). 23A. E. Raevskaya, A. L. Stroyuk, and S. Y. Kuchmiy, “Preparation of Colloidal CdSe and CdSjCdSe Nanoparticles from Sodium Selenosulfate in Aqueous Polymers Solutions,” J. Colloid Inrerfuce Sci. 302, 13341 (2006). 24 I. H. Ryu, B. G. Choi, J.-W. Yoon, K. B. Shim, K. Machi. and K. Hamada, “Synthesis of CaMo04 Nanoparticles by Pulsed Laser Ablationin Deionized Water and Optical Properties,” J. Lumin., 124, 67-70 (2006). 25J. H. Ryu, J.-W. Yoon, C. S. Lim, and K. B. Shim, “Microwave-Assisted Synthesis of Barium Molybdate by a Citrate Complex Method and Oriented A gregation,” Marer. Res. Ball., 40, 146&76 (2005). k . Mazzocchia, C. Aboumrad, C. Diagne, E. Tempesti, J. M. Herrmann. and G. Thomas. “On the NiMo04 Oxidative Dehydrogenation of Propane tn Propene: Some Physical Correlation with the Catalytic Activity,” Caral. Lefr., 10, 181-91 I).
X. Huang, J. Ma, P. Wu, Y. Hu, J. Dai, Z. Zhu, H. Chen, and H. Wang,
“Hydrothermal Synthesis of LiCoPO4 Cathode Materials for Rechargeable Lithium Ion Batteries,” Marer. Lert., 59, 578-82. 0
91
This Page Intentionally Left Blank
Chromium-Doped Forsterite Nanoparticle Synthesis by Flame Spray Pyrolysis Takao Tanit and Shu Saeki Inorganic Materials Laboratory, Toyota Central R&D Laboratories Inc., Aichi 480-1 192, Japan
Takenobu Suzuki and Yasutake Ohishi Department of Future Industry-Oriented Basic Science and Materials, Toyota Technological Institute, 468-851 1 Nagoya, Japan
Synthesis of chromium-doped forsterite (Mg2Si04:Cr) nanoparticles by flame spray pyrolysis (FSP) was investigated. The morphology, crystalline phase, and photoluminescence of the products were evaluated. Crystalline Mg2Si04:Cr nanoparticles of several 10 nm in diameter were obtained, although a small amount of the submicrometer-sized particles and the unreacted MgO phase existed. The roduct powder showed electron-spin resonance signals from C& and photoluminescence typical for Cr4' in Mg2SiO4, sufgesting that a part of the Cr4' ions were incorporated into Si4 sites by FSP. On the other hand, the effects of excess Si02addition on the structural and optical characteristics of Mg2Si04:Cr were examined. Addition of excess Si02up to 20 mol% did not influence these characteristics of the products. Further addition of excess Si02 (60-100 mol%) enhanced the formation of the amorphous phase and resulted in the emission from Cr3+ in the amorphous phase in addition to an emission from Cr" in Mg2Si04.
Flame spray pyrolysis (FSP)'- is an elegant process to prepare metal oxide nanoparticles in one step by spraying and combusting precursor solutions where metal salts are dissolved in organic solvents. Product particles are formed by evaporation, condensation, and oxidation of the metal species (e.g., atoms or oxide clusters) in the gas phase. High-flame temperatures and short residence time in the hot zone by FSP can result in highly crystalline nanoparticles, which are appropriate for dispersants of the nanoparticle-polymer (or glass) composites. However, it is not clear whether the short processing time is enough for crystallization of Mg2Si04and/or incorporation of cr4+ into si4+ sites or not. This paper investigates the potential to apply the FSP process to Mg2Si04:Cr nanoparticle synthesis. The structural and optical characteristics of the products are evaluated. In addition, the effects of excess Si02 addition on these properties are examined because it may control Mg2Si04:Cr crystal size as seen in ZnO/SiO2 synthesis by FSP" where the ZnO crystal size was varied by addition of SiOz.
I. Introduction
F
ocus is placed on new-type optical waveguides where functional nanoparticles are dispersed in glass or polymer. Yoshida et al.' prepared Ti02 nanoparticles by a reversed micelle method and dispersed them in polyimide to control the refractive index of the polyimide waveguide. A waveguide where semiconductor nano-crystals are dispersed in quartz glass is proposed for an optical signal amplifier with controllable amplification wavelength by selecting the nano-crystals.2 Chromium-doped forsterite (Mg2Si04:Cr) is a lasting material whose activity was first reported in 1988,3 and Mg2Si04:Cr single crystals are applied commercially to tunable near-infrared (IR) lasers, whereas, if Mg2Si04:Cr nanoparticle-polymer (or glass) composites are obtained, they may be advantageous in terms of flexibility of shape and a low production cost compared with single crystals and can be applied widely to laser, amplifier, and other optical devices. Sol-gel synthesis is a well-known route for preparation of Mg2Si04:Crnanoparticles.4-6Although the sol-gel process seems appropriate to synthesize homogeneous nanoparticles, it primarily needs expensive metal alkoxide precursors and complicated steps for synthesis, increasing a process cost. In addition, its low process temperature can result in low crystallinity and thus low optical properties of the product. J. Ballat-ntrihuting
editor
Manuscnpt No. 21560. Received March 7, 2006; approved October 30. 2006. This work is supported in part by MEXT, the Private University High-Tech Research Center Program (2lX&20IO) and JSPS.KAKENH1 (17560322). 'Author to whom correspondence should be addressed. e-mail: tdnitakd@mosk. tytlahs.co.jp
Synthesis Methods for Powders
11. Experimental Procedure
A stoichiometric composition of MgzSi04:Cr is Mg2(Sio.w. Cro.01)04in this study. Synthesis was carried out for five compositions with various amounts of excess Si02 addition (X= 0, 0.1, 0.2, 0.6, and 1.0 in molar ratio). All powders are labeled by the amounts of excess Si02 addition ( X ) . Here, X = 0.1 corresponds to the composition of Mg2(Si0.99Cro,ol)04+0. 1 SOz. Magnesium acetate tetrahydrate (Wako, S grade), hexamethyldisiloxane (HMDSO, Wako, S grade) and chromium acetate (Wako, chemical grade) were used as Mg, Si, and Cr sources, respectively. These precursors were dissolved in a mixture of methanol (Wako, S grade) and I-butanol (Wako, S grade) with a volume ratio of methanol/butanol = 9/1, resulting in precursor solutions of 1.O mol/L in total metal ion concentration. Synthesis was carried out using a spray flame reactor." A pump supplied the precursor solution at a feed rate of 12 cm'/ min to the nozzle where the solution was dispersed into droplets by oxygen at 11 L/min. The droplets were ignited by supporting methane/oxygen flames surrounding the nozzle. The total flow rates of methane and oxygen were 2 L/min each. Air at 50 L/min was supplied through a porous metal plate ring for excess oxidant. The product particles were collected on a Gore-Tex filter with the aid of a vacuum pump. The particle morphology was observed by transmission electron microscopy (TEM; Nihon Denshi, JEM2000EX, 200 kV, Akishima, Japan). The compositional analysis was carried out by energy-dispersive X-ray spectroscopy (EDS; Hitachi, S3600N, Tokyo, Japan). The specific surface area (SSA) was measured by nitrogen adsorption (Micro Data, 4232). The crystalline phase of the product was identified by X-ray diffraction (XRD: Rigaku, RINT-TTR, 50 kV, 300 mA, Tokyo, Japan) at 93
0 Forsterite Periciase
20
30
40
50
60
70
2 Wdeg. (CuKa) Fig. 3. X-ray diffraction spectrum of the stoichiometric Mg2Si04:Cr powder. Forsterite (Mg2Si04)and periclase (MgO) as well as an amorphous phase (a broad peak at 20 = 20'-40") are observed. Fig. 1. Transmission electron microscopic image of the stoichiometric Mg2SiO4:Cr powder. Polyhedral aggregates of nanoparticles and a small amount of submicrometer-sized spherical particles are observed.
20 (CuKa) = 20"-70" and scan speed = Z"/min. The bonding state of metal ions was evaluated by IR spectroscopy (Nicolet, Avantar 360, Waltham, MA). The state of Cr ions was examined by electron spin resonance (ESR: Bruker, ESP300E, 100 mW, Tsukuba, Japan). The measurement was carried out at 9.6 GHz (X-band) in the field range of &I000 G with modulation of 100 kHz at 25°C. The near-IR photoluminescence (PL) spectrum of the product was measured at I-nm intervals in the Wavelength range of 820-1700 nm. The sample was excited by an 800-nm laser beam which was generated from a CW Tisapphire laser (Coherent, 890, Santa Clara, CA) and chopped mechanically at a frequency of 97 Hz. The emission from the sample was dispersed by a single monochromator (blaze wavelength: 1.0 pm, grating: 600 grooves/mm, resolution: 3 nm) and then detected by an InGaAs photo detector (Hamamatsu Photonics, H58S2-11, Hamamatsu, Japan) and a lock-in amplifier (NF, L15640,Yokohama, Japan).
In. Results and Discussion (I)
Stoichiometric Mg2SiO&r Synthesis ( A ) Morphology: Figure 1 shows a TEM image of the stoichiometric Mg2Si04:Cr powder (A' = 0). Polyhedral aggregates of nanoparticles, observed typically in flame-made powders,13 were obtained, while a small amount of submicro-
0
0.5
1 EnergylkeV
1.5
2
Fig.2. Energy-dispersive X-ray spectra of (a) the aggregates of the nanoparticles and (b) the large particles in the stoichiometric Mg2Si04:Crpowder. The MgjSi ratio is similar for both particles, indicating no segregation of the Zn and Si species.
94
meter-sized spherical particles (large particles) also existed. The nanoparticle diameter observed by TEM was about 20-50 nm, which was roughly consistent with the BET-equivalent average primary particle diameter (dBET:32 nm) estimated from the measured SSA (58 m*/g) and solid density (MgzSi04: 3.21 x lo' kg/m3)14assuming solid spherical Mg2SiO4particles. Such inhomogeneous morphologies have been obtained in the case of ZnO/Si02 synthesis by FSP usin zinc acetate and Si02 sol as Zn and Si precursors, respectiveIy.e2 In that case, the Si precursor did not evaporate in the flame and the particle was formed in the li uid phase in the dispersed droplets, as seen in spray pyrolysis?' resulting in gas phase-made aggregates of ZnO nanoparticles and liquid phase-made large Si02 particles, whose composition was confirmed by EDS. On the other hand, the EDS spectra (spot size = about 100 nm) of the aggregates of the nanoparticles (Fig. 2(a)) and the large particles (Fig. 2(b)) showed that the Mg/Si ratio was similar for both particles in this study, indicating no segregation of the Zn and Si species. Different from the previous study, the large particles may be formed irregularly because of the inhomogeneous thermal history of the aerosol in the spray flame. ( B ) Crystalline Structure: Figure 3 shows an XRD pattern of X=O. The main product was forsterite (Mg2Si04: PDF34-0189), suggesting that crystallization of Mg2Si04 primarily takes place during FSP. On the other hand, periclase (MgO: PDF4S-0946) and an amorphous (a broad peak at
12uo
1100
1000
900
800
700
Wavenumberlcm-'
Fig.4. Infrared (IR) spectrum of the stoichiometric Mg2Si04:Cr powder. The IR spectra of Mg2Si04, MgO, and SiO2 are shown also for comparison. The absorptions at 870 and 1000 cm-' typical for Mg2Si04 are observed. Progress in Nanotechnology: Processing
800
900
lo00 1100 1200 1300 1400 1500 Wavelengthlnm
Fig. 5. Normalized photoluminescence spectrum of the stoichiometric Mg2Si04:Cr powder. Emission at 1200 nm typical for Mg2Si04:Crare observed, suggesting that Cr ions substitute not Mg2+ but Si4+ sites in Mg2Si04as Cr4'.
2e=20°40") phase were also observed. In the FSP process, MgO and Si02nuclei can be formed independently from the gas phase during cooling because of different boiling points (MgO: 3600"C, Si02: 295O0C)," followed by crystallization of MgO with Si02 forming MgzSi04. A part of MgO and Si02 nuclei may be cooled down before crystallization of Mg2Si04because of the inhomogeneous thermal history of the aerosol. Figure 4 shows the IR spectrum of X = 0. The IR spectra of Mg2Si04, MgO, and Si02 are also shown in the figure for comparison. Absorptions at 870 and 1000 cm-l typical for Mg2Si04 (SiLO-Mg bonding)I6 were observed, whereas an absorption at 1050 cm-l typical for S O 2 (Si-0-Si bonding)16 was not seen, indicating that the Si4' and Mg2+ ions are linked or influenced by each other. The result suggests that the amorphous phase (Fig. 3: XRD) is not Si02 but magnesium silicates. (C) Emission: Figure 5 shows the normalized PL spectra of X = 0. Emission at 1200 nm typical for Mg2Si04:Cr4was observed, while there were no emissions at low wavelengths (e.g., 70&1100 nm), which have been seen for Cr3+ substituted with Mg2+ in Mg2Si02." or MgSi03.1s The result suggests that a part of the Cr ions substitute not Mg2' but Si4+ sites in Mg2Si04 as Cr4' by FSP. The ESR spectrum of X = 0 showed an ESR signal at 800 G (Fig. 6) although it was very weak and noisy. Budil et al." studied ESR of a Mg2Si04:Cr single crystal and reported that Cr4' showed broad ESR signals at 600 or 800 G, whereas Cr3+ showed sharp ESR signals at 15W3200 G for X-band measurement at room temperature. Although it is difficult to compare directly the ESR spectra of the single crystal with those of the powder sample, the ESR signal in this study
Fig. 7. Transmission electron microscopic images of the product powders with various amounts of excess SiOz addition (X) to
[email protected])04.Addition of excess SiOz primarily does not influence the particle morphology.
may have originated from Cr4' in Mg2Si04particles because of its similar position and shape to those of Cr4' in the single crystal, su gesting that a part of the Cr4' ions were incorporated into Si". sites by FSP, qualitatively consistent with the PL result.
( 2 ) Addition of Excess Silica Figure 7 shows TEM images for X = 0.1, 0.2, 0.6, and 1.0. As seen for X = 0 (Fig. I), polyhedral aggregates of nanoparticles
x=o.1
0 Forsterite 0 Periclase
0
. : x .-I 111
X=0.6
A
20
0
200
400
600
800
1000
Magnetic Field (G) Fig. 6. The electron spin resonance (ESR) spectrum of X = 0. The ESR signal was seen at 800 G although it was very weak and noisy.
Synthesis Methods for Powders
30
40 50 2 Wdeg. (CuKa)
x=1.o
-
60
L
70
Fig.8. X-ray diffraction spectra of the product powders with various amounts of excess SiOz addition (X)to Mg2(Si0.99Cr~ 0 1 ) 0 4 . Compared with that for the stoichiometric MgzSi04:Cr powder (Fig. 3), the peak heights from Forsterite (Mg2SiO4)and periclase (MgO) lower whereas the broad peak is intensified as the amount of excess SiOz increases. While, no silicon-rich crystalline phases such as enstatite (MgSi03) are seen.
95
sized particles as well as an unreacted MgO phase existed. The product powder showed electron spin resonance signals from Cr4’ and PL typical for Cr4’ in Mg2Si04,suggesting that a part of the Cr4’ ions were incorporated into Si4+sites by FSP. On the other hand, addition of excess Si02 up to 20 mol% (X=O.l and 0.2) did not influence the structural and optical characteristics of the products. On the other hand, further addition of excess Si02 ( X = 0.6 and 1.0) enhanced a formation of the amorphous phase and resulted in emission from Cr3+ in the amorphous phase in addition to emission from Cr4+ in Mg2Si04,although the particle morphology did not change so much. The crystal size of Mg2Si04 was not controlled by addition of excess Si02. Future study will be focused on evaluation and improvement of emission efficiency by optimizing the Cr-doping amount for X = 0. Fig. 9. Normalized photoluminescence spectra of the product powders with various amounts of excess SiOzaddition ( X ) to Mg2(Si0.9&r0.01)04. For X=O.1 and 0.2, the emission spectra are similar to that for the stoichiometric Mg2Si04:Crpowder (Fig. 5), whereas further addition of excess Si02 (X= 0.6 and 1.O) results the emission from Cr3+ (broad emission at 850-1 100 nm) in addition to the emission from Cr4’.
and a small amount of submicrometer-sized spherical particles were also observed for all powders. In FSP, particle morphology is influenced mainly by the kinds of precursors and solvents used as well as the combustion conditions (e.g., O2 flow rate), and these are the same for all powder syntheses in this study, which could be the reason why addition of excess SiOz did not influence the particle morphology. Figure 8 shows XRD patterns for X = 0. I, 0.2, 0.6, and 1.0. Compared with that for X = 0, the peak heights from Mg2Si04 and MgO decreased, whereas the broad peak was intensified as the amount of excess SiOz increased. On the other hand, no silicon-rich crystalline phases such as enstatite (MgSi03: PDF190768) were seen for all powders. The phase diagram2” shows that the melting point of Mg2Si04 (1888°C) is higher than that of MgSi03 (1 557°C) and that the phase changes “liquid -+ Mg2Si04+liquid -+ Mg2Si04+MgSi03” during cooling in the composition range of Mg/Si = 2/1 to 1/1, indicating that the Mg2Si04formation is preferred during cooling. A rapid cooling rate of the FSP process can result the formation of the “Mg2Si04+amorphous” phase from the “Mg2Si04+liquid” phase without crystallization of MgSi03 regardless of the amounts of excess SO2. The IR spectra showed Si-O-Mg absorptions at 870 and 1000 cm-l as seen for X = 0, although the peak heights decreased as the excess SiOz increased, suggesting that the amorphous phase is also magnesium silicates, while the crystal sizes (dXm) of Mg2Si04calculated from the full width at half maximum of the (131) peak were 31,40,37,43, and 28 nm for X = 0, 0.1, 0.2, 0.6, and 1.0, respectively. Considering experimental errors, the dXRD of Mg2Si04 did not change so much for all powders, suggesting that the crystal size is not controlled by addition of excess Si02 in this study, different from ZnO/Si02 synthesis.’I Figure 9 shows normalized PL spectra for X = 0.1, 0.2, 0.6, and 1.0. For X=O.l and 0.2, the PL spectra were primarily similar to that for X = 0. On the other hand, broad emission at 850-1100 nm was seen in addition to the emission at 1200 nm for X = 0.6, suggesting existence of Cr3+ ions substituting Mg2+ sites as discussed before (Section III(c)). The broad peak was intense for X = 1.0 where the amorphous phase increased, suggesting that C?’ ions are located mainly in the amorphous magnesium silicate.
IV. Summary The MgzSi04:Cr nanoparticles of several 10 nm in diameter were prepared by FSP, although a small amount of submicrometer-
96
Acknowledgments The authors are grateful to Dr. H. Nozaki (Toyota CRDL) for ESR measurement .
References ‘M. Yoshidd, M. Ldl, N. D. Kumar, and P. N. Prasad, “TiOZ Nano-ParticleDispersed Polyimide Composite Optical Waveguide Materials Through Reverse Micelles,” J. Muter. Sci.,32, 4047-51 (1997). ’Japanese Patent, Kokai Number 199664894. ’V. Petricevic, S. K. Gayen, R. R. Alfano, K. Yamagishi, H. Anzai, and Y. Yamaguchi, “Laser Action in Chromium-Doped Forsterite,” Appl. Phys. Lett., 52, 104lL2 (1988). 4D. G . Park, J. M. Burlitch, R. F. Geray. R. Dieckmann. D. B. Barber, and C. R. Pollock. ‘ % G e l Synthesis of Chromium-Doped Forsterite,” Chem. Muter., 5, 518-24 (1993). ’P. S. Devi, H. D. Gafney. V. Petricevic, and R. R. Alfano, “Synthesis and Spectroscopic Properties of Cr4+ Doped Sol-Gels;’ J. Nan-Cryst. Solids,203, 78x3 f 1996). ~~
6‘ P. S. ’Devi, H. D. Gafney, V. Petricevic, R. R. Alfano, D. He, and K. E. Miyano. “Sol-Gel Svntheses and Soectroscooic Characterization of ChromiumDoped Silicates a i d Germanates,’: Chem. haw.,12, 1378-85 (2000). ’M. Sokolowski, A. Sokolowska, A. Michalski. and B. Gokieli, “The “lnFlame-Reaction’’ Method for A1203Aerosol Formation,’’ J. Aerosol. Sci., 8.2 19-
30 (1977).
‘R. M. Laine, T. Hinklin, G. Williams, and S. C. Rand, “Low-Cost Nanopowders for Phosphor and Laser Applications by Flame Spray Pyrolysis”; pp. 5CKLlO in Metustuhle. Mechanically Alloyed and Nunocrystalline Materiuls, PIS 1 und 2 , Materials Science Forum 343-3, Edited by J. Eckert. H. Schlorh, and L. Schultz. Trans Tech Publications Ltd., Zurich Uetikon. Switzerland, 2000. 9L. Madler, H. K. Kammler, R. Muller, and S. E. Pratsinis, “Controlled Synthesis of Nanostructured Particles by Flame Spray Pyrolysis,” J. Aerosol. Sci., 33, 369-89 (2002). ”A. U. Limaye and J. J. Helble, “Morphological Control of Zirconia Nanoparticles Through Combustion Aerosol Synthesis,” J. Am. Ceram. Sac., 85, 112732 (2002).
“T. Tani. L. Midler. and S. E. Pratsinis, “Zinc Oxide/Silica Composite Nanoparticle Synthesis by Flame Spray Pyrolysis,’’ J. Muter. Sci., 37, 4627-32
(2002).
”T. Tani, K. Takatori, and S. E. Pratsinis, “Evolution of the Morphology of Zinc Oxide/Silica Particles Made by Spray Combustion,” J. Am. Ceram. Soc., 87,
365-70 (2004).
”S. E. Pratsinis, “Flame Aerosol Synthesis of Ceramic Powders.” Prog. Energy Comhu7t. Sci.. 24. 197-219 (1998). I4D. R. Lide, CRC Hundbook of Chemistry und Physics, 81st edition, CRC Press, Boca Raton, FL, 2000. ”G. L. Messing, S. C. Zhang, and G. V. Jayanthi, “Ceramic Powder Synthesis by Spray-Pyrolysis,” J. Am. Cerum. Soc., 76, 2707-26 (1993). 16R. A. Nyquist and R. 0. Kagel, lnfrured Spectra oJ lnorgunic Componud,v. Academic Press Inc., Orlando, FL, 1971. ”R. Moncorge, G. Cormier, D. J. Simkin, and J. A. Capobianco, “Fluorescence Analysis of Chromium-Doped Forsterite (Mg2Si04),”IEEE J. Quantum Electron.. 27, 114-20(1991). “R. Moncorge, M. Bettinelli, Y. Guyot, E. Cavalli, J. A. Capobianco, and S. Girard, “Luminescence of Ni[2]+ and Cr’+ Centres in MgSiO, Enstatite Crystals,” J. Phys.-Condes. Mufrer, 11, 683141 (1999). I9D. E. Budil, D. G . Park, J. M. Burlitch, R. F. Geray, R. Dieckmann, and J. H. Freed, “9.6-Ghz and 34-Ghz Electron-Paramagnetic-Resonance Studies Chromium-Doped Forsterite,” J. Chem. Phys., 101. 353848 (1994). ’OR. S. Roth, Phase Equilibria Diugrams. Vol. 111, Fig. 10369. The American Ceramic Society, Westerville, OH, 2001.
Progress in Nanotechnology: Processing
Formation of AI2O3-TiC Composite Nano-Particles Synthesized from Carbon-Coated Precursors Hisashi Kaga*’+?$ and Rasit Koc* Department of Mechanical Engineering and Energy Processes, Southern Illinois University at Carbondale, Carbondale, Illinois 62901
The formation of nano-sized alumina-titanium carbide ( A 1 2 0 ~ TIC) composite powders from a carbon-coated titanium dioxide-aluminum (TiOIAI) mixture was investigated. The carbon-coated TiOIAl mixture altered the mechanism of the reaction, compared with standard mixtures, to produce highquality nano-sized AI2O3-TiC powders. Data synthesized from intermediate temperatures indicate that these products form via Ti203 and A13Ti. TEM images of the A120TTiC powders showed fine size (50-100 nm), narrow size distribution, and lack of agglomeration. DSC data for the carbon-coated TiOl A1 mixture showed a single endothermic and four successive weak exothermic reactions as the carbon coating moderated the heat release during the reaction.
A
I. Introduction
ceramic materials are used in a great variety of applications. Among these advanced ceramics, aluminatitanium carbide (AI2O3-TiC) composites have been used as cutting tools and magnetic head slider substrates for a hard disk drive. This application is aided by the combination of hightemperature strength, high wear resistance, and good corrosion resistance.’-’ For these applications as with other ceramic materials, A1203-TiC powders with a homogenous chemical composition, fine particle size, narrow particle size distribution, and loose agglomeration are required for the manufacture of the final components.8 Several different methods exist for synthesizing AI2O3-TiC composite powders, each resulting in various particle size distributions, morpholog , state of agglomeration, Y*9 Efficient production of chemical purity, and stoi~hiometry.~high-quality AI2O3-TiC powder begins with low-cost, high-purity raw materials. Titanium dioxide (Ti02), aluminum (Al), and carbon (C) powders are commonly used for an aluminothermic reaction to produce AI2O3-TiC powders in industry. The aluminothermic reaction produces large amounts of powder from inexpensive precursor materials. However, there is presently no commercial process for the production of submicrometer AI2O3-TiC powders. The arrangement and particle sizes of Ti02, Al, and C in the mixture must be controlled in order to produce high-quality, low-cost AI2O3-TiC composite powders. The carbon coating of Ti02 or a TiO2-AI mixture can be a very effective way to control precisely the reaction path and rate of the aluminothermic reaction for the production of superior AI2O3-TiC composite powders.
The carbon coati:f,Rrocess was developed and patented by Koc and Glatzmaier for the production of TIC and silicon carbide. The advantage of the coating process is the intimate contact between reactants. This intimate contact allows reactions to occur closer to the thermodynamic predictions than conventional mixtures. The industrial production of AI2O3-TiC powders generally occurs as per reaction. 3Ti02 + 4A1+ 3C -+ 2A1203
+ 3TiC
This process is widely used because of the inexpensive starting materials, which can easily be obtained, and the reaction does not require high temperatures (1200°C).12-14Synthesis via reaction (1) is generally considered to be divided into two stages, shown in the following reactions”-”:
DVANCED
T. ksmann--wntrihuting editor
Manuscript No. 21949. Received June 28. 2006; approved September 18,2006. Research for this paper was performed at Southern Illinois University at Carbondale, sponsored by the U S . Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Industrial Technologies, Advanced Industrial Materials Program, under contract DE-AC05-960R22464 with Lockheed Martin Energy Research Corporation/UT-Batelk. ‘Member, American Ceramics Society. ‘Author to whom correspondence should be addressed. E-mail:
[email protected] ‘Present address: National Institute of Advanced Industrial Science and Technology (AIST) 2266-98 Anagahora, Shimo-shidami, Moriyama-ku, Nagoya 463-8560, Japan.
Synthesis Methods for Powders
Ti
+ C + TIC
(3)
In reaction (2), molten aluminum reduces the surrounding solid Ti02 particles to titanium. The heat released in reaction (2) provides the required energy for reaction (3), in which the reduced titanium reacts with free C to form TIC. Although the formation of AI2O3-TiC composites generally follows the above sequence, the effect of C-coated starting materials has not been investigated. This paper describes the formation of AI2O3-TiC from Ccoated precursor materials. The effectiveness of the C-coated precursor was investigated in terms of the formation of A1203TIC compared with standard mixtures. The C-coated precursor provides a low-cost method to produce high-purity nano-sized AI2O3-TiC composite powders. 11. Experimental Procedure
(1) Starting Materials The starting Ti02 powder (P-25, Degussa Corp., Ridgefield Park, NJ) has a high specific surface area (51 m2/g), and Xray diffraction (XRD) showed that both anatase and rutile phases were present. Propylene gas (C3H6) was used as the C source for the C coating process. Carbon black (Monarch 880, Cabot, Waltham, MA) and Al (41000, Alfa Aesar, Ward Hill, MA) powders with specific surface areas of 220 and 1 m2/g, respectively, were used as starting powders for mixtures. Three different starting mixtures were used to prepare three precursors: fully coated, partially coated, and fully mixed. Figure 1 shows schematic drawings of the three precursors. ( A ) Preparation of C-Coated Ti021AI Mixtures (Fully Coated): The fully coated precursor was prepared by coating a mixture of Ti02 and Al. First, Ti02 and Al powders at the desired weight percent ratios were thoroughly mixed in ethyl alcohol by an attritor (HD01-A100-9, Union Process, Akron, OH) at 140 r.p.m. for 4 h in air. The ratio of the powders to ethyl 97
Fig. 1. Schematic illustrations of the three precursors investigated in this research.
alcohol was 2.1 by volume. After drying in a vacuum furnace (1400E, VWR Scientific, Philadelphia, PA) for 24 h, approximately 100 g of the Ti02/AI mixture was placed in a stainlesssteel vessel (10 cm inner diameter x 35 cm length) and subjected to the carbon coating process as explained el~ewhere.'*~'~ The steel chamber was purged with argon and evacuated. The chamber was then filled with C3H6 gas to a pressure of 275 kPa and heated to 550"C, the temperature at which C3H6decomposes to C, HZ,and CH4. After reaching 550"C, the system (powder, gas, and steel vessel) was allowed to equilibrate for 60 min. The hydrocarbon gas was then released to atmospheric pressure and refilled to 275 kPa. C, released from the hydrocarbon, coated the surfaces of the Ti02/A1mixture. After approximately 5 min, the remaining CH4 gas was released and the vessel was refilled with fresh hydrocarbon gas. This process was repeated until approximately 9.3 wt% of C was deposited, which required about 20 cycles (each cycle was 10 min). The weight percentage of the C in the precursor was determined using themogravimetric analysis. As the source of C is a hydrocarbon gas, impurities commonly found in C black, e.g., iron, are not introduced into the system. As the starting powders also have a high surface area, these precursors should result in high-purity, high-surface-area AI2O3-TiC powders. ( B ) Preparation of C-Coated Ti02 and A1 Mixtures (Partially Coated): The partially coated precursor was prepared by coating Ti02 powders as above, and then mixing this coated powder with Al powder. The C-coated Ti02 powders were mixed with Al in the attritor for 1 h in air. They were not mixed with ethyl alcohol to prevent removal of the uniform C coating of TiO2, which can occur with wet mixing. The partially coated precursor was prepared in order to prove the effectiveness of the C coating. If the Ti02 is uniformly coated by C, the C coating will act as a barrier and inhibit the reaction between Ti02 and Al, which is the first step of the aluminothermic reaction, and prevent the overall reaction of Eq. (1) from occurring. ( C ) Preparation of Ti02,Al, and C Black Mixtures (Fully Mixed): The fully mixed precursor was prepared using a method similar to that used in industry, with no C coating attempted. The appropriate weight percent ratios of TiO2, Al, and C black were mixed in ethyl alcohol in the attritor for 4 h in air. After drying, the specific surface area of the mixed precursor was measured to be 50 m2/g. The mixed precursor was intended for comparison with the coated precursors.
(2) Synthesis of A120J-TiC Powders All three precursors were reacted in a tube furnace (CTF 17/75/ 300, Carbolite, Sheffield, U.K.) with an inner diameter of 7 cm; 10 g of a given precursor was placed in a graphite crucible (6.5 cm inner diameter x 20 cm long) inside the furnace. The AlzO3T i c powders were then produced by promoting the aluminothermic reaction. The fully coated precursors were heated for 2 h at 600", 800", IOOO", 1200", and 1500°C to investigate the reaction mechanisms. The partially coated and mixed precursors were synthesized at 1000", 1200", and 1500°C for 2 h. All samples were heated in a flowing argon atmosphere; the heating and cooling rates were 4"C/min. 98
(3) Characterization of Synthesized A120TTiC Powders XRD (DMAX-B, Rigaku, Tokyo, Japan) of powders was carried out on a lab X-ray source using CuKa radiation for phase identification of samples. Each powder sample was scanned from 28 = 20"-80" at a scanning speed of 4"/min. The specific surface areas of the powders were measured using a BET gas adsorption surface area analyzer (Gemini 2360, Micromeritics, Norcross, GA) with 0.34.4 g of each sample. Samples were degassed at 175" for 4 h before the BET multipoint specific surface area was measured based on the amounts of gases absorbed with the corresponding pressures. TEM (FA7100, Hitachi Inc., Tokyo, Japan) imaging was used to examine the C coating on fully and partially coated precursors, as well as to examine the morphology of reacted powders. Samples were prepared and imaged on holey C formvar grids. DSC was utilized to describe the formation mechanism of AI2O3-TiC powders. Pellets of each precursor were prepared by single action, uniaxial pressing in a hardened steel die (inner diameter = 31 mm). The pellets were broken to accommodate the size restriction of the DSC apparatus (Labsys TG-DTA/DSC, Setaram Corp., Caluire, France). These samples were heated to 1200°C in a Pt crucible at 20"C/ min in flowing argon gas. Each experiment was repeated to ensure consistent results. III. Results and Discussion
( I ) C Coating Precursors A very uniform coating resulted from the carbon coating process. Figure 2(A) shows the TEM micrograph of the fully coated precursor. It shows two large particles in the lower region, which are Al, while the smaller dark particles are Ti02. Both Al and Ti02 particles are uniformly surrounded by the slightly lighter area, which is the C coating. BET specific surface areas of the uncoated Ti02/AI mixture and Ti02were measured to be 35 and 51 m2/g, respectively. After being coated with 9.3 wt% C, the surface areas were 31 and 45 m2/g, respectively; the reduced specific surface area indicates an increase in particle size. The TEM and BET results indicate that a uniform C coating can be obtained on the Ti02/AI mixture by using a hydrocarbon gas. Figure 2(B) shows the TEM micrograph of the partially coated precursor, and a higher magnification on the coated region is also presented. This image was obtained before Al was mixed with C-coated Ti02, and All Ti02particles were uniformly coated by C. Figure 2(C) shows the TEM micrograph of the fully mixed precursor. The dark particles indicate Al (large) and TiOz (small), as indicated in the image. The lighter particles in the upper middle portion indicate C black. Unlike fully and partially coated precursors, the mixed precursor has very little contact area between the C and TiO2 or Al. ( 2 ) Results of Synthesis ( A ) Fully Coated Precursor: Figure 3 shows the XRD patterns of the fully coated precursor and its products after synthesis at various temperatures. The diffraction pattern for the precursor exhibits peaks corresponding to Ti02 and Al. The sample synthesized at 600°C shows the same XRD pattern as the precursor because the aluminothermic reaction does not take place at this temperature. However, at 800"C, above the melting temperature of Al (660"C), the diffraction pattern shows intermediate phases, such as AI3Ti and TizO3, as well as the desired final product of AI2O3.The initial reaction between Al and Ti02 leads to the formation of A1203,A13Ti, and Ti203according to the following reaction: Eq. (4). 3TiO2 + 5AI + A1203
+ A13Ti + Ti203
(4)
The two intermediate phases Ti203 and A13Ti are considered to form A1203 and TIC in the presence of C according to the following equation: Eq. (5).
I STi203 + A13Ti + 4C -+ 1.5A1203 + TIC
(5)
Progress in Nanotechnology: Processing
Fig.2. TEM micrograph of the three precursors. (A) Fully coated precursor; The carbon coating is visible as a translucent fringe surrounding the A1 and TiOz particles. (B) Partially coated precursor; Ti02 particles are shown coated with carbon and a higher magnification on the coated region was also presented. A1 powder was not yet mixed with the coated TiO2. (C) Fully mixed precursor; Al, TiOz. and carbon black particles are indicated.
However, at IOWC, there is an intermediate-phase Ti203 but not AI3Ti. It is believed that Ti203 reacts with C to form TIC and CO(g) according to: Ti203 + 5C -+ 2TiC
+ 3CO(g)
(6)
This was supported by the fact that the total C content decreased by about 10% after synthesis as COk) was produced. As the total amount of COk) was approximately two orders of magnitude lower than that of the solid phase, reaction (1) is not affected by the gaseous species, as reported by Choi and Rhee13 Fig. 3 also includes the BET surface areas of the fully coated precursor and the resulting products. At 800"C, the surface area is slightly higher than that of the 600°C product and the precursor due to intermediate phases. At 1000°C,the surface area decreased from that at 800°C because most intermediate phases transformed into the final products of A1203 and TIC. At both 1200" and 1500"C, intermediate phases were all transformed to A1203 and TIC. The decrease in the surface area of the 1500°C products relative to that from 1200°C is caused by Synthesis Methods for Powders
sintering or particle growth in the powder at higher reaction temperatures.'5316Figure 4 shows a TEM micrograph of reaction products from the fully coated precursor at 1200°C. Particles range in size from 50 to 100 nm and are mostly spherical with a few edges. ( B ) Partially Coated Precursor: Figure 5 shows the XRD patterns of the partially coated precursor synthesized at 1200" and 1500°C. At both temperatures, there exist unreacted and intermediate phases such as Ti02, Ti305, and unknown Ti,O,,, indicating very little A1203-TiC formation. The C coating acted as a barrier to Ti02-A1 particle contact, preventing the initial reaction of Ti02 and Al. In other words, the Ccoated Ti02inhibited reaction with A1 and the aluminothermic reaction did not take place at these temperatures. The BET surface areas were 30.4 and 2.8 m2/g for 1200" and 15OO"C, respectively. The high surface area at 1200°C is due to the presence of unreacted TiOz and C, while the low surface area at 1500°C is due to reaction between Ti02 and C. The C-coated Ti02 reacted with C to form TIC and CO(g), via Ti407 and Ti305 at higher temperatures, as Al did not reduce it to titanium.16 99
0
r
I
UI
e!
0
20
30
A
.
40
1200°C. 30.4d/g
.l* 50
281'
O A e
60
70
1
80
Fig. 5. X-ray diffraction pattern synthesized from the partially coated precursor along with the specific surface areas of each.
I?
? 20
IJ
p o 00
30
40
-
oyQooy I
-
Precursor, 31m2/g
I
50
28 / "
60
70
80
Fig.3.
X-ray diffraction patterns of fully coated precursor and its synthesized products along with the specific surface areas of each.
( C ) Fully Mixed Precursor:
No extra phases, except Al2O3 and Tic, were observed to be synthesized at IOOO", 1200", and 1500°C. However, the results of the surface area showed that a low specific surface area < 1 mZ/gwas obtained for all samples. The powder obtained was much larger than that obtained from the fully coated precursor. As a result, the proposed processing method is advantageous for the formation of nano-sized A1203-TiC composite powders.
(3) Effect of C Coatings To understand the effect of the carbon coating and the formation mechanism, all three precursors were examined by the DSC. The heat flow as a function of temperature was recorded, and the DSC curves are shown in Fig. 6. All DSC curves show an absorption peak at about 670°C, corresponding to the melting of Al. In the Ti02-AI-C system, it is believed that Ti02 and A1 react initially to form A1203 and Ti as shown in reaction (2)?,1*,13 In other words, TiOz and Al must be in contact for the first step of the aluminothermic reaction to occur. In the case of the partially coated precursor, the C coating inhibits the reaction between Ti02 and Al by preventing their contact. The DSC data for the partially coated precursor had two minor exothermic reactions, with peak maxima at 801 and 866°C. This confirms the XRD results that the aluminothermic reaction
500 600 700 800 900 I000 1100 1200 TemperatureP C
Fig. 6. DSC patterns for A1203-TiC formation from three different
precursors. was incomplete. This is explained by the fact that the Ti02 was uniformly coated by C, which provides a suitable method for controlling the reaction path and rate. The mixed precursor had a single large exothermic reaction with a maximum at 941°C. Clearly, the mixture of Al, Ti02, and C released its heat at once with no control of the reaction rate due to the absence of a C coating. This large exothermic reaction sinters the powder as indicated by a low specific surface area (< 1 m2/g) in the final product. For the fully coated precursor, there were four successive weak exothermic reactions, with peak maxima at 823", 871", 968", and 1092°C. The C coating acts as a barrier to control the reaction path and rate when heat is released by the reaction between Al and Ti02. This leads to low heat-released exothermic reactions and a high surface area of the final product. From these results, coating the Ti02/AI mixture with C provides an effective means to control the reaction path and rate. This control of the reaction leads to the production of AI2O3-TiC composite powders with nano-sized powders.
IV. Conclusion
Fig. 4. TEM micrograph of the reaction product from the fully coated precursor after 2 h at 1200°C in flowing Ar.
100
AI2O3-TiC composite powders were synthesized using three different precursors. The fully coated precursor showed that the products had only two phases, A1203and TIC, which formed via intermediate phases of Ti203, A13Ti, and gaseous CO. The partially coated precursor showed that the reaction did not achieve completion due to the C coating preventing a reaction between the Al and Ti02. The mixed precursor showed that the synthesized powder had a very low surface area, even though A1203and T i c were produced. These results show that there are three advantages of using the fully coated precursor instead of mixed precursors, which are commonly used in industry. First, the C-coated Ti02/AI mixture leads to an intimate contact of the reactants and can be a very effective way to control the reaction path and rate. Second, the C coating inhibits the partial sintering of particles by moderating the magnitude of the exothermic reProgress in Nanotechnology: Processing
action. Third, it produces high-purity A1203-TiC composite powders because the hydrocarbon gas provides a non-contaminated C source. Consequently, the new process results in a very complete reaction, yielding high-purity and nano-sized A1203TIC powders that meet the exact requirements for the manufacture of ceramic materials and composites.
Acknowledgment The authors wish to thank Dr. Geoffrey A. Swift and Zach Crothers for contributions to this manuscript.
References ‘A. G. King, “Ceramics for Cutting Tools,” Am. Ceram. Soc. Bull., 4 3 , 3 9 5 4 3 (1965). ’R. P. Wahi and B. Ilschner, “Fracture Behaviour of Composites Based on A1203and TIC,” J . Marer. Sci., 15, 875-85 (1980). ’R. A. Culter. A. V. Virkar, and J. B. Holt, “Synthesis and Densification of Oxide Carbide Composites,” Ceram. Eng. Sci. Proc., 6, 715-28 (1985). 4S. J . Burden, “Comparison of Hot-Isostatically-Pressed and Uniaxially HotPressed A1203-TiC Cutting Tool,” Am. Ceram. Soc. Bull., 67, 1003-5 (1988). ’ S . Adachi, T. Wada. T. Mihara, Y. Miyamoto, and M. Koizumi, “HighPressure Self-Combustion Sintering of Alumina-Titanium Carbide Ceramic Composite,” J . Am. Ceram. Soc., 73, 1451-2 (1990).
Synthesis Methods for Powders
“R. A. Culter. A. C. Hurford. and A. V. Virkar, “Pressureless-Sintered Al2O3T i c Composite,” Marer. Sci. Eng. A , 105/106, 183-92 (1988). ’A. Krell and P. Blank. “Tic-Strengthened A1203 by Powder Tailoring and Doping Procedures,” Muter. Sci. Eng. A , 161, 295-301 (1993). ‘M. N. Rdhaman. Ceramic Processing and Sintering, ch. 2, pp. 3 8 4 2 . Marcel Dekker Inc, New York. 1995. 9C. R. Bowen and B. Derby, “The Formation of TiC/A1203Microstructures by a Self-propagating High-Temperature Synthesis Reaction,” J . Marer. Sci., 31, 3791-803 (1996). ‘OR. Koc and G. Glatzmaier, “Process for Synthesizing Titanium Carbide, Titanium Nitride, and Titanium Carbonitride, US Pat.5, 417, 952 (1995). “ G . Glatzmaier and R. Koc, “Method for Silicon Carbide Production by Reacting Silica With Hydrocarbon Gas, US Pat.5, 324, 494 (1994). I2T. D. Xia, Z. A. Munir, Y.L. Tang, W. J. Zhao, and T. M. Wang, “Structure Formation in the Combustion Synthesis of AI2O3-TiC Composites,” J . Am. Ceram. Sac.,83, 507-12 (2000). I3Y, Choi and S.-W. Rhee, “Effect of Precursors on the Combustion Synthesis of TiC-AIZ03 Composite,” J . Mater. Res., 9, 17614 (1994). I4M. P. Borom and M. Lee, “Rapid Rate Sintering of AI2O3-TiC Composites for Cutting-Tool Application,” Adv. Ceram. Mafer., 3, 3 8 4 (1988). I5R. Abramovici, “Composite Ceramics in Powder or Sintered Form Obtained by Aluminothermal Reactions,” Muter. Sci. Eng., 71, 313-9 (1985). I6C. R. Bowen and B. Derby, “Modelling of Self-propagating High-Temperature Synthesis Reactions,” Br. Ceram. Proc., 50, 2%38 (1993). I7C. R. Bowen and B. Derby, “Self-Propagating High Temperature Synthesis of Ceramic Materials,” Br. Ceram. Trans., 96, 25-30 (1997). “R. Koc and J. Folmer, “Synthesis of Submicron Titanium Carbide Powders,” J . Am. Ceram. Sue., 80, 952-6 (1997). I9G. A. Swift and R. Koc, “Formation Studies of TIC From Carbon Coated Ti02,” J . Mafer. Sci., 34, 308S93 (1999).
101
This Page Intentionally Left Blank
SYNTHESIS OF Sm0.5Sr0.~Co03-, AND La0&-0.4Co03-,NANOPOWDERS BY SOLUTION COMBUSTION PROCESS Narottam P. Bansal National Aeronautics and Space Administration Glenn Research Center Cleveland, OH 44 135 Zhimin Zhong QSS Group, Inc. NASA Glenn Research Center Group Cleveland, OH 44 135
ABSTRACT Nanopowders of S ~ O . S S ~ O . ~ C (SSC) O O ~and - , Lao.$3-0.&003-, (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 considered' as the premium power generation devices in the fbture 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 50O-80O0C, 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*'. To enhance the oxygen ion conductivity of the electrolyte at the reduced temperature, La~-,Sr,Ga~-yMg,O, (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 Sm&30.~CoO3-, (SSC), L ~ O . ~ S ~ O . ~(LSC), COO~-~ Lao.~Sro.2Coo.2Feo.~03-, (LSCF) will be used to substitute Lal-,SrYMn03-,(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
Synthesis Methods for Powders
103
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-" 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>l o 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 t e ~ h n i q u ebecause ~-~ 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)~(98 % purity), Co(N03)2.6H20 (97.7 % purity) and glycine (NH2CH2COOHY99.5 % 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 log 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:
104
0.6 La(N03)3 + 0.4 Sr(N03)2 + Co(N03)2 + 3.2 H2NCH2COOH + (1.8 - x/2) O2 + La0.6Sr0.4Co03-,+ 6.4 C02 + 8 H 2 0 + 3.9 N2
(1)
0.5 Sm(N03)3 + 0.5 Sr(NO3)2 + Co(NO3)z + 3.2 H2NCH2COOH + (1.95 - x/2) 0 2 + Smo.~Sro,sC003.,+ 6.4 C02 + 8 H20 + 3.85 N2
(2)
Progress in Nanotechnology: Processing
1
lear red solution; heat at -80 "C; concentrate to -2M metal nitrate basis
beaker preheated to 300-400"C
Black powder; heat treat 700-1300 "C,2 h each, in air
(SEM) Figure 1.-Flow chart for solution-combustion synthesis of La0.6Sr0.4CoO~~ and Sm0.5Sr0.5CoO~~ nanopowders
indicating the formation of C02, N2, and H20 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 (0.02"/28 step, time per step 0.5 or 1 s) in the 28 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.
Synthesis Methods for Powders
105
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 SrC03. 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.
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. SrC03 phase was also observed in the as-synthesized powder and after calcination at 700 "C. An unknown peak at 32" (probably Sr3Co206.13, 83-375) appeared for the powder calcined at 800 and 900 "C. Formation of the perovskite phase, Lao.~SrO.4C003-~, is completed above 1000°C as observed by XRD results in Fig. 3. The asprepared SSC powder showed the presence of Sm203,Co304, and SrC03 phases. The desired Smo.~Sr0.~CoO3., perovskite phase emerged as the major phase after the powder was calcined at
106
Progress in Nanotechnology: Processing
Synthesis Methods for Powders
107
700 "C. Secondary phases such as Sr3Co206.13remained even after the powder was heat treated at 1100 "C. Perovskite phase-pure Smo.5Sro.~CoO3-,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 Smo.5Sro.5Co03.x and Lao.~Sro.4CoO3.,powders made by solution-combustion synthesis after heat treatments at various temperatures in air Heat treatment Crystalline phasesa Temp. Time ("C) (h) As -- L ~ , , ~ S ~ O . ~ CSrC03 OO~-~, L%.6Sr0.4CoO3-x synthesized System
1000
2
1100
2
Average grain size (Wb 12
S ~ O . ~ S ~ O , ~Sr3C0206.13, C O O ~ - low ~ , intensity
38
Sm&ro.5Co03-x,Sr3C0206.13, low intensity
41
peak at 32" 20
peak at 32" 20
1200 smo.5sro.5coo3-x 2 1300 2 Smo.5SrosCoO3-x "Phases in decreasing order of peak intensity bCalculatedfrom Scherrer formula using FWHM of XRD peak in 47-48" range of 20. The SEM micrographs of Lao.~Sro.&o03., and Smo.&-o,5CoO3-,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
108
Progress in Nanotechnology:Processing
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.
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'' :
where t is the average particle size, h the wave length of Cu K, radiation, B is the width (in radian) of the XRD diffraction peak at half its maximum intensity, and 0 B 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
Synthesis Methods for Powders
109
B~
=
B~~ - B~~
(4)
where BM and Bs 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.
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
110
Progress in Nanotechnology: Processing
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 depends'. 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.sSr0.sCo03-~ (SSC) and La0.&,.&003-~(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, A&. Muter., 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. Muter. 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, Muter. Lett., 10, 6- 12 (1990). 6. M. Marinsek, K. Zupan, and J. Maeek, Ni-YSZ Cermet Anodes Prepared by CitrateNitrate 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 LaGa03 Based Oxide Ion Conductor. I. Doped SmCo03 as a New Cathode Material, J. Electrochem. Soc., 145 [9], 3177-3 183 (1998).
Synthesis Methods for Powders
111
8. R. D. Purohit, S. Saha, and A. K. Tyagi, Nanocrystalline Thoria Powders via GlycineNitrate Combustion, J. Nuclear Mater., 288 [l], 7-10 (2001). 9. T. Ye, Z. Guiwen, Z. Weiping, and X. Shangda, Combustion Synthesis and Photoluminescence of Nanocrystalline Y2Og Eu Phosphors, Mater. Res. Bull., 32, 50 1 (1997). 10. K. Prabhakaran, J. Joseph, N. M. Gokhale, S. C. Sharma, and R. Lal, Sucrose Combustion Synthesis of LaxSql-,$ln03 (x 5 0.2) Powders, Ceram. Int., 31 [2], 327-33 1 (2005). 11. B. D. Cullity, Elements of X-Ray Diffraction, TdEdition, Addison-Wesley, Reading, MA, p. 284 (1978).
112
Progress in Nanotechnology: Processing
COLLOIDAL PROCESSING AND SINTERING OF NANO-Zr02 POWDERS USING POLYETHYLENIMINE (PEI) Yuji Hotfa'*, Cihangir Duran"2, Kimiyasu Sato' and Koji Watari' 1
National Institute of Advanced Industrial Science and Technology, Advanced Sintering Technology Group, Advanced Manufacturing Research Institute, Anagahora 2266-98, Shimoshidami, Moriyama-ku, Nagoya, Japan 2GebzeInstitute of Technology, Department of Materials Science and Engineering, PK 141,41400, Gebze, Kocaeli, Turkey *Correspondingauthor
ABSTRACT A stable colloidal suspension is important for fabricating dense samples with uniform microstructure using colloidal processing methods. Aqueous nano-Zr02 suspensions were prepared using polyethylenimine (PEI) as a dispersant. PEI adsorption on nano-Zr02 surfaces was influenced by PEI content and suspension pH. The isoelectric point (IEP) shifts from pH 7 at 0 wt% PEI to pH 10.3 at 3 wt% PEI. Stable suspensions had mean particle sizes in the range of 100 to 150 nm and sedimentation rates less than 0.4 &h, as compared to 2-5.5 pm and 10-50 mm/h for unstable suspensions. Electrostatic interactions, hydrogen bonding and PEI conformation were found to be controlling mechanisms on the colloidal stability of the suspensions. The amount of PEI adsorbed on nano-ZrO2 surfaces was characterized using Thermogravimetric analysis (TG) and Fourier transform infrared spectrometer (FT-IR). The densification behavior of samples containing 3 wt% PEI at pH 7.1 sintered for 4 h at 1000 to 1300 "C was characterized. Relative density was found to increase rapidly from 54 % at 1100 "C to 92 % at 1200 "C and finally to 98 % at 1300 "C. Therefore, pellets were sintered at 1300 "C for 4 h to quantitatively correlate the processing conditions such as effect of pH and PEI content with densification. INTRODUCTION The dispersion of ceramic powders in liquid is of importance in the colloidal processing methods such as slip casting. These methods have been shown to be superior to conventional dry pressing in terms of controlling density and uniform microstructure evolution in the green and sintered states." * If aggregates form, suspension stability and subsequently sintered properties are severely degraded. Therefore, aggregation must be prevented during colloidal processing3.
Synthesis Methods for Powders
113
Nano-particles can form aggregates in solution due to van der Waals attractive forces at short interaction distance^.^. Therefore, sufficiently large stabilizing forces such as electrical double-layer repulsion or steric interactions should be used to create an energy barrier to inhibit aggregation. Polyethylenimine (PEI) has been used as a dispersant for various ceramic powders and has been shown to enhance stability of ceramic powders in ~ a t e r ~When - ~ . PEI is dissolved in a neutral or acidic solution, proton adsorption results in the protonation of the amine groups. Therefore, positively charged PEI easily adsorbs on the negatively charged ceramic surfaces, which provides an electrosteric effect that prevents aggregation of the ceramic powders5. In this study, the effects of PEI content and pH on the dispersion of nano-Zr02 powders in water were characterized. The objectives were to investigate both PEI adsorption mechanisms on nano-ZrO2 particles and to define the pH range in which colloidally stable nano-Zr02 suspensions can be prepared to fabricate dense sintered ceramics. The effects of pH and PEI content on the suspension properties were characterized by measuring zeta potential, particle size and sedimentation rate. PEI adsorption on nano-Zr02 powders was evaluated from thermogravimetric analysis (TG) and FT-IR measurements. Green samples were used to compare densification behavior at 1300°C as a function of pH and PEI content. EXPERIMENTAL PROCEDURE 3 mol% yttria-stabilized nano-ZrO2 powders with average particle size of 50 to 75 nm were purchased from Aldrich. Surface charge of nano-ZrO2 particles was modified with polyethylenimine (PEI) (MW 10000, Anhydrous, Wako Pure Chemical Ind., Ltd., Japan). PEI was added at concentrations ranging from 0 to 3 wt% with respect to the dry weight of nano-ZrO2. 1 vol% nano-ZrO2 suspensions at various PEI contents (0 to 3 wt%) and pH values were prepared. First, PEI was dissolved in distilled water and then nano-ZrOz powder was added. Suspensions were ultrasonicated at 120 Watts for 10 min. pH was adjusted using reagent-grade HC1 and NaOH (Wako Pure Chemical Ind., Ltd., Japan). Then, the suspensions were stirred for 6 h and finally centrifuged at 5000 rpm for 1 second to remove bigger particles from the suspension. Zeta potential was manually measured by applying 50 V (Model 502, Nihon Rufuto Co., Ltd, Japan). Particle size distribution was measured by using a laser particle size analyzer (Horiba LA-920, Japan). Sedimentation behavior of suspensions was characterized with a pulsed near infrared light (Turbiscan ma 2000, Formulaction, France). A clarified region at the top, and the sediment region at the bottom were characterized as a function of time. In sedimentation kinetics analysis, thickness of the clarification region from the top was considered and light transmitted at 3 % intensity was used.
114
Progress in Nanotechnology: Processing
The amount of PEI adsorption was determined from Thermogravimetric Analysis (TG) (Seiko Instruments SII, SSC/5200). Suspensions were centrifuged at 25000 rpm for 15 min to separate sediment and supernatant. The sediment was washed twice with distilled water to remove any excess (free) PEI. After final centrifugation, the sediment was dried under vacuum at 100°C for 2 h before analysis. TG of as-received ZrO2 powder was chosen as a reference. Weight loss between 200-600°C was used in the adsorption calculations. The surface characterization of nano-ZrO2 and PEI-modified nano-ZrO2 powders was studied by FT-IR (Perkin Elmer, Spectrum GX, USA) after the samples were mixed with KBr powder. IR spectra of as-received PEI and 25 wt% aqueous PEI solutions at various pH values were measured after each solution was sandwiched as a thin layer between two CaF2 plates. Green bodies were fabricated by slip casting using gypsum molds. Densification behavior as a function of sintering temperature was first tested on the samples containing 3 wt% PEI prepared at pH 7.1 from 1000 to 1300 "C with an isothermal hold of 4 h. PEI burn out and sintering were performed at the same heating cycle as the pellets were heated at 100 "Ck. Densities were calculated using the Archimedes principle. The relative density was found to increase sharply from 54 % at 1100 "C to 92 % at 1200 "C and finally to 98 % at 1300 "C. Therefore, pellets were sintered at 1300 "C for 4 h to quantitatively correlate the processing conditions such as effect of pH and PEI content with densification. Microstructure observations were carried out by a scanning electron microscopy (SEM) (Model JSM5600N, JEOL, Japan). RESULTS AND DISCUSSION Figure 1 shows the effect of pH and 80 PEI on zeta potential (5) of the nano-ZrO2 60 suspensions. Electrophoresis properties of 40 > ZrO2 in an aqueous solution determine the E 20 suspension stability. Attractive London-van .-cd z0 o der Waals forces should be overcome by 0 a -20 Q c . repulsive forces such as electrostatic or N -40 polymeric to prepare stable suspensions'. In -60 other words, high zeta potential value induces -80 sufficiently high surface charge, which causes strong repulsive double-layer force6. Figure 1 Zeta potential plots of suspensions as a function of pH and PEI content; ( 0 ) 0 wt% PEI, Nano-ZrOZ suspension with 0 wt% PEI has an (+) 0.5 wt% PEI, (A) 1 wt% PEI, (m) 2 wt% PEI, isoelectric point (pHiep) at nearly pH 7. PEI and ( 0 ) 3 wt% PEI. interacts with nano-ZrO2 surface and shifts the pHiep to more alkaline region as the amount of PEI content is increased. The pHiep increases from pH 7 at 0 wt% PEI to pH 10.4 at 3 wt% PEI. This
4-4
b)
Synthesis Methods for Powders
115
can be attributed to the fact that PEI is a cationic polyelectrolyte and addition of a strong acid to the PEI-containing solution neutralizes -NH- basic groups, resulting in a positively charged polymer skeleton according to the following reaction';
As a result, positively charged -NH;- groups can adsorb on negatively charged nano-ZrOz surfaces by an
30
0
/
electrostatic attraction, which causes the PHjep to shift to the alkaline region. Furthermore, nano-ZrO2 powders have various negative zeta potentials in the alkaline -72 region due to the different pHiep values such as mV at 0 wt% PEI compared to as low as -9 mV at 3 wt% PEI at pH 11. The zeta potential curves for the suspensions with PEI have a tendency to approximate the original curve of the suspension without PEI at highly acidic (e.g., pH 3) and highly alkaline (e.g., pH
c=
12) solutions. It was reported that the degree of PEI dissociation (a)increases with decreasing pH, that is, fully dissociated PEI at pH 2 (a=1) and undissociated
2000
1800
I600
Wavenumber, cm-'
1400
2000
0 0
25
0-
E 2o
2
2
15
?g
0
10.p;-
10
8
0
pHiep even though the particle size decreases slightly. The particle size measurements were carried out right after the sample preparation step and then the suspensions were subjected to the sedimentation tests for 24 h. Aggregation of the particles with time takes place due to the instability of the suspensions. In this case, the gravitation force accelerates the settling of the aggregates. However, the stability range in nano-Zr02 suspensions is successfully extended up to pH 9.1 with 3 wt% PEI addition (Fig. 4c). This can be attributed to the adsorption and conformation of PEI in the region where the 5 is positive. The PEI molecules undergo a transition from a compact coiled to a stretched conformation with increasing a (or, alternatively, decreasing pH) due to an electrostatic repulsion between the neighboring ionized sites. Therefore, there happens a strong electrostatic repulsion between both positively charged PEI sites and nano-Zr02 powders at pH 2 h
No microemulsion formed
Boldface indicates conditions used for zirconia synthesis. The aqueous phase was 1M zirconyl nitrate hydrate; the chamber volume is 285 mL.
2.2k0.3 MPa was reached, after which additional liquid C02 was injected into the reactor to further increase the pressure, while keeping the temperature within the stated range. The liquid+vapor CO2 was observed through sapphire windows in the vessel. After ca. 11 min of C 0 2 addition, at a pressure of about 7.6 MPa, the liquid+vapor became supercritical (the critical point of scC02 is at T, = 7.5 MPa and T, = 32°C). Continued addition of liquid C02 increased the pressure at 0.5 MPa/min to achieve experimental pressures of 13.9, 15.3, or 17.3 MPa; the temperature of the reactor reached 70"-72°C. This aqueous/ scC02 microemulsion was stable for > 2 h at pressures above 13.9 MPa at 70°C. For safety reasons, the highest pressure used in the reaction vessel was 17.3 MPa. The precipitating agent, either 0.04 moles of ammonia in a 14 mol/L solution or 3 mL of urea (14 mol/L), was then injected into the vessel via a three port injection. An excess of precipitating agent was used to neutralize the acidic nature of the water-scC02 mixture.20The pressure declined quickly by about 0.3 MPa after the addition. The reaction mixture rapidly turbidified, most likely owing to the rapid formation of nanoparticles within the microemulsion droplets. Over the course of 40 min, the nanoparticles aggregated and/or settled to the bottom of the reaction tank, leaving a clear "supernatant." After 1 h continuous stirring, the vessel was vented. Venting was performed slowly in order to prevent the entrainment of the synthesized powder in the expelled gas. The powder, consisting of a zirconia precursor (amorphous hydrous zirconia),*' was collected from the bottom of the vessel. The powder was repeatedly (>3 x ) washed with 50 mL ethanol and filtered, until the color changed from dark-yellow into pale yellow. Finally, the powder was dried overnight at room temperature (25°C). The zirconium oxide nanoparticles were then synthesized from the amorphous hydrous zirconia precursor particles by calcining the dried powder at 450"C, then at 550"C, and finally at 650"C, for 2 h at each temperature. Thermal analysis of precursor calcination was performed at a heating rate of IO"C/min in air (the same rate and conditions used in the synthesis
9
calcinations above), using thermal ravimetric (TGA) and differential thermal analyses (DTA).' The sizes of the zirconia nanoparticles were characterized by transmission electron microscopy (TEM) analysis, and X-ray diffraction analysis (XRD) was used to determine the crystal structure of Zr02 nanoparticles. Electron Spectroscopy for Chemical Analysis (ESCA) was used to determine the surface elemental composition of the amorphous hydrous zirconia.
III. Results ( 1 ) Microemulsion Stability Table I shows the effect of [Zr4'(aq)]/[PFPE] molar ratio and the reaction conditions (pressure and temperature) on the microemulsion stability of l mol/L aqueous zirconium in scCO2 in our reaction chamber. Temperatures above 65°C were necessary to obtain a stable zirconium hydrate nitrate(aqueous)/PFPE/ scC02microemulsion system at a [Zr4'(aq)]/pFPE] molar ratio of 0.65 (i.e., with 4.3 mL of solution), and stable microemulsions were found at pressures between 13.9 and 17.3 MPa. No stable microemulsions could be obtained at higher ratios of water to surfactant, at any of the temperatures and pressures studied. ( 2 ) Sizes and Morphologies of Amorphous Hydrous Zirconia Precursor Particles and Zirconia Nanoparticles After determining the conditions for microemulsion stability, amorphous hydrous zirconia precursor particles were synthesized by adding ammonia to stable water/ZrO(NO& microemulsions in scCO2. The precursor particles were then calcined in zirconia at S O T , as described above. TEM photographs of typical zirconia nanoparticles calcined from precursors prepared at three different pressures are shown in Fig. 1. Reactions at each pressure were performed at least twice, and different runs gave qualitatively similar results, in terms of particle shape and size. Particles formed from precursors made at the lowest pressure studied, 13.9 MPa, were especially ellipsoidal, with lengths
Fig. 1. Transmission electron microscopy (TEM) micrographs and electron diffraction pattern of zirconia (ZrOz) nanoparticles synthesized in a [Zr4'(aq)]/PFPE/scCOz microemulsion by adding aqueous ammonia, followed by calcination at 550°C. Reaction pressure: (a) 13.9 MPa, (b) 15.3 MPa, and (c) 17.3 MPa.
130
Progress in Nanotechnology: Processing
Fig. 2. Long and short axes of zirconia nanoparticles calcined at 550"C, vs. scCOz pressure during precursor synthesis. The trend toward smaller and more spherical particles with increasing synthesis pressure correlates well with changes in microemulsion droplet morphology reported in the literature. The lines shown are "guides to the eye."
Fig. 3. Transmission electron microscopy (TEM) micrograph of precursors and the Zr4' concentration was 1M. The operation conditions are the same as those in Fig. l(a).
of 21-72 nm along the major axis and from 16 to 37 nm along the minor axis. Precursors synthesized at higher pressures gave smaller and more spherical final products, with typical dimensions shrinking to 10-30 nm at 15.3 MPa and to 4-13 nm at 17.3 MPa. Figure 2 summarizes the effect of reaction pressure on the dimensions of the final zirconia nanoparticles. Figure 3 shows a TEM photograph of a sample of the amorphous hydrous zirconia precursor (i.e., before calcination) that was calcined to the zirconia shown in Fig. I(a). The TEM images show that the calcined zirconia particles are slightly smaller than the precursors, as expected. A tabulation of sizes of particles measured and marked in Figs. 1 and 3 is presented in Table 11.
( 3 ) Analysis of Amorphous Hydrous Zirconia Precursor Figure 4 shows the thermal gravimetric/differential thermal analysis (TGA/DTA) of the zirconia precursor particles prepared at 17.3 MPa, using a heating rate of 10"C/min in an air atmosphere. The TGA curve shows two major weight-loss stages. One stage ranges from room temperature to 210°C; there is about a 10% weight loss over this temperature range. In the DTA plot in Fig. 4, there is one broad endothermic peak, at
139"C, and three exothermic peaks at 303", 347", and 449°C. The TGA curve shows that major weight losses were associated with both endothermic and exothermic effects. The endothermic valley over the temperature range from room temperature to 210°C corresponds to the 10% weight decrease in the sample over the same temperature range. The two exothermic peaks at -303" and at 347°C are associated with the decomposition of amorphous hydrous zirconia precursor, or of trapped surfactant or solvent. The third exothermic peak at 449°C indicates the crystallization of tetragonal zirconia, as shown by XRD measurements (vide infra). Zirconia precursor powders synthesized in the supercritical C 0 2 reverse microemulsions were calcined at temperatures of 300", 450", 550", and 650"C, separately. XRD measurements were made at room temperature after slowly cooling each of these powders in air, Fig. 5. The XRD patterns were essentially the same for ZrOl precursors prepared at any of the three pressures studied (13.9, 15.3, and 17.3 MPa). Also evident from Fig. 5: XRD of zirconia precursors did not show any evidence of crystallinity after two hours at the calcination temperature of
Table 11. Sizes of Zirconia Nanoparticles Labeled in Figs. 1 and 2, in Nanometers Fig. I(a)
Fig. 2
No.
Major axis
Minor axis
Fig. l(b)
Fig. l(c)
Major axis
Minor axis
1 2 3 4 5 6 7 8 9 10
63.00 48.00 36.00 46.00 41.00 22.00 41.00 46.00 29.00 51.00
27.00 23.00 29.00 27.00 21.00 20.00 22.00 34.00 27.00 29.00
22.22 11.11 14.81 16.67 29.63 22.22 14.81 18.52 15.74 20.37
34.09 36.36 16.14 29.32 30.68 37.50 24.77 28.86 55.68 25.00
18.18 20.45 13.18 26.59 25.00 25.68 23.86 18.64 32.95 23.64
42.30+ 10.94
25.90k4.13
18.61f4.98
9.26 12.96 9.26 12.96 10.19 6.48 4.63 4.63 14.81 10.74 7.59 8.33 11.11 12.96 5.56 9.43 3.12
3 l.84+9.93
22.82 f5. I9
11
12 13 14 15 Average
Synthesis Methods for Powders
+
131
h
40
P
v
w 20
2
f
E n O
f
e
1L-J-
550'C t
t
t *
I
I
450°C
L
20
c
01
0
i
139°C
$
I
100
I 200
I
I
I
400 500 Temperature ("C) 300
I
600
+
700
Fig. 4. Thermal gravimetric/differential thermal analysis (TGA/DTA) curves of zirconia precursors (17.3 MPa synthesis pressure), calcined in air.
300°C. The tetragonal phase was observed after 2 h at the calcination temperature 450°C (cf. JCPDS card #17-0923) and then the monoclinic phases were found after calcining for 2 h at either 550" or 650°C (cf. JCPDS card #Ol-0750).
( 4 ) Particles Obtained With Urea As the Precipitating Agent At high temperature and pressure, ammonia solution and carbon dioxide can react to form urea. It is not likely that any significant urea was formed in our reactions, however, as urea is not formed at temperatures below 180"C, even at very high C 0 2 Nonetheless, we explored the effects of using urea as a precipitating agent in the synthesis of zirconia precursors, to further rule out the possibility that urea is the active reactant in our synthesis. SEM photographs of zirconia nanoparticles synthesized using urea are shown in Figs. 6(a) and (b). The other reaction conditions were the same as those in Figs. ](a) and (c), respectively. Zr02 particles formed using urea are highly heterogeneous, with sizes from microns to tens of microns, and with irregular shapes. Although urea formation is not an important side reaction in our synthesis, a urea precursor, ammonium carbamate, may be present in the reaction. To determine whether ammonium carbamate or other nitrogen-containing species are present in significant quantities, ESCA was carried out on hydrous zirconia precursors, synthesized using ammonia as the precipitant. The results are shown in Fig. 7. In addition to zirconium and oxygen, fluorine and carbon were found (from the surfactant), but nitrogen was undetectable. ESCA on hydrous zirconia prepared using urea also showed no nitrogen.
IV. Discussion ( I ) Effect of Reaction Pressure on Zirconia Precursors and Zirconia Nanopavticle Products There are many factors, including the pressure and temperature of scC02, solute concentration in aqueous phase, water-tosurfactant molar ratio, and surfactant choice, that can affect the properties of scC02 microemulsions and that would therefore influence the characteristics of the particles synthesized in the microemulsion. In particular, synthesis of uniform nanoparticles requires a stable microemulsion. Microemulsion droplets can take on a variety of sizes and shapes, depending upon the amount of water, surfactant type, and ionic strength, e t ~ In. scCO2 ~ ~ inverse microemulsions, it 132
300°C
E
-20 w
20
I
I
30
40
50 20 (degrees)
60
0
Fig. 5. X-ray diffraction (XRD) analysis of zirconia precursors (either 13.9, 15.3, or 17.3 MPa synthesis pressures)calcined at 300", 450", 550",
and 650"C, bottom to top.
has been shown that decreasing pressure increases droplet size, at a fixed surfactant Concomittant with the increase in droplet size, an increasing deviation from sphericity is expected-nly by forming increasingly elongated droplets can the fixed amount of surfactant be accommodated at the droplet surfaces. These changes in microemulsion droplet morphology parallel the changes that we observed in the final (calcined) particle morphology. As shown in Figs. l(a 100 nm, respectively. The sharp diffraction peaks in the XRD pattern would indicate a larger crystallite size than that obtained from TEM results. The small crystallite size shown by the CDF TEM image is most probably that of Inz03, while the large ones belong to SrIn204.The crystallite size of SrIn204being larger than 100 nm (Fig. 3(b)), the electron beam cannot pass through these large crystallites. Hence, the corresponding electron diffraction pat-
Fig. 1. Scanning electron microscopy micrograph and size distribution histogram of as-prepared sample spray pyrolysis-2.
138
t
o
'
I
10
'
I
.
20
I
.
30
I
.
40
I
.
50
I
'
60
I
.
70
1
1
ao
2(e) Fig. 2.
X-ray diffraction spectra of as-prepared sample spray pyrolysis-2.
tern obtained shows a mixture of In203and SrInz04.XRD results, on the other hand, mainly show a SrIn204phase, and very little Inz03phase. It can also be seen from XRD results that the diffraction peaks due to In203 are much broader, indicating a smaller crystallite size, whereas that from SrIn204 are very sharp, indicating a larger crystallite size. For PL measurements, it was noted that the as-prepared, uncalcined samples did not show any PL emission. The samples were therefore postcalcined at 1000°C for 4 h, for better PL efficiency. The sample post-calcined at 800°C for 4 h shows PL emission owing to better crystallinity. PL measurements were carried out under ultraviolet excitation (266 nm) from a frequency-doubled NdYAG laser and detected by a PMT. Figure 4 shows the PL spectra of Pr-doped SrIn204samples with three different doping concentrations 0.05%, 0.5%, and 5.0%. The emission spectra show narrow bands typical of rare-earth ions. The most prominent bands were seen in the blue and red region at 492,605, and 619 nm. Some less intense bands were also seen in the green and red region. The emission features are all due to different transitions from Pr. The blue emission that peaked at 492 nm is due to transition from 3Po-+3H4.The 605 emission line is due to the transition 'D2-+'H4,and the emission at 619 is the 'Po+ 3H6 transition.' Apart from these three main emission bands, the weak band located at 548 nm is due to the transition 3PO+3H5.The emission bands at 654, 691, and 719 nm can be ascribed to transitions from the 3Po+3F2.3,4of Pr. It is to be noted that the feature observed at 532 nm is the second harmonic frequency of the incident laser. The rare-earth elements have partially filled shells off electrons that give rise to narrow localized electronic transitions that occur at wavelengths ranging from the far-infrared to the vacuum-ultraviolet. It is also well known that the emission characteristics of rare-earth elements are relatively weakly influenced by the host material. This is because the 4felectrons of the rareearth ions are highly localized due to shielding by the outer filled shells of 5p and 5s electrons, and hence their optical transitions are atomic-like even when the ion is in a crystalline solid. The intensity of PL emission varied with varying concentrations of Pr. The highest PL intensity was observed in the case of SP-2 with a Pr doping of 0.5%. The PL intensity for SP-3 (0.05% Pr) was much less and it was the weakest for SP-1 (5% Pr). The small PL intensity in the case of SP-3 can be explained by the low doping level of Pr, which leads to fewer Pr ions available for emission. In the case of SP-I, the large number of Pr sites could possibly lead to cross-relaxation between the Pr ions and thus a decrease in the PL intensity." The parameters for PL measurements for the three samples with different doping concentrations were kept identical so that a comparison of the PL intensities could be made. Progress in Nanotechnology: Processing
Fig. 3. Centered dark-field transmission electron microscopy images of sample spray pyrolysis-2 (a) as-prepared in thin section; (b) post-calcined at 800°C for 4 h; (c) shows the diffraction pattern of (a); (d) analysis of the diffraction pattern.
and made up of very small nanocrystallites. CDF TEM images of the agglomerates by TEM show the smallest size of the crystallites to be about 10 nm, which grow to > 100 nm by calcination at 800°C. Sharp, blue, green, and red luminescence bands were observed in these Pr-doped SrInz04 samples. The PL intensity varied with varying concentrations of Pr. The highest PL intensity was observed in the case of samples with a Pr doping of 0.5%.
0.60 0.55 0.50 -
0.45 0.40 4 0.35 5 0.30 L ‘ b j,0.25 -
Q
.-c
5
References
c ; 0.20 -
a
J
0.15 -
0.10 -
0.05,-
0.00 -0.05
300
sp-2 sp-3 sp-1 I
400
.
I
500
.
I
600
.
I
700
.
800
Wavelength (nm)
Fig. 4. Photoluminescence spectra of the praseodymium-doped strontium indate samples with varying doping concentrations.
IV. Conclusion Spherical, luminescent, Pr-doped SrIn204 particles were obtained by an ultrasonic SP process. SP is a simple and effective technique for obtaining phosphor particles at a low cost. The main advantages of this technique are the high purity, spherical shape, and uniform size character of the powder obtained. XRD and electron diffraction results show that SrIn204phase is generated at a furnace temperature of about 800°C. The average powder size as obtained from SEM results was about 580 nm. TEM results show the spherical particles to be polycrystalline, Synthesis Methods for Powders
‘R. Balda, J. Ferna’ndez, I. Sae’z de oca’riz, M. Vcda, A. J. Garci’a, and N. Khaidukov, “Laser Spectroscopy of Pr3+ Ions in LiKYI-,Pr,FS Single Crystals,” Pb s Rev. B, 59, 9972-80 (1999). ‘M. Malinowski, M. F. Joubert, and B. Jacquier, “Dynamics of the IR-to-Blue Wavelength Upconversion in Pr’+-Doped Yttrium Aluminum Garnet and LiYF, Cr stals,” Pbys. Rev. B, SO, 12367-74 (1994). &H. i.Dieke, “Chapter 13: Terbium Tb,” in Spectra and Energy Levels of Rare Earth Ions in Crystals, Edited by H. M. Crosswhite, and H. Crosswhite. Interscience, New York, 1968. 4S. E. Dali, V. V. S. S. Sunder, M. Jayachandra, and M. J. Chockalingan, “Synthesis and Characterization of Aln204 Indates, A = Mg. Ca, Sr, Ba,” 1. Mater. Sci. Lett., 17, 619-23 (1998). ’F. S. Kao and T. M. Chen, “A Study on the Luminescent Properties of RedEmitting Praseodymium-Activated Srlnz04 Phosphon,” 1.Solid Sfare Cbem., 156. 8 4 7 (2001). 6F. S. Kao, “A Study on the Luminescent Properties of New Green-Emitting Srln204:xTbPhosphor,” Muter. Cbem. Pbys., 76. 295-8 (2002).
’S. Jain, D. J. Skamser, and T. T. Kcdas, “Morphology of Single-Component Particles Produced by Spray Pyrolysis,” Aerosol Sci. Tecbnol., 27, 575-90 (1997). *Y. L. Song, S . C. Tsai. C. Y. Chen, T. K. Tesng, C. S. Tasi, J. W. Chen, and Y. D. Yao, “Ultrasonic Spray Pyrolysis for Synthesis Of Sphericdl Zirconia Particles,” J. Am. Ceram. Soc., 87 [lo] 186471 (2004). 9A. Gurav, T. T. Kodas, T. Pluym, and Y. Xiong, “Aerosol Processing of Materials,” Aemyol Sci. Techno/., 19,411-52 (1993). ‘ON. K. Teddy and K. T. R. Reddy, “Growth of Polycrystalline SnS Film by Spray Pyrolysis,” Thin Solid Films, 325, 4-6 (1998). “T. T. Kodas, “Generation of Complex Metal Oxide by Aerosol Processes: Superconducting Ceramic Powders and Films,” Angew. Chem. Int. Ed. Engl. Adv. Muter., 28 [6] 794-806 (1989).
139
I2Y. C. Kdng, H. S. Roh, H. D. Park, and S. B. Park, “Optimization of V U V Characteristics and Morphology of BaMgA110017:Eu2+Phosphor Particles in Spray Pyrolysis,” Ceram. Inf., 29, 41-7 (2003). ”S. M. Abrarov, Sh. U. Yuldashev, B. B. Lee, and T. W. Kang, “Suppression of the Green Photoluiminescence Band in ZnO Embedded into Porous Opal by Spray Pyrolysis,” J . Lwnin., 109, 25-9 (2004). I4Y. C. Kang, H. S. Roh, and S. B. Park, “Preparation of Yz03: Eu Phosphor Particles of Filled Morphology at High Precursor Concentration by Spray Pyrolysis,” Adv. Mater., 12 [6] 451-3 (2000).
140
I5M. Abdullah, F. Iskandar, S. Shibdmoto, T. Ogi, and K. Okuyama, “Preparation of Oxide Particles with Ordered Macropores by Colloidal Templating and Spray Pyrolysis,” Acta Mafer.,52. 5151-6 (2004). 16F. Iskandar, L. Gradon, and K. Okuyama, “Control of the Morphology of Nanostructured Particles Prepared by the Spray Drying of a Nanoparticles Sol,” J . CoNoid Inreface Sci., 265, 29C303 (2003). I7J. Hegarty, D. L. Huber, and Y. M. Yen, “Fluorescence Quenching by Cross Relaxation in LaF3:Pr3+,” Phys. Rev. B, 25, 5638-45 (1982).
Progress in Nanotechnology: Processing
Nano-Blast Synthesis of Nano-size Ce02-Cd203 Powders Oleg Vasylkiv’ ICYS, National Institute for Materials Science 1-1, Namiki, Tsukuba, Ibaraki 305-0044, Japan
Yoshio Sakka MEL. National Institute for Materials Science 1-2-1, Sengen, Tsukuba, Ibaraki 305-0047,Japan
Valeriy V. Skorokhod Institute for Materials Science, NASU, 3, Kiev, 03680, Ukraine
-
)
-
CeC13 7 H z 0 and GdC13 6 H 2 0 that were dissolved in water were precipitated with urea (NHzCONHz) to produce matrix agglomerates for three-component nano-reactors. Mixing hexamethylenetetramine with dilute nitric acid resulted in the formation of well-dispersed nano-particles of cyclotrimetilene & &trinitramine & c ( (RDX) in the solvent. Nano-reactors were produced by impregnating the nano-C&N606 into the matrix agglomerates of an intermediate complex of cerium and gadolinium compounds. Blast initiation of the C&I&O, resulted in extremely rapid detonation and gaseous products formation at temperatures of 2000”-5000”C, which were compressed into a volume nearly equal to the initial volume of each RDX nano-particle. Multiple “nano-blasts” occurred in the volume of each nano-reactor. The impact of the blast waves led to fragmentation of the surrounding matter. The evolution of a large volume of gaseous products dissipated the heat of the process and limited temperature increase, thus reducing the possibility of local sintering among the primary particles. The shortterm high temperature generated during the blasts enhanced the solid solubility of the metal oxides. Uniform aggregates of 22-74 nm consisting of 6-14 nm crystallites of gadolinia in ceria solid solution were synthesized. I. Introduction
M
ULTI-METAL
oxide and composite ceramic nano-size powders enable quality improvement and differentiation of product characteristics at scales currently unachievable with commercially available coarse powders. Fabrication of nano-powders with uniform morphology and precise stoichiometry is the key to realizing high-performance devices based on nano-structured metal oxide ceramics for a wide range of application^.'-^ Particle agglomeration is a natural result of the dominant effect of interparticle forces when the particle size is less than 1 pm. Agglomeration refers to adhesion of particles to each other because of van der Waals forces of attraction, which is significantly larger in nano-particles. If weak forces hold the nano-size particles together, the agglomerates are referred to as soft agglomerates. These agglomerates can be easily redispersed in a suitable liquid medium or in the dry state. In contrast, strong forces due to dissolution-reprecipitation during synthesis and post-synthesis treatment at the contact regions, which form G . Messmg--contributing editor
Manuscript No. 21334. Received January 6, 2006; approved January 23. 2006. Presented at the 9th International Ceramic Processing Science Symposium, Coral Springs, FL, Jan. 8-11.2006. ‘Author to whom correspondence should be addressed. e-mail: oleg.vdsylkiv@nims. wjp
Synthesis Methods for Powders
necks, and (or) subsequent solid necking due to sintering (high-temperature calcination) result in aggregates or hard agglomerates.”16 For these reasons, nano-sized powders readily agglomerate during processing. Once the nano-powder is agglomerated, the strength of the dried agglomerate is too high to realize the benefits of the nano-size primary crystallites. The critical parameter, which had been chosen for controlling and optimization in this study, was the mean secondary aggregate size of the powder after synthesis, post-synthesis treatment, and after calcination. Single- and multi-component nano-powders have been synthesized from aqueous and non-aqueous solutions. A typical procedure involves several sequential steps: (1) preparation of single- or multi-component starting solutions of metal salts (usually aqueous solutions); (2) preparation of aqueous solutions of different precipitants (reductants); (3) reductive decomposition of the starting single- or multi-component solution to obtain the precipitant, colloidal suspension, or gel of the desired end-product phase or intermediate multi-component product; (4) separation of the end (intermediate) product; (5) deagglomeration of the synthesized (precipitated) powder prior to calcination; and (6) synthesis of the end-product powder via calcination, i.e., thermal decomposition of the intermediate products. 1-8*17-24 Processing of multi-component non-agglomerated nano-powders has proven to be extremely challenging and often results in a non-homogeneous multiphase compound with a poor morphol0gy..2-24Typically, nucleation, growth, aggregation, and agglomeration of the first component occur within seconds under very mild conditions. The nucleation of the second component often starts at a higher temperature and requires more time and (very often) different pH. The final product of such “co-precipitation” is a non-homogeneous composite powder, nano-crystalline in nature, but in fact consisting of micrometer-sized agglomerates with very poor morphology and composition h~mogeneity.~ To achieve the desired solid solution, such a multi-component composite powder would require an excessive calcination temperature and unnecessarily prolonged holds with no realistic prospect of achieving a fine-grained s t r ~ c t u r e . ” ” ” ~ ~ The present study was aimed at establishing a new method of preparing nano-size, agglomerate-free, cerium-gadolinium oxide (CGO) ceramic powders with precise morphology and chemical composition. During the past decade, Ce02-based materials have been intensively investigated as catalysts, structural and electronic promoters of heterogeneous catalytic reactions, and oxygen-ion conducting solid electrolytes in electrochemical solid oxide fuel Cyclotrimethylene trinitramine ( R D X b i s a colorless crystalline solid with a density of 1.8 g/cm3. Its structural formula is hexahydro-1,3,5-trinitro1,3,5-triazine or (CH2-NNo2)3.9-’6’2527At temperatures lower than 1IO’C, it is extremely 141
Table I. Characteristics of Cyclotrimethylene Trinitramine (cfi~606)zs-27 Molecule size 0.48 nm Density 1.8 g/cm3 Temperature of thermal detonation 233°C Detonation rate 8,350 m/s Pressure at front of blast wave 33.8 GPa Generated heat during explosion 1.3 kcal/g 0.9 L/g Gas product volume
stable. Decomposition of RDX starts at about 170"C, melting at 2WC, and exploding at 233°C (Table I). Cyclotrimethylene trinitramine is chemically or otherwise energetically unstable. Usually, explosions involve a rapid and violent oxidation reaction with a sudden violent release of mechanical, chemical energy, accompanied by generation of a high temperature and release of extremely hot gases. It causes pressure waves in the local medium in which it occurs.1~1632s27 11. Experimental Procedure
For multiple nano-blast synthesis of CGO, CeCI3. 7H20 and GdC13.6H20 (both 99.9% pure from Wako Pure Chemicals Co., Osaka, Japan) were weighted and separately dissolved in doubly distilled and deionized water at a concentration of 0.1M. The starting materials were used as received without further purification. The initial amount of cerium and gadolinium compounds varied according to the concentrations of both ceramic oxides in the resulting solid solution. To produce complex bi-component intermediate agglomerates, urea (NH2CONH2) (Wako Pure Chemicals Co.) was used as a precipitation agent. NHzCONHz was dissolved in deionized water at a concentration of 2M per I-xM of CeCI3. 7H20 and xM GdCI3.6H20. Two beakers with a total volume of the NH2CONH2aqueous solution of 300 mL (200 mL for cerium chloride and 100 mL for gadolinium chloride solutions) were prepared. Nano-reactors comprising complex threecomponent intermediate agglomerates were synthesized as follows: (I) To produce matrix agglomerates for the threecomponent nano-reactors, nucleation of cerium oxide in the aqueous solution was conducted by spraying 200 mL of urea aqueous solution into the cerium chloride aqueous solution (both solutions were heated to 60°C) under the stirring conditions of 1600 rpm. After precipitation, the suspension was stirred at the prescribed temperature for 10 h. Gadolinium complex was nucleated by spraying gadolinium chloride aqueous solution into a rapidly stirred (I600 rpm) suspension of the as-synthesized cerium oxide. Because of the existence of the residual non-reacted urea, decomposition began within the first minute from the start of the gadolinium chloride solution spraying. After 30 min, 100 mL of urea solution was added by spraying into the stock suspension. Subsequent stirring at 80°C for 5 h was conducted to finalize synthesis and homogenize the suspension. Finally, the product was washed with water, followed by re-dispersion of the agglomerates of cerium and gadolinium intermediate compounds in ethanol (C2H50H, 99.5% reagent grade, Kanto Chemicals, Japan) using an ultrasonic horn (Model USP-600, Shimadzu, Kyoto, Japan). (2) To produce cyclotrimethylene trinitramine (C3H6N606), hexamethylenetetramine was dissolved in deionized water at a concentration of 0.1M. Concentrated (-93%) nitric acid (from Wako Pure Chemicals Co.) was added to urotropin solution. Mixing urotropin with nitric acid led to the formation and re cipitation of cyclotrimethylene trinitramine ( c ~ H ~ N ~ o?5-27 ~ )- . Mixing dissolved hexamethylenetetramine with diluted nitric acid caused the formation of well-dispersed nano-particles of the C&N606 in the Solvent. 142
Fig. 1. TEM micrograph of as-synthesized cyclotrimethylene trinitramine's nano-particles.
(3) Nano-reactors were produced by colloidal impregnation of the C3H6N606 nano-particles (see Fig. 1) into the matrix agglomerates of an intermediate complex of cerium and gadolinium compounds. (4) Subsequently, the threecomponent intermediate agglomerates i.e., nano-reactors, were separated from the supernatant by centrifugation (10 000 rpm for 15 min). The powder was then washed once with distilled and deionized water to remove the residual reaction by-products. Washing with water was followed by two-time ethanol washing with subsequent centrifuging, re-dispersing in fresh ethanol, and then a final slow evaporation of the ethanol ( T = 70°C) using a drying oven. Multiple "nano-blast" calcination/deagglomeration of ceriagadolinia solid solution was conducted by extremely rapid heating of the nano-reactors to the temperature of thermal detonation Of C3H6N606. Cyclotrimethylene trinitramine and nano-reactors were tested for decomposition using thermogravimetric and differential thermal analysis (TG-DTA) (Thermo Plus 2 Rigaku, TG8120, Rigaku, Tokyo, Japan). The reference used for the DTA was an AI2O3sample, the sample containers in the equipment were alumina, and an atmosphere O& was used. Heating rates of 20"C/min were used. The TG-DTA experiments were used to determine the dependence of calcination conditions versus heating rate. The particle-size distribution was analyzed by the dynamic light scattering method (DLS) using a laser particle-size analyzer (Model LSPZ-100, Otsuka Electronics, Osaka, Japan). A very small amount of each powder (15 mg) was dispersed in the distilled water for the analysis. Observation via TEM (Model JEM-2100-F, JEOL, Tokyo, Japan) operated at 200 kV was used to determine the powder morphology. Phase identification of the powders and distribution of the components into each aggregate were determined by nano-area energy dispersion X-ray spectroscopy analyzer (TEM-EDX), and from X-ray diffractometry data (XRD) (Model RINT 2000, Rigaku). The XRD profiles were recorded using CuKu radiation under 40 kV and 300 mA at room temperature.
III. Results and Discussion A general diagram of the processing pathway is shown in Fig. I . A well-dried powder composed of the threecomponent interProgress in Nanotechnology: Processing
Table 11. Powder Aggregale/Agglomerate Size Distribution for Three-Component Intermediate Agglomerates Cerium and Gadolinium Compounds with and without c3H,&o6, as Synthesized after Multiple "Nano-blast" Calcinations/Deagglomeration, and, Finally, after Subsequent Non-Isothermal Calcination up to 450°C with 30 min Holds Composition
Engineered nano-reactors(nm)
After multiple blast treatment (nm)
After calnnation up to 450°C (nm)
28-740 30-1260
1W7
22-74 230-360
GdzoCesoOl95 with C3H6Nb06 GdJ2eanOl0,without C2H,N,0~
mediate agglomerates with a relatively poor non-homogeneous morphology was filled into an alumina container for further thermal treatment. An important factor behind this methodology is to prevent the ignition of the impregnated particles of C3H6N606at 180°C by ultra-rapid heating of nano-reactors through thermal detonation temperature of 233°C (Table I). The thermal detonation temperature is the temperature at which spontaneous multiple ruptures of the N-NOz bonds occur. The blast initiation of cyclotrimethylene trinitramine begins in nanosize regions that are capable of accumulating the heat and transferring it into chemical energy, thus starting the blast reaction. Cleavage of the N-N02 chemical bonds requires less energy for isolated molecules (or clusters) than for molecules located in the s/g) forms bulk of the solid. Extremely rapid detonation ( gaseous products with a temperature of 2000"-5000"C compressed into a volume equaling the initial volume of each cycbtrimethylene trinitramine particle. Multiple nano-explosions start within the volume of each nano-reactor. The instantaneous power of each explosion (i.e., the expansion of compressed gases from the volume equaling the initial volume of each exploded particle of cyclotrimethylene trinitramine) is 500 MW/g.25-27The impacts of the blast waves lead to the fragmentation of the surrounding matter of matrix agglomerates. The rapid evolution of a large volume of gaseous products during combustion dissipates the heat of the process and limits temperature increase, thus reducing the partial sintering among the primary particles. This gas evolution also limits inter-particle contact, resulting in a less-agglomerated product. Multi-blast deagglomeration of the nano-powder occurred due to the highly energetic impact of the blast waves, while the short-term high temperature generated during the explosions enhanced the solid solubility of one component into the other. Nano-size cerium and gadolinium oxide composite and, very soon afterward, a solid solution of the gadolinium oxide in the cerium matrix oxide were synthesized. Utilizing this "nano-blast" calcination
-
-
.
:.
100; 100
700 600
! '. ;
-
0 Time (sec) Fig. 3. Thermogravimetric and differential thermal analysis (TG-DTA) analysis of the thermal detonation and explosion of RDX (heating rate = 20"C/min, dashed-dotted line, temperature; dashed line, DTA; solid line, TG).
technique, we produced a cerium-gadolinium oxide (CGO) powder with an average primary crystallite size of 6 - 14 nm, an aggregate size distribution of 22-74 nm (see Table II), uniform morphology, and precise stoichiometry (Fig. 2). We analyzed the thermal decomposition under rapid heating of both cyclotrimethylene trinitramine itself (Fig. 3) and of cyclotrimethylene trinitramine distributed within the matrix complex precursor agglomerates of cerium and gadolinium compounds (Fig. 4). This enabled us to identify the blast decomposition conditions of the C3H6N606nano-particles synthesized by the described technique and colloidally impregnated into intermediate precursor agglomerates. In addition, this allowed us to confirm them by comparison with the blast decomposition conditions of the C3H6N606itself (Fig. 5 ) The results for TG-DTA of the thermal blast of the C3H6N606 itself at a heating rate of 20"C/min are shown in Fig. 3. Three stages of thermal decomposition could be identified. At the heating rate of 20"C/min, the ignition of the C3H6N606 started at around 180°C. About 202'-205"C is the melting point of cyclotrimetilen trinitramin. A slight endother-
_______..._.------
75
480
Fig. 2. General diagram of the processing pathway.
Synthesis Methods for Powders
-50
520
600 Time (sec)
560
640
150
680
Fig. 4. Thermogravimetric and differential thermal analysis (TG-DTA) analysis of multi-blast calcination of preliminary engineered nano-reactors (heating rate = ZO"C/min, dashed-dotted line, temperature; dashed line, DTA; solid line, TG).
143
Fig. 5. TEM micrograph of porous cerium-gadolinium intermediate
agglomerates.
mic peak could be detected irrespective of the extremely short melting time (2 s for the particles of the C3H6N6O6filled into the container) prior to the multiple blasts. The thermal detonation and multiple blast decomposition of the impregnated C3H6N6O6 particles occurred at -233°C. Just at the beginning of the ignition reaction, the TG analysis detected a significant increase in the mass of the sample (approximately 10.8 weight % at a heating rate of exactly 20”C/min). This is explained by the capturing of external oxygen from the neighboring space by the reacting species. The ignition instantly (within nano-seconds) transformed into thermal detonation, i.e., the C&N606 exploded. Even the temperature as detected by the thermocouple of the TG/DTA system momentarily increased by about 100°C.
Fig. 6. TEM micrograph of cerium-gadolinium oxide nano-aggregates that led to nano-blast synthesis from nano-reactors. 144
Fig.7. TEM micrograph of cerium and gadolinium oxide aggregates calcined with no blast treatment.
Figure 4 shows TG-DTA analysis results for the total summarized explosion of impregnated C3H6N6O6particles during the thermal decomposition of three-component intermediate complex agglomerates i.e., nano-reactors under heating at a rate of 2O”C/min. The reaction results observed here are not, however, limited to this particular rate. The strong exothermal peak detected by differential thermal analysis at a similar temperature and time as for C3H6N606itself (seen in Fig. 3) confirms the Occurrence of multiple blasts of the C3H6N606of nano-particles distributed into the volume of the nano-reactors. Table I1 shows the size distribution of the powders aggregate/ agglomerate for the three-component intermediate agglomerates of cerium and gadolinium compounds with and without C3H6N606,as synthesized, after multiple “nano-blast” calcination/deagglomeration; and, finally, after subsequent non-isothermal calcination up to 450°C. As-synthesized agglomerates with a very wide size distribution of 30-1260 nm were impregnated with separately synthesized particles of cyclotrimetilene trinitramine. Washing of such three-component agglomerates and subsequent ultrasonic deagglomeration reduced the size of the agglomerates to 28-740 nm. Multiple nano-blast calcination reduced the average aggregate size to 1 8 4 7 nm. Nano-aggregates of gadolinia solid solution in ceria matrix synthesized by the multiple “nano-blast” calcinations/deagglomeration technique are shown in the TEM micrograph in Fig. 6. After the nano-blast treatment, the powder was non-isothermally calcined up to 450°C for 30 min to remove the products of explosive decomposition of cyclotrimetilenetrinitramine.
Fig. 8. XRD pattern of ceria-gadolinia solid solution synthesized by multiple nano-blasts of engineered nano-reactors.
Progress in Nanotechnology: Processing
Thus, such treatment acts to preserve both the powder’s compositional homogeneity and morphology. The same calcination treatment was conducted for the agglomerates of cerium and gadolinium oxides that were not embedded with explosive particles, but were preliminarily ultrasonically deagglomerated. From the data listed in Table 11, and TEM micrograph of cerium and gadolinium oxide aggregates calcined without the blast treatment (see Fig. 7), we can conclude that the aggregates of the composite oxide produced were much coarser and the size distribution was wider (230-360 nm). Figure 8 shows XRD patterns of the ceria-gadolinia solid solution produced by multiple nano-blast synthesis. All XRD peaks were attributed to the Gd20Ce8001.95 solid solution. Moreover, these peaks were relatively broad, indicating that the powder was composed of very fine crystallites.
IV. Conclusion The fabrication of nano-powders with uniform morphology and precise stoichiometry is the key to realizing high-performance devices based on nano-structured metal oxide ceramics and metal-ceramic composites for a wide range of applications. Here, we demonstrate a new processing technique that is based on engineering multi-component nano-reactors with subsequent multiple “nano-blast” calcination/deagglomeration. Multiple nano-blasts of impregnated particles of C3H6N6O6deagglomerate the powder due to the highly energetic impacts of the blast waves. The solid solubility of one component into the other is enhanced by the extremely high local temperature generated during the nano-explosions. We produced a nano-size agglomerate-free ceria-gadolinia powder with excellent morphology and an average aggregate size of 48 nm.
Acknowledgments This study was performed through Special Coordination Funds for Promoting Scienceand Technology from the Ministry of Education, Culture, Sports, Science, and Technology of the Japanese Government.
References ’H. Gkiter, “Nanocrystalline Materials: Basic Concept and Microstructure,” Acfa Mafer., 48, 1-29 (2000). ’S. Tjong and H. Chen. “Nanocrystalline Materials and Coatings,” Mafer. Sci. En Res., 45, 1 4 8 (2004). K ‘ . Johnston and P. Shah, “Making Nanoscale Materials with Supercritical Fluids,” Science, 303, 482-3 (2004).
Synthesis Methods for Powders
’T.Zhang, J. Ma, L. Kong. P. Hing, and J. Kilner, “Preparation and Mechanical Properties of Dense Ceo8Gdn.202-a Ceramics,” Solid Stare Ionics, 167. 1 9 1 4 (2~4). 0 . Vasylkiv, T. Kolodiazhni, Y. Sakka, and V. Skorokhcd, “Synthesis and Characterization of Nanosize CeriaGadolinia Powders,” J . Ceram. SOC.Jpn., 113, 1014 (2005). 60.Vasylkiv and Y. Sakka, “Synthesis and Colloidal Processing of Zirconia Nanopowder,” J. Am. Ceram. Soc., 84,2489-94 (2001). ’0. Vasylkiv, Y. Sakka. Y. Maeda, and V. Skorokhod, “Nano-Engineering of Zirconia-Noble Metals Composites.” J. Eur. Cerum. Soc., 24. 469-73 (2004). *N.-L. Wu, S.-Y. Wang, and 1. A. Rusakova, “Inhibition of Crystallite Growth in the Sol-Gel Synthesis of Nanocrystalline Metal Oxides.” Science, 285. 1375-7 (1999). 9G. Ulrich, “Flame Synthesis of Fine Particles,” Chem. Eng. News, 62. 22-9 (1998). “ S . Pratsinis, “Flame Aerosol Synthesis of Ceramic Powders,” Prog. Energy Combust. Sci.,24. 1977219 (1998). “C. Sorensen, W. Hageman, T. Rush, H. Huang, and C. Oh, “Aerogelation in a Flame Soot Aerosol,” Phys. Rev. Left, 80, 1782-5 (1998). ”R. Purohit, B. Sharma, K. Pillai, and A. Tyagi, “Ultrafine Ceria Powders Via Glycine-Nitrate Combustion,” Mafer. Res. Bull., 36, 271 1-21 (2001). ’A. Varma and J.-P. Lebrat, “Combustion Synthesis of Advanced Materials,” Chem Eng. Sci., 41, 2179-% (1992). ‘9.Tillotson, L. Hrubesh, R. Simpson, R. Lee, R. Swansiger, and L. Simpson, “SolCel Processingof Energetic Materials,” J. Non-Crys. Solidr. 225,358-63 (1998). ”T. Tillotson, A. Gash, R. Simpson, J. Hrubesh, L. Jr. Satcher, and J. Poco, “Nanostructured Energetic Materials Using SokGel Methodologies,” J. NonCr s f Solidr. 285, 338-15 (2001). ‘M. Kuklja, “Thermal Decomposition of Solid Cyclotrimethylene Trinitramine,” J . Phys. Chem. B, 105, 1015942 (2001). ”M.-P. Pileni, “The Role of Soft Colloidal Templates in Controlling the Size and Shape of Inorganic Nanocrystals,” Nut. Mater., 2, 145-50 (2000). “J. Millman, K. Bhatt, B. Prevo, and 0. Velev, “Anisotropic Particle Synthesis in Dielectrophoretically Controlled Microdroplet Reactors,” Naf. Mater., 4, 98-102 (2005). I9J.-S.Lee and S.-C. Choi, “Crystallization Behavior of NanoCeria Powders by Hydrothermal Synthesis Using a Mixture of HzOzand NH40H,” Mafer.Left., 58, 39&3 (2004). ”0. Vasylkiv and Y. Sakka, “Synthesis and Sintering of Zirconia Nanopowder by Non-[sothermal Decomposition from Hydroxide.” J . Ceram. SOC.Jpn., 109, 500-5 (2001). ”0. Vasylkiv and Y. Sakka, “Non-Isothermal Synthesis of Yttria-Stabilized Zirconia Nanopowder Through Oxalate Processing: I, Characteristics of (Y-Zr) Oxalate Synthesis and its Decomposition,” J. Am. Ceram. Soc., 83, 2196-202 (2000). 220. Vasylkiv, Y. Sakka, and H. Borodians’ka, “Non-Isothermal Synthesis of Yttria-Stabilized Zirconia Nanopowder Through Oxalate Processing: I, Morpholog Manipulation,” J. Am. Ceram. Soc., 84, 2484-8 (2001). ‘3Z. Tianshu. P. Hing, H. Huang, and J. Kilner. “Ionic Conductivity in the ce02-Gdz0, System (O.OS 99%, Merck) was introduced as a solvent and stabilizer for zirconium and titanium precursors. Tnethanolamine (HOCH2CH&N; > 98%, Merck) was used to dissolve lead acetate trihydrate in iso-propanol at 80°C under reflux. The processing scheme opted is illustrated in Fig. 1. The PZT sol was maintained at 60"-70°C until a transparent resinous gel was formed. The gel was dried between 100" and 160°C for 10 days continuously. The temperature was increased up to 250"C, and a swollen mass was obtained. This brownish black mass of PZT precursor was ground in an agate mortar to a fine powder and heat treated at various temperatures from 350" to 500°C for 1.5 h to study its crystallizing behavior and other characteristics.
( I ) Characterization The phase chemistry of the PZT gel powder was characterized by X-ray diffraction (XRD; Model JDX-9C, JEOL, Tokyo, Japan) analysis at room temperature, with CuKu radiation and a nickel filter. Differential thermal analysis (DTA) and thermo gravimetric analysis (TGA) were conducted (Model; STA 409C, Netzsch, Selb, Germany) in air atmosphere at a heating rate of S"C/min up to 830°C. The molecular structures of the PZT precursor gel and powders were interpreted by using Fourier transform infrared spectroscopy (FTIR; Model 470-FTIR Nexus, Thermo Nicolet, Madison, WI), in the wave range from 4000 to 500 cm-' and a resolution of 8 cm-I. The particle size and grain morphology were evaluated by using a transmission electron microscope (TEM; Model CX200, JEOL) working at an accelerating voltage of 200 kV. The TEM sample was prepared by dispersing a small drop of the ultrasonicated suspension onto a copper grid precoated with an amorphous carbon film.
1II. Results and Discussion
( I ) XRD Analysis The XRD results of PZT gel powders processed at various temperatures are shown in Fig. 2. The precursor gel exhibited amorphous behavior below 25OoC, which indicates that a polymeric network exists holding the metal ions together. The decomposition of lead acetate started above 300°C and lead oxide was observed to exist in a free state at 35OoC,as shown by the XRD plot. At 430"C, several peaks from perovskite lattice planes were observed, which were accompanied by two small pyrochlore 147
I
I
ZI(OC~H~ complex )~ dissolved in 2-propanol solution RefluxinglNp at 60°C
I
Ti (OC3H7)4 dissolved in 2-propanol solution
I
Refluxing/6O0C Pb(CH3C00)2.3H20dissolved in
Refluxing/BO"C
PZT precursor sol
I
I
Aging and drying
1
I
I
Gelation
1
Calcinations
PZT perovskite nano-powder Fig. 1. Scheme for the preparation of lead zirconate titanate (PZT) nano-powders.
peaks. Thus, it may be concluded that the perovskite phase started to form at 430°C. The XRD plot of the gel powder treated at 500°C identified a single-phase perovskite. But, in order to investigate the exact perovskite formation temperature, further experimentation was carried out by using the FTIR and DSC/TG analytical techniques, which also helped to study the nature of reactions taking place at various temperatures. The XRD pattern of PZT gel powder heat treated at 500°C exhibited broad peaks indicative of fine crystalline particles. The PZT average crystallite size was
A
Perovskite
+ Pyrochlore
500°C c
.-2 -ma,
430°C
c
a
*
250°C I
20
30
I
I
40 50 2theta "degrees"
I
60
70
Fig. 2. X-ray diffraction plots of lead zirconate titanate gel powders calcined at various temperatures. 148
T = 0.9h/Bc ose ~ where T is the average particle size in angstroms, B is the width of the peak at half the peak height in radians, his the wavelength in angstroms, and OB is the Bragg angle in degrees. Das et al.14 and Bose and BanerjeeIs have produced nano PZT powders at 450°C by using inorganic metal precursors. ' ~ reported nano-crystalline PZT perovskite C. Liu et ~ 1 . have formation at 650°C by precipitating the PZT recursor powder from an organic solution, and D. Liu et d.' have produced nano-PZT powders at 450°C by using an ethylene glycol solvent system. The present paper illustrates nano-crystalline phasepure PZT perovskite formation at 470°C by using organic metal precursors and a TEA/iso-propanol solvent system following a specific process, which has not been reported before. The lowtemperature perovskite crystallization can be attributed to the decomposition and combustion reactions associated with the organics and TEA complexes.
(2)
350°C
[r
calculated to be approximately 14.5 nm by using Scherrer's equation
FTIR Analysis
Figure 3 shows the FTIR spectra of the PZT gel and powder heat treated at various temperatures at a rate of 5"C/min. Many absorption bands related to the acetate anions were clearly observed, i.e. the 1407 and 1550 cm-' bands from 80°C gel may be attributed to the symmetric and asymmetric (COO-) vibrations, respectively, while the medium bands at 1339 and 910 cm-l may be attributed to the deformation vibration of CH, groups and the stretching mode of C-C bonds, respectively. The medium Progress in Nanotechnology: Processing
-=!
t
10 Ot-\
910
3318 1550
v
-0 a
2400°C
A f
lii
t
a
1076
I-
n
m +: ._
-'"
5E
!-
-60 -50
0
I 4000
I
3500
I
3000
I
I
I
2500 2000 1500 Wavenumber (cm-')
I
1000
I 1
500
Fig. 3. The Fourier transform infrared spectra of lead zirconate titanate (PZT) gel and powders calcined at different temperatures.
and weak intensity peaks at 1016, 1157, and 1260 cm-' may be attributed to the stretching vibration of the C-N bonds of TEA. A sharp band at 1076 cm-' was assigned to the C-0 stretchin mode of iso-propanol. Two weak bands at 2870 and 2964 cm- $ may be attributed to the CH3 asymmetrical stretching and C-H stretching from various methyl groups present among the precursors and solvents, respectively. The broad shallow band at 3318 cm-', approximately, may be attributed to the OH- group (hydrogen bond) present in isopropyl alcohol, TEA, and chemically combined water in lead acetate.I8 As predicted by FTIR data, heat treatment at a higher temperature was required to decompose the acetate and other organic ligands present in the precursor gel. The gel powder heat treated at 250°C exhibited peaks from acetate ligands, with lower amplitudes indicative of their partial decomposition. The temperature was raised further to remove the residual carbon associated with the acetate, methyl, and amine ligands. Above 400°C the PZT gel powder exhibited no association of acetate ligands which indicated that they were decomposed completely and residual carbon, if any, was oxidized. Traces of 1407 and 1550 cm-' bands were observed up to 430°C, and they disappeared completely at 500°C. A hump at 980 cm-l was observed in the spectrum of 430" and 500°C powders, which is an indication of formation of a new structure, i.e perovskite. The FTIR data exhibited perovskite crystallization onset at 430°C, which continued up to 500°C. The actual crystallization temperature of perovskite was determined from TG/DTA analysis.
( 3 ) TGIDTA Thermal Analysis PZT gel prepared at 80°C was characterized for thermal behavior. The TG/DTA results are shown in Fig. 4. The DTA curve had two main exothermic peaks at 334" and 422"C, respectively, and some small exothermic peaks associated with these. The TG curve had three major weight loss steps at 240", 320", and 350"C, respectively. The DTA curve indicates that the PZT precursor gel decomposed exothermally with a sharp peak at 334°C. This exotherm may be assigned to the major decomposition reaction associated with TEA complexes. Some minor peaks observed just before 334°C peak may be associated with the burning and oxidation of propoxy, butoxy, and acetate ligands associated with metal ions. The formation of free lead oxide occured in this temperature range, as can be observed in the XRD plot. Comparison with FTIR results indicated that the intensity of the peaks from TEA ligands decreased as the temperature increased and at 350°C, the peaks disappeared. Therefore, the weight loss associated with the exotherm at 334°C indicates the TEA decomposition reaction. These decomposition reactions were accompanied by a drastic weight loss in the temperature range between 240" and 320°C. Synthesis Methods for Powders
200
400
600
800
Temperature /"C Fig. 4. Thermo gravimetric (TG)/differential thermal analysis (DTA) of lead zirconate titanate resinous gel prepared at 80°C.
Weight loss continued from 320" to 422"C, which was accompanied by another exothermic peak on the DTA curve. The exotherm may be attributed to the crystallization of the pyrochlore phase, as evidenced by the previous XRD results. The weight loss during this stage resulted from the decomposition reactions of residual organics. Substantial heat energy was released by these reactions and the crystallization of the perovskite phase was finally observed by a small exothermic shoulder close to 470°C on the DTA c ~ r v e . ~ Finally, ~ . * ~ at the completion of decomposition reactions and PZT perovskite formation, no further weight loss or chemical activity was observed.
( 4 ) TEM Observations The PZT gel powder heat treated at 500°C was subjected to TEM microstructural analysis to determine the average particle size. Figure 5 shows the TEM image of the as-dispersed PZT nano-powder. The image reveals that the nano-particles were agglomerated at some locations with an average particle size of less than 15 nm. This is also in good agreement with the XRD line-broadening results (14.5 nm). The individual particles exhibited a spherical morphology. The selected area electron diffraction (SAED) confirmed the crystallinity of the as-obtained PZT nano-particles.
IV. Conclusion Nano-crystalline (< 15 nm) PZT powders have been synthesized successfully by using sol-gel processing. The formation of PZT
Fig. 5. Transmission electron microscope image of lead zirconate titanate nano-powder heat treated at 500°C. Inset is the selected area diffraction pattern.
149
nano-particles was discussed, interpreted, and verified by using FTIR along with XRD and TG/DTA analytical techniques altogether by adopting a specified procedure. Exothermic decomposition of the organics and TEA complexes resulted in the evolution of gases (NH3, C02, H20) and substantial heat energy, which crystallized a single-phase perovskite at 470°C. The PZT particle size was measured by using TEM and compared with XRD line-broadening results. The as-obtained nano-powders may be suitable for lower temperature sintering and to produce a denser bulk material. The volatility of lead oxide that occurs at higher sintering temperatures may be minimized, and enhanced physical and piezoelectric properties may be obtained as well. Acknowledgment The TEM investigation work conducted by Dr. M. Farooque is gratefully acknowledged.
References ‘B. Jaffe, R. S. Roth, and S. Marzullo, “Piezoelectric Properties of Lead Zirconate Titanate Solid-Solution Ceramics,” J. Appl. Phys., 25 [6] 809-10 (1954). ’W. Aiying, P. M. Vilarinho, I. M. M. Salvado, and J. L. Baptista, “SolGel Preparation of Lead Zirconate Titanate Powders and Ceramics: Effect of Alkoxide Stabilizers and Lead Precursors,” J . Am. Ceram. Soc., 83 [6] 137%85 (2000). ’Y. Matsuo and H. Sasaki, “Formation of Lead Zirconate-Lead Titanate Solid Solutions,” J. Am. Cerum. Soc., 48, 289-91 (1965). 4S. S. Chandrateriya, R. M. Fulrath, and J. A. Pask, “Reaction Mechanisms in the Formation of PZT Solid Solutions,” J. Am. Ceram. Soc., 64 171422-5 (1981). ’H.Hirashima, H. Onishi, and M. Nagakowa, “Preparation of PZT Powders from Metal Alkoxides,” J. Non-Crysr. Solids., 121, 4 0 4 4 (1990). 6Q. F. Zhou, H. L. W. Chan, and C. L. Choy, “Nanocrystalline Powders and Fibers of Lead Zirconate Titanate Prepared by the SolGel Process,” J. Mufer. Process. Techno!.,63, 281-5 (1997).
150
’H.M. Cheng, J. M. Ma, B. Zhu, and Y. H. Cui. “Reaction Mechanisms in the Formation of Lead Zirconate Titanate Solid Solutions Under Hydrothermal Conditions,” J. Am. Ceram. Soc., 76 (31 625-9 (1993). *A. M. Bruno and J. A. Eiras, “Preparation of Coprecipitated Ferroelectric Ceramic Powders by Two-Stage Calcinations,” J. Am. Cerum. Soc.. 76 [I I] 27366 (1993). 9W. D. Kingery, H.K. Bowen, and D. R. Uhlmann, In Introduction ro Cerumics, 2nd ed, pp. 469-77. Wiley, New York, 1976. ‘OJ. Ryu, J. J. Choim, and H. E. Kim, “Effect of Heating Rate on the Sintering Behavior and the Piezoelectric Properties of Lead Zirconate Titanate Ceramics,” J. Am. Ceram. Soc., 84 [4] 9 0 2 4 (2001). “I. M. Aiying Wu,M. Salvado, P. M. Vilarinho, and J. L. Baptista, “Lead Zirconate Titanate Prepared from Different Zirconium and Titanium Precursors by Sol-Gel,” J. Am. Ceram. Soc., 81 [lo] 2 6 4 M (1998). ”G. Yi, Z. Wu, and M. Sayer, “Preparation of Ph(Zr,Ti)O, Thin Films by SolGel Processing: Electrical, Optical and Electro-Optic Properties,” J. Appl. Phys., 64 5 2717-24(1988). Jr. Meyer, T. Shrout, and S. Yoshikawa, “Lead Zirconate Titanate Fine Fibers Derived from Alkoxide-Based Sol-Gel Technology,” J. Am. Cerum. Soc., 81 [4] 861-8 (1998). I4R. N. Das, A. Pathak, and P. Pramanik, “Low-Temperature Preparation of Nanocrystalline Lead Zirconate Titanate and Lead Lanthanum Zirconate Titanate Powders Using Triethanolamine,” J. Am. Ceram. Soc., 81 [I21 3357-60 (1998). ”S. Bose and A. Banerjee, “Novel Synthesis Route to Make Nanocrystalline Lead Zirconate Titanate Powder,” J . Am. Ceram. Soc., 87 [3] 487-9 (2004). I6C. Liu, B. Zou, A. J. Rondinone, and Z. J. Zhang, “SolGel Synthesis of FreeStanding Ferroelectric Lead Zirconate Titanate Nanoparticles,” J. Am. Chem. so;? 173,43445 (2001). D. Liu, H. Zhang, W Cai. X. Wu, and L. Zhao, “Synthesis of PZT Nanocrystalline Powder by a Modified SolGel Process Using Zirconium Oxynitrate as Zirconium Source,” J. Mater. Chem. Phys., 51, 186-9 (1997). ‘*K. Kitaoka, H. Kozuka, and T. Yoko, “Prepartion of Lead Lanthanum Zirconate Titanate Fibers by SolGel Method,” J. Am. Ceram. Soc., 81 [5] 1189-96 (1998). I9A.Towata, H. J. Hwang, M. Yasuoka, and M. Sando, “Seeding Effects on the Crystallization and Microstructure of S o l 4 e l Derived PZT Fibers,” J. Mater. Sci.,35, 4009-13 (2000). ’k.Kitaoka, H. Kozuka, and T. Yoko, “Prepartion of Lead Lanthanum Zirconate Titanate Fibers by SolGel Method,” J. Am. Ceram. Sac., 81 [5] 1189-96 0 (1998).
“2.
Progress in Nanotechnology: Processing
Synthesis of AIN Nanopowder from y-AI2O3 by Reduction-Nitridation in a Mixture of NH3-C3H8 Tomohiro Yamakawa,+ Junichi Tatami, Toru Wakihara, Katsutoshi Komeya, and Takeshi Meguro Graduated School of Environment and Information Sciences, Yokohama National University, Yokohama 240-8501, Japan
Kenneth J. D. MacKenzie School of Chemical and Physical Sciences, Victoria University of Wellington, Wellington, New Zealand
Shinichi Takagi and Masahiro Yokouchi Kanagawa Industrial Technology Research Institute, Evina 243-0435, Japan
Aluminum nitride ( A N ) powders were synthesized by gas reduction-nitridation of y-Al2O3 using NH3 and C3Hs as the reactant gases. A N was identified in the products synthesized at 1100"-1400"C for 120 min in the NH&&18 gas flow confirming that AIN can be formed by the gas reduction-nitridation of y-A1203. The products synthesized at 1100°C for 120 min contained unreacted y-A120J. The 27Al MAS NMR spectra show that AI-N bonding in the product increases with increasing reaction temperature, the tetrahedral A104 resonance decreasing prior to the disappearance of the octahedral A106 resonance. This suggests that the tetrahedral A104 sites of the y-A1203 are preferentially nitrided than the A106 sites. A N nanoparticles were directly formed from y-AI203 at low temperature because of this preferred nitridation of A104 sites in the reactant. A N nanoparticles are formed by gas reduction-nitridation of yA1203 not only because the reaction temperature is sufficiently low to restrict grain growth, but also because y-A1203 contains both A104 and A106 sites, by contrast with a-A1203 which contains only A106.
A
I. Introduction
In recent years, the synthesis of AIN nanoparticles has active1 been pursued because of their properties. Plasma syndirect nitridation using NH3 gas?0321vapor phase synthesis,22323 and electron beam heating24have been proposed. More recently, Suehiro et ui.2s,26used gas reduction-nitridation in the AI2O3-NH3-C3H8system to synthesize AIN particles at lower nitridation temperatures which are thermodynamically advantageous compared with the AI2O3-C-N2 system. It has also been reported that AlN nanoparticles can be synthesized from 6, y-A1203.27The formation mechanism of AIN from transition alumina has not yet been elucidated. Studies on the synthesis mechanism of AIN from A1203have been carried out, based on phase analysis and microstructural observations using X-ray diffraction (XRD), transmission electron microscopy (TEM), and scanning electron microscopy (SEM). 27Al MAS NMR has also been used to investigate the formation mechanism of oxynitride ceramics such as A10N28*29 and SiAION.3@-33 NMR spectroscopy should also provide useful information on the formation of AIN by gas reduction-nitridation of y-A1203.The purpose of the present work is to investigate the formation mechanism of nano AlN particles from a transition alumina (y-A1203), by gas reduction-nitridation using several analytical techniques including 27AlMAS NMR.
L U M I N U M NITRIDE (AIN) ceramics
have attracted considerable attention as IC substrates, packages, heat-sinks, and fillers, because of their high intrinsic thermal conductivity (330 W/mK), high electric insulation (> lOI4 R . cm), and low thermal ex ansion coefficient (3.2 x 10-6/K), which is close to that of Si?,' Several investigations have been conducted on the synthesis of AIN by various techni ues, including direct n i t r i d a t i ~ n , ~ , ~ chemical va or deposition,' and carbothermal reduction-nitridation.8-1aIn particular, A1203-C-N2 system has found industrial application, as the morphology of the products can be controlled because of the endothermic reaction. Furthermore, it is well known that the AIN powder synthesized by this technique is easy to be densified the sintered body. However, a longer reaction period and higher temperature is needed to fully nitride aA1203.In order to synthesize AIN at lower temperatures and shorter times, several researchers have investigated the use of various A1203 polymorphs as starting materials. They reported that transition aluminas are easier to be nitrided than aA1203.'2-'4 L. Klein--contributing editor
Manuscript No. 20406. Received April 12, 2005; approved July 19, 2005. 'Author to whom correspondenceshould be addressed. e-mail:
[email protected] Synthesis Methods for Powders
11. Experimental Procedure
Commercial nanocrystalline y-A1203powder (AKPGOI 5, Sumitom0 Chem. Co., Tokyo, Japan) was used as the starting material. The main characteristics of the raw powder are summarized in Table I. The raw AI2O3powder was weighed into an AI2O3 boat, placed in an electric furnace with a high-purity AI2O3work tube, and fired to 700°C at a heating rate of 5"C/ min in Ar gas (99.999% purity) to eliminate oxygen in the system and remove the surface water of the y-A1203.We initially confirmed that the characteristics of the y-AI2O3did not change during the pre-heating treatment up to 700°C. Heating was continued in a flowing gas mixture (4 Ljmin) of NH3 (99.999% purity) and 0.5 vol% C3Hs (99.99% purity). The sample was heated to the reaction temperature of 1100°-1400"C at a rate of 8"C/min and held for &120 min before being cooled in NH3. After the heating, the sample was removed the reactor in shortterm contact with ambient air, it was vacuum-encapsulated the samples. The phases present in the products were identified by X-ray diffractometry (RINT2500, Rigaku, Tokyo, Japan) using CuKa radiation operated at 50 kV and 300 mA. Their morphologies were observed with a transmission electron microscope (JEM2000FX, JEOL, Tokyo, Japan). The specific surface areas of the 151
Table 1. Characteristics of the Raw Powder Characteristics
Value
Purity (%) Crystalline phase Specific surface are (m2/g) XRD crvstallite size (nm) Transition temperature to C L - A I ~("C) O~ \
I
1281'
+Calculated from a diffraction peak y-A1203 (440). 'Determined by differential thermal analysis.
powders were measured by the single-point of Brunauer-EmattTeller (BET) method (Quantasorb, Quantachrome, Boynton Beach, FL). The nitridation ratio was calculated from weight change by oxidation of the resultant powder in air with therm~ g r a v i m e t r y(TGA; ~ ~ TG-8 120, Rigaku, Tokyo, Japan). Oxygen contents were measured by OxygenjNitrogen analyzer based on infrared absorbance technique (EGMA-650, HORIBA, Tokyo, Japan) after correcting calibration curve of standard samples. The solid-state 27Al MAS NMR spectra (Unity 500 spectrometer, Varian, Palo Alto, CA) were acquired at 11.7 T, operating at 130.244 MHz and a 5 mm Doty MAS probe with a spinning speed of 1@12 kHz. A 1 ps x/10pulse for solution was used with a recycle delay time of 1 s, and the spectra were referenced to Al(HzO);+.
IIl. Results and Discussion Figure 1 presents the XRD patterns of the products synthesized at 1100"-1400"C for 120 min. The products formed at 1100°C consist chiefly of AIN and residual unreacted y-A1203. Singlephase A1N can be obtained over 1200°C. Figure 2 shows the XRD patterns of the products synthesized at 1400°C without holding the sample at the maximum temperature, in Ar and NH3-0.5 vol% C3H8 atmospheres. Although most of y-AI203 fired in Ar was transformed into a-A1203 because of a high temperature, u-A1203was not identified in the sample fired at the NH3-C3Hg gas mixture. It was suggested that )'-A1203 should be directly converted to AIN. The nitridation behavior of y-Al O3 is consistent with that reported in a previous study,265 however, its detailed mechanism are, as yet, unexplained. Figure 3 shows the temperature dependence of the nitridation ratio at 1100"-14OO"C for 120 min. Nitridation ratio significantly
-p
0 AlN
0
Fig. 2. XRD pattern of the products synthesized at 1400°C (a) NH3C3H8 gas mixtures and (b) Ar atmosphere.
increased as the reaction temperature increased, consistent with the XRD results. A nitridation ratio of 80% was achieved by firing at 1200°C. Even in samples fired at high temperatures for long soaking times, complete nitridation was not achieved, possibly because of the formation of aluminum hydroxide species by reaction of AIN with high surface area and the moisture in the air. However, the presence of such thin surface oxide layer on the AIN particles could not be confirmed by XRD?7,36 Figure 4 shows TEM photographs of the products synthesized at 1200"-140O0C for 120 min, together a photograph of the initial material (Fig. 5(a)). All the fired products consist of nanoparticles, but in the sample synthesized at 1400"C, these nanoparticles are seen to have sintered during primary particle formation and coalescence (Fig. 5(c)). Figure 5 shows the specific surface area and BET average particle size of the product produced by firing at 1100"-1400"C for 120 min. The size of the nanoparticles increases with increasing temperature. The specific surface area was estimated to be 71.8-24.2 m2/g, the average BET particle size being 2G76 nm, an appreciable change by comparison with the initial powder. This result is in good agreement with the TEM observation.
2o 10
w .
50
2O/ifegree
70
80
Fig.1. XRD pattern of the products synthesized at (a) llOO°C, (b) 1200"C, (c) 1300°C, and (d) 1400°C for 120 min.
152
70
2Wdegrce
t
01 1100
I
I
1200 1300 Temperature ("C)
1
Fig. 3. Nitridation ratio of the products obtained from firing at various temperatures for 120 min.
Progress in Nanotechnology: Processing
Fig. 5. Specific surface area and Brunauer-Ematt-Teller particle size of the products synthesized at various temperatures for 120 min.
Table 11. Oxygen Contents and Lattice Constants of AW Nanopowder Lattice constants Reaction conditions
140OoC-120 min 1400T-60 rnin 14OOoC-120 rnin
Oxygen contents ( ~ 1 % )
2.8 3.2 5.7
a
(A)
3.1110 3.1 110 3.1 109
c
(4
4.9794 4.9792 4.9730
than that prepared in our previous and are comparable with commercial grade of AIN powder. Figure 6 shows the 27Al MAS NMR spectra of samples reacted at 1100"-14OO0C for 120 min, together with the spectrum of the initial y-A1203,which shows the typical tetrahedral A104 resonance at 66 ppm and the octahedral AIOd resonance at 12 pprn (Fig. 4(a)). As nitridation proceeds, the AI-N resonance at 114 ppm appears and increases in intensity. The broadening and asymmetry of this peak to higher fields in the spectra of the products formed at 1100" and 1200°C suggests the formation of a transient aluminum oxynitride species which typically appears at 80-100 ppm, depending on the N:O This result suggests that nitridation proceeds via these intermediate
Fig.4. TEM photographs of (a) the raw material and the products synthesized at (b) 1200°C and (c) 1400°Cfor 120 min.
Oxygen contents and lattice constants are summarized in Table 11. AIN nanoparticles synthesized at 1200°C for 120 rnin and 1400°C for 60 and 120 rnin contain 5.6, 3.2, and 2.8 wt% oxygen contents, respectively. It was found that AIN lattice constants of c-axes were increased along with the synthesis period. It is well known that a change in c-axis lattice parameter is caused by dissolution of oxygen in A1N.37Thus, these results suggest that residual oxygen in AIN nanopowder is reduced by reduction nitridation. XRD analysis shows the presence of AIN only, with no y-A1203detectable in samples fired at 1200°C for 120 min, even though the nitridation ratio indicates that oxygen still remains in these samples. The oxygen contents in samples Prepared at 1400°C for 60-120 min are mainly from the surface oxygen layer because of lattice contents were almost the same. However, resultant AIN possessed lower low oxygen contents
Synthesis Methods for Powders
(4 so 0 -60 Chemical shift (pprn) w r t Al(H20)63t
100
-100
*'A1 MAS NMR spectra of (a) the raw material and the products synthesized at (b) 110o"C, (c) 1200"C, and (d) 1400°C (*denotes
Fig.6.
spinning sidebands).
153
AILCLN products. The tetrahedral A104 sites of the y-alumina appear to be more susceptible to nitridation than the octahedral sites, which persist in the sample fired at 1200°C even after the tetrahedral resonance has disappeared. All the AIL0 resonances claim on have given way to the AIN peak by 1400°C. Min et the basis of quantum mechanical modeling that structural models for both y-A1203 and AlON with vacancies located at the tetrahedral sites are more unstable than those with vacancies located at the octahedral sites. Ydmaguchi and Yanagida4’ have reported that spinel phases such as y-AI2O3are stabilized not by AI3+ but by nitrogen. Thus, it appears that the A104 in the spinel are preferentially converted to AIN4, thereby stabilizing the tetrahedral sites and preventing their thermal conversion to A106 (the u-Al2O3structure contains only octahedral Al). It is well known that the bond length of A 1 4 in the A104 site is shorter that that of A106 site.42743Moreover, the covalent bond of AI-N is stronger than that of A 1 4 in A106 or A104. However, previous research on defective spinel structure model of y-A1203 utilizing the several simulations has demonstrated that cation vacancies are preferentially located at the octahedral Beside, nitrogen atoms preferentially replace oxygen atoms in A104 tetrahedron!’ These results show that the energy cost for a modest occupation of tetrahedral sites by nitrogen is small. From the above consideration, we suggest that results of preferentially nitridation of A104 sites are corresponding to previous model by simulations. Nitridation of y-AI2O3by the gas reduction-nitridation process is thus seen to be because of the particular susceptibility of the tetrahedral A104 site to nitridation, forming the tetrahedral crystal structure of AIN. Reaction at such low temperatures also results in restricted grain growth, producing nanosize AIN particles. When oxygen dissolves into the AIN lattice, vacancies form at Al sites and coalesce to generate A106 stacking faults. The residual oxygen in the initially formed products of this study may therefore occur in conjunction with stacking faults in the AIN. It seems therefore that the formation of AIN from y-AI2O3 progresses initially by the rapid conversion of A104 into AIN4, followed by the gradual nitridation of the residual A106 units.
IV. Conclusion AIN nanopowders were synthesized from y-A1203by gas reduction-nitridation over 1200”C, and that nitridation involved grain growth. Despite the temperature of the reaction, aA1203 was not generated. The tetrahedral A104 sites of the y-AI203 are nitrided to tetrahedral AIN, in preference to the octahedral A106 sites, but eventually all the AIL0 units are converted to nitride. The relatively low reaction temperature of the process enables AIN nanoparticles readily to be formed from y-AI203.
References ’V. A. Virkar, T. B. Jackson, and R. A. Cutler, “Thermodynamic and Kinetic Effects of Oxygen Removal on the Thermal Conductivity of Aluminum Nitride,” J. Am. Ceram. Soc., 72 [ 1 I] 203 1 4 2 (1992). ’Y. Baik and R. A. L. Drew, “Aluminum Nitride: Processing and Applications Key,” Eng. Muter., 122-, 553-70 (1996). ’I. Kimura, K. Ichiya, M. Ishii, N. Hotta, and T. Kitamura, “Synthesis of Fine AIN Powder by a Floating Nitridation Technique Using an N2/NH3Gas Mixture.” J. Mater. Sci. Lett., 8, 3 0 3 4 (1989). 4K. Komeya, N. Matsukaze, and T. Megro, “Synthesis of AIN by Direct Nitridation of Al alloys,” J Ceram. Soc. Jpn., 101, 1319-23 (1993). ’K. G . Nickel, R. Riedel, and G. Petzow, ‘Thermodynamic and Experimental Study of High-Purity Aluminum Nitride Formation from Aluminum Chloride by Chemical Vapor Deposition,” J . Am. Ceram. Soc., 72, 1804-10 (1989). ‘J. R. Park, S.-W. Rhee, and K.-H. Lee, “Gas-Phase Synthesis of AIN Powders from AIC13-NH3-N2.” J . Mater. Sci., 28, 57-64 (1993). ’R. Riedel and K . 4 . Gaudl, “Formation and Characterization of Amorphous Aluminum Nitride Powder and Transparent Aluminum Nitride Film by Chemical Va or Deposition.” J . Am. Ceram. Soc., 74, 13314 (1991). Bachelared and P. Joubert, “Aluminum Nitride by Carbothermal Nitridation,” Marer Sci. Eng., A109, 247-51 (1989).
ak.
154
’P. Lefort and M. Billy, “Mechanism of AIN Formation Through the Carbothermal Reaction of A1203in Flowing N2 Atmosphere,” J . Am. Cerom. Soc., 76. 2295-9 (1993). ”H.-K. Chen, C.4. Lin, and C. Lee, “Kinetics of the Reaction of Carbon/Alumina Powder Mixture in a Flowing Nitrogen Stream,” J . Am Ceram. Soc.. 77. 17534 (1994). “R. G. Odonnell and M. B. Trigg. “The Mechanism of Conversion of A1203to AIN Via Carbothermal Synthesis,” Micron, 25, 575-9 (1994). ”A. Tsuge, H. Inoue, M. Kasori, and K. Shinozaki, “Raw Material Effect on AIN Powder Synthesis from A1203Carbothermal Reaction,” J . Marer. Sci., 25. 235941 (1990). I3Y. W. Cho and J. A. Charles, “Synthesis of Nitrogen Ceramic Powders by Carbothermal Reaction and Nitridation: Part 3 Aluminum Nitride,” Matrr. Sci. Tecknol., 7,495-504 (1991). I4D. Qian, T. Meguro, Z. Xiong, J. Xie, and K. Komeya, “Effect of AIN and Raw Materials on Synthesis of AIN Powder by Carbothermal Reaction-Nitridation,” J. Mater. Sci. Soc. Jpn., 32, 82-6 (1995) (in Japanese). ‘’K. Baba, N. Shohata, and M. Yonezawa, “Synthesis and Properties of UItrafine AIN Powder by R F Plasma,’’ Appl. Pkys. Lett., 54, 2309-1 1 (1989). 16M. Iwata. K. Adach, S. Furukawa, and T. Amakawa, “Synthesis of Purified AIN Nan0 Powder by Transferred Type Arc Plasma,” J . Fhys. D:Appl. Pkys., 37, 1?:-7 (2004). S. Yu, D. Li, H. Sun, H. Li, H. Yang, and G. Zou, “Microanalysis of Singlephase AIN Nanocrystals and AIN-AI Nanocomposites Prepared by DC Arc-Discharge,” J . Crysral Growth, 183, 284-8 (1998). ’*F.J. Moura and R. J. Munz, “Vapor-Phase Synthesis of Nanosize Aluminum Nitride Particles Using a Two-Stage Transferred Arc Reactor.” J. Am. Ceram. Soc., 80 [9] 2425-8 (1997). ”Y. Sakka, H. Okuyama, T. Uchikoshi, and S. Ohno, “Characterization of Degraded Surfaces of Al Ultrafine Powders,” NonoStruct. Mater., 5, 577-88 (1995). ’OY. Qiu and L. Gao, “Nitridation Reaction of Aluminum Powder in Flowing Ammonia,” J . Eur. Ceram. Soc., 23, 2015-22 (2003). ”Y. Qiu and L. Gao, “Novel Way to Synthesis Nanocrystalline Aluminum Nitride from Coarse Aluminum Powder,” J . Am. Ceram. Soc., 86 [q 1214-6 (2003). ”M.-C. Wang, N.-C. Wu, M . 3 Tasi, and H.-S. Liu, “Preparation and Characterization of AIN Powders in the AIClrNH3-N2 System,” J . Cryst. Growth., 216,69-79 (2000). =B. C. D. Lello, F. J. Moura, and I. G. Solorzano, “Synthesis and Characterization of Nano-Scale Aluminum Nitride Produced from Vapor Phase,” Marer. Sci. Eng. C, 15, 67-9 (2001). “S. Iwama, K. Haydkawa, and T. Arizumi, “Ultrafine Powders ofTiN and AIN Produced by a Reactive Gas Evaporation Technique with Electron Beam Heatin ” J . Crystal Growth. 56, 265-9 (1982). ”T. Suehiro, J. Tatami, T. Megro, S. Matsuo, and K. Komeya, “Synthesis of Spherical AIN Particles by Gas-Reduction-Nitridation Method,” J. Eur. Ceram. Soc., 22, 5 2 1 4 (2002). ’q.Suehiro, J. Tatami, T. Megro, K. Komeya, and S. Matsuo, “Aluminum Nitride Fibers Synthesized from Alumina Fibers Using Gas-Reduction-Nitridation Method,” J . Am. Ceram. Soc., 85 [3] 715-7 (2002). 27T.Suehiro, N. Hirosaki, R. Terao, J. Tatami, T. Megro, and K. Komeya, “Synthesis of Aluminum Nitride Nanopowder by Gas-Reduction-Nitridation Method,” J . Am. Ceram. Soc., 86 [6] 10468 (2003). =. Y. I.Kim, M. A. Sriram, P. H. Mcmichael, P. N. Kumta, B. L. Phillips, and S. H. Risbud, “New Molecular Precursors from the Reaction of Hydrazine and Aluminum Alkoxide for the Synthesis of Powders in the A I a N System,’’ J. Pkys. Ckem. B., I01 I241 4689-96 (1997). -M. E. Smith, “Observation of Mixed AI(0, N)4 Structural Units by ”AI M a ic Angle Spinning NMR,” J . Phys. Ckem., 96, 1444-8 (1992). C. Ekstroem, K. J. D. MacKenzie, V. V. White, 1. W. M. Brown, and G . C. Barris, ‘Volatile Products Formed by Carboreduction and Nitridation of Clay Mixtures with Sillica and Elemental Silicon,” J . Mater. Ckem., 96,1225-30 (1996). ”K. J. D. MacKenzie, R. H. Meinhold, G. V. White,C. M. Sheppard, and B. L. Sherriff, “Carbothermal Formation of p’-Sialon from Kaolinite and Halloysite Studied by 29Si and 27AISolid State MAS NMR,” J. Muter. Sci., 29 [lo] 261 1-9 (1 ~ 4 ) . K. J. D. MacKenzie, “Solid State Multinuclear N M R A Versatile Tool for Studying the Reactivity of Solid Systems,” Store lonics, 172, 383-88 (2004). 33T.Brauniger, P. Kemp ens, R. K. Harris, A. P. Howes, K. Liddell, and D. P. Thompson, “A Combined F4N/Z7AlNuclear Magnetic Resonance and Powder Xray Diffraction Study of Impurity Phase in p’-Sialon Ceramics,” Solid Stare Nucl. Ma n Reson., 23 [I-21 62-76 (2003). 4 . . V . Nicolaescu, G. Tardos, and R. E. Riman, “Thermogravimetric Determination of Carbon. Nitrogen, and Oxygen in Aluminum Nitride,” J . Am. Ceram. Soc., 77 [9] 2265-72 (1994). ”N. Hashimoto, Y. Sawada, T. Bando, H. Yoden, and S. Deki. “Preparation of Aluminum Nitride Powder from Aluminum Polynuclear Complexes,” J. Am. Cerum. Soc., 74 [6] 12824 (1991). xE. Ponthieu, P. Grange, B. Delmon. L. Lonnoy, L. Leclercq, R. Bechara. and J. Grimblot, “Proposal of a Composition Model for Commercial AIN Powder,” J . Eur. Ceram. Soc., 8, 23341 (1991). 37G. A. Slack, “Nonmetallic Crystals with High Thermal Conductivity,” J . Ph s Ckem. Solids.. 34. 321-35 (1973). ‘K. J. D. MacKenzie and M. E. Smith, Mulrinuclear Solid Stare N M R 1100°C and time > 10 h. Monazite/xenotime coatings heat treated at temperatures up to 1100°C and time 100 h are not hermetic.
(the green line) is shown for comparison with the actual (Y,La)monazite/xenotime pattern (the black line). Prominent monazite and xenotime rings are identified with the d-spacings. As expected, some of t h e d-spacings were slightly shifted in the (Y,La)-monazite solid-solution. Both monazite and xenotime are present in the (Y,La) monazite/xenotime coating in roughly equal amounts. A thin film of amorphous AIP04 was present at the fiber/ coating interface (Fig. 6). Trace amorphous Alp04 was previously observed in fiber coatings made from rod-shaped rhabdophane particles, and was inferred to form by reaction of slight excess coating phosphorous with A1203 in the f i b e r ~ . ' ~ In coatings made with 1 wt% alumina doped sols, AIP04 films were not observed at the fiber-coating interface (Fig. 8). SEM observations show significant variations in coating thick184
-
ness 2 pm to (Fig. 9).
-
50 nm along and between fiber filaments
(4) Coating Grain Growth Lap04 and( Y, La) PO41 YPOd Grain growth of (Y,La)-monazite/xenotime coatings heat treated at I 100"-1300°Cfor I , 10, and 100 h is compared with LaP04 coatings made from similar precursors3' (Figs. 1C13). TEM and SEM observations show slower grain growth rates in (Y,La)-monazite/xenotime coatings (Figs. 10 and 1 I). The grain growth kinetics was evaluated using SEM data from Table 11, the results of which are shown in Fig. 12. The growth kinetics was represented by an equation of the form4': d - do = k( T)f'
(2) Progress in Nanotechnology: Processing
Table 11. Average Grain Size (nm) in Monazite (M) and Monazite/Xenotime (M/X) Coatings T 1ooo"c
1 IOWC
1300°C
1200°C
I
M
MIX
M
MIX
M
MIX
M
MIX
Ih 10 h 100h
35 45 48
37 41 45
53 76 93
47 63 75
140 200 256
96 130 156
268 3 14 597
157 213 302
of [In(d-d,)-nIn(t)] vs. l/Tfor monazite and (Y,La)-monazite/ where do is the initial and d is the final grain size, k(T) is a temxenotime grain growth are shown in Fig. 13. Q for monazite perature dependent growth constant, r is time, and n is a growth grain growth was 135 and 109 kJ/mol for monazite/xenotime exponent (Fig. 12). The growth exponents for monazite (0.1@ grain growth. These values were smaller than those found 0.23) were consistently larger than those for monazite-xenotime for CeP04 monazite (192 kJ/mol) and ErP04 xenotime (159 mixtures (0.084.12). All measured growth exponents were kJ/mol). much smaller than those observed for Ce-monazite (0.5)50and Er-xenotime (0.33).5' YP04 xenotime sintering rates are reported to be lower than those. for Lap04 monazite.22 Coating porosity and thin film constraints may be at least partly responsible ( 5 ) Coated Fiber Strength for the low growth exponent^.^^,^^ Also, two-phase. mixtures The cause of fiber strength degradation during fiber coating has may alter the grain growth rates by grain boundary pinning or been inferred to be high temperature stress corrosion from coatZener But in this particular situation the Zener ing precursor decom osition Froducts trapped in the coating model was not applied due to the following reasons: Experi(Table I and Fig. 3). '2-15339*5 s9 For rod-shaped La-rhabdop mental evidence shows that during heat treatment, the YP04 hane, high as-coated fiber strength correlated with low precursor phase forms a solid solution with the LaP04 phase. Powders weight loss at high temperatures. As-coated fibers made with heat treated at 1000°C for 100 h had 16.2 at.wt% of Y dissolved equiaxed La-rhabdophane precursor had high ascoated in LaP04. Also, powders heat treated at 1400°Cfor 1 h had 33.3 strength despite significant precursor weight loss above at.wt% of Y in LaP04.45The volume fraction of LaP04 will 1100°C39(Table I, Figs. 3 and 14). However, unlike rod-shaped therefore vary with the heat treatment time and temperature. In precursor, the fibers coated with equiaxed monazite had a conaddition, the grain growth study show that both the YP04 and siderable strength loss after long-term heat treatment for 100 h LaP04 phase grow whereas in the Zener model the inclusion is at 1100°C. Similar to equiaxed La-rhabdophane coated fibers, assumed to be inert and its size is fixed. Finally the kinetics of monazite/xenotime as-coated fibers made with the equiaxed growth of xenotime are slower than that of m ~ n a z i t e . ~ " ~ ~ ' ~ ' (Y,La)-rhabdophane ~' precursors had high as-coated strength k(7) is expressed by the usual Arrhenius equation: despite significant precursor weight loss above 1100°C (Table I, Figs. 3 and 14). However, their response to long-term heat treatment was rather similar to the rod-shaped pure La-mona(3) zite coatings. These fibers retained their high strength after heat treatment for 100 h at 1100°C (Fig. 14).'3s'4339 Nextelm where Q is an activation energy, R is the gas constant, T is 720 fibers previously coated with rod-shaped LaP04 had the temperature ("K) and c is a rate constant. Arrhenius plots strengths of 1.7550.05 GPa after 110o"C/100 h heat-treatment (Fig. 14). The corresponding values for equiaxed monazite and (Y,La)-monazite/xenotime mixtures were 1.38 kO.06 GPa and 1.76k0.05 GPa. The corresponding strength of uncoated fiber was 1.80&0.07 GPa. SEM and TEM studies of the coated fibers heat treated at 1100"C/100h showed the equiaxed monazite coatings to densify
?
-
Fig.12. Isothermal grain growth in thick equiaxed LaP04 and (Y,La)P04 fiber coatings heat treated at 1000"1300"C for 1-100 h. Growth exponents are shown on the graph for each heat-treatment temperature.
Membranes, Films, and Coatings
Fig.13. Arrhenius plots of equiaxed LaP04 and (Y,La)P04 grain growth for 1, 10, 100 h showing the calculated activation energies. 185
Fig. 14. Tensile strengths of equiaxed Lap04 coated Nextel 720 fibers; (a) as-coated fiber and (h) heat treated at 1 lOO”C/lOOh.
after heat treatment at IlOO°C/lOO h, whereas the monazite/ xenotime coating given a similar heat treatment does not (Fig. 11). In principle dramatic grain growth rate changes are expected as the rate controlling mechanism changes from coarsening by surface diffusion in porous non-hermetic coatings to grain boundary and lattice diffusion in denser coatings. This transition is diagnostic of open porosity transitioning to closed porosity in the coatings. This transition was not observed for either monazite or monazite/xenotime grain growth exponents or activation energies from 1000” to 1300°C (Figs. 12 and 13), although TEM observations of coatings heat treated at 1 100”C/ 100 h showed the monazite coatings to be at least locally hermetic and monazite/xenotime coatings to be non hermitic (Fig. 11). A tentative explanation for the strength loss after long-term heat treatment for coated fiber made from equiaxed Larhabdophane may relate to the high surface area and high sintering rate of equiaxed particles (Figs. 10-12). Equiaxed particles may adsorb more precursor decomposition products than rods of the same diameter. These desorb at higher temperatures. Some decomposition products may be trapped in coating pores adjacent to the fiber, and cause growth of fiber surface flaws by stress corrosion cracking driven by intergranular residual stress. The two-phase monazite/xenotime coatings was non hermetic (Fig. 11). Hence the less rapid densification of the two-phase coatings may trap a smaller amount of decomposition products adjacent to the fiber. Therefore, strength degradation by the two-phase coating was less severe.
IV. Conclusions A single precursor two-phase monazite-xenotime mixture (Y,La)PO4.0.7Hz0 was made and used to coat Nextel 720 fiber tows. The precursor was characterized by TGA/DTA and XRD studies before fiber coatings. The precursor transformed to a mixture of a (Y, La)-monazite and xenotime at -950°C. Coatings heat treated in-line at 1100°C formed a mixture of monazite/xenotime with a 13 nm grain size. Grain growth of two-phase monazite/xenotime coatings was slower than that of pure monazite coatings. The growth exponents for monazite (LaP04) were consistently larger than those for monazite-xenotime [(Y,La)P04/YP04]mixtures. As-coated fiber strength was not degraded by the monazite/xenotime coating derived from nanosized equiaxed particles, despite significant weight loss from the coating precursor at high temperature.
-
186
After heat treatment at I lO0”CjlOO h, the monazite/xenotimecoated fiber retained strength, while the monazite coated fiber was degraded in strength. Coated fiber strength loss may be related to the relative densification rates of the two coatings.
References ‘P. E. D. Morgan and D. B. Marshall, “Ceramic Composites of Monazite and Alumina,” J. Am. Ceram. Soc., 78, 1 5 5 3 4 3 (1995). *K. A. Keller ef a/., “Effectiveness of Monazite Coatings in Oxide/Oxide Composites After Long Term Exposure at High Temperature,” J. Am. Ceram. Sac., 86, 325-32 (2003). ’P.-Y. Lee, M. Imai, and T. Yano, “Effects of Sintering Condition on Mechanical Properties of Monazite-Coated Alumina-Fiber/Alumina-Matrix Composites Fabricated by Hot-Pressing,’’ J. Ceram. Soc. Japan, 112, 29-34 (2004). 4R. J. Kerans, R. S. Hay. T. A. Parthasarathy, and M. K. Cinibulk, “Interface Design for Oxidation Resistant Ceramic Composites,” J. Am. Ceram. Sac., 85, 25!9-632 (2002). F. W. Zok, “Developments in Oxide Fiber Composites,’’ J. Am. Ceram. Soc., 89, 330%24 (2006). 6A. G. Evans and J. W. Hutchinson, “Overview No. 120, The Thermomechanical Integrity of Thin Films and Multilayers.” Acta Mefall. Mafer., 43, 2507-30 c l y . E. E. Boakye, P. Mogilevsky, J. Welter, R. S. Hay, and R. J. Kerans, “Monazite Coatings on SIC Fibers I: Fiber Strength and Thermal Stability,” J. Am. Ceram. Soc. 89 [ I I], 3475-80 (2006). ‘D. J. Srolovitz and S. A. Safran, ‘Capillary Instabilities in Thin Films. I. Ener etics,” J . Appl. Phys., 60, 247-54 (1986). ’D. J. Srolovitz and S. A. Safran, “Capillary Instabilities in Thin Films. 11. Kinetics,” J. Appl. Phys., 60, 255-60 (1986). ‘OK. T. Miller, F. F. Lange, and D. 9. Marshall, “The Instability of Polycrysd i n e Thin Films: Experiment and Theory,” J. M a w . Res., 5, 151-60 (1990). “E. Werner, “The Spheroidization of Thin Plates,” Acfa. Mefall.,37, 2047-53 y9). E. Boakye, R. S. Hay, and M. D. Petry, “Continuous Coating of Oxide Fiber Tows Using Liquid Precursors: Monazite Coatings on Nextel 720,” J. Am. Ceram. So:? 82. 2321-31 (1999). E. E. Boakye, R. S. Hay, P. Mogilevsky, and L. M. Douglas. “Monazite Coatings on Fibers: 11, Coating without Strength Degradation,” J. Am. Ceram. Soc., 84, 2793-801 (2001). I4E. E. Boakye and P. Mogilevsky, “Fiber Strength Retention of La and Ce-P04 Coated Nextel’“ 720,” J. Am. Ceram. Soe., 87,314-6 (2004). ”R. S. Hay and E. Boakye, “Monazite Coatings on Fibers: 1, Effect of Temperature and Alumina Doping on Coated Fiber Tensile Strength,” J. Am. Ceram. Soc., 84, 2783-92 (2001). 16D. M. Wilson, “New High Temperature Oxide Fibers.” pp. 1-12 in High Temperalure Ceramic Matrix Composifes, Edited by W. Krenkel and R.H.N. Schneider. Wiley-VCH, Weinheim, Germany, 200 I . I’D. M. Wilson and L. R. Visser, “High Performance Oxide Fibers for Metal and Ceramic Composites,” Composites A, A32, 1143-53 (2001). “D.-H. Kuo and W. M. Kriven, “Characterization of Yttrium Phosphate and a Yttrium Phosphate/Yttrium Aluminate Laminate,” J. Am. Ceram. Soc., 78, 31214 (1995). I9D. 9 . Marshall, J. 9. Davis, P. E. D. Morgan, and J. R. Porter, pp. 27-36, Key Engineering MareriaLr. Trans Tech Publications. Switzerland, 1997.
Progress in Nanotechnology: Processing
2”P. E. D. Morgan and D. B. Marshall, “Functional Interfaces for Oxide/Oxide Composites,” Muter. Sci. Eng.,A162, 15-25 (1993). ”Y. Hikichi, T. Ota, K. Daimon, T. Hattori, and M. Mizuno, “Thermal, Mechanical, and Chemical Properties of Sintered Xenotime-Type RP04 (R = Y, Er, Yb. or Lu),”J . Am. Ceram. SOC.,81, 2 2 1 H (1998). 22D.Bregiroux, S. Lucas, E. Champion, F. Audubert, and D. Bernache-Assollant, “Sintering and Microstructure of Rare Earth Phosphate Ceramics R D O 4 with R E = La, Ce, or Y,” J. Eur. Cerum. Soc., 26, 2 7 9 4 7 (2006). ”K. B. Alexander, P. F. Becher, S. B. Waters, and A. Bleier, “Grain Growth Kinetics in Alumina-Zirconia (CeZTA) Composites,” J. Am. Cerum. Soc., 77, 93z446 (1994). J. D. French, M. P. Harmer, H. M. Chan, and G . A. Miller, “Coarsening Resistant Dual Phase Interpenetrating Microstructures,” J . Am. Cerum. SOC.,73. 250910 (1990). 2SF. F. Lange and M. M. Hirlinger, “Grain Growth in Two-Phase Ceramics: Al O3 Inclusions in 2102,’’ J. Am. Cerum. Soc., 70, 827-30 (1987). “L. Bo, S. Liya, L. Xiaozhen, Z. Shuihe, W. Chunfang, and L. Wenjing, “Monazite Coatings on Short Alumina Fibers Using Layer-by-Layer Assemby Techni M,”Muter. Sci. Eng., A354,3 2 6 3 2 (2004). q7V. Buissette, M. Moreau, T. Gamin, and J.-P. Boilot, “Luminescent Core/ Shell Nanoparticles with a Rhabdophane LnP04.xH20 Structure: Stabilization of Ce’+-Doped Compositions.” Adv. Funct. Muter., 16, 351-5 (2006). 28V.Buissette, M. Moreau, T. Gamin, and J.-P. Boilot, “Colloidal Synthesis of Luminerent Rhabdophane LaP04:Ln3+.xH20(Ln = Ce, Tb, Eu; x = 0.7) Nanocrystals.” Chem. Muter., 16, 3767-73 (2004). 29 H.-K. Jung, J . 3 . Oh, S. Seok, and T.-H. Lee, “Preparation and Luminescence Properties of LaPO4:Er,Yb Nanoparticles,” J . Lumin., 114, 2005 (2005). ’OS. Lucas, E. Champion, D. Bernache-Assolant, and G. Leroy, “Rare Earth Phosphate Powders ReP04.nH20 (Re = La, Ce, or Y) - Part 11. Thermal Behavior,” J. Solid-SfufeChem., 177, 1312-20 (2004). ”S. Lucas, E. Champion, D. Bregiroux, D. Bernache-Assolant, and F. Audubert, “Rare Earth Phosphate Powders ReP04.nH20 (Re = La, Ce, or Y) - Part 1. Synthesis and Characterization,” J. Solid-Sfare Chem., 177, 1302-1 I ( 2 W ) . ”T. Masui, H. Tategaki, S. Furukawa, and N. Imanaka, “Synthesis and Characterization of New Environmentally-Friendly Pigments Based on Cerium Phosphate,” J. Cerum. Soc. Jupun. 112, 646-9 (2004). ”J. M. Nedelec, D. Avignant, and R. Mahiou, “Soft Chemistry Routes to YPOCBased Phosphors: Dependence of Textural and Optical Properties on Synthesis Pathways,” Chem. Muter., 14,651-5 (2002). ’‘J. M. Nedelec, C. Mansuy, and R. Mahiou. “Sol-Gel Derived YPO4 and LuPO4 Phosphors, a Spectroscopic Study,” J . Mol. Str., 651653, 165-70 (2003). ”H. Oncda, H. Nariai, H. Maki, and I. Motooka, “Syntheses of Various Rare Earth Phosphates from Some Rare Earth Compounds,” Mafer. Chem. Phys., 73, 19-23 (2002). 36K. Rajesh, P. Mukundan, P. K. Pillai, V. R. Nair, and K. G. K. Warner, “High-Surface-Area Nanocrystalline Cerium Phosphate through Aqueous SolGel Route,” Chem. Mater., 16, 2 7 W 5 (2004). 37K. Rajesh e f ul., “Synthesis of Nanocrystalline Lanthanum Phosphate for Low Temperature Densification lo Monazite Ceramics,” Muter. Letf.,58,1687-91 (2004). ”E. E. Boakye, P. Mogilevsky, and R. S. Hay, “Synthesis of Spherical Rhabdophane Particles,” J . Am. Cerum. Soc., 88,274M (2005).
Membranes, Films, and Coatings
39 E. E. Boakye, R. S. Hay, and P. Mogilevsky, “Spherical Rhabdophane Sols 11. Fiber Coating,” J . Am. Ceram. Soc., 90 [5]. 1580-8 (2007). 40R.S. Hay, J. R. Welch, and M. K. Cinibulk. “TEM Specimen Preparation and Characterization of Ceramic Coatings on Fiber Tows,” Thin Solid Filmy, 308-309, 389-92 (1997). 4’M. K. Cinibulk, J. R. Welch, and R. S. Hay, “Method for Preparation of TEM Specimens of Coated Fibers,” J. Am. Cerum. Soc., 79, 2 4 8 1 4 (1996). 42M. D. Petry, T. Mah, and R. J. Kerans, “Validity of Using Average Diameter for Determination of Tensile Strength and Weibull Modulus of Ceramic Filaments,” J . Am. Ceram. Soc., 80, 2 7 4 1 4 (1997). 47E.E. Boakye, R. S. Hay, and P. Mogilevsky, “Spherical Rhabdophane Sols 11. Fiber Coating,” J . Am. Ceram. Soc., 90, 1580.8 (2007). %. Min, K. Daimon, T. Ota, T. Matsubara, and Y.Hikichi, “Synthesis and Thermal Reactions of Rhabdophane-(Yb or Lu),” M u m . Res. Bull., 35,219%205 (2ow. “P. Mogilevsky, E. E. Boakye, and R. S. Hay, “Solid Solubility and Thermal Ex ansion in LaP04-YP04 System,” J. Am. Cerum. Soc., 90, 1899-907 (2006). ‘M. K. Carron, C. R. Naeser, H. J. Jr. Rose,and F. A. Hildebrand, “Fractional Precipitation of Rare Earth with Phosphoric Acid,” U. S . Geol. Surv. Bull., 1036, 253-275 (1958). 41 E. E. Boakye, R. S. Hay, and P. Mogilevsky, “Spherical Rhabdophane Sols II. Fiber Coating,” J . Am. Ceram. Soc., 90 [5], 158C-8 (2007). “G. F. Fair. R. S. Hay, and E. E. Boakye, ”Precipitation Coating of Monazite on Woven Ceramic Fibers: 1. Feasibility,” J. Am. Cerum. Soc., 90, 44-55 (2006). 49J. W. Martin, R. D. Doherty, and B. Cantor, “Microstructural change due to Grain Boundary Energies”, pp. 3 0 7 4 6 in Srubility of Microstrucfure in Metallic Sysfems, Edited by R. W. Cahn. University Press, Cambridge, 1997. 9. Hikichi, T. Nomura, Y. Tanimura, S. Suzuki, and M. Miyamoto. “Sintering and Properties of Monazite-Type CeP04,” J . Am. Cerum. Soc.. 73, 3 5 9 4 6
(1990).
”I. A. Bondar, A. I. Domanski, L. P. Mezentseva, M. G. Degen, and N. E. Kalinina, “A Physicochemical Study of Lanthanide Orthophosphate,” Russ. J. Inorg. Chem., 21, I 1 2 6 9 (1976). ”M. Stech, P. Reynders, and J . Rcdel, “Constrained Film Sintering of Nanocrystalline TiOZ.” J. Am. Cerum. Soc., 83, 1889-96 (2000). ”M. N. Rahaman, Cerumic Processing undSintering. Marcel Dekker Inc., New York, 1995. “N. Maazi and N. Rouag, ‘Consideration of Zener Drag Effect by Introducing a Limiting Radius for Neighbourhocd in Grain Growth Simulation,” J. Crystul Growth, 243, 361-9 (2002). S5E.Nes, N. Ryum, and 0. Hunderi, “On the Zener Drag,” Acra Metall., 33, 11-22 (1985). 56M.N. Rahaman, R. E. Dutton, and S. L. Semiatin, “Effect of Solid Solution Additives on the Densification and Creep of Granular Ceramics,” Acru Muter., 45, 3017-28 (1997). ”N. Moelans, B. Blanpain, and P. Wollans, “Pinning Effect of Second-Phase Particles on Grain Growth in Polycrystalline Films Studied by 3-D Phase Field Simulations,” Acfu Muter., 55, 2173-82 (2007). ’*E. E. Boakye, R. S. Hay, M. D. Petry, and T. A. Parthasarathy, “ZirmniaSilica-Carbon Coatings on Ceramic Fibers,” J Am. Cerum. Soc., 87, 1%7-76 (2004). 59R. S. Hay, E. E. Boakye, and M. D. Petry, “ElTect of Coating Deposition Temperature on Monazite Coated Fiber,” J . Eur. Ceram. Soc., 20, 589-97 (2000). 0
187
This Page Intentionally Left Blank
Template-Free Self-Assembly of a Nanoporous TiOz Thin Film Yanfeng Gaot and Masayuki Nagai Advanced Research Laboratories, Musashi Institute of Technology, Tokyo158-0082, Japan
Won-Seon Seo Korea Institute of Ceramic Engineering & Technology (KICET), Seoul, Korea
Kunihito Koumoto Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan
We report a nanoporous TiOl thin film prepared using a supersaturated aqueous solution containing peroxotitanium complex ions. The film morphology can be regulated by chemical kinetics, which was partially controlled by solution conditions such as the concentrations of starting materials, pH values, and the temperatures of the solutions. Porous films with various morphologies from particulate to curved sheet shaped were prepared on different substrates including Si, polymers, and glass. Porous microstructures of films permitted us to prepare a crack-free film 2-3 pm in thickness. These films were still amorphous under the present treatment conditions. Dye-sensitized solar cells using annealed (anatase) films of different morphologies as electrodes yielded conversion efficiency ranging from 1.3% to 3.1%. Improvement in performance may be achieved by either increasing the film thickness or inducing crystallization in solutions. 1. Introduction OROUS Ti02 with
P
a tailored pore size of the order of nano to micrometers has attracted considerable attention because of potential applications including controlled drug delivery, photo catalysis, energy conversion, filtration, and biomedical membranes.' Although a variety of template-based approaches2 have been developed for the production of porous Ti02 thin films, the creation of high-quality structures, preferably over large areas, uniformly and at a low cost, is still a challenging problem.' Inspired by procedures used by living creatures in nature, where high-performance materials (magnets in magnetotactic bacteria, ferritine, teeth, bone, shells, etc.) are produced by self-assembling of highly selective structure^,^ we have developed a template-free, one-step method for the preparation of a nanoporous continuous Ti02 thin film over a large area. One feature of this process is gas generation during film deposition. The release of gas during film deposition leaves voids among self-linked colloid particles and acts as a template for the self-assembled construction of a networked porous film, and the size of the voids can be tailored by controlling the rate of gas generation, which is associated with the reaction conditions, such as temperature and/or pH of the solution.
N. Padture-contributing editor
Manuscript No. 21846. Received May 29,2006 approved November 9,2006. This work was financially supported by Murata Science Foundation and a specific fund from Musashi Institute of Technology. 'Author to whom correspondence should be addressed. e-mail:
[email protected] Membranes, Films, and Coatings
Another feature of this process lies in the control of the film morphology by simply regulating the solution conditions (temperature, pH), which enables us to obtain deposits with various morphologies, from a sphere shape to a nanosheet shape. When used as an electrode in a dye-sensitized solar cell (DSSC), a nanoporous film (-2 pm in thickness) with a unique morphology shows an efficiency of 3.1 YOunder illumination by AM 1.5 (- 100 mW/cm2) simulated sunlight. Our method is different from traditional chemical solution approaches. It is based on the solution chemistry and the solid-liquid interface reaction. We began with the preparation of a transparent peroxotitanium solution. The dissolution of H2Ti03 resulted in the formation of peroxotitanate complexes represented by [TiO(02)(OH)2]2- at high pH (> 10). The existence of hydroxyl (OH) and peroxo groups in the chemical composition of the peroxotitanate complex suggests that low temperatures and high pH values are favorable to stabilize the peroxotitanate complex. On increasing the temperature (to room temperature) or lowering the pH value (to 2.4), OH and peroxo groups were released, and a TiOz-based solid phase was pre~ipitated.~ Under these conditions (room temperature, pH = 2.4), the formation of a sphere-shaped particulate film was confirmed by both SEM and atomic force microscope (AFM) observations! The fundamentals of this process have been developed in our previous research. This work, however, is not a simple extension of the previous reported work, but provides further scientific insights toward understanding the fundamentals of the process. The film deposition process in a supersaturated solution is controlled by several factors, in which the solution physical chemistry, nucleation, and crystal growth are involved. Understanding how the reaction proceeds is a prerequisite for regulation of the deposition conditions, such as pH, temperature, and additives, such that it is possible to control the film properties in terms of morphology, roughness, adhesion of films to substrates, and crystalline phases along with their crystallinity, sizes, and orientation. The examples can be seen in studies involving the crystal growth and design of artificial minerals such as CaC03, BaS04, silica, etc.' Figure 1 shows the schematic mechanism proposed for the formation of a nanoporous Ti02 film. Firstly, the nucleation and growth of TiO2-based films were performed by adsorption of clusters grown in the solution or through the heterogeneous mode under different conditions (a). The bubbles that are evolved from the solution are present among separated islands of deposits during the initial stage of nucleation and growth. The release of these gas bubbles opens up pathways among deposits, acting as a negative template for the construction of foam architecture (b). Oxygen is generated partially from the condensation process as shown in Eq. (4) (see Section III), and partially from the decomposition of hydrogen peroxide (H202).
-
189
Fig. 1. Schematic description of the formation mechanism for a nanoporous TiOz thin film.
In fact, the formation of bubbles is observed during the deposition process, which plays a significant role in the construction of a porous structure. Control of the size and the quantity of pores can be achieved by controlling the chemical kinetics of a series of reactions. When the evolution rate of oxygen gas is slower than the rate of particle growth and densification, the oxygen gas that is generated from the solution and sealed among colloid particles should create voids or a continuous path, permitting gas to escape from the substrate surface to the solution. The rates of gas generation, film deposition, and solid formation in the bulk solution are all related to the solution conditions, such as the concentrations of the reactants, pH values, and the temperatures of the solutions. By controlling these parameters, we can manipulate the size of the pores or even fabricate a dense film. 11. Experimental Procedure
( 1 ) Materials and Method The raw chemicals in this study were mainly titanic acid (H2Ti03, 97%, Mitsuwa Chem., Osaka, Japan), ammonia (NH3, 28% in water, Kishida, Osaka, Japan), hydrogen peroxide (H202,30% in water, Kishida), and deionized water (resistivity > 18.4 MQ . cm, Millipore, Tokyo, Japan). First, we prepared a transparent peroxotitanium complex solution by adding H2Ti03 to a solvent mixture containing an appropriate amount of NH3 and H202. The beaker, along with the chemicals in it, was cooled in an ice bath during this operation. The detailed description of this process has been published previously? The deposition solution contained 5 mM of Ti4+,and the solution pH was adjusted to 1.0-2.0 by adding an appropriate amount of HN03. cm, Shinetsu, Tokyo, Japan) The p-Si (resistivity: 5-10 glass with transparent conductive oxides (resistivity: 10 R cm, Asahi Glass, Tokyo, Japan), or polyethylene terephthalate (film type with a thickness of 50 pm, UNITIKA, Osaka, Japan) of 1 cm x 1 cm was used as a substrate. Before use the substrates were cleaned ultrasonically in acetone (99.5%, Kishida), ethanol (99.5%, Kishida), and deionized water. After drying at 5 0 T , it was exposed to ultra-violet light (184.9 and 253.7 nm; low-pressure mercury lamp, 200 W, PL21-200, 15 mW/cm2 for 254 nm, SEN Lights Co., Osaka, Japan) for 5 min. Deposition was conducted by soaking the substrates in the prepared solution containing 5 mM Ti4' at 20"-95"C for several hours to 48 h. The substrates were hung vertically in the solution with the glass side against the beaker wall. Subsequently, the substrate was rinsed with deionized water three times, and then dried at 50°C in air.
-
( 2 ) Integration into Solar Cells and Characterization In order to increase the adhesive property of the Ti02 film to the substrate, a thin layer of dense anatase Ti02 was grown on the F:Sn02 (FTO, < 12 Q/square, Asahi Glass) substrate by the liquid-phase deposition method before the substrate was treated in a peroxotitanium complex solution. Briefly, the FTO substrate was soaked at 90°C for 2 h in an aqueous solution containing 0.05M of (NH4)2TiF6 (W%, Stella Chemifa, Osaka, Japan) and 0.015M of H3BO3 (99.5%, Showa Chemical, Tokyo, 190
Japan); the starting pH of the solution was regulated to 2.8 by adding an appropriate amount of HN03.5b Porous Ti02 films were then produced repeatedly in fresh peroxotitanium complex solutions (5 mM Ti4+) at 95°C for 12 h on transparent conductive substrates under different pH. The thicknesses of films deposited five times were 2-3 pm. After crystallizing to anatase by annealing at 500"C, these films were immersed subsequently in a 0.5 mM Ruthenium535-bisTBA (N-719, chemical name: cis-bis (isothiocyanato) bis (2,2'-bipyridyI4,4'-dicarboxylato)ruthenium(I1):bis-tetrabutylammonium, produced by Solaronix, Aubonne, Switzerland)-ethanol solution for 12 h at room temperature. A Pt layer coated on a transparent conductive plate was used as a counter-electrode. A small amount of electrolyte solution, Iodolyte TG-50 (Solaronix), was loaded into the gap of two electrodes by the capillary effect. Simulation solar light from a Xenon lamp had an intensity of 100 mW/cm2 (AM1.5) (Solar Simulator: Model YSS-50, Yamashida Denso, Tokyo, Japan). The distance from light exit to cell was 12 cm. The area of the Ti02 electrode was 0.28 cm2. An evaluation system (R0240A, Advantest, Tokyo, Japan) was used to measure the current-voltage (J-V) characteristics of the solar cells.
(3) Characterization Other Properties The film morphology was observed with a field-emission scanning electron microscope (FE-SEM, JSM-6700F, JEOL, Tokyo, Japan) and a scanning probe microscope (SPM, SPI3800N, Seiko, Tokyo, Japan) operated to obtain AFM graphs of the films. The scanning frequency for SPM was 1-2 Hz. The calcined samples were also characterized with a transmission electron microscope (TEM, JEM4010, 400 kV, point-to-point resolution: 0.15 nm, JEOL). The phase composition was characterized using X-ray diffraction (XRD, RINT2100, 40 KV, 30 mA, CuKa, Rigaku, Tokyo, Japan) with a graphite monochromator. The scan was operated in the out-of-plane mode at a rate of I"/min. Transmittance was measured using an ultraviolet visible spectrophotometer (V-570, JASCO, Tokyo, Japan). The chemical compositions were investigated using an X-ray photoelectron spectroscope (Esca3200,8 kV, 30 mA; Pass Energy: 75 eV, Shimazu, Kyoto, Japan). A trace amount of Au was sputtered to the surface as a reference. The X-ray source was MgKa and all spectra were referenced to the Au4f signals. Raman spectra were measured using a JASCO NRS-2100 system (laser power: 10 mW, scan speed: 100 cm-I min-I). III. Results and Discussion Understanding the reaction equations is fundamental to regulate kinetically the reaction for obtaining the desired morphologies of films. The formation of peroxotitanium (IV) from either (titanium) Ti metal powders or Ti4+ has been reported to release protons and to produce a cationic complex ion of peroxotitanium (Eq. (I))?' These Ti species are hydrolyzed immediately in water, resulting in turbid solutions. For obtaining a homogeneous solution for film preparation, an additional amount of acid is usually added, and the solution temperature is usually maintained below or close to 0°C. In the present study, we selected a relatively stable, solid Ti(IV) chemical, titanium acid (H2Ti03), Progress in Nanotechnology: Processing
as the starting material for Ti(1V). Although H2Ti03is difficult to dissolve in water even in 30% H202, the addition of a large amount of ammonia shifts the reaction to the right-hand side (Eq. (l)), promoting the dissolution of H2Ti03 (Eq. (2)). The as-prepared solution is composed of a peroxotitanium complex in the anionic form (Eq. (2)). The component of peroxotitanium (IV) may change with a decrease in the solution pH, as a result of a condensation reaction among the peroxotitanium species. However, these condensation reactions may probably produce either H+ or OH- depending on the charge forms of peroxotitaniums,6b which are cationic or anionic depending on the solution P H . ~ ~ Therefore, the effect of solution pH on the condensation reactions is complex. Previous studies have suggested that protons can play the role of a reactant when the solution pH is decreased from -9 to a pH range of 24.’ At pH below 1, a mononuclear complex with a formula of Ti(02)’+ is formed. Condensation of this peroxotitanium species may release protons. Thus, increasing the solution pH may lower the rates of hydrolysis/condensation reactions toward the formation of Ti02-based precursors (Eq. (4)). The condensation of peroxotitanium complexes may precipitate various titanium oxide hydrates with or without residue of peroxo groups. One of these deposits has been determined to be Ti01.4(02)0.5(OH)0.2 when deposition was carried out close to pH 2.4?
Ti4++ H2Oz + Ti(Oz)(OH),-,
(4-n)+
+ nHt
TiO(OH),+HzO2+2OH-@ [TiO(02)(OH)2]2-+2H20 (2)
Ti(O2)”
+ nH20
@
Ti0,T(02),(OH),
+ 2 H+ + 0 2
(4)
The stability constant of peroxotitanium (IV) in different acidic media was found to decrease with increasing solution temperature,6b and both the decomposition of peroxotitanium to Ti (IV) ions and the hydrolysis reaction of Ti (IV) to solid phases are endothermal reactions.’ Thus, one can hope that treatment in solutions at relatively high temperatures can accelerate the formation of solids. On the basis of these reaction equations, we proposed to optimize the film microstructures by controlling the growth kinetics, i.e. by regulating the solution conditions: pH and temperature. Figure 2 shows scanning electron micrographs for the Ti02 precursor films prepared at pH 1.8 in solutions with different temperatures for various times along with those of anatase films obtained by annealing the as-deposited films. All the films show porous characteristics with pore sizes of tens to about 100 nm and an approximate thickness of 100-200 nm. The pore size increased with increasing solution temperatures. It seems that the solution temperature has significant effects on the film morphology; on increasing the soaking temperatures, typically from 20°C (Fig. 2(B)) to 30°C (Fig. 2(C)) to 50°C (Fig. 2(D)), the number of cracks decreased and a homogeneous porous film was obtained (Fig. 2(D)). Crack generation is associated with stress distribution in the film during annealing. Both substrate properties and film thickness affect the stress distribution. There should be, therefore, a certain critical thickness, beyond which films tend to form cracks during drying or heating.8a It is probable that the constructed porous structure itself can relax the stress and prevent cracks from generating during annealing. In fact, we have obtained films about 2-3 pm in thickness without apparent cracks. Compared with those in the as-deposited film prepared at room temperature (Fig. 2(A)), the cracks became large after annealing at 600°C (Fig. 2(B)). However, the film thickness remained almost unchanged. This finding suggests that the densification in the direction perpendicular to the substrate is not obvious, implying that the porous structure was not destroyed even after annealing, Membranes, Films, and Coatings
Fig. 2.
Scanning electron microscope photographs of TiOz thin films prepared under different conditions: (A) as deposited at room temperature (-20°C) for 48 h; (B) annealed at 600°C of film A; (C) deposited at 30°C for 48 h and annealed at 600°C for 2 h, (D) deposited at 50°C for 48 h and annealed at 600°C for 2 h. All the scale bars are 0.1 pm.
which was also the case for those prepared under other conditions. From the cross-sectional graph in Fig. 2(D)2, it can be seen that pores were present not only in the superficial layer but also throughout the film. At low temperatures (Figs. 2(B) and (C)), the formation of aggregates is not noticeable. However, at a relatively higher temperature (50”C), the adhesion of particles can be clearly observed. These colloid particles 10-30 nm in diameter linked with each other, enclosing voids of various sizes. The difference in deposit microstructures implies the various formation mechanisms. As we have discussed above, the rate for the solid formation was accelerated with increasing reaction temperature. The induction period obviously became short at relatively high temperatures; precipitates were formed in all solutions as deposition progressed on. A large amount of precipitates in the bulk solution enables films to grow by attachment of tiny particles from the solution at relatively high temperatures. To investigate the effects of solution pH on the film morphology, films were prepared at the same temperature of 95°C for 24 h in solutions at pH = 1.0, 1.5, and 2.0, respectively. As shown in Fig. 3, void sizes can be varied by regulating the solution pH, whose length increased with decreasing pH. At pH 1.0, thin films were constructed by distorted thin nanosheets of a TiOz precursor, forming a homogeneous surface with interconnected nanovoids. These nanosheets seem to have grown from a 191
inn
I.
PET film
I - -
300
500
700
900
1100 1300 1500 1700 1900
Wavelength / nm Fig. 5. UV-V spectra of Ti02 films deposited on polyethylene terephthalate (PET) films (conditions: pH 1.0, pH 1.5, pH 2.0; [Ti]5 mM, 95°C at pH 2 for 12 h).
Fig. 3. Scanning electron microscope photographs of TiOz precursor films deposited at 95°C for 24 h in 5 mM peroxotitanium complex solutions with different pH (Al, A 2 pH 1.0; BI, B 2 pH 1.5; C1, C2: pH 2.0).
central point, extending into a flower-like architecture (Fig. 3(A2)), which were also observed in a cross-sectional SEM photograph. This growth behavior may imply that the film was formed by a heterogeneous nucleation mechanism, where preferable growth on pre-existing nuclei resulted in competitive growth from one point to various directions. The aspect ratio of nanosheets decreased obviously on increasing the pH to 1.5 and 2.0. These SEM photographs also suggest the changes in surface roughness, increasing with decreasing pH. Further characterization of the porous film was performed with an AFM. The film surface clearly demonstrated porous characteristics; most of the pores were nanometers in size (Fig. 4). For the measured areas of 2 pm x 2 pm, the quantitative surface roughness (represented using root mean square) was
11, 21, and 51 nm for the film obtained at pH 2.0, pH 1.5, and pH 1.0, respectively. The relatively rough surface of that obtained at pH 1.0 may lead to the problem of poor contact of the film to the electrode when the film is applied to solar cells. It was reported that there were three kinds of pores in the TiOz powders prepared by the hydrolysis of titanium tetraisopropoxide in the presence of HNO, and NH40H under ultrasonic irradiation: fine intra-a gregated pores, larger interaggregated pores, and voids.8' The existence of voids and inter-aggregated pores in both the as-deposited and calcined films is obvious according to the present research results (SEM micrographs), whereas direct observation of intra-aggregated pores such as mesopores is limited to the resolution of SEM. Adsorption edges of ultra-violet spectra became sharp with increasing pH, suggesting that the surfaces of films became smooth and there was an increase in film density, which is in accord with the results of surface roughness and SEM observation. Note that the transmittance of sample pH 1.0 is only half of that of pH 2.0 (Fig. 5). The decrease in transmittance in the visible range may be due to the pores-resulted light scattering. A similar result was also reported for a film synthesized by the sol-gel method.8c For the as-deposited film, the crystallinity was characterized by both XRD (not shown here) and Raman spectroscopy. No crystalline phases such as anatase or rutile were detected by either XRD or Raman for the samples prepared at pH 1.5 and pH 2.0, suggesting that the film deposited under the present conditions was in an amorphous state. For integration into a solar cell, the film needs to be annealed for crystallization. For film prepared under pH 1.O, a small peak at about 51C520 cm-' was observed, which may be assigned to crystallized titaniarelated forms. However, almost no corresponding peaks were detected by XRD. Thus, the crystallization in this case may be very weak.
Fig.4. Atomic force microscope images ofTiOz thin films deposited under different pH values.
192
Progress in Nanotechnology: Processing
Fig.6. Scanning electron microscope graphs of Ti02 powders after annealing of those obtained under different pH; (A) pH 1.0, (B) pH 1.5, (C) pH 2.0; the scale bars are 1 pm.
For integration into solar cells, the samples were crystallized by annealing at different temperatures. Crystallization into anatase Ti02at 25Oo-3WC was confirmed by XRD. In this study, all the samples were calcined at 500°C for 30 min to increase their crystallinities. To collect sufficient amounts, the precipitates in solutions were collected for characterization by SEM and TEM. Figure 6 shows the SEM photographs for the samples after annealing. From the SEM graphs, the morphologies of the deposits changed slightly compared with those observed for Ti02 films on substrates. Highly porous Ti02 aggregates were obtained for that synthesized under pH 1.0 (Fig. 6(A)). Although the difference in sheet dimensions is visible, a single deposit collected from solutions is also sheet shaped (Fig. 6(A)), similar to that in thin films (Figs. 3(A1) and (A2)). The pores obviously became small when the pH increased to 1.5. Isolated particles were formed when the pH was further increased to 2.0. This result suggests that the formation of inter-aggregated and voids is closely related to the solution conditions, which may have effects on both the surface physical chemistry of colloid particles in solutions and the chemical compositions of precursor precipitates. Figure 7 shows the TEM photographs of Ti02 powders after annealing at 500°C. It seems that the crystals prepared under pH 1.0 are thin plate shaped, with particle sizes from several nanometers to about 50 nm in length (Fig. 7(A)). In a magnified graph (Fig. 8), the crystalline lattice fringes of the anatase (101) plane have clearly been observed. The sizes of powders obtained at pH 1.5 appear to be the same as those obtained at pH 1.O, but these powders seem to be aggregated from tiny nanoparticles (Fig. 7(B)). The particles obtained at pH 2.0 were found to be spindle shaped, which were also observed to be formed through aggregation (Fig. 7(C)).
Figure 9 shows the J-V curves of DSSC using the nanoporous Ti02 thin films as electrodes. The best result was obtained for the TiOz film prepared at pH 1.5, which gave a short-circuit current density (J,) of 13 mA/cm2, a fill factor (FF) of 37%, an open-circuit photovoltage (V0J of 662 mV, and a conversion efficiency of 3.1%. A device using a film prepared at pH 2.0 gave a conversion efficiency of 2.2%. Among three sample films, the film prepared at pH 1.0 showed poorer properties than the other two, which may be partially due to the large surface roughness of the film, and partially due to lowered transparency (see Fig. 5). Compared with V , of the DSSCs derived from pH 1.0 to pH 1.5 films, the J, increases largely from 3.9 to 13 mA/ cm2, more than a twofold increase. The V,, fluctuates between 662 and 632 mV, but the F F increased for the low-efficiency curves (0.37 vs 0.56). Accordingly, the increase in the overall conversion efficiency is entirely due to an increase of J,. The film obtained at pH 2.0 showed the largest value of V , = 730 mV. However, F F is obviously lower than those obtained for the film obtained at pH 1.5. In some cases, we also observed a rapid decrease in J, of a closed circuit, which may be caused by the unexpected reverse electron transport that weakened the charge separation or the limited redox reaction rate that reduced electron mobility. Table I lists the results of this study and those from representative studies reported in the l i t e r a t ~ r e . ~A- ' device ~ with the electrode prepared at pH 1.5 demonstrated conversion efficiencies similar to that obtained by electrosynthesis" or hydrothermal crystallization." Note that either electrosynthesis or hydrothermal crystallization was conducted using a suspension containing P-25 Ti02 nanoparticles.lO*" The film obtained by electrosynthesis was annealed at 450°C. Considering the film thickness, the conversion efficiency of the present film ( 2-3 N
Fig. 7. Transmission electron microscope graphs ofTiOz powders after annealing of those obtained under different pH; (A) pH 1.0, (B) pH 1.5, (C) pH 2.0; the scale bars are 50 nm.
Membranes, Films, and Coatings
193
GA-20 > GA- 10, as shown in Fig. 5(a). In the case of the GA-30 membrane, however, the gas permeance showed considerable stability in comparison with that of y-Al203 under hydrothermal conditions. For a single dopant, 30 mol% containing Gaz03 in y-A1203 was found to be the most effective under hydrothermal conditions at 500"C, as shown in Fig. S(a). The effects of La203 doping on the hydrothermal stability of yA1203 as well as GA-30 membranes were then studied, and the results are summarized in Fig. 5(b). The H2 permeance of LA-6 membranes increased gradually, approaching that of the m-A1203 porous support within 20 h in steam. It has generally been assumed that the presence of Laz03 on the y-A1203surface might immobilize the surface ions and thereby reduce sinterin of the material and impede the phase transition to a-A1203?g9 It is interesting to note, however, that La203-doped A1203 membranes have higher heat resistance than A1203membranes in the absence of steam.' The situation is totally different under hydrothermal conditions. In the case of LGA-10, the permeance increased to a slightly higher degree than that of the LGA-6 sample. In the case of LGA-15, a very large difference was observed between the LGA-6 sample and LGA-15. The results revealed that increasing the amount of Laz03in the sol from 10% to 15% in the GA-30 system was not advantageous for fabrication of a hydrothermally stable intermediate layer. During hydrothermal treatment, two distinguishable phenomena were observed: a drastic increase in permeance for 2 h and very slow progress of subsequent increases upon introduction of a stream containing 75% steam. After the first 2 h, the permeance of LGA-6 membrane had increased only ca. 2% in the presence of steam. This can be considered a remarkable result, as both the thermal and hydrothermal stability of the yA1203membrane was insufficient at I500°C."
( 4 ) Meso-Pore Structure The N2 permeation and the calculated pore size distribution (PSD), obtained by permporometry, for a supported as-prepared mesoporous y-A1203membrane are shown in Fig. 6(a) ( 0 ) . The calculated PSDs show the distribution of the total number of pores within a certain range. In the case of y-A1203,it can be seen that the permeation at large Kelvin radii was almost Progress in Nanotechnology: Processing
I
.-nc L-
2 0.8 0
al 0 C
m
kl n
0.6 0.4 0.2
0.0 0
2
4
6
2 10 0 Kelvin diameter [nm]
8
4
6
8
10
Fig.6. Pore size distribution of (a) mesoporous y-Alz03,(b) GA-30, (c) LA-6, and (d) LGA-6 membrane before ( 0 ) and after steam treatment (0) for 20 h measured by a nano-permprometer.
zero, and that a strongly increased permeation was observed at around 4-4.5 nm. This implies that no pores or only a few pores were present with Kelvin radii larger than 4.5 nm. The calculated maxima was found at rk = 4.5 nm. The pure mesoporous y-A1203membrane showed an average pore size somewhat larger than those of GA-30, LA-6, and LGA-6 membranes. The pore size distributions of all doped as-prepared samples had a rather narrow distribution. Figures 6(bHd) ( 0 ) clearly show a sharp pore size distribution with an average Kelvin radius of 3.0, 3.0, and 3.2 nm for as-prepared GA-30, LA-6, and LGA-6 membranes, respectively. From the above permeability results, it was clear that permeability increases due to increases in pore size as steam exposure time increases. In the case of mesoporous y-A1203, the average pore size increased and the pore size distribution (Fig. 6(a), 0) broadened with prolonged heat exposure in the presence of steam. These results agree very well with those of Gallaher and Liu." A general trend was observed, i.e., the PSD shifted toward larger pore radii with prolonged steam exposure at 500°C for the y-A1203,GA-30, and LA-6 samples. Figure 6(b) (0)shows that the GA-30 membrane retained a somewhat smaller pore size (- 3.5 nm), however, after hydrothermal treatment. To the best of our knowledge, no permporometry measurement has yet been performed on doped y-A1203-based composite-supported membranes after hydrothermal treatment. Laz03-doped y-A1203,especially, might be an interesting system, as several studies have already reported on its high thermal stability using powder calcinations. In this study, we observed that the pore size of an LA-6 membrane shifted from 2.1 to 5.5 nm after hydrothermal treatment, as shown in Fig. 6(c) (0). It seems that the LA-6 sample was not effective for pore stabilization under hydrothermal conditions at 500°C. In the case of Membranes, Films, and Coatings
the LGA-6 sample, on the other hand, the pore size was unchanged after hydrothermal treatment. A sharp pore size distribution with an average Kelvin radius of 2.4 nm was found for the LGA membrane, as shown in Fig. 6(d) (0).
Fig.7. Scanning electron microscope images of top surface view of (a) y-AIzO, and (b) LGA-6 after hydrothermal treatment. 219
-
10.0
-n
9.5
- I
I
m
'?
N
-'E
"'O
9.0 8.5
. 0 T-
o) 0 C
8.0
E
7.5
m
n
G
7.0
0
10
20 30 Timelh
40
50
0
2
4
6
8
0
Kelvin diameterlnm
(a) Effect of steam on the time course of Hz permeance for an LGA-6 membrane on the a-AIz03porous support at 500°C for 50 h. (b) Pore size distribution of (a) mesoporous LGA-6 membrane before ( 0 ) and after hydrothermal treatment (0)for 50 h measured by a nano-permprometer. Fig. 8.
(5) Microstructure Very smooth morphology was observed only in the case of asprepared y-A1203,GA-30, and LGA-6 samples. It was apparent that a large number of Al2O3particles agglomerated on the surface after modification with La203. Because of these changes in the nanostructure of the membranes, doping with La203 alone was not an appropriate strategy for improving the hydrothermal stability of y-A1203. The sol showed a quite viscous appearance with the addition of Ga2O3 content up to 30 mol% to AI2O3. Even after the addition of 6 mol% La203to GA-30, the sol remained viscous, and thus appropriate for membrane fabrication on the a-AI2O3porous support. The surface of the steam-exposed LGA-6 membrane was smooth and crack-free, whereas morphology with spherical grains was found on the y-AI2O3 membrane, as shown in Figs. 7(a) and (b). The observed gas permeation was consistent with morphological changes in the case of the y-A1203sample after hydrothermal treatment. The rough morphology observed after hydrothermal treatment, in the case of LGA-10 and LGA-15 membranes, might be due to high viscosity and significant shrinkage during hydrothermal treatment for 20 h. ( 6 ) Durability of LGA-6 Membranes The H2 gas permeance results suggest that the most critical period was the first few hours of hydrothermal treatment. Our fabricated mesoporous LGA-6 membrane showed only a small change within the first 2 h. The membrane appeared to be relatively stable thereafter. The higher concentration of steam appears to have an increased mineralizing effect. It was not clear, however, whether a truly stable plateau had been reached under these conditions or whether the kinetics of the transformation had only slowed to the point where much longer exposure times might be necessary to observe significant changes in the membrane structure. To determine the durability of the LGA-6 membrane, the gas permeation performance was examined for more than 20 h in the presence of 75% steam. The LGA-6 membrane proved able to withstand a N2 to steam ratio of 1.3 at 500°C for more than 50 h with no sign of degradation, as shown in Fig. 8(a). The LGA-6 membrane was found to have high durability at 500°C in the presence of 75% steam. Figure 8(b) shows the pore size distribution of an LGA-6 sample after hydrothermal testing for more than 50 h. The pore radius did not shift after hydrothermal treatment. As expected, the pore size distribution shows good consistency with the gas permeance of the LGA-6 sample. As the membrane prepared in this study has a very narrow pore size distribution, with the average pore radius in the range of 3.W.5 nm, the contributions of ordinary diffusion and viscous Poiseuille flow to the gas fluxes were negligible 220
compared with those of Knudsen flow. Gas permeation through the y-AlzO3 membrane deviated from the Knudsen diffusion mechanism after hydrothermal treatment. The observed gas permeation was consistent with Knudsen diffusion in the case of the LGA-6 sample after hydrothermal treatment, as was expected based on its stable retention of a crack-free microstructure. We have demonstrated that the addition of combined 6 mol% La203 as well as 30 mol% Ga203 (referred as LGA-6) can improve the hydrothermal stability of y-AI2O3membranes. As stated above, the incorporation of well-dispersed La203into a solid solution of GA-30 might enhance the stabilization of pore growth under hydrothermal conditions. With respect to yA1203, certain stabilization conditions of the tests need to be recorded accurately: (i) the exact temperature of the test (thermodynamics and kinetics), (ii) the duration of the test (kinetics), and (iii) the type of atmosphere (dry or wet). For temperatures below I O O O T (5OO0C,for example), the conditions were not too severe, and some stabilization could be observed and reported. As it was obvious that an additive must remain on the surface in order to decrease the rate of surface diffusion, large ions were required to prevent dissolution into the bulk. Based on these considerations, lanthanides were a logical choice. Indeed, it was found that the addition of 6 mol% La203increases the thermostability of y-A1203.9.'9 GA-30 is widely recognized as a steam resistance catalyst,6 and we also observed considerable hydrothermal stability in the GA-30 membrane. The high steam tolerance was therefore considered to be due to the incorporation of well-dispersed La203into the solid solution of GA-30, which was thought to enhance the stabilization of pore growth under hydrothermal conditions.
IV.
Conclusions
The thermal and hydrothermal stability of mesoporous membranes was studied in terms of gas permeability, pore structure, and surface microstructure. The permeability and characterization results showed that the hydrothermal stability of the La203doped Ga2O3-AI2O3(referred to here as LGA-6) mesoporous membrane was greatly improved in comparison with mesoporous y-A1203, while the pore system remained highly ordered. The fabricated mesoporous membrane exhibits an extraordinarily high hydrothermal stability at 500°C for a very long service period. The ability of a fabricated mesoporous membrane to withstand steam is advantageous for many reaction systems, and offers possibilities for better exploiting the benefits of a membrane reactor configuration. The hydrothermal test indicates that the membrane prepared is viable for use as an intermediate layer for fabricating a microporous gas separation membrane. Progress in Nanotechnology: Processing
References ‘E. Kikuchi, “Membrane Reactor Application to Hydrogen Production,” Cafal. Today, 56.97-101 (2000). ’K. Jarosch and H. I. de Lasa, “Novel Riser Simulator for Methane Reforming Using High Temperature Membranes,” Chem. Eng. Sci., 54, 1451-60 (1999). ’8. N. Nair, T. Okubo, and Nakao, “Structure and Separation Properties of Silica Membrane,” Membrane, 25 (21 73-85 (2000). 4R. M. De Vos and H. Verweij, “High-Selectivity, High-Flux Silica Membranes for Gas Separation,” Science, 279, 1710-1 (1998). ’K. L. Yeung, J. M. Sebastian, and A. Varma, “Mesoporous Alumina Membranes: Synthesis, Characterization, Thermal Stability and Nonuniform Distribution of Catalyst,” J. Memhr. Sci., 131. 9-28 (1997). 6M. Haneda, Y. Kintaichi, T. Mizushima, N. Kakuta, and H. Hamada, “Structure of Ga2O3-AI2O3Prepared by SolGel Method and its Catalytic Performance for NO Reduction by Propene in the Presence of Oxygen,” Appl. Caral. B, 31, 8192 (2001). 7Md. H. Zahir, K. Sato, and Y. Iwamoto, “Development of Hydrothermally Stable SolGel Derived La20,-Doped Ga2O3-AI2O3 Composite Mesoporous Membrane,” J . Membr. Sci., 247, 95-101 (2005). *K. Shimizu, A. Satsuma, and T. Hattori, “Selective Catalytic Reduction of NO by Hydrocarbons on Ga2O3/AI203Catalysts,” Appl. Caral. B, 16, 319-26 (1998). 9Y. S. Lin and A. J. Burggraaf, “Preparation and Characterization of HighTemperature, Thermally Stable Alumina Composite Membrane,” J. Am. Ceram. Soc., 74 [I]219-24 (1991).
Membranes, Films, and Coatings
‘‘I. S. Church, N. W. Cant, and D. L. Trimm, “Stabilization of Aluminas by Rare Earth and Alkaline Earth ions,” Appl. Caral. A., 101. 105-16 (1993). “G. R. Gallaher and P. K. T. Liu, “Characterization ofCeramic Membranes I . Thermal and Hydrothermal Stabilities of Commercial 40 A Membranes,” J. Membr. Sci., 92, 2 9 4 4 (1994). ”C. H. Chang, R. Gopalan, and Y.S. Lin. “A Comparative Study on Thermal and Hydrothermal Stability of Alumina, Titania and Zirconia Membranes.” J. Membr.
Sci.,91, 2 7 4 5 (1994).
”A. Nijmeijer, H. Kruidhof, R. Bredesen, and H. Verweij, “Preparation and Properties of Hydrothemally Stable y-Alumina Membranes,” J. Am. Ceram. Soc., a,$’] 136-40 (2001). R. S. A. de Lange, J. H. A. Hekkink, K. Keizer, and A. J. Burggraaf, “Formation and Characterization of Supported Microporous Ceramic Membranes Prepared by SolGel Modification Techniques,’’ J. Memhr. Sci.,99, 57-75 (1995).
I5T. Tsuru, T. Hino, T. Yoshioka. and M. Asaeda, “Permporometry Characterization of Microporous Ceramic Membranes,” J. Membr. Sci., 186. 257-65 (2001). 16H. Schaper, E. B. M. Doesburg, and L. L. van Reijen, “The Influence of Lanthanum Oxide on the Thermal Stability of Gamma-Alumina Catalyst Sup ports,” Appl. Coral., 7,21 1-20 (1983). I7V. G. Hill, R. Roy, and E. F. Osborn, “The System AluminaGallia-Water,” J . Am. Ceram. S a c , 35, 136-42 (1952). “9. Beguin, E. Garbowski, and M. Primet, “Stabilization of Alumina by Addition of Lanthanum,” Appl. Cafal., 75, 119-32 (1991). I9Y. S. Lin, C. H. Chang, and R. Gopalan, “improvement of Thermal Stability of Porous Nanostructured Ceramic Membranes,” Ind. Eng. Chem. Rex, 33, 8% 70 (1994).
221
This Page Intentionally Left Blank
Behavior of Silicon Oxycarbonitride Thin Films Derived from Poly(Urea)Methyl Vinyl Silazane Tsali J. Cross' and Rishi Raj University of Colorado at Boulder, Department of Mechanical Engineering, Boulder, Colorado 80309
Tsali J. Cross, Somuri V. Prasad, and David R. Tallant Sandia National Laboratories,* Materials and Process Sciences Center,Albuquerque, New Mexico 87185
A process for deposition of silicon oxycarbonitride films from poly(urea)methyl vinyl silazane (PUMVS) by spin coating precursor solutions onto a substrate, followed by polymerization, cross-linking and pyrolysis has been developed. The crosslinked polymer films (350 nm thick), deposited on variety substrates (e.g., silicon, sapphire, zirconia), were pyrolyzed in nitrogen or ammonia environments either in a hot isostatic press or in a tube furnace. Their microstructure was characterized using infrared and Raman spectroscopy. The tribological (friction and wear) behavior was evaluated in dry nitrogen and air with 50% relative humidity using a unidirectional linear wear tester in a ball-on-disk configuration. Wear surfaces, transfer films and wear debris were analyzed by scanning electron micrograph (SEM)/energy dispersive spectroscopy (EDS).
Introduction Silicon oxycarbonitride (SiCNO) ceramics are promising materials for microelectronics and optoelectronics, owing to high-temperature oxidation resistante, tunable band gap characteri~tics,~'~ adjustable transparency in the visible and IR and high-temperature thermal stability.6 Their refractory nature makes this class of materials attractive for tribo-
'
'Sandia is a multiprogram laboratory operared by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-
~~AL~~OIIO. '
[email protected]@sandia.gov 02006 The American Ceramic Society
Membranes, Films, and Coatings
logical applications in harsh environments. ',7-9 SiCNO films can be deposited by various techniques, including CVD," CVD bias-assisted hot filament," RF PECVD," electron cyclotron resonance PECVD,I3 pulsed laser d e p ~ s i t i o n , 'ion ~ beam sputtering,' RF sputtering," plasmajet CVD," reactive magnetron sputtering, and via silazane polymer-derived ceramic (PDC) routes. 1,19-23 Wear and friction (static or dynamic) limits the reliability of silicon microlectromechanical system (MEMS) devices, because of its high-wear and -friction coefficient. Surface modifications such as hard coatings, including Sic and W, solid lubricants, films deposited by atomic layer deposition, and adsorbed organic molecules have been explored for reducing friction and wear of
''
223
MEMS However, there is a limited selection of wear-resistant MEMS materials with low friction that can withstand high-temperature and corrosive environments. Deposition of SiCNO films by the polymeric, silazane-PDC routes, where liquid precursors are cast as thin films on a substrate and pyrolyzed in situ in the ceramic state, has additional benefits such as cost and ease of application. However, studies on the fundamental understanding of the mechanisms of friction and wear of the materials are limited. PDCs are refractory materials that have structure at the nano-scale and have high strength and hardPDCs are processed from liquid organometallic precursors by cross-linking the polymers into an infusible solid followed by controlled pyrolysis. PDCs remain inert and highly stable, both mechanically and chemically in harsh environ~nents.~~ Typical examples of PDCs include SiCNOs, and silicon oxycarbide (SiOC). The microstructure of PDCs depends on precursor synthesis, polymerization, degree of cross-linking, the temperature at which the cross-linked part is pyrolyzed, as well as the environment of the pyrolysis process. Polymer-derived silicon carbonitride typically contains phases of silicon atoms bonded to carbon and nitrogen in sp3 configurations, as well as phases of carbon in aromatic sp2 configuration^.^^ These ceramics retain their amorphous microstructure above 1400°C,32 where nucleation of S i c and Si3N4 phases is limited. It is speculated the carbon phases help to reduce coarsening of S i c and Si3N4 phases by physically and thermodynamically impeding their growth. The presence of carbonrich phases also presents an opportunity to tailor their microstructure and chemistry for improved friction and wear performance.
The objective of the current study was to synthesize SiCNO PDCs into thin films, characterize their microstructure and chemistry, and evaluate their tribological behavior in various environments.
Materials and Methods Synthesis and Processing of Thin Films
Thin films were processed with the liquid monomer poly(urea)methyl vinyl silazane (PUMVS, trade name CerasetTM,Kion Corp. Huntingdon Valley, PA). The chemical structure and a schematic representation of the polymerization steps through vinyl groups are shown in Fig. 1. The precursor for SiCNO films consisted of PUMVS, 30 wt% acetonitrile, and 5 wt% photoinitiator, 2,2 dimethoxy-2-phenyl-acetophenone (DMPA, Aldrich Inc., St. Louis, MO) that was polymerized with ultraviolet radiation. Acetonitrile served to improve precursor wettability and the dissolution of DMPA. Acetonitrile was selected as an appropriate solvent, based on its volatility with PUMVS and DMPA, and wetting behavior after spin coating. Tetrahydrofouran and heptane solvents also generated pinhole-free coatings after spin coating, but acetonitrile was selected because it is less toxic. The liquid monomer, PUMVS, was also used to spin coat and was thermally polymerized, cross-linked, and pyrolyzed in order to observe any effects that may result from the solvent and photoinitiator, DMPA. Films were deposited on commercially available wafers of (100) silicon (boron doped), (0001) sapphire (A1203), and partially yttria-stabilized zirconia (YZTP). The wafers were first cleaned ultrasonically
z -Si I
I
- CH =CH2
+tycHYH2 -Si
-
-Si
-
z- 1
H
(a)PUMVS
(b) Vinyl group polymerization
Fig. 1. (a) MoLecuLar structure of the monomer f i r poly(urea)rnethyl vinyl silazane (PUMVS), and (b) its polymerization through vinyl gfoups.
224
Progress in Nanotechnology: Processing
Fig, 2. Flow chart ofpolymer derived Silicon oycarbonitride thin film deposition process, (I) Thefilm is deposited by spin coating. (2) The coating polymerization was initiated either by ultraviolet or thermal energy. (3) The polymerized coating is pyrolyzed either in a hot isostatic press in nitrogen, or in a @wing tube fimace in nitrogen or ammonia gas.
in acetone, followed by methanol. They were then subjected to UV ozone cleaning for 15 min. Cleaning procedures, involving standard piranha solution (3: 1, H2SO*:H202) to first remove organics and a 48% hydrofluoric acid solution to remove native oxides, were found not to have an effect on film uniformity or its chemistry. Precursor solutions were dispensed using a syringe, and 0.2 pm syringe filter, with enough fluid to cover the entire 12 mm x 12 mm-diced wafers. The wafers were then spun at 4000 rpm for 30 s. The precursor films were polymerized either thermally at 350°C or by exposing to W radiation with a dose of 11.4 J/cm2 at 354 nm. They were cross-linked at 400°C under nitrogen in either a flowing tube furnace or in a hot isostatic press under 13.79 MPa (2000 psi) of isostatic pressure. One batch of cross-linked films were pyrolyzed in a hot isostatic press and flowing tube furnace in nitrogen, and another batch in a flowing tube furnace in ammonia. A flow chart of the steps in thin film synthesis is shown in Fig. 2. Sample inventories including the precursor, cross-linking, and pyrolysis
Membranes, Films, and Coatings
conditions are shown in Table I (nitrogen pyrolysis) and Table I1 (ammonia pyrolysis). Characterization
The structure and chemistry of the films on silicon were evaluated by Fourier transform infrared (FTIR), attenuated total reflectance (ATR), and microRaman spectroscopy. Raman spectra were obtained using the 458 nm line of an argon laser. The microscope accessory to the Raman system focused the laser beam on the samples, collected a 180" backscatter and transported the backscattered light to a triple spectrograph. A charge-coupled device detector detected photons dispersed by the spectrograph. We removed broad (fluorescence) background features from the resulting Raman spectra using a polynomial fitting technique. For FTIR, a spectrometer, fitted with a horizontal sample placement, W light exposure, and modified reading photoaccessory that collected data in transmission, was used. ATR spectra were obtained from an additional, separate
225
Table I.
Sample Inventory of Thin Films Cross-linked and Pyrolyzed in Nitrogen on Different Substrates Under Various Conditions Precursor, polymerization, cross-linking, and pyrolysis in nitrogen
Sample
Precursor
Cross-linking and pyrolysis Polymerization environment
FILM 1 PUMVS with 30 wt% Acetonitrile, 5wt% DMPA
W cured in open air or UV cured under nitrogen
Flowing tube furnace: nitrogen gas
FILM 2 PUMVS with 30 wt% Acetonitrile, 5wt% DMPA
W cured in
Hot isostatic press: nitrogen
FILM 3 Pure PUMVS
open air
gas
Thermally cured All processed in flowing tube in a glove box furnace under with 25 ppm nitrogen H 2 0 content
Cross-linking temperature control
Pyrolysis temperature control
FTIR results
HR1: 1200°C/h, HR2: 300"C/h, Dwelll: 90 rnin Dwell2: 60 or at 400°C 180 rnin at
Si bonds: Si-0, Si-C, Si-CH3 looooc C bonds: C-H HR1: 800"C/h, HR2: 40"C/h, Si bonds: Dwelll: 100 rnin Dwell2: 600 rnin Si-0, Si-C, at 400°C, at 1000°C Si-CH, C bonds: 14 MPa C-H HR1: 1200"C/h, HR2: 30O0C/h, Si bonds: Dwelll: 90 min Dwell2: 60 min Si-0, Si-C, Si-CH3 at 350°C at 825°C C bonds:
Raman results
Highly ordered graphene phases Highly ordered graphene phases
C-H FILM 4 PUMVS with on ZrOz 30wt% Acetonitrile, 5wt% DMPA
W cured in open air
Hot isostatic press: nitrogen gas
HR1: 800"C/h, Dwelll: 100 rnin at 400"C, 14 MPa
HR2: 40"C/h, Dwell2: 600 rnin at 1000°C HR3: 40"C/h, Dwell3: 600min at 1250°C
PUMVS, poly(urea)methyl vinyl silazane; DMPA, 2,2 dimethoxy-2-phenyl-acetophenone.
Table 11.
Sample Inventory of Thin Films Cross-linked and Pyrolyzed in Ammonia on Different Substrates Under Various Conditions Precursor, polymerization, cross-linking, and pyrolysis in ammonia
FILM 5 PUMVS with 30 wt% acetonitrile, 5wt% DMPA
W cured
FILM 6 PUMVS with
UV cured
30 wt%
acetonitrile, 5 w t % DMPA
226
Flowing in open air tube furnace: ammonia gas
Flowing in open air tube furnace: ammonia gas
HR2:
Si bonds: Ordered Si-0, graphene Si-N phases C bonds:
HR1 : 12OO"C/h, Dwelll: 90 min at 400°C
30O0C/h, Dwell2: 60 min at 1000°C
NO CCOSS-
HRl :
linking step
1200"C/h, Dwell2: 90 min at 1000°C
C-H Si bonds: No carbon Si-N detected C bonds:
C-H
Progress in Nanotechnology: Processing
horizontal sample mount with a diamond, singlebounce plate, and photoaccessory that was added on to the FTIR spectrometer. All infrared spectra from films were collected with FTIR in transmission, with the exception of FILM 4, where transmission through zirconia substrate was not possible and it was necessary to acquire FILM 4's spectrum with ATR. Film surfaces and cross-sections were examined in the SEM to assess the topography, substrate-film interfaces, and porosity in the films. Cross-sections of films were prepared by dicing the coated wafers after carefully scribing with a diamond-tipped pen at the back.
Tribological Testing Friction and wear tests were performed using a ballon-disk linear wear tester (Fig. 3). A Si3N4 ball (3.175 mm diameter) was used as the counterface. Normal load was applied by means of deadweights. Friction force was measured with a strain gage load cell. Measurements were made at three applied normal loads (i) 98 mN, which corresponded to "low" Hertzian contact stresses of 0.42, 0.58, and 0.62 GPa on silicon, zirconia, and sapphire substrates, respectively, (ii) an "intermediate" normal load of 294 m N corresponding to contact stresses of 0.61, 0.84, and 0.89 GPa on silicon, zirconia, and sapphire substrates, respectively, and (iii) a "high" normal load of 519mN, corresponding to contact stresses of 0.73, 1.01, and 1.08 GPa on silicon, zirconia, and sapphire substrates, respectively. The device was housed in an environmental chamber with provision to
Fig.3. A photograph of the linear wear tester employed for fiiction measurements.
Membranes, Films, and Coatings
monitor dew point and oxygen content. Mixing dry air with air passed through water bubblers provided the desired relative humidity in the test chamber. Measurements were made in a dry nitrogen environment (< 10 pprn 0 2 and < 100 pprn H 2 0 ) and in air with 50% relative humidity. Tests were run in unidirectional mode for 1000 cycles over a track distance of 1.6 mm at a sliding speed of 3.3 mm s-'. Results
Microstructure and Chemistry FTIR and ATR Spectroscopy: FTIR spectra of films before and after photoinitiated polymerization are shown in Fig. 4. The spectra were obtained on UVcured films in a nitrogen purged environment (Fig. 4b) as well as in air (Fig. 4c) in order to investigate the role of oxygen. Photoinitiated polymerization occurred through the vinyl groups as illustrated in Fig. l b where a reduction in the carbon-carbon double bonds (1600cm-') and increase in the methylene groups was evident. The bands near 2300cm-' (Fig. 4a) are because of nitrile groups within the solvent, acetonitrile. The solvent evaporates with the addition of heat generated from exposure to ultraviolet radiation, as shown by the disappearance of the nitrile bands in Figs. 4b and c. Homopolymerization of silazanes has been well characterized by Kroke et ~ 2 1 and . ~ ~others and are shown to polymerize through vinyl groups as well, along with hydrosilylation reactions at elevated temperatures. The presence of the strong Si-0 peak at 1100 cm-' for polymerizations performed in air shows that oxygen becomes chemically incorporated into the polymer network itself. FTIR spectra of the films after UV-initiated polymerization in an open air, and a nitrogen purged, low oxygen environment are shown in Fig. 5. The spectrum in Fig. 5c (FILM 3) is also representative of a 100% concentration PUMVS film, cross-linked at 400°C and pyrolyzed at 825°C in a nitrogen-purged glove box with 25 ppm H z O content, after etching the native oxide on the silicon substrate with 48% HF for 5 min. Because of the near absence of Si-N (900cm-') or N-H (3384cm-') bonds, it is believed that the thin films pyrolyzed between 700" and 1000°C in nitrogen are primarily composed of a material consisting of Si-0 (1098 cm-'), and Si-0-C bonds, as indicated by the peak between 800 and 850cm-' in Fig. 5c.
227
Fig. 4. Fourier transform infiared spectra of (a) poly(urea)methyl vinyl sihzune with 30 wt% acetonitrile and 5 wt% 2,2 dimethoxy-2phenyl-acetophenone, and (b) afer it was Wphotoinitiatedpolymerization in a nitrogen-purged, low oxygen environment, and (c) afer W photoinitiated polymerizution in air.
Films deposited on zirconia were further annealed at 1250"C, but their composition was not easily identifiable from its ATR spectrum, because of the heavily convoluted peak around 1100-1600 cm-' (Fig. 5d). Films deposited on silicon or sapphire evaporated almost entirely when annealed at 1250°C. FTIR spectra of films cross-linked and pyrolyzed in ammonia are shown in Fig. 6. Cross-linking of films at 400°C (Fig. 6a) before pyrolysis between 700" and 1000°C in ammonia, resulted in broadening of the base at 900cm-' of the large Si-0 peak at around llOOcm-' out to 900cm-', shown in Fig. 6b. The breadth of the underlying band around 3 100-3500 cm-', indicates the presence of 0-H groups. Thin films cross-linked and pyrolyzed in ammonia are therefore speculated to comprise of complex Si-0, Si-C, 0-H, and possibly Si-N bonds. The films that were directly pyrolyzed in ammonia, without cross-linking, showed the presence of N-H bonds at 3384 cm-', and a strong peak at 900 cm-' (Fig. 6c). The breadth of the under-
228
lying band around 3100-3500 cm-', indicates the presence of 0-H groups as well, but not conclusively. This suggests noncross-linked films are comprised of primarily SiN, bonds. Raman Spectroscopy: All films that were pyrolyzed in a flowing tube furnace in nitrogen, as well as HIPed films that were cross-linked under pressure, yield a Raman spectrum with only two peaks corresponding to the D and G bands, shown in Fig. 7. These Raman bands are characteristic of stacked sheets of six-membered, aromatic (graphene) carbon rings,34 similar to those in graphite. The D peak is typically located near 1350cm-', while the G peak is usually located near 1600 The shifts in these peaks, from 1350 and l6OOcm-', found in the spectra from films shown in Fig. 7, could be because of incorporation of silicon or nitrogen into the graphene Some Raman spectra, obtained from locations both on and off wear tracks, indicated reduction of graphene domain size,
Progress in Nanotechnology: Processing
Fig. 5. Typical Fourier transform infiared spectra o f j l m s at dzffient stages o f j l m processing. All films were processed in a nitrogen atmosphere and spectra are shown (a) after cross-linking at 400°C (6) after 600°C (c) (FILMS I , 2, 3) after 1000°C beat treatments (I and 3 b), and (d) after a 1250°C heat treatment (FILM 4).
presumably by crushing. No carbon Raman bands were found in the Raman spectrum for the ammonia-pyrolyzed film, that was not cross-linked before pyrolysis. We compared the Raman spectra of the PDC films in Fig. 7 to spectra obtained from samples of a phenolic composite pyrolyzed at temperatures from 375" to 2700°C. One of the Raman spectra of the phenolic composite samples, treated at 980°C is shown in Fig. 7. The Raman spectra of the pyrolyzed phenolic samples show the evolution of graphene-related Raman bands as graphite-like domains increase in size with increasing exposure temperatures. The spectrum of the film crosslinked (400°C) and pyrolyzed (1000°C) in ammonia is, except for enhancement in intensity and an upshift in frequency of its D peak, generally consistent with the spectrum of the 980°C phenolic composite. Enhancement in intensity and an upshift in frequency of the D peak is consistent with nitrogen incorporation into the graphene structure, based on Raman spectra of a pyrolyzed, nitrogen-containing polymer.36 The spectrum of the film cross-linked and pyrolyzed (1000°C) in nitro-
Membranes, Films, and Coatings
gen is intermediate in Raman band profile between spectra of phenolic composite samples treated at 980°C and 1200°C. However, its G peak is upshifted in frequency (at 1615 cm-' from about 1600 cm-' in spectra of pyrolyzed phenolic composites). Wang et aL3* attribute an upshift in G band frequency specifically to boron incorporation into the graphene matrix but more generally to graphene sheets adjacent to layers incorporating intercalant atoms. The Raman spectra in Fig. 7 of the ammonia and nitrogen pyrolyzed and cross-linked films are generally consistent with the formation of stacked graphene ring structures formed by thermal charring, as occurs in a phenolic resin matrix. The observed shifts in their D and G peak frequencies are because of incorporation of noncarbon atoms in the graphene matrix. Topography and Cross-Sectional SEM: A SEM of typical film surface, regardless of pyrolysis conditions, is shown in Fig. 8a. The topography was measured by optical profilometry. The average surface roughness (4)was
229
600
850
1100 1350 1600 1850 2100 2350 2600 2850 3100 3350 3600 3850
Wavenumber (cm-’)
Fig. 6 Typical Fourier transform inpared spectra o f j l m at diferent stages o f j l m processing. Alljlm were processed in an ammonia atmosphereand spectra are shown (a) after cross-linkingat 400°C in a flowing tubefitrnace in ammonia, (6) (FILM 5) after cross-linking and 1OOOOC pyrolysis, and (c) (FILM c;) after 100O”Cpyrolysisonly, without a cross-linking heat treatment.
6.7nm. Film thickness was measured from the SEM micrographs of the cross-sections; the average film thickness was 350 f40 nm. Films also appear to adhere well with all the substrates without any interfacial porosity; a typical cross-section of the film on silicon is shown in Figs. 8a and b. Films that were not cross-linked before pyrolysis were found to be porous (Fig. 8b) whereas cross-linking before pyrolysis resulted in dense films (Fig. 8c). This observation suggests preparation of a dense ceramic film requires a minimum degree of cross-linking before pyrolysis.
Tribology Fig. 7. Typical Raman spectra of silicon oxycarbonitridefilm cross-linked andpyrolyzed in nitrogen (FILMS I, 2) and ammonia (FILM 5). Also shown is a comparison to carbon referencefiom a phenolic char annealed at 980°C. Narrow bands are indicative of extended graphene structure.
230
Typical friction traces (friction coefficient as a hnction sliding cycles) of films processed in nitrogen environment (FILMS 1, 2) are given in Fig. 9. These measurements were made in dry nitrogen environment
Progress in Nanotechnology:Processing
Fig. 3. Friction coeficient versus number of sliding cycLes by a linear wear tester in a d?y nitrogen environment with contact stress 0.42 GPa tested on film deposited on silicon. Tbejiction traces are representative ofjilms cross-linked and pyrohzed in a nitrogen environment (FILM I), that were (a) polymerized in open air, and (b) polymerized and processed in a nitrogen-purged glove box
(FILM 3).
Fig 8. Scanning electron microscope micrograph of silicon OxycarbonitriAfilms on silicon substrates: (a) topography of a vpicaljlm, (6) cross-section of a Jilm that was cross-linked and pyrolyzed (FILM 5), and (r) cross-section of a pyrolyzed film that was not cross-linked (FILM G). A l l j l m were processed in ammonia.
at an applied load of 98 mN. FILM
3 was processed in
the glove box including the spinning and polymerization steps, whereas FILM 1 was spun and polymerized in air followed by cross-linking and pyrolysis in nitrogen. The film that was spun and polymerized in open lab air had an extremely short wear life (Fig. 9a). SEM micrographs of the wear scar clearly reveals generation
Membranes, Films, and Coatings
of large amount wear debris (Fig. lOa), partial exposure of substrate silicon, and transfer of the debris to the counterface ball (Fig. 1Oc). The EDS analysis on wear surface in Fig. 10b showed predominantly silicon and oxygen peaks. This behavior was representative of all the films that were spun and polymerized in lab air, irrespective of the subsequent processing steps in nitrogen. By contrast, films processed in the nitrogen glove box all the way beginning with spinning showed a steady state friction coefficient of 0.21 0.01 throughout the 1000cycle test. Films cross-linked and pyrolyzed in ammonia showed a progressive decrease in friction coefficient with sliding distance. As shown in Table 11, the films were spun and polymerized in lab air before cross-linking and pyrolysis in an ammonia atmosphere. Figure 11 shows a typical friction trace in dry nitrogen environment at a contact stress of 0.42 GPa on a silicon substrate. SEM micrograph of a wear scar on FILM 5 is shown in Fig. 12a. A higher magnification micrograph of the wear scar shows portions of the film inside the wear scar curling up. Figure 12c corresponds to the wear surface of the counterface ball, where in the presence of large number of rolled up debris can be seen. Friction behavior of films (pyrolyzed in nitrogen and ammonia atmospheres) in humid air is shown in
231
1.01
0.8
,
I
,
.
I
I
I
I
800
1000
i
0
200
400
600
Cycles
Fig. 11. Friction coeficient versus number of sliding y l e s by a Linear wear tester in a d y nitrogen environment with contact stress 0.42 GPa on a silicon substrate. Friction traces are representativeof
(FILM 5).
Deep grooves on the ball surface and ploughing of the silicon substrates are also evident in the SEM micrographs in Fig. 14. These results imply that the films, irrespective of the processing conditions, wore out almost instantly exposing the underlying substrate to silicon nitride ball counterface. Particulate wear debris at the sliding interface might have been responsible for the observed friction coefficients that are higher than those of the bare silicon. Discussion
Fig. 10. Scanning electron micrographs of a wear track at (a) low and (b) high magnification, and (c) the Si@*counte$ace ball aft.. a linear wear test performed in d y nitrogen on a j l m cross-linked and pyrolyzd in nitrogen (FILM 2), and deposited on silicon.
Fig. 13. It can be seen that the friction coefficients are high (- 0.8). It should also be noted from Table I11 that the friction coefficients are in fact higher than those measured on bare silicon substrates. Subsequent SEM analyses did not reveal wear debris roll-ups either on the wear surface or on the counterface ball (Figs. 14a and b).
232
A summary of the friction behavior for all films at various contact pressures and environments is shown in Table 111. There appears to be a strong correlation between film density, and pyrolysis atmosphere, and the friction and wear behavior. The degree of cross-linking before pyrolysis affected the rate of diffusing gas, and density of the films. Nitrogen pyrolysis produced a microstructure consisting of an amorphous Si-0, Si-C matrix surrounding graphene structures whose ordering is consistent with their thermal exposure, as revealed by FTIR (Fig. 5) and Raman spectroscopy (Fig. 7). Pyrolysis in ammonia produced a microstructure comprised of an amorphous matrix of Si-0, 0-H, Si-C, and possibly Si-N bonds surrounding graphene nano-clusters, as revealed by FTIR (Fig. 10) and Raman spectroscopy (Fig. 7). The
Progress in Nanotechnology: Processing
Fig. 13. Friction coe$cient versus number of sliding cycles by a linear wear tester in air with 50% relative humidity environment with contact stress 0.42 GPa. The friction traces are representative offilms deposited on silicon that were cross-linked andpyrolyzed in either a hot isostatic press or flowing tube&rnace in a nitrogen environment (FILMS I , 2), and cross-linked and pyrolyzed in a flowing tube&rnace in ammonia (FILM 5).
Si-CH3 E SiZ-NH
+ NH3 -+=Si-NH2 + CH4
+ H20
-+
= Si-NH2
= Si2-0 + NH3
Fig. 12. Scanning ekctron micrographs of a wear track at (a) low and (6) high magnification, and (c) the Sifl4counteface ball ajer a linear wear testpefomed in dry nitrogen on a film cross-linked andpyrolyzed in ammonia (FILM 5), and deposited on silicon.
retention of Si-N or Si-0 bonds during pyrolysis is anticipated from reactions with ammonia or with latent water, respectively as follows:
Membranes, Films, and Coatings
(a)
(b)
Several films were prepared to investigate the influence of the processing parameters, including cross-linking, pyrolysis environment, and the temperature cycle on the composition of the films, specifically nitrogen and oxygen contents. Subtle changes in their FTIR spectra indicated that the dehydrogenation reactions for the films are successful when the duration of crosslinking dwell times at 350°C are longer than 90min, and when the films are pyrolyzed at 700°C for longer than 15 min. However, their FTIR spectra of pyrolyzed films did not differ from those that were shown in either Fig. 5c or Fig. bb. Further, an observable change in the FTIR spectra was not shown even with a 95%Ar, 5%H2 mixture of gas that was employed during pyrolysis. The films with the lowest friction (in dry nitrogen environment) were obtained by pyrolyzing in an ammonia atmosphere, where, presumably more nitrogen may have been incorporated within the film during pyrolysis, based on the reaction in equation (a). Their
233
Table 111. Summary of Typical Friction Results Summary: friction and wear of films after 1000 cycles COF in dry nitrogen
Sample
COF low contact stress (GPa): Substrate 0.42 < ts < 0.62
Films pyrolyzed in nitro en Bare zro2 ~ 0 % 0.8f0.1 : Bare A1203 COF: 0.25 f0.02 Bare Si COF: 0.5 f 0 . 2
FILM FILM FILM FILM FILM
1
Si
2 2 3 4
A1203
A2O3 Si
Zr02
COF: COF: COF: COF: COF:
0.7 k 0.1 0.7k0.2 0.7 f0.1 0.21 kO.01 0.25 f 0.02
Films pyrolyzed in ammonia FILM 5 Si COF: 0.1 fO.01 COF: 0.22f0.02 FILM 5 A 1 2 0 3 FILM 6 Si COF: 0.7 f0.1
COF intermediate contact stress (GPa): 0.61 < ts < 0.89
COF: 0.9 f0.05
COF: 0.75 f0.05 COF: 0.65 f0.2
COF: 0.75 f0.05 COF: 0.25 k 0.1 20% RH in Air COF: 0.6 k0.1 COF: 0.7f0.2 COF: 0.7 f0.3 COF: 0.7 f0.1
COF: 0.25 f0.02
COF: 0.4 f0.02
COF: 0.67 f0.03
COF: 0.1 fO.01 COF: 0.45f0.05
COF: 0.5 f0.05 COF: 0.6f0.1
COF: 0.75 & 0.05 COF: 0.8 rfr 0.02 COF: 0.7 f0.1
low-friction behavior is a result of the formation of sheat-reducing roll-ups. However, these films do not have inherent resistance to wear, and the low friction is because of the formation of roll-ups. All the films showed high-friction and -wear behavior in humid environments, because of ploughing of the film from the formation of a harder third body phase of unknown composition, probably Si-0. It is speculated that nitrogen within films may improve its hardness and stiffness, while oxygen may deteriorate its mechanical properties, as would be expected for amorphous Si-N versus Si-0 based ceramics, in general. Residual stresses within the films may also play an important role, helping to facilitate roll-up formation. Further characterization of the mechanical properties was not pursued, because the films did not have the desired friction and wear response, particularly for microsystem applications. Although there were differences in the ordering of the carbon phases within nitrogen and ammonia pyrolyzed films, its effects on tribological behavior are not understood. However, it is clear that the graphene/ graphitic regimes did not dominate the tribological be-
234
COF high contact stress (GPa): 0.73 < ts < 1.08
COF in 50% RH in air COF low contact stress (GPa): 0.42 < a < 0.62
havior, because high friction was observed in humid environments. The lubricity of graphite in air is related to its lamellar, hexagonal sttucture. It is generally accepted that the ease of slip of the graphite basal planes increases from intercalation of adsorbed water and oxygen.37’38 Ongoing research on bulk polymer-derived SiCNO ceramics showed that the bulk materials exhibited friction coefficients of 0.2f0.25 in both dry and humid environments and they were extremely resistant to wear.39 This ongoing study revealed that pyrolyzing the material at temperatures greater than 1200°C was essential to achieve the desired tribological behavior. The high-temperature anneal has the effect on bulk SiCNO specimens in increasing its hardness and stiffness. Therefore, it was decided that higher temperatures were required to improve the tribological behavior of films. It is difficult to subject the films to high-temperature anneals because of the sublimation of the film and the reaction with the substrate. Despite these challenges, films were deposited on zirconia substrates and annealed at 1250°C. Unfortunately, these films were extremely thin, because of evaporation or sublimation during the
Progress in Nanotechnology: Processing
0.8 -
0.0
0
-Annealed
200
film on zirconia
400 600 Cycles
800
1000
Fig. IS. Friction coefficient versus number of sliding ycles by a linear wear tester in a dry nitrogen environment with contact stress 0.42GPaforfilms deposited on zirconia substrates. Friction trace is typical f i r (FILM 4).
Conclusion
We have developed a process for deposition of SiC-
NO films from PUMVS. Films that were spun and po-
Fig. 14. Scanning electron micrographs of the (a) wear track and (6) Si&dcountprf.ce ball a j e r a linear wear test peformed in air with 50% relative humidity on a film cross-linkedandpyrolyzed in ammonia (FILM 5), and deposited on silicon.
anneal, which made it difficult to characterize them with
ATR infrared spectroscopy. EDS revealed that the film contained Zr, 0, and C, but not silicon or nitrogen. So, we are not sure if this particular film indeed belonged to the family of SiCNO PDCs. However, its friction coefficient remained low (0.25 0.02) and steady throughout the 1000-cycle test (Fig. 15). Furthermore, the wear scars were so shallow that they could not be identified with certainty even with a high-resolution SEM. “Low-temperature” pyrolysis techniques such as laser pyrolysis,’ and other may eventually be employed to apply thin film PDC’s onto low-melting substrates, and help alleviate any effects from thermal mismatch. However, before these engineering issues may be addressed, a better understanding of the relationship between the desirable chemistry, microstructure, and tribology of PDCs should be gained.
Membranes, Films, and Coatings
lymerized in air, and then pyrolyzed in nitrogen wore through immediately in dry nitrogen environment, which produced high friction (0.8). However, films that were spun, polymerized, and pyrolyzed (in nitrogen) in a low-HzO glove box, had reduced friction coefficients. Films pyrolyzed in ammonia showed an initial coefficient of friction of 0.2, which decreased progressively with sliding distance. The mechanism of low friction is because of the formation of wear debris roll-ups. All films wore through in humid environments without forming friction-reducing roll-ups. Cross-linking was the important processing parameter in affecting the density and composition of the films. Films deposited on zirconia substrates and annealed at 1250°C showed promising tribological behavior. References R. Riedel, H.-J. Kleebe, H. Schonfelder, and F. Aldinger, “A Covalent Micro/ Nano-Composite Resistant to High-Temperature Oxidation,” Nature, 374 526-528 (1995). A. Badzian, T. Badzian, R. Roy, and W. Drawl, “Silicon Carbonitride, a New Hard Material and Its Relation to the Confusion About ‘Harder Than Diamond’ C3N4,” Thin Sol. Films, 354 148-153 (1999). D. H. Zhang, Y. Gao, J. Wei, and Z. Q. Mo, “Influence of Silane Partial Pressure on the Properties of Amorphous SiCN Films Prepared by ECRCVD,” Thin Sol. Films, 377-378 607-610 (2001).
235
F. J. Gomez, P. Prieto, E. Elizalde, and J. Piquerasa, “SiCN Alloys Deposited by Electron Cyclotron Resonance Plasma Chemical Vapor Deposition,” Appl. Phys. L e n , 69 773 (1996). 5. G. Soto, E. C. Samano, R. Machorro, and L. Cota, “Growth of S i c and SiCJ, Films by Pulsed Laser Ablation of S i c in Ar and N2 Environments,” /. Vac. Sci. Technol., A, Vac. Surf: Films, 16 1311-1315 (1998). 6. H. J. Seifert, J. Peng, H. L. Lukas, and F. Aldinger, “Phase Equilibria and Thermal Analysis of Si-C-N Ceramics,” J Alloys Camp., 320 251-261 (2001). 7. H.-J. Krauss and G. Mom, “Laser Pyrolysis of Polysilazane A New Technique for the Generation of Ceramic-like Coatings and Structures,” Kq Ens Muter., 206213 467-470 (2002). 8. P. Jedrzejowski, J. Cizek, A. Amassian, J. E. Hemberg-Saphieha, J. Vlcek, and L. Martinu, “Mechanical and Optical Properties of Hard SiCN Coatings Prepared by PECVD,” Thin Solid Films, 447-448 201-207 (2004). 9. E. Bertran, E. Martinez, G. Viera, J. Farjas, and P. Roura, “Mechanical Properties of Nanometric Structures of SilSiC, C/SiC and ClSiN Produced by PECVD,” Dia. Rel. Muter., 10 1115-1 120 (2001). 10. A. Bendeddouche, R. Berjoan, E. Beche, T. Merle-Mejean, S. Schamm, V. Serin, G. Taillades, A. Pradel, and R. Hillel, “Structural Characterization of Amorphous SiCxNyChemical Vapor Deposited Coatings,” /. App. Phys., 81 [91 6147-6154 (1997). 11. 2. Gong, E. G. Wang, G. C. Xu, and Y. Chen, “Influence of Deposition Condition and Hydrogen on Amorphous-to-Polycrystalline SiCN Films,” Thin SolidFilms, 348 [l-21 114121 (1999). 12. H. Sachdev and P. Scheid, “Formation of Silicon Carbide and Silicon Carbonitride by RF Plasma-CVD,” Dia. Rel. Muter., 10 1160-1 164 (2001). 13. D. H. Zhang, Y. Gao, J. Wei, and 2. Q. Mo, “Influence of Silane Partial Pressure on the Properties of Amorphous SiCN Films Prepared by ECRCVD,” Thin Solid Films, 377-378 607-610 (2000). 14. T. Tharigen, G. Lippold, V. Riede, M. Lorenz, K. J. Koivusaari, D. Lorenz, S. Mosch, P. Grau, R. Hesse, P. Streubel, and R. Szargan, “Hard Amorphous CSiJy Thin Films Deposited by RF Nitrogen Plasma Assisted Pulsed Laser Ablation of Mixed GraphitelSi3N4-Targers,” Thin Solid Films, 348 103-1 13 (1999). 15. 2. He, G. Carter, and J. S. Colligon, “Ion-Assisted Deposition of C N and Si-GN Films,’’ Thin SolidFilms, 283 90-96 (1996). 16. K. B. Sundaram and J. Alizadeh, “Deposition and Optical Studies of Silicon Carbide Nitride Thin Films,” Thin Solid Films, 370 151-154 (2000). 17. J. Wilden, A. Wank, M. Asmann, J. V. R. Heberlein, M. 1. Boulos, and F. Gitzhover, “Synthesis of Si-C-N Coatings by Thermal Plasmajet Chemical Vapour Deposition Applying Liquid Precursors,” Appl. Oqanomct. Chon., 15 841-857 (2001). 18. T. Berlind, N. Hellgren, M. P. Johansson, and L. Hultman, “Microstructure. Mechanical Properties, and Wetting Behavior of Si-C-N Thin Films Grown by Reactive Magnetron Sputtering,” Surf Coat. Technol., 141 145-1 55 (2001). 19. Y. W. Bae, H. Du, B. Gallois, K. E. Gonsalves, and B. 1. Wilkens, “Structure and Chemistry of Silicon Nitride and Silicon Carbonitride Thin Films Deposited from Ethylsilazane in Ammonia or Hydrogen,” Chon. Muter., 4 478483 (1992). 20. J. Bill and D. Heimann, “Polymer-Derived Ceramic Coatings on ClC-Sic Composites,”/. Eur. Ceram. SOL.,16 1115-1120 (1996).
4.
236
21. M. R. Mucalo, N. B. Milestone, 1. C. Vickridge, and M. V. Swain, “Preparation of Ceramic Coatings from Pre-Ceramic Precursors: Part 1 S i c and “Si3N&i2N20’ Coatings on Alumina Substrates,” /. Muter. Sci, 29 44874499 (1994). 22. G. Motz and G. Ziegler, “Simple Processibiliry of Precursor-derived SiCN Coatings by Optimised Precursors,” Kq Ens Muter., 20621 3 475-478 (2002). 23. Y. D. Blum, R. M. Platz, and E. J. Crawford, “Glass Strengthening by Polymer-Derived Ceramic Coatings,” /. Am. Ceram. Soc., 73 [ l ] 170-172 (1990). 24. R. Maboudian, W. R. Ashurst, and C. Carraro, “Tribological Challenges in Micromechanical Systems,” Tribol. Lett., 12 95-100 (2002). 25. S. A. Henck, “Lubrication of Digital Micromirror Devices,” Tribal. Lett., 3 239-247 (1997). 26. T. M. Mayer, J. W. Elam, S. M. George, P. G. Kotula, and R. S. Goeke, “Atomic-Layer Deposition of Wear-Resistant Coatings for Microelectromechanical Devices,” Appl. Phys. Lett., 82 2883-2885 (2003). 27. L.-A. Liew, W. Zhang, L. An, S. Shah, R. Luo, Y. Liu, T. Cross, M. L. Dunn, V. Bright, J. W. Daily, and R. Raj, “Ceramic MEMS New Materials, Innovative Processing and Future Applications,” Am. Ceram. SOC.Bull., 80 [5] 2530 (2001). 28. R Riedel, G. Passing, H. Schonfelder, and R. J. Brook, “Synthesis of Dense Silicon-Based Ceramcis at Low Temperatures,” Nature, 355 [20] 7 1 4 7 1 7 (1992). 29. J. Bill, J. Seitz, G. Thurn, J. Durr, J. Canel, B. Z. Janos, A. Jalowiecki, D. Sauter, S. Schempp, H. P. Lamparter, J. Mayer, and F. Aldinger, “Structure Analysis and Properties of Si-C-N Ceramics Derived from Polysilazanes,” Phys. Stam Solidi A, [166] 269-296 (1998). 30. R. Riedel, “Advanced Ceramics from Inorganic Polymers”; pp. 1-50 in MateriuL Science and Technology: A Comprehensive Treatment, Vol. 178. Processing of Ceramics, Pan II. Ed., R. J. Brook. VCH, Wurzburg, Germany, 1996. 31. J. Bill, J. Schumacher, K. Miller, S. Schempp, J. Seia, J. Durr, H. P. Lamparter, J. Golczewski, J. Peng, H. J. Seifert, and F. Aldinger, “Investigations into the Structural Evolution of Amorphous SiCN Ceramics from Precursors,” 2 Metallkd, 94 [4] 335-351 (2000). 32. S. R. Shah and R. Raj, “Mechanical Properties of a Fully Dense Polymer Derived Ceramic made by a Novel Pressure Casting Process,” Acta Mater., 50 [16] 4093-4103 (2002). 33. E. Kroke, Y.-L. Li, C. Konetschny, E. Lecomte, and C. Fasel, “Silazane-Derived Ceramics and Related Materials,” Mater. Sci. Eng. R. A Rev. /., 26 97199 (2000). 34. Y. Wang, D. C. Alsmeyer, and R. L. McCreery, “Raman Spectroscopy of Carbon Materials: Structural basis of observed spectra,” Chem. Muter., 2 557-563 (1990). 35. A. C. Ferrari and J. Robertson, “Interpretation of Raman Spectra of Disordered and Amorphous Carbon,” Phys. Rev. B, 61 [20] 14095-14107 (2000). 36. D. Tallant Polyacrylonitrile, personnel communication. 37. M. Hokao, S. Himnaka, Y. Suda, and Y. Yamamoto, “Friction and Wear Properties of GraphitelGlassy Carbon Composites,” Wear, 237 54-62 (2000). 38. P. J. Bryant, P. L. Gutshall, and L. H. Taylor, “A Study of Mechanisms of Graphite Friction and Wear,” Wmr, 7 118-126 (1964). 39. T. Cross, S. V. Prasad, D. R. Tallant, and R. Raj. MANUSCRIPT IN PROGRESS.
Progress in Nanotechnology: Processing
Synthesis and Tribology of Carbide-Derived Carbon Films A. Erdemir* and A. Kovalchenko Argonne National Laboratory, Energy Technology Division, Argonne, Illinois 60439
C. White, R zhu, A. Lee, and M. J. McNallan Department of Civil and Materials Engineering, University of Illinois, Chicago, Illinois 60607
B. Carroll and Y. Gogotsi Department of Materials Science and Engineering and A. J. Drexel Nanotechnology Institute, Drexel University, Philadelphia, Pennsylvania 19104
Carbide-derived carbon (CDC) films are produced at atmospheric pressure on the surfaces of carbide-based ceramic materials and coatings by a high-temperature chlorination process. These nanoporous carbon films contain carbon nanoonions and amorphous carbon, and may contain nanocrystalline diamond and graphite as well, depending on the synthesis conditions. The combination of such diverse carbon phases in one material or coating provides unique and potentially useful properties for a wide range of engineering applications. In this paper, we will present the results of a comprehensive study on the tribological behavior of these films. The friction coefficient of C D C in open air is comparable with that of graphite and is typically in the range of 0.15-0.25. However, the friction coefficients of C D C tend to decrease with decreasing humidity. In dry nitrogen, its friction coefficient is -0.1 or less. Such behavior is in contrast to that of crystalline graphite, which normally exhibits low friction at high humidity, but high friction at low humidity or in vacuum. The friction coefficient of C D C becomes increasingly lower under heavier loads; however, increasing sliding velocity does not seem to affect its frictional behavior significantly. Using a hydrogenation process that removes residual chlorine from the C D C film, the friction coefficients of CDC can be further lowered to values as low as 0.03. In an attempt to understand some of the underlying mechanisms, we carried out comprehensive chemical and structural studies of the sliding surfaces as well as bulk films and correlated these findings with the friction and wear behavior of C D C films.
Introduction Carbide-derived carbon ( C D C ) is a novel nanostructured material that holds promise for numerous ‘
[email protected] 02006 The American Ceramic Society
Membranes, Films, and Coatings
industrial applications such as catalysis, hydrogen storage, molecular sieves, rolling, rotating, and sliding tribological systems.lP2It is produced o n the surfaces of metal carbides such as S i c a n d TIC by a high-tem erature chlorination process at atmospheric pressure?”Speci fically, the carbide-forming element is selectively etched out in a halogen-containing carrier gas at temperatures
237
ranging from 600°C to 1200°C. The carbon atoms left behind rearrange themselves and form the layer of CDC. While pure chlorine can be used, the typical carrier gas composition for chlorination of the specimens described in this paper consists of Ar and small amounts (i.e., 3-10%) of Clz. Hydrogen may also be added to this mixture to control the reaction kinetics and, hence, the growth rates and structure of the CDC films. Because of the higher thermodynamic stability of metal chlorides in comparison with CCl, within the range of 600-10OO0C, chlorine reacts preferentially with the metallic part in carbides and leaves carbon behind as a thin film. Recent electron microscopic studies have revealed that CDC may contain not only amorphous carbon and turbostratic graphite but also nanocrystalline diamond and nested fullerene nanostructures such as nanotubes and nano-onion~.~ The process used in the production of CDC is unique mainly because of its ability to produce diamond-structured carbon at atmospheric pressure and temperatures below lOOO"C.* This process does not require the creation of a gas discharge plasma or other forms of high-energy activation that are needed for the synthesis of diamond films. Unlike diamond and diamond-like carbons (DLC), CDC can be grown in the thickness range from 1 to more than 100 pm at very high growth rates. Because of their high internal stresses and very low deposition rates, diamond and DLC films may not be grown very thick. In most instances, the DLC films are less than 5 p m thick, while diamond could be up to 2Opm in thickness. Because of the very low growth rates, it may take a very long time to deposit such carbon films on suitable substrates. The growth rate of CDC films is very high and takes only a few hours to produce films that are tens of micrometers thick. As C D C is derived from carbidebased substrates, there is no significant change in the original dimensions of the work pieces. At or near the interfaces between CDC and carbide-based substrates, a chemically graded interface provides a very smooth transition from a high hardness value near the CDC/SiC interface to a lower value at the external surface. This graded interface reduces the susceptibility of CDC to spallation and delamination in comparison with diamond or DLC films, which are vulnerable to spallation due to the build up of very high internal stresses at or near the interface. Because of its unique nanostructure, CDC possesses very attractive physical, mechanical, and tribological properties. In particular, recent systematic studies in
238
our laboratory have confirmed that CDC can afford very low friction and wear to sliding tribological interfaces. Specifically, depending on the test conditions, this coating can provide friction coefficients of 0.03-0.3 and wear rates of 10-9-10-7 mm3/N m.5-8 Therefore, C D C holds promise for large-scale applications in rotating equipment industries where S i c is used as a higher end seal material. In recent years, CDC films were further optimized to yield much improved friction and wear properties. Specifically, the growth kinetics of these films were thoroughly examined in gas mixtures including Ar, C12, and H2 at temperatures ranging from 600°C to 1200°C. As a result of these systematic studies, the growth conditions that can lead to the formation of an ideal film were determined. Extensive mechanical and tribological studies of the optimized films have indicated a close relationship between the growth conditions and the performance of these coatings.'-* Specifically, the hardness of the CDC layers can be increased by reducing the temperature of the treatment and blending the chlorinating gas mixture with hydrogen. In this paper, we concentrate mostly on the mechanical and tribological properties of highly optimized CDC films on Sic. Furthermore, we explore the role of hydrogen in the friction and wear behavior of the CDC films. During our studies, we used Raman spectroscopy and scanning electron microscopy (SEM) to determine the microstructure and chemistry of the CDC films and correlated these findings with the friction and wear data obtained under a wide range of test conditions.
Experimental Details Synthesis of CDC Films The C D C films can be produced on the surfaces of all kinds of carbide-based bulk ceramics and their thin coatings. One of the most common carbide ceramics is S i c (HexoloyTM, Saint Gobain Ceramics, Niagara Falls, NY), which we have used extensively in our previous studies over the years. Figure l a shows a schematic representation of the chlorination process for which a short description is provided below, but more details are available in recent papers by Gogotsi and colleagues14 The interface between CDC film and Sic substrate is physically rugged and chemically graded, as shown in Fig. Ib. The S i c substrates used in our study were about 6 mm thick and prepared in disk and/or square shapes.
Progress in Nanotechnology: Processing
Fig. 2. Photographs of (a) as-received and (6) carbide-derived carbon treated at 1000°C S i c disks showing a black carbon layer on the exposed suface. Fig. 1. (a) Schematic representation of the chlorination process used in the production of carbide-derived carbon (CDC) on S i c and (b) illustration of the graded inteface between CDC and Sic.
The nominal diameter of the disks was about 30mm, while squares were roughly 25.4 mm x 25.4 mm. One of the faces of these flat samples was highly polished to an RMS surface finish of better than 0.01 pm. Chlorination of disk and square S i c substrates was performed in gas mixtures consisting of various ratios of
Membranes, Films, and Coatings
C12:Hz and Ar at 1000-1100°C for 3-6h. The gases used in the tube furnace were supplied from cylinders of pure C12, pure Ar, and H2. They were passed through anhydrous CaSO4 and concentrated HZS04 to remove any traces of water vapor. The typical gas mixture used in our study included 9% C12-4.5% H2-86.5% Ar (by volume). The photographs in Fig. 2 show one of the S i c disks before and after C D C treatment. The disk changed its color from gray to black after chlorination. The coating thickness ranged from 20 to 1OOpm and
239
average. Further details of the test method used in the tribological testing of CDC films can be found in ASTM G99, Section 3 (Metals Test Methods and Analytical Procedures, Volume 3.02, Wear and Erosion; Metal Corrosion).
Results Microstructure and Mechanical Properties Fig. 3. Scanning electron microscopy image showing the morphology of a carbide-derived carbon-coated su face at high magnifcation. The su face becomes ve’y rough ajer the chlorination treatment.
depended on the process parameters, such as temperature, gas flow rate, gas mixture, and duration. Figure 3 shows the physical condition of the as-produced CDC on S i c surface. As is clear, the surface of the CDC layer is rather rough after chlorination and removal of the outer powdery material. Such a rough surface might cause additional frictional forces during tribological testing; therefore, before subjecting CDC to tribological tests, we performed a very brief polishing on these surfaces. Some of the polished samples were hydrogenated by annealing at 800°C for 8 h in 5% H2/95% Ar. The treated specimens were analyzed by micro-Raman spectroscopy using an Ar ion laser (514.5 nm line) at x 500 magnification with a spot size of 2 pm. Post-treatment polishing of some samples to a mirror finish was performed manually on a polishing wheel using an alumina suspension.
Some of the chlorinated S i c samples were sectioned and mounted on an SEM stage for microscopic examination. Figure 4 shows the cross-sectional morphology of a typical CDC film produced in a Cl-Hz gas mixture at 1000°C. As is clear, the film on top has a distinct microstructure, and the interface is rugged, as shown in Fig. I b. The columnar morphology typical of most physical vapor-deposited films is not visible on CDC films, and except for some polishing marks, the overall microstructure is very dense and featureless. The polishing marks are most likely due to the fact that CDC is softer than S i c and hence can readily be scratched during the polishing of the harder Sic. The CDC film shown in Fig. 4 is about 50 pm thick and appears to be strongly bonded to its substrate. The microcracks shown in this photomicrograph were not there originally, but generated from a Vickers indentation close to the interface. Because of the graded interface and strong bonding between CDC and Sic, the crack propagated into the CDC and away from the interface. The cracks deflected toward the CDC instead of following the interface because CDC is under a tensile stress. This is supported by
Friction a n d Wear Tests The friction and wear properties of CDC films and as-received Sic samples were evaluated in a ball-on-disk machine under a wide range of test conditions. The contact load varied between 2 and ION, while the sliding velocity was in the range of 10-1 5 cm/s. Tests were performed in a number of environments, including dry nitrogen and open air (with 17-32% relative humidity). The total sliding distance varied between 120 and 180 m, depending on the wear track diameters. The test balls (9.55 mm in diameter) were made of sintered Si3N4 and had a surface finish of better than 0.01 pm centerline
240
Fig. 4. Cross-sectional micrograph of a carbide-derived carbon film grown on S i c The microcrack shown was caused by a Vickers indentation near the interjace.
Progress in Nanotechnology: Processing
0.8 nm. Therefore, the softness of these films may have, in part, been due to the porous nanostructure. The hardness of the transition layer between CDC and S i c was much higher, that is, 20-35 GPa. Films produced at 600-800°C were two to three times harder and more resilient than the films produced at or above 900°C.
Tribological Properties Fig. 5. Typical Raman spectrum of a carbide-derived carbonfilm used in our studies. This particular sample was chlorinated in 2:l Cl*:H,for 6 h at 1000°C.
numerous ripples and cracks in the outermost layer of CDC seen in Fig. 3. Detailed studies by Raman spectroscopy revealed two principal peaks for CDC films, as shown in Fig. 5. One of the peaks was centered at around 1340cm-', while the other was at 1590 cm-'. The position of these two peaks is close to that of D and G bands of graphite. The D band is downshifted, while the G band is upshifted compared with microscrystalline graphite. A much higher intensity of disorder-induced D band compared with G band, corresponding to in-plane vibrations of carbon atoms in graphite, suggests a nanocrystalline and disordered structure of CDC. Hence, based on the Raman results, we concluded that the CDC was highly graphitic. However, separate transmission electron microscopy studies at very high resolutions revealed a truly nanocomposite microstructure for CDC. Specifically, these studies have shown that the microstructure is composed of amorphous carbon, turbostratic graphite, carbon nano-onions, and nanocrystalline diamond phases3,* Diamond nanocrystals were found within the transition zone between S i c and CDC in some samples, while the graphitic phase was more dominant in the outer layer of CDC. In general, the structure and composition of the films depend on the synthesis conditions, such as temperature and environment. Detailed mechanical characterization studies by nanoindentation have revealed that the C D C films were rather soft, with typical hardness values ranging from 1 to 1.5 GPa. The elastic modulus of these films was also very low; that is, in the range of 8-12GPa. Such a low hardness and resilience may have been due to the highly graphitic nature of CDC, as revealed by Raman studies. It is important to remember that CDC films are porous at the nanoscale with a pore size of 0.7-
Membranes, Films, and Coatings
Tribological tests were performed in a pin-on-disk machine under a wide range of test conditions. Initially, we ran several tests with base (untreated) S i c in different environments and compared its friction coefficients with those of the as-grown CDC. Figure 6 shows the typical friction coefficients of S i c and CDC during sliding against the Si3N4 balls. The friction coefficient of base S i c is around 0.8 in open air and fluctuates rather heavily, but that of the C D C is around 0.25 and is very steady. These friction values are typical of SiCbased ceramics and pyrolytic graphite in open air. When tests were run in dry nitrogen, the friction coefficient of C D C became much lower, that is, less than 0.1. However, the friction coefficient of bulk S i c increased to 0.9. Figure 7 shows the wear tracks formed on the surface of a C D C film during tests in air and dry nitrogen. As is clear, the size of the wear track formed during testing in open air is much larger. As is apparent from Fig. 3, the physical roughness of as-grown CDC is rather high. The typical R, values of
Fig. 6 Friction coefficients of base S i c and carbide-derived carbon films in open air and d y nitrogen. Thefiiction coefficient of S i c in dty nitrogen was around 0.9.
241
Fig. 7. Typical sizp of wear tracksformed on carbide-derived carbon during tests in open air and dry nitrogen.
such films were in the range of 0.3-0.4pm. By progressively polishing the surface of such films, we were able to lower the R, values to less than 0.1 pm. When we repeated some of the tribological tests on such polished CDC films, we obtained as high as 40% reductions in friction (depending on the test environment). Figure 8 shows the effect of mechanical polishing on the frictional behavior of CDC films in open air. It is well known that physically rough surfaces can cause high friction and severe wear losses in most sliding contacts. Specifically, if the sliding surfaces are very rough, a high level of mechanical interlocking can take place between surface asperities, and this condition leads to high frictional losses. A prime example is the inherently rough surface finish of microcrystalline diamond films, which can cause very high friction and severe wear losses during sliding contacts. In fact, recent systematic
Fig. 8. Effect ofpolishing on the friction coeficient of carbidederived carbon in laboratory air.
242
studies have demonstrated the existence of an almost linear correlation between surface roughness and friction and wear coefficients of such diamond films. In general, the higher the surface roughness, the greater the friction and wear losses.9 Earlier studies by Erdemir and colleague^'^-'^ have determined that hydrogen plays an important role in the friction and wear behavior of diamond and DLC films. Specifically, DLC films grown in hydrogen-rich plasmas of a plasma-enhanced chemical vapor deposition system exhibited friction coefficients less than 0.01 in dry nitrogen. Based on these findings, we hypothesized that CDC films should also be able to provide very low friction coefficients after a hydrogenation treatment. Through a systematic study, we determined that a postprocess annealing of CDC in 5% H2+95% Ar at 800°C for 8 h was very effective. This hydrogen treatment decreased the IDIIG ratio of Raman spectra of CDC films. This change suggested that the CDC films had become more graphitic during hydrogen treatment, but a considerable disorder still existed. Hydrogen also removes residual chlorine trapped in pores of CDC. Figure 9 summarizes the results of friction tests on CDC films before and after the hydrogenation treatment. As is clear, the friction coefficients of as-grown CDC films are much higher in both air and dry nitrogen than those of the CDC films subjected to a post-hydrogenation treatment. The friction coefficients of the as-grown CDC in air and dry nitrogen are around 0.18 and 0.14, respectively, but after hydrogen treatment the value decreases to 0.07.
Fig. 9. Friction coe8cients of carbide-derived carbon films before and after the hydrogenation treatment.
Progress in Nanotechnology: Processing
Fig. 10. Effect of contact load on thejiction coeficient of carbide-&rived carbon films.
In another series of tests, we investigated the effects of load on the friction of CDC films. Figure 10 shows the frictional behavior of hydrogen-treated C D C under 2 and 10 N loads. Specifically, it shows that under higher loads, the friction coefficient becomes even lower, that is, 0.03. Also note that the initial friction coefficient is lower and the time required to reach the steady-state friction value is shorter under the heavier load. The wear of sliding ball and disk surfaces was very difficult to quantify after wear tests. In all cases, a shallow wear track was formed on the CDC-covered disk side, and its size was strongly dependent on the test environment. The tracks formed in open air wete much wider than those formed in dry nitrogen and had smooth surfaces as shown in Fig. 7. The wear on the Si3N4 ball side was, in most cases, minimal (partially due to the short duration of the sliding tests) and hard to quantify. A typical 3D image of one of the wear scats formed on a Si3N4 ball is shown in Fig. 11 (a). A dark contrasting transfer layer was often observed on some of the sliding ball surfaces, especially after tests in dry nitrogen (see Fig. 11 (b)).
Discussion From the results of the tribological tests presented above, we can conclude that CDC has the ability to lower drastically the friction coefficients of Sic-based ceramics. In separate studies, we obtained similar results from TIC, WC, and other carbide-based ceramics. Such
Membranes, Films, and Coatings
Fig. 11. (a) Three-dimensional (30) and (6) optical images of Si@d ball sufaces afier a sliding test in d y nitrogen. 3 0 image suggests that not much wear had occurred, but the optical photograph reveals the presence of a transfir hyer on and around the contact spot. The width of wear scar shown is 300 pm.
a reduction in friction can be primarily attributed to the low-shear strengths of the phases that make up CDC. Such phases can easily shear during sliding contacts to provide low friction. The CDC films used in our tests were largely made of a graphitic phase. Graphite is a well-known solid lubricant and is able to reduce the friction of sliding tribological interfaces.l 4 At the same time, graphitic carbon in CDC forms a network struc-
243
ture that does not contain large planar crystals of graphite that would break under load applied during friction tests. Among the other phases that make up CDC, we feel that nano-onions with their great ability to carry a load and roll over under sliding contact situations may have also played a significant role in lowering the frict i ~ n Nanocrystalline .~ diamond phases are also known to provide low friction when present on a sliding surface.15 Overall, the nanostructured nature of the CDC is a key to the impressive frictional properties shown in Fig. 6. As for the dramatic effects of the test environment on friction, studies to date have confirmed that a very strong relationship exists between the frictional behavior of most carbon films (diamond, DLC, or graphitic carbon films) and the species within the test environmerits. 10,15-19 Some of these studies”-”,” have shown that, if a film is hydrogen free or poor, then very high friction coefficients may result (especially during tests in inert or vacuum environments). It is now generally believed that the surface termination states of these carbon films are extremely important for their frictional behavior. If their sliding surfaces are fully terminated by hydrogen, then these films may provide fairly low friction coefficients. Conversely, if their surfaces are chemically active and contain dangling bonds, then the friction coefficients of these films may become very high. In addition to strong covalent bonding, other types of short- and long-range chemical and/or physical interactions (van der Waals forces, electrostatic and n-n* attractions, capillary forces, etc.) may exist between sliding carbon surfaces and contribute to overall friction.’ ‘-13 We believe that most of the frictional interactions mentioned above are relevant to the behavior of the CDC films tested in our studies. In dry and inert test environments, strong covalent and relatively weak n-n* interactions as well as van der Waals forces may act at the sliding interfaces and control the frictional behavior of CDC films. The n-n* interactions are relevant to the sliding interfaces of graphite or graphitic carbons, such as CDC. Also, the frictional behavior of the CDC films tested in open air could be largely controlled by the graphitic phase, and their friction coefficients (i.e., 0.2) measured in air are typical of graphite. However, when these films are tested in dry nitrogen, their friction coefficients become even lower; this behavior is not the case for pure graphite, which generally provides low friction in moist air. For such an observation, we provide the following explanation.
244
As CDC results from selective etching of metal atoms in carbides by C12, some portion of C12 is retained within the growin CDC structure (either in the atomic or bonded form)!’ Also, the small amounts of hydrogen within the carrier gas can act in a similar fashion. Such possibilities may serve the same purpose as water molecules do in a graphitic structure by simply overcoming or reducing the detrimental effects of the “ S O called” n-n* interactions. Another halogen gas, fluorine, is often used in the making of fluorinated graphite, which provides very low friction and possesses friction properties that are relatively insensitive to large fluctuations in relative humidity.21 Also, fluorine doping of DLC films has substantial beneficial effects on the wear behavior of such films.22 We believe that a similar mechanism may have been responsible for the relatively lower friction coefficients of the CDC films in dry nitrogen. Energy-dispersive X-ray analysis of CDC has verified the presence of some Cl in the structure, but the detection of hydrogen was not possible by this method. Nanopores in CDC also allow accumulation of humidity due to capillary forces and termination of dangling bonds by functional groups throughout the coating. Chlorine transport through the growing CDC film has been identified as the rate-controlling step for growth of thick CDC layers on The activation energy for CDC growth is consistent with the chlorine being chemically adsorbed on the surfaces of the nanoscale pores in the CDC. Only a small fraction of the chlorine in the CDC can be removed by an 8-h heat treatment in N2 at 800°C, while a similar treatment in Ar-5% H2 removes 90% of the chlorine. In the presence of hydrogen, the thermodynamic driving force for formation of HCl contributes to the removal of chlorine from C D C , and excess hydrogen is available to replace chlorine atoms on adsorption sites on the CDC surfaces. As for the much lower friction coefficients (the lowest of all measured) of hydrogen-treated CDC films in dry nitrogen, we provide the following mechanistic explanation. Within the mostly amorphous and disordered graphitic microstructure of CDC films (especially near the top), many voids, defects, and active atomic and/or molecular sites may contain dangling or unsatisfied (T bonds. The existence of such bonds at sliding contact interfaces is undesirable (especially in inert and vacuum environments) as they can easily cause adhesive interactions during sliding motion. During the hightemperature hydrogenation, we believe that most of these bonds are effectively eliminated by the hydrogen
Progress in Nanotechnology: Processing
atoms. Such a situation will undoubtedly lead to a higher degree of chemical passivation of the carbon atoms at or near the CDC surface. When such surfaces are rubbed against other materials (Si3N4 in our case), they cause much lower adhesion (hence, friction). Some of the hydrogen may diffuse or migrate well into the structure and remain there as interstitials, while others may eliminate active 6 bonds within the film as well. Remember that C-H bonding is very strong (i.e., stronger than single C-C bonds), and that thermal desorption of hydrogen from carbon does not occur below 700°C.23 In short, elimination of the possibility of strong covalent bond interactions at sliding CDC interfaces is a major reason for the very low friction behavior of CDC films in dry nitrogen. This explanation is consistent with the mechanistic model proposed for the superlubricity of highly hydrogenated DLC
films.’ 1,1324
Conclusions
The results of our study have further confirmed that CDC films are self-lubricating and capable of providing very low friction coefficients to sliding surfaces, especially in dry nitrogen. In humid air, the friction coefficients are somewhat higher but still three to four times lower than the base Sic. High-temperature hydrogenation treatment has a dramatic effect on the frictional behavior of C D C films. It causes significant reductions in friction, especially during tests in dry nitrogen and under heavier loads. The as-grown C D C samples also yielded relatively lower friction in dry nitrogen. In open air, the friction coefficients of both the as-received and hydrogen-treated samples were similar (i.e., around 0.2). Based on these experimental observations, we proposed that the frictional behavior of CDC films in dry nitrogen was largely controlled by their subnanometer porosity and the presence of onion-like structures at sliding interfaces. Hydrogen treatment may have effectively reduced or eliminated dangling bonds, removed chlorine trapped in pores, and hence, provided the very low friction coefficients observed in dry nitrogen. The frictional behavior of these films in laboratory air was largely dominated by the adsorbed water molecules and/ or oxygen-containing hnctional groups covering the surface of CDC.
Membranes, Films, and Coatings
References Y. G. Gogotsi, J.-D. Jeon, and M. J. McNallan, “Carbon Coatings on Silicon Carbide by Reaction with Chlorine-Containing Gases,” J Muter. Chon., 7 [9] 1841-1848 (1997). 2. D. A. Ersoy, M. J. McNallan, and Y. Gogotsi, “Carbon Coatings Produced by High Temperature Chlorination of Silicon Carbide Ceramics,” Muter. Res. Inn., 5[2] 55-62 (2001). 3. Y. Gogotsi, S. Welz, D. A. Ersoy, and M. J. McNallan, “Nature, Conversion of Silicon Carbide to Crystalline Diamond-Structured Carbon at Ambient Pressure,” Nacure, 41 1[6835] 283-287 (2001). 4. S. Welz, Y. Gogotsi, and M. McNallan, “Neucleation, Growth, and Graphitization of Daimond Nanocrysrals During Chlorination of Carbides,” j.Appl. Phys., 93[7] 4207-4214 (2003). 5. D. A. Ersoy, M. J. McNallan, Y. Gogotsi, and A. Erdemir, “Tribological Properties of Carbon Coatings Produced by High Temperature Chlorination of Silicon Carbide,” Tribol. Trans., 43 809-81 5 (2000). 6. M. J. McNallan, D. Ersoy, R Zhu, A. Lee, C. White, S. Welz, Y. Gogotsi, A. Erdemir, and A. Kovalcbenko, “Nano-Structured Carbide Derived Carbon (CDC) Films and their Tribology,” j. Tringhua Sci. Tccbnol., 10 699-703 (2005). 7. 8. Carroll, Y. Gogotsi, A. Kovalchenko, A. Erdemir, and M. J. McNallan, “Effect of Humidity on the Tribological Properties of Carbide-Derived Carbon (CDC) Films on Silicon Carbide,” Tribol. Lett., 15 [l] 51-55 (2003). 8. A. Erdemir, A. Kovalchenko, M.J. McNallan, S. Welz, A. Lee, Y. Gogotsi, and 8. Carroll, “Effects of High-Temperature Hydrogenation Treatment on Sliding Friction and Wear Behavior of Carbide-Derived Carbon Films,” Surf Coat. Tccbnol., 188 588-593 (2004). 9. A. Erdemir, “Wear of Diamond and Diamond-Like Carbon Films,” J Eng. Tn’b~l.,216 387-400 (2002). 10. A. Erdemir and C. Donnet, “Tribology of Diamond, Diamond-Like Carbon and Related Films,” Modcrn Tribolo~Handbook. ed., 8. Bhushan. CRC Press, B o a Raton, FL, 871-899, 2000. 11. A. Erdemir, 0. L. Eryilmaz, and G. Fenske, “Synthesis of Diamond-Like Carbon Films with Superlow Friction and Wear Properties,” j. Vac. Sci. Tecbnol., A18[4] 1987-1992 (2000). 12. A. Erdemir, “Superlubricity and Wearless Sliding in Diamond-like Carbon Films,” M a w . Res. SOC.Symp. Proc., 697 391-403 (2002). 13. A. Erdemir, “The Role of Hydrogen in Tribological Properties of DiamondLike Carbon Films,” Surf Coat. Technof.,146-147 292-297 (2001). 14. A. Erdemir, “Solid Lubricants and Self-Lubricating Films,’’ Handbook of Modern Tribology, CRC Press, B o a Raton, FL, 787-818, 2001. 15. A. Erdemir, G. R. Fenske, A. R. Krauss, D. M. Gruen, T. McCauley, and R. T. Csencsits, “Tribological Properties of Nanocrystalline Diamond Films,” Surf Coat. Technol., 121 565-572 (1999). 16 S. V. Pepper, “Effect of Electronic Structure of the Diamond Surface on the Strength of the Diamond-Metal Interface,” J. Vuc. Sci. Technol., 20[3] 643646 (1982). 17. D. S. Kim, T. E. Fischer, and 8. Gallois, “The Effects of Oxygen and Humidity on Friction and Wear of Diamond-like Carbon Films,” Surf Coat. Tcchnol.. 49 537-542 (1991). 18. M. N. Gardos, Tribiological “Fundamentals of Polycrystalline Diamond Films,” Surf Coat. Techno[., 113[3] 183-200 (1999). 19. C. Donnet, “Recent Progress on the Tribology of Doped Diamond-like and Carbon Alloy Coatings: A Review,” Su$ Coat. Tecbnol., 101 180-186 (1998). 20. A. Lee, “Effects of Processing on Tribological Properties of Carbide Derived Carbon on Silicon Carbide,” Ph. D. Thesis in Materials Engineering, University of Illinois at Chicago, Chicago-Illinois, December, 2005. 21. R. L. Fusaro and H. E. Sliney, “Graphite Fluoride [(CF x)n]-A New Solid Lubricant,” ASLE Tram., 13 56-65 (1970). 22 C. Donnet, J. Fontaine, A. Grill, V. Patel, C. Jahnes, and M. Belin, “WearResistant Fluorinated Diamond-like Carbon Films,” Surf Coat. Technol.,9495 531-536 (1997). 23 C. Su and J. C. Lin, “Thermal Desorption of Hydrogen from the Diamond C(100) Surface,” Surf Sci., 406 149-166 (1998). 24 A. Erdemir, “Design Criteria for Superlubricity in Carbon Films and Related Microstructures,” Tribol. Inr., 37 [7] 577-583 (2004). 1.
245
This Page Intentionally Left Blank
Nanotubes, Nanorods, and Nanowires
This Page Intentionally Left Blank
DESIGN, FABRICATION AND ELECTRONIC STRUCTURE OF ORIENTED METAL OXIDE NANOROD-ARRAYS Lionel Vayssieres National Institute for Materials Science, International Center for Young Scientists Tsukuba, Ibaraki, Japan 305-0044
ABSTRACT Materials chemistry has emerged as one of the most consistent fabrication tool for the rational delivery of high purity functional nanomaterials, engineered from molecular to macroscopic scale at low cost and large scale. An overview of the major achievements and latest advances of a recently developed growth concept and low temperature aqueous synthesis method for the fabrication of purpose-built large bandgap metal oxide semiconductor materials and oriented nano-arrays are presented. Important fundamental insights of direct relevance for semiconductor technology, optoelectronics, photovoltaics, and solar hydrogen generation are revealed by in-depth investigations of the electronic structure of metal oxide nanostructures with new morphology and architecture carried out at synchrotron radiation facilities. INTRODUCTION The controlled fabrication of well-defined and well-ordered one-dimensional large bandgap semiconductor nanomaterials such as nanorods, nanowires and nanobelts as well as their large scale manufacturing at low cost remain is of crucial importance to unfold the exciting and promising future of nanodevices. In addition to the rational and economical manufacturing of nanostructures, a better fundamental knowledge of their electronic structure as well as their physical, chemical, electrical and interfacial and structural properties is necessary to hlly reveal and exploit their fascinating potentials. To achieve such ambitious targets, a new approach for the rational development of semiconductor nanostructures has been developed. A strategy based on the chemical and electrostatic lowering of the water-oxide interfacial energy of the systems' allowed a direct thin film growth by heteronucleation onto various substrates (amorphous, poly-, or single-crystalline) of large physical areas. Such approach has the capability to generate functional nanomaterials at large scale low cost and low temperature. For instance, crystalline metal oxide nanostructures consisting of oriented multi-dimensional arrays featuring building blocks of controlled morphologies, sizes, aspect ratios and orientations at nano-, meso-, and micro-scale are genuinely fabricated without template, surfactant, undercoating, or applied external field from the hydro1 sis-condensation of aqueous metal salts and complexes at mild temperatures, ca. below 100°C . In-depth investigations of the electronic structure of such novel nanostructures have been carried out by x-ray spectroscopies at synchrotron radiation facilities. Such studies include x-ray photoelectron3 spectroscopy as well as soft x-ray absorption and emission4 measurements, including polarization dependent' and energy dependent resonant inelastic x-ray scattering6 experiments. The electron structure mapping results reveal important fundamental knowledge of orbital character and symmetry, bandgap, and quantum confinement effects of direct relevance for semiconductor technologies, optoelectronics, gas sensors, photovoltaics and photocatalysis for solar hydrogen generation.
Y
Nanotubes, Nanorods, and Nanowires
249
FABRICATION METHOD Our strategy to generate large-area of advanced nano and micro-particulate thin films at low cost and large scale is a bottom-up aqueous chemical growth approach7 that is wellsupported by a classical thermodynamic model, monitoring the nucleation, growth, and aging processes via the experimental control of the interfacial free energy of the system'. RESULTS AND DISCUSSION Such a strategy has been well-illustrated on the size control of magnetite nanoparti~les~ over an order of magnitude, i.e. between 1 and 10nm. This concept and synthetic method allows the design and the creation of metal oxide nanomaterials with novel morphology, texture, and orientation which enables to probe, tune, and optimize their physical properties". Particulate thin films and 3-D arrays are obtained by direct growth onto various substrates from the condensation of aqueous precursors at low temperatures. Such an approach to material synthesis offers the ability to generate anisotropic nanoparticles and to control their orientation onto substrates. Such an approach has been successfully applied for the growth of advanced nano and microparticulate metal oxide materials such as three-dimensional arrays of ferric oxide nanorods' for photovoltaic12 and phot~catalytic'~ applications, nanocomposite arrays of trivalent iron and chromium corundum oxidesI4 as well as oriented nanorod-I5, microrod-I6 and mi~rotube-'~ arrays of zinc oxide for photovoltaic, optoelectronic devices and gas sensors". Ferroma netic 3-D array of metallic iron nanorods for magnetic devices", 2-D arrays of chromium oxideYo for non-linear optics and ma netoelectrics applications, 3-D oriented arrays of tin dioxide nanorods with square cross-section2 F for gas sensing and photocatalytic applications have also been fabricated by such a method. An example of the variety of metal oxide structures that can be obtained with such an approach is given in figure 1.
'
Figure 1 Field-emission scanning electron microscopy images of crystalline oriented arrays of Sn02 (a); ZnO (b); and a-Fe2O3 (c) grown onto various substrates by aqueous chemical growth
250
Progress in Nanotechnology: Processing
Figure 2 (a) Polarization-dependent x-ray absorption spectra of ZnO (Wurtzite structure) 3-D arrays consisting of isotropic (spherical) and anisotropic (rod) morphology. The inset illustrates the XAS experimental geometry, where a-, b-axes define the sample surface plane, the c-axis is along the growth direction of the ZnO rods, E is the polarization of incoming photons and B indicates the angle of incidence with respect to the sample surface: 10' (E // c, lines) or 90' (E I c, dots); (b) Calculated polarization dependent 0 1s XAS of wurzite bulk ZnO, where energy is referred to valence-band maximum22;(c) Experimental and calculated x-ray absorption spectra of ZnO rods. The calculations of the XAS spectra include the DOS and the probability of transition to the Opx+y and Opz states5. Synchrotron radiation studies of the electronic structure of such materials have been carried out at beamline 7.01 at the Advanced Light Source (ALS), Lawrence Berkeley National Laboratory. Such beamline is equipped with a 99-pole, 5-cm period undulator and spherical grating monochromator. For instance, the determination of orbital symmetry and orbital character of the conduction band as well as the bandgap of 11-VI semiconductor ZnO (wurtzite structure) has been investigated. Strong polarization effects have been recorded on oriented
Nanotubes, Nanorods, and Nanowires
251
nanorod arrays (fig. 2a) compared to spherical nanoparticle~~. The measured XAS peaks are confirmed by the calculated polarization dependent 0-xy and 0 - z 1sXAS spectra22of bulk ZnO (Fig. 2b, 2c). The XAS has relatively strong in-plane character at energies of around 11, 17, 25 and 29 eV, and out-of-plane character at energies of around 11, 14, 20, and 32 eV. The broad measured XAS between peak a1 and peak a2 is not seen in the calculated spectrum. It may partly be broadening and/or excitonic effects, but may also arise from native donor-like defects. Direct assessment of the bandgap by probing the occupied (by XES) and unoccupied levels (by XAS) of ZnO nanostructures and single crystals has also been carried out4. The XES spectra of bulk and nanostructured ZnO are displayed together with the corresponding XAS spectrum in Fig. 3. The 0 K emission spectrum reflects the 0 2p occupied states (valence band), and the 0 1s absorption spectrum reflects the 0 2p unoccupied states (conduction band). In the photon energy region of 530-539 eV, the x-ray absorption can be mainly assigned to the 0 2p hybridized with Zn 4s states. In the region of 539-550 eV the spectrum is mainly attributed to 0 2p hybridized with Zn 4p states. Above 550 eV, the contribution is mainly coming from 0 2 p Z n 4d mixed states. Stronger s-p-d hybridization was revealed in nanostructured ZnO since the contributions of features at 520 eV and 523 eV are enhanced. A well defined band gap can be observed between the valence-band maximum and conduction-band minimum. Our absorption-emission spectrum yields the fundamental band-gap energy of 3.3 eV, which is in agreement with the 3.4 eV found for bulk ZnO.
Figure 3 Oxygen x-ray absorption-emission spectrum reflected conduction band and valence band near the Fermi level of ZnO nanoparticles in comparison with bulk Zn04. The electronic structure of quantum rod arrays of hematite has also been investigated6. Their Fe 2p absorption spectrum is displayed in figure 4a. The spectral shape appears very similar to previous XAS measurements conducted on polycrystalline or single-crystal samples. The typical spectrum shows the spin-orbit interaction of the 2p core level that splits the L2 (2~112)and L3 (2~312)edges, and the p-d and d-d Coulomb and exchange interactions that cause multiplets within the edges. The ligand field splitting of 3d transition metals, being of the same order of magnitude as p-d and d-d interactions (1-2 eV), gives a 1.4-eV-energy splitting between the tZg (xy,yz, xz) and eg (x2-y2, 3zz-r2)orbitals. Figure 4b shows the Fe L-emission spectrum recorded with a higher photon-energy excitation (ca 750 eV). The spectral shape shows two peaks originating from the transitions of 3d orbitals to 2 ~ 1 1 2and 2~312core levels. A branching ratio (LdLa) of 0.8 is found for the a-Fe2O3 bundled nanorod arrays, which appears substantially larger than that of the single crystal. The RIXS spectrum recorded at the Fe L-edge of a-Fe203
252
Progress in Nanotechnology: Processing
nanorods is shown in figure 4c. Several energy-loss features are clearly resolved. The low energy excitations, such as the strong d-d and charge-transfer excitations, are identified in the region from 1 to 5 eV. The 1-eV energy-loss features originate from multiple excitation transitions. The 4.1 and 6.4 eV excitation originates from charge transfer between oxygen 2p and iron 3d orbitals. The 2.5-eV excitation, which corresponds to the bandgap transition of hematite, appears significantly blue shifted compared to the reported 1.9-2.2 eV bandgap of single-crystal and polycrystalline samples as suggested above by the higher LB/La branching ratio observed in the L-emission spectrum (figure 3b) of the nanorods. Such direct observation of a substantial (0.30.6 eV) bandgap increase is successfully attributed to a 1-D (lateral dimension) quantum confinement effect in the designed bundled ultrafine nanorods of hematite (fig. 4d). Such a finding strongly suggests that these designed quantum rod-arrays would meet the bandgap and band edge requirements for solar hydrogen23generation without an applied bias. Further studies aiming at developing novel low-cost visible-active semiconductor photocatalysts are currently under investigation in our laboratories.
Figure 4 Fe 2p absorption (a), Fe L emission (b) and energy-dependent resonant inelastic x-ray scattering spectra (c) of a-Fe203 quantum rod-arrays (d). The inset (c) shows a schematic representation of the radiative de-excitation process for the two core excitatiom6
Nanotubes, Nanorods, and Nanowires
253
CONCLUSION The ability to hierarchically design and fabricate, at low cost, large three-dimensional arrays of transition metal oxides consisting of anisotropic nanoparticles with controlled orientations and aspect ratios onto various substrates without the use of templates, surfactants, or applied external fields contributes not only to the development of smart and functional nanomaterials for optoelectronic, photoelectrochemical, gas sensors and magnetic devices, but also to a better fundamental understanding of orbital character and symmetry contributions, bandgap and quantum confinement effects of one-dimensional anisotropic nanostructures by indepth studies of their electronic structure by soft x-ray spectroscopies at synchrotron radiation facilities. REFERENCES 1
L. Vayssieres, “On the thermodynamic stability of metal oxide nanoparticles in aqueous solutions”, h t . J. Nanotechnol. 2005, 2(4), 41 1-439 2 L.Vayssieres, “On the design of advanced metal oxide nanomaterials”, Int. J. Nanofechnol. 2004, I(I-2), 1-41 3 A. Henningsson, A. Stashans, A. Sandell, H. Rensmo, S. Sodergren, L.Vayssieres, A. Hagfeldt, S. Lunell, H. Siegbahn, “Proton insertion in Polycrystalline W03 studied with electron spectroscopy and semi-empirical calculations”, in Advances in Quantum Chemistv, edited by E. J. Brandas and E. Brandas (Academic Press, 2004), Vol. 47, pp. 23-36 4 C. L. Dong, C. Persson, L. Vayssieres, A. Augustsson, T, Schmitt, M. Mattesini, R. Ahuja, C. L. Chang, J.-H. Guo “The electronic structure of nanostructured ZnO from x-ray absorption and emission spectroscopy and the local density approximation”, Phys. Rev. B 2004, 70(19), 195325 ’J.-H. Guo, L.Vayssieres, C. Persson, R. Ahuja, B. Johansson, J. Nordgren, “Polarizationdependent soft-x-ray absorption of highly oriented ZnO microrods” J. Phys. : Condens. Matter 2002,14(28), 6969-6974 6 L.Vayssieres, C. Sathe, S. M. Butorin, D. K. Shuh, J. Nordgren, J.-H. Guo, “One-dimensional quantum-confinement effect in a-Fe2O3 ultrafine nanorod arrays”, Adv. Mater. 2005, I 7(19), 2320-2323 7 L.Vayssieres, “Designing ordered nano-arrays from aqueous solutions”, Pure Appl. Chem. 2006, 78(9) 1745-1751 ‘L. Vayssieres, “PrCcipitation en milieu aqueux de nanoparticules d’oxydes: ModClisation de l’interface et contr8le de la croissance”, PhD. Thesis, UniversitC Pierre et Marie Curie, Paris, France (1995), pp. 1-145 ’L.Vayssieres, C. Chaneac, E. Tronc, J.P. Jolivet, “Size tailoring of magnetite particles formed by aqueous precipitation: An example of thermodynamic stability of nanometric oxide particles”, J. Colloid Interface Sci. 1998,205(2), 205-212 “L.Vayssieres, “Advanced semiconductor nanostructures”, C. R. Chimie 2006,9(5-6), 69 1-701 “L.Vayssieres, N. Beermann, S.-E. Lindquist, A. Hagfeldt, “Controlled aqueous chemical growth of oriented three-dimensional nanorod Arrays: Application to iron(II1) oxides”, Chem. Mater. 2001, I3(2), 233-235 12 N. Beermann, L.Vayssieres, S.-E. Lindquist, A. Hagfeldt, “Photoelectrochemical studies of oriented nanorod thin films of Hematite”, J. Electrochem. SOC.2000, I47(7), 2456-246 1
254
Progress in Nanotechnology: Processing
13
T. Lindgren, H. Wang, N. Beermann, L.Vayssieres, A. Hagfeldt, S.-E. Lindquist, “Aqueous photoelectrochemistry of hematite nanorod-array”, Sol. Energy Muter. Sol. Cells 2002, 71(2), 23 1-243 14 L.Vayssieres, J.-H. Guo, J. Nordgren, “Aqueous chemical growth of aFe203-aCr203 nanocomposite thin films”, J. Nunosci. Nunotechnol. 200 1,1(4,, 385-388 15 L.Vayssieres, “Growth of arrayed nanorods and nanowires of ZnO from aqueous solutions”, Adv. Muter. 2003,15(5), 464-466 16 L.Vayssieres, K. Keis, S.-E.Lindquist, A. Hagfeldt, “Purpose-built anisotropic metal oxide material: 3D highly oriented microrod-array of ZnO”, J. Phys. Chem. B 2001, 105(17), 33503352 17 L.Vayssieres, K. Keis, A. Hagfeldt, S.-E. Lindquist, “Three-dimensional array of highly oriented crystalline ZnO microtubes”, Chem. Mater. 2001, 13(12), 4395-4398 18 J. X. Wang, X. W. Sun, Y. Yi, H. Huang, Y. C. Lee, 0. K. Tan, and L.Vayssieres, “Hydrothermally grown ZnO nanorod arrays for gas sensing applications”, Nanotechnology 2006,17(19), 4995-4998 ”L.Vayssieres, L. Rabenberg, A. Manthiram, “Aqueous chemical route to ferromagnetic 3D arrays of iron nanorods”, Nuno Lett. 2002,2(12), 1393-1395 20L.Vayssieres,A. Manthiram, “2-D mesoparticulate arrays of a-Cr203”, J. Phys. Chem. B 2003, 107(12), 2623-2625 21 L.Vayssieres, M. Graetzel, “Highly ordered Sn02 nanorod-arrays from controlled aqueous rowth”, Angew. Chem. Int. Ed. 2004, 43(28), 3666-3670 F2 C. Persson, C.L. Dong, L. Vayssieres, A. Augustsson, T, Schmitt, M. Mattesini, R. Ahuja, J. Nordgren, C.L. Chang, A. Ferreira da Silva, J.-H. Guo “x-ray absorption and emission spectroscopy of ZnO nanoparticles and highly oriented ZnO microrod arrays”, Microelectronics J. 2006,37(8), 686-689 23Solur Hydrogen and Nanotechnology, L.Vayssieres (Editor), Proceedings of SPIE-The International Societyfor Optical Engineering, SPIE Press Ltd., 2006, Vol. 6340, 336 pages
Nanotubes, Nanorods, and Nanowires
This Page Intentionally Left Blank
ELECTROSPINNING OF ALUMINA NANOFIBERS Karin Lindqvist, Elis Carlstrom IVF Industrial Research and Development Swedish Ceramic Institute Argongatan 30 SE-43153 Molndal, Sweden Anna Nelvig, Bengt Hagstrom IFP Research Argongatan 30 SE-431 53 Molndal, Sweden ABSTRACT A straightforward manufacturing route of alumina nanofibers was established by the electrospinning of a commercial alumina sol mixed with polyethylene oxide. The molecular weight and concentration of the polymer were varied as well as the spin parameters in order to find suitable fabrication settings. The results showed that the polymer content should be at least 27 vol% in the green fibre in order to avoid brittle fibres and the molecular weight of PEO should be at least 400 000 g/mol. The shear and elongation rheological properties of the suspensions were measured. A contraction flow method for measuring the elongational viscosity was evaluated and found to be suitable for highly shear thinning systems with high viscosity. The fibre diameters of calcined samples were in the range 400-700 nm. The electrospinning should be performed in a dry environment facilitating better control of the fibre deposition. INTRODUCTION Electrospinning is a convenient and simple process to obtain ultra thin ceramic fibres. In its simplest form a ceramic precursor suspension is placed in a syringe and subjected to a high electric potential visa-vi a grounded collector electrode. A schematic drawing of the process is shown in figure 1. When a voltage is applied to the syringe the pending spherical drop at the needle tip is changed into a conical shape (Taylor cone) and, upon a further increase in applied voltage (producing an increased charge density in the suspension), the electrostatic repulsive forces will overcome the surface tension of the suspension and a fine charged jet is ejected from the Taylor cone towards the grounded collector. During its flight the jet is highly elongated and the suspension medium is evaporating leaving a fibrous non-woven mat on the collector. The “green” fibres so produced are calcined into ceramic fibres at elevated temperature. Powder dispersions, alkoxide solutions or colloidal sols can be mixed with polymers to create the ceramic precursor suspension. During the last few years a number of different ceramics have been spun into sub-micron fibres by the electrospinning technique[13. The rheological properties of the precursor suspension turn out to be important for the electrospinning process and for the formation of uniform fibres. Since the electrostatic forces available are comparatively weak the counter acting viscous forces have to be limited in order to create thin fibres. This will set an upper bound on viscosity. At lower viscosities surface tension will become a dominant factor and capillary break up of the liquid thread into droplets (Rayleigh instability) may produce electro spraying instead of electrospinning. To this end it is common practise to use a soluble polymer in the precursor suspension to regulate its viscoelastic
Nanotubes, Nanorods, and Nanowires
257
properties. Several authors [2-41 have emphasised the importance of the elongational viscosity for the formation of uniform fibres. It was shown that the Rayleigh instability could be completely suppressed if a critical value of the elastic stress in the jet was exceeded. Yu [4] used an extensional rheometer measuring the filament break-up of the sample. Stading [5] developed a measurement system for shear thinning fluids that measured the elongational viscosity in contraction flow. This method gives the possibility to control the strain rate during the measurement, which is an advantage compared to measurements based on filament break-up. The parameters controlling the final diameter of electrospun fibres have been the subject of extensive research. Sigmund [ l ] modified the theory for prediction of the final jet radius to account for higher bulk electrical conductivity in ceramic-polymer suspensions. According to the theory the final green fibre diameter is dependent on the surface tension, the volume flow rate of the suspension, the volume current through the jet, the electrical conductivity and the electric field strength. A few papers report on the fabrication of electrospun aluminium oxide fibers. Azad [6] made transparent alumina fibers from aluminium pentadionate mixed with PVP in an acetone-ethanol solution. Larsen et al. [7] reported on the fabrication of alumina fibers from aged sol prepared from aluminium di-sec-butoxide ethylacetoacetat. The objective of this study was to establish a robust straightforward process for electrospinning of alumina fibers with controlled fibre morphology. There are commercially available ceramic sols commonly used as high refractory binders. These materials have high purity, small particle size, are delivered as stable colloidal water based dispersions and could therefore be suitable as raw material for electrospinning. No papers have to our knowledge been published on manufacturing of ceramic nanofibers from commercially available sols. Several water-soluble polymers could be candidates for facilitating the formation of ceramic fibers. Polyethylene oxide (PEO) is commonly used in electrospinning and was chosen in this study. The molecular weight and concentration of the polymer have been varied as well as the spin parameters in order to find suitable fabrication settings. The shear and elongation rheological properties of the suspensions have been measured. A contraction flow method for elongational viscosity was evaluated to see if this method was suitable for electrospinning suspensions. The fibre diameters were measured on calcined samples. EXPERIMENTS A commercially available colloidal alumina sol, Nyacol AL20 (Nyacol Nan0 Technologies Inc., USA) with a primary particle size of 50 nm was used as the ceramic precursor. Polyethylene oxide with molecular weights (Mw) of 100,000 (look), 400,000 (400k) and 900,000 (900k) (Sigma Aldrich, Sweden) were dissolved in the sol by magnetic stirring so that a suspension of polymer and sol was obtained. The conductivity (CDM210, Radiometer Analytical, France) was measured on the alumina sol as received and with the polymer dissolved. The suspensions were rheologically characterized by measuring the steady state shear viscosity and the phase shift in oscillatory shearing in a cup and bob (CC25) measurement system (StressTech, Rheologica, Sweden). The elongational viscosity was estimated through the measurement of the pressure drop over a specially designed contraction flow nozzle (Reologen i Lund, Sweden) that produces a constant elongational strain rate along its centre line, see figure 2. The measurement of elongational viscosity requires a rather strong shear thinning behaviour of the fluid in order to subtract the contribution from shear flow components to the measured pressure drop.
258
Progress in Nanotechnology:Processing
Electrospinning was performed in the experimental set up schematically shown in figure 1. Initial experiments were performed in ambient atmosphere with the electrospinning parameters set to 20 kV (high voltage power supply ES5OP from Gamma High Voltage, Ormon Beach, FL, USA), a syringe-collector distance of 20 cm; a syringe of 0.8 mm and a flow rate of 0.056 ml/min. A grounded platinum foil was used as collector electrode. Electrospinning for fibre diameter measurements were performed in a Plexiglas box under flowing nitrogen gas with parameters set to obtain a single well defined Taylor cone at the tip of the syringe needle. The parameters for each system are shown in table 1. The fibres were calcined at 700°C in air. XRD was used to identify the alumina phases obtained after calcinations. The specific surface area was measured (Gemini 2300, Micromeretics) on green and calcined fibres. Samples for fibre diameter measurements were prepared by sedimentation of calcined and crushed fibres on a sample holder. The fibres were manually measured from SEM micrographs.
Figure 1. Schematic of the electrospinning Figure 2. Measuring system for contraction process flow measurement. Table 1. Parameters for electrospinning of fibers for diameter measurements. Sample Composition Voltage Collector distance Field strength (kV) (cm) (kV/cm) A 3wt%400k 10 15 0.67 B 6wt%400k 20 15 1.33 C 3wt%900k 10 10 1 D 3wt%900k 10 10 1
Flow rate (mumin) 0.05 0.05 0.005 0.01
RESULTS The results from the shear viscosity measurements and frequency sweep measurements are shown in figures 3 and 4.
Nanotubes, Nanorods, and Nanowires
259
Figure 3. Shear viscosity vs. shear rate for polyethylene oxide dissolved in alumina sol.
Figure 4. Phase angle vs frequency for PEO dissolved in alumina sol Figure 3 shows that the suspension with 5 wt% PEO lOOk was Newtonian and had a low shear viscosity. No fibres were obtained with this sample but spherical particles from electrospraying, see figure 5. Severe dripping during handling also made this suspension unsuitable for electrospinning and no further analyses were made. The suspensions with PEO 400k and PEO 900k had higher shear viscosities and were shear thinning. Suspensions with higher concentrations of PEO 400k or 900k could not be dissolved properly in the sol due to very high viscosities. Oscillatory measurements in shear were performed to obtain the viscoelastic properties and the results are shown in figure 4. A phase shift of 90 degrees between strain and stress means that the material is completely viscous. The lower the phase shifts the more elastic
260
Progress in Nanotechnology: Processing
is the suspension. Although the shear viscosity is moderate for the sample with 3wt% PEO 900k the elasticity is high at high frequencies indicating the importance of molecular weight for this property. Preliminary electrospinning experiments were performed in ambient environment without controlling humidity. The fibre formation process was erratic and fibres were deposited not only on the collector electrode but also on surrounding surfaces. The fibres were in fact “standing up” forming a 3-dimensional cotton like morphology in the applied electric field and fell back (more or less) on the collector when the voltage was turned off. Turning on the voltage again resulted in the “erection” of the fibres. It seems as the fibres were polarised in the field. The formation of 3-dimensional cotton like structure has so far not been recorded when pure polymer solutions are electrospun to fibres. However, Azad [8] obtained this type of behaviour when processing zirconia and ceria fibres in ambient atmosphere. Samples with 2wt% polymer in the alumina sol had a lot of broken fibres when studied in SEM, see figure 6. The sample was very brittle, which is believed to be due to the low polymer content (20 vol% in the green fibre). An addition of 3 wt% polymer resulted in more continuous fibres and no broken fibres were found in samples with 6wt % polymer added, see figure 7.
Figure 5. 5 wt% PEO 100k.Spherical particles from electrospraying, green sample.
Figure 6. 2wt% PEO 900k. Green broken Fig 7.6wt% PEO 400k green fibres. fibres due to low polymer content.
Nanotubes, Nanorods, and Nanowires
261
10000 lu
5
W/o
400k -3 900k -+3 400k t .6
U
a
2
1000
b 0
-
lu
0 .+
10
lu
cn
K 0 rl
1 1
10
100
1000
Extension rate (s-1)
Figure 8. Elongational viscosity vs. extension rate for PEO alumina sol suspensions.
Figure 9. Results from XRD measurements on calcined fibres of 6wt%PEO 400k pressed into a thin disc shape. The elongational viscosity and fibre diameters were measured on the systems shown in table 1. Figure 8 shows the results from the measurements of elongational viscosity with the results corrected for the shear viscosity that has an influence on the registered force during the measurements. Only one extension rate could be accurately measured for the system with 3 wt% PEO 400k due to the low shear viscosity and the limited shear thinning of that system. There is a much larger difference between the behaviour of the systems in extension than in shear. For a Newtonian fluid the elongational viscosity is three times (Trouton's ratio) the shear viscosity. In
262
Progress in Nanotechnology: Processing
this case, when the molecular weight of PEO is sufficiently high, the elongational viscosity is 50100 times larger than the shear viscosity and it can be assumed that the suspensions are strain hardening in elongation. Such behaviour will stabilise the filament and prevent capillary break UP.
Very little information about the actual elongational strain rates during electrospinning is available in the literature. Yu measured the jet thinning of aqueous PEO solutions from the Taylor cone up to the point where the whipping instability sets in. Based on the flow rate and the diameter progression along the spin line the strain rate can be measured in this region. Data from Yu indicate strain rates in order of magnitude around 10 s-'. Unfortunately, the strain rates in the whipping instability region are not known since it is difficult to assess the point along the fibre where stretching ceases. . The elongational rheometer used in this study cover a wide range of strain rates around 10 s'l so it seems reasonable to assume that it provides viscosity data in the relevant strain rate regime. Figure 9 shows that the main phase obtained when calcining to 700°C was y alumina. There is no amorphous phase in the sample but the broad peaks show that the grain size is very small, about 30 nm. The fibre diameters of calcined samples obtained from measurements on SEM micrographs are given in Table 2. A frequency diagram is shown in figure 10. All of the systems had uniform fibres without branches when the electrospinning parameters were set to give a single Taylor cone, which is required to get proper whipping of the jet. Comparing system C and D, having the same composition and electric field strength shows that an increased flow rate resulted in larger fibre diameter. Figure 11 shows a SEM micrograph of system D with 3 wt% PEO 900k.
Figure 10. Calcined fibres diameter frequency diagram
Nanotubes, Nanorods, and Nanowires
263
Figure 11. 3 wt% PEO 900k calcined fibres electrospun in Plexiglas box under flowing nitrogen. Table 2. Results from measurements of conductivity, fibre diameter and specific surface area. Sample Composition Conductivity Average Standard Confidence Specific surface (mS/cm) fibre deviation interval area (m2/g) (95%) diameter (nm) (nm) Alumina sol 2.2 3wt% lOOk 2.0 A 3wt%400k 2.0 42 1 f 11 166.9 112 B 6wt%400k 1.8 557 f 47 165.6 330 33.3 (green) f 8 C 3wt%900k 1.9 526 97 D 3wt%900k 1.9 710 f 16 158.5 160 The value of the conductivity was independent of the molecular weight of the PEO and seems to be influenced mainly by the concentration of the sol in the sample. The specific surface area of the green sample of 6wt% PE0400k was only 33.3 m2/g while the calcined samples had much higher values. The large specific surface area of the calcined samples is mainly due to the porous crystal structure of the y - alumina. The surface of the alumina is “hidden” by the polymer before calcination of the sample.
DISCUSSION The choice of a commercial alumina sol as a source for the ceramic phase of electrospun nanofibres resulted in a straightforward preparation of the spinning suspension. The choice of polymer, molecular weight and concentration is crucial for the formation of uniform fibres. The initial electrospinning in this study showed that suspensions with PEOlOOk did not form continuous fibres but resulted in spherical particles due to the Rayleigh instability. Higher molecular weight or concentration of the polymer resulted in uniform fibres. However, too low polymer content gave very brittle fibres and green fibres broke during handling. It can therefore be concluded that a minimum amount of polymer, in this study about 27 vol%, is necessary to obtain continuous and usable fibres.
264
Progress in Nanotechnology: Processing
The fixed initial spinning conditions often resulted in several simultaneous Taylor cones at the tip of the syringe needle. Spinning was also discontinuous in the sense that the cones appeared and disappeared in a stochastic fashion and drops were accumulated and released periodically at the needle tip. It was soon realised that in order to have one single active Taylor cone and a continuous spinning without dripping it was necessary to tune the spinning conditions for each suspension. To avoid multiple jets and dripping, as compared to the initial settings, it was necessary to decrease the flow rate from the syringe pump andor increase the field strength by increasing the voltage or decreasing the distance to the collector. The parameter combinations in Table 1 resulted in a single well defined Taylor cone and continuous spinning without dripping. There is a processing window for each suspension allowing a stable and well defined spinning. It was noted that the ambient humidity greatly affected the deposition of fibres. By using nitrogen blanket the relative humidity was brought close to zero and the fibres were then directed and deposited on the collector. Humidity may affect the surface resistivity of surrounding objects made from otherwise insulating materials like wood, paper and plastics via adsorption and it can be speculated that a low humidity will concentrate the electrical field lines from the charged syringe towards the grounded collector plate. However, the deposition was still 3-dimensional in the sense that a “cotton like” structure was built up from the collector surface, see figure 12.
Figure 12. Morphological feature of green alumina fibres electrospun under flowing nitrogen. This type of highly porous structure may be interesting for certain applications like high temperature filtration. This type of structure is not seen when pure polymer solutions are electrospun. In this case a 2-dimensional non-woven mat is formed on the collector electrode. This type of 2-dimensional structure was also recorded when alumina particles were dispersed in aqueous PEO solutions and electrospun. According to the theory [ 11 of predicting the green fibre diameter an increased flow rate and /or a decreased field will increase the fibre diameter. The bulk conductivity and the surface tension of the suspension will also have an influence on the fibre diameter. In the systems studied the conductivity is almost constant and the differences in surface tension will
Nanotubes, Nanorods, and Nanowires
265
also be very small. The fibre diameters as influenced by the electrospinning parameters were not studied in a systematic manner. However, some observations regarding the possible effect of elongational viscosity can still be made. Comparing system A (3wt% PEO 400k) and B (6wt% PEO 400k) with the same flow rate but B having twice the field strength one would expect B to show a smaller diameter than A if the elongational viscosity had no effect. In fact, B shows thicker fibres. The higher elongational viscosity due to the higher polymer concentration in B might explain this. A similar argument is found by comparing A and C (3wt%PEO 900k). C having higher field strength and significantly lower flow rate than A, both should work in the direction of thinner fibres, still C shows thicker fibres. It should be noted that the fibre diameters were measured on calcined samples obtained by sedimentation on a sample holder. This sample preparation is not possible to use for green fibres that would dissolve during the sedimentation. However, the shrinkage of the calcined fibres will be the same for all suspensions containing 3wt%PEO regardless of the molecular weight. For the system containing 6wt% the shrinkage will be larger; that is the green fibres were coarser than measured on the calcined samples. This further stresses the influence of the elongational viscosity as stated above. The rheological measurements showed that there is a much larger difference in the extensional behaviour between the systems than in the shear behaviour. Contraction flow measurements are suitable for highly shear thinning systems with a rather high viscosity. This method would have been more suitable if the polymer suspension contained polymer of higher molecular weights or higher concentrations. Extensional rheology measurements based on filament break up is probably more suitable and also possible to use on solutions with polymers of low molecular weight and concentration. CONCLUSIONS We have shown that a straightforward manufacturing of alumina nanofibers is possible by the electrospinning of a commercial alumina sol mixed with polyethylene oxide. The polymer content should be at least 27 vol% and then have a molecular weight of at least 400 000. The concentration range should be 3-6 wt% of polymer in the sol. The electrospinning should be performed in a dry environment facilitating better control of the fibre deposition. The manufactured fibres had diameters in the range 400-700 nm and further optimisation can reduce these values. REFERENCES Sigmund, W., et al., Processing and Structure Relationships in Electrospinning of 1. Ceramic Fiber Systems. J. Am. Ceram. SOC., 89(2), 395-407.(2006). Daga, V., Helgeson, M., and Wagner, N., Electrospinning of Neat and Laponite-Filled 2. Aqueous Poly(ethy1ene oxide) Solutions. Journal of Polymer Science: Part B: Polymer Physics, 44, 1608- 16 17.(2006). Shenoy, S.L., et al., Role of chain entanglements on fiber formation during 3. electrospinning of polymer solutions: good solvent, non-specific polymer-polymer interaction limit., Polymer, 46, 3372-3384.(2005). Yu, J.H., Fridrikh, S.V., and Rutledge, G.C., The role of elasticity in the formation of 4. electrospun fibers. Polymer, 47,4789-4797.(2006). Stading, M. and Bohlin, L. Contraction flow for characterization of extensional properties 5. of semi-solid materials. in XZVth Znternational Congress on Rheology. 2004. Seoul, Korea: The Korean Society of Rheology.
266
Progress in Nanotechnology: Processing
6. 7.
8.
Azad, A.-M., Fabrication of transparent alumina (A1203) nanofibers by electrospinning. Materials Science and Engineering: A, 435-436,468-473.(2006). Larsen, G., et al., A method for making Inorganic and Hybrid (Organic/Inorganic) Fibers and vesicles with Diameters in the Submicrometer and Micrometer range via Sol-gel Chemistry and Electrically Forced Liquid Jets. J. Am. Chem: Soc., 125( 11541155).(2003). Azad, A.-M., Matthews, T., and Swary, J., Processing and characterization of electrospun Y203-stabilized Zr02 (YSZ) and Gd203 -doped CeOa (GDC) nanofibers. Materials Science and Engineering: B, 123,252-258.(2005).
Nanotubes, Nanorods, and Nanowires
267
This Page Intentionally Left Blank
ZnO Nanofiber and Nanoparticle Synthesized Through Electrospinning and Their Photocatalytic Activity Under Visible Light Haiqing Liu,'3$,5Jinxia Yang,$ Jianhe Liang,$ Yingxing Huang,$ and Chunyi TangS kollege of Chemistry and Materials Science, Fujian Normal University, Fuzhou 350007, China $Key Laboratory of Polymer Materials of Fujian Province, Fuzhou 350007, China
In this paper, we fabricate ZnO nanofibers and nanoparticles through electrospinning precursor solution zinc acetate(ZnAc)/ cellulose acetate(CA) in mixed-solvent N,N-dimethylformamidelacetone. Depending on the posttreatment of precursor ZnAc/CA composite nanofibers, both ZnO nanofibers and nanoparticles were synthesized after calcination of precursor nanofibers. The morphology and crystal structure of the ZnO nanofiber and nanoparticle were characterized by scanning electron microscopy, transmission electron microscopy, atomic force microscopy, and X-ray diffraction. It was found that the mean diameter of the ZnO nanofiber and nanoparticle was ca. 78 and 30 nm, respectively. The photodegradation of dye molecules such as Rhodamine B and acid fuchsin catalyzed by the ZnO nanofiber and nanoparticle was evaluated under the irradiation of visible light. Both morphological ZnO species showed strong photocatalytic activity. However, the ZnO nanofiber in the form of nanofibrous mats showed much higher efficiency than the nanoparticle although the latter has a smaller size than the former. The porous structure of ZnO nanofibrous mats is believed to improve the contacting surface areas between the catalyst and the dye molecules, while the aggregation of ZnO nanoparticle in the solution lowers the photocatalytic efficiency. I. Introduction
T
HE ZnO is an important
semiconductor material with a band gap of 3.37 eV and an excitation binding energy of 60 meV. Owing to its nontoxicity and strong oxidizing power to organic pollutants under visible light irradiation, ZnO has been applied as a photocatalyst in environmental cleaning.' However, it is well known that the photocatalytic activity of ZnO strongly correlates with the size of the ZnO specimen. That is, the larger the size of ZnO the higher the recombination rate of photoinduced electron-hole pairs at or near its surface, resulting in lower photocatalytic activity or vice versa. Owing to quantum confinement of nanostructured material, the separation rate of photo-induced charge carriers in semiconductor photocatalysts can be significantly increased by reducing its size to the nanometer range, and therefore enhance its quantum efficiency and photocatalytic activity. ZnO in the form of a nanoparticle exhibits strong photocatalytic oxidation to organic pollutants; but it is realized that the nanoparticle photocatalyst may repollute the treated water due to difficult recovery. Moreover, the nanoparticle aggre ation during aging may lower the photocatalytic efficiency. To overcome the drawbacks of ZnO nanoparticles, ZnO nanowires and nanoplatelets supported on an aluminum film have been synthesized. It is reported that they
B
exhibit a high surface-to-volume ratio and are very stable against aggregation, and they show higher efficiency than conventional nanoparticle in degrading organic pollutants."6 However, the yields of a nano-ZnO film from custom methods are relatively low and cannot meet the needs of photocatalytic application on a large scale. Therefore, new strategies have to be developed to prepare large quantities of ZnO films with high surface areas. It is well documented that a fibrous membrane composed of nanofibers exhibits large surface areas and porous which makes it a very promising morphology for photocatalyst. Among many methods such as hydrothermal and vapor- hase transport used for the preparation of ZnO nanofibers,bl1° electrospinning is the only method able to yield abundant quantities of continuous nanofibers for practical applications. The electrospinning technique has been applied to prepare a one-dimensional nanofiber with diameter down to tens of nanometers and length up to several centimeters, and the nanofibers are often collected as nanofibrous mats.' Although e-spinning has been used mainly for the fabrication of organic polymeric nanofibers, ceramic metal oxide nanofibers have also received considerable interest because of their potential applications in areas including photovoltaic devices, piezoelectric material, gas sensors, and solar cells.12 Metal oxide nanofibers are usually prepared by electrospinning a precursor metal salt solution with the help of a proper polymer, followed by calcination to decompose the polymer completely and turn metal salt into metal ~ x i d e . ' ~So . ' ~far, the only precursor system reported for the e-spun ZnO nanofiber is zinc acetate (ZnAc)/poly(vinylalcohol) in water. The diameter of a ZnO nanofiber as obtained is over 100 nm." Up to now, the photocatalytic activity of ZnO nanofiber has not been examined experimentally. In this work, we demonstrate the preparation of a ZnO nanofiber from a novel spinning solution system, i.e., N,N-dimethylformamide (DMF)/ acetone as a solvent, ZnAc as a precursor, and cellulose acetate (CA) as a fiber template. From this solution system, ZnO nanoparticles smaller than 40 nm in diameter are obtained by direct calcination of ZnAc/CA composite nanofibers, whereas ZnO nanofibers with diameter down to 23 nm are synthesized by calcination of Zn(OH)2/cellulosecomposite nanofibers, which is obtained from the hydrolysis of ZnAc/CA composite nanofibers in base solution. Additionally, for the first time, the photocatalytic activity of ZnO nanofibers is evaluated and compared with that of ZnO nanoparticles by measuring the photo-degradation of some dye molecules under visible light irradiation.
11. Experimental Procedure
J. Ninwontrihuting editor
(I)
Manuscript No. 23749. Received September 15,2M)7; approved December 9, 2007. The authors are grateful for the financial support by the Initiative Fund for the Returned Overseas Chinese Scholar administered by the State Education Ministry, and the Key Project of Natural Science Foundation of Fujian Province (Grant No. E0620001). 'Author to whom correspondence should be addressed. e-mail: haiqing.liu@gmdi~.com
Nanotubes, Nanorods, and Nanowires
Materials Zinc acetate dihydrate (Sinopharm Chemical Reagents Co., Shanghai, China) was recrystallized in distilled water. CA with a degree of substitution (DS) of 2.45 and an M , of 3.0 x lo4 was Obtained from Eastman. All Other chemicals were as received. 269
(2) Preparation of ZnO Nanofiber Transparent spinning solutions containing 6 or 10 wt% ZnAc were prepared by adding ZnAc to 20 wt% CA in a 2.1 (v/v) DMF/acetone solvent mixture, followed by magnetic stirring at ambient temperature. They were placed in a syringe with an 18-gauge stainless spinneret. The feeding rate was 5 pL/min, monitored by a syringe pump (TS2-60, Longer Precision Pump Co. Ltd., Baoding, China). An electrode was clamped on the spinneret and connected to a power supply (DW-P303-IAC, Tianjin Dongwen High Voltage Plant, Tianjin, China). A grounded counter electrode was connected to a collector aluminum foil, which was placed 10 cm away from the orifice. The electric field was maintained at 10 kV. The ZnAc/CA composite nanofibrous membranes on the grounded collector were dried in a vacuum oven at 90°C for 5 h. One part of the ZnAc/CA composite nanofibrous membrane was calcinated at 500°C in air for 5 h. The other part of the membrane was hydrolyzed in 0.1 N NaOH aqueous solution for 24 h at ambient temperature to transform the ZnAc/CA composite nanofibers into Zn(OH)2/cellulose nanofibers, and then washed with distilled water completely. The hydrolyzed membrane was vacuum dried at 50°C for 5 h. Subsequently, it was calcinated in the same way as the original ZnAc/CA composite nanofibrous membrane.
(3) Characterization The viscosity of the spinning solution was measured by a rotational viscometer (NDJ- 1, Shanghai Precision & Scientific Instrument Co. Ltd., Shanghai, China) at 30°C. Fourier transform infrared (IT-IR) spectra were collected on a Nicolet Avatar 360 (Nicolet, Pittsfield, MA) spectrometer in KBr form. The structure of the ZnO nanofiber was characterized by an X-ray diffractometer (Xpert MPD Pro, Philips, the Netherlands) with CuKu radiation. The diameter and morphology of the ZnO nanofiber and the nanoparticle were observed by scanning electron microscopy (SEM) (JSM-6300LV, JEOL), transmission electron microscopy (TEM) (JEM2010, JEOL), and atomic force microscopy (AFM) (NanoScope Ma).
( 4 ) Photocatalytic Activity of ZnO Nanofiber A lab-made photochemical reactor was set up according to reference.I6 A 500 W tungsten lamp was used as the visible light source. A light filter was placed between the lamp and the dye solution to only allow visible light (h>420 nm) to pass through. Five milligrams of ZnO nanofiber or nanoparticle was added to a cylindrical glass vessel containing 10 g of 45 ppm acid fuchsin, Amido black 10B, or 6 ppm Rhodamine B (RhB) aqueous solution, respectively. At a certain time interval, the degradation reaction was stopped and the solution was centrifuged. The dye concentration in the supernatant was measured on a UV-vis spectrometer (Lambda 850, PerkinElmer, Waltham, MA). Parallel degradation reactions under the same conditions were conducted for varied time intervals. Only one measurement was made for each batch. Two control experiments were also conducted to test: ( I ) the adsorption of dye on the ZnO nanofiber in darkness and (2) photodegradation of the dye when exposed to visible light without the presence of catalyst ZnO.
111.
Results and Discussion
( 1 ) Morphology and Structure of ZnO Nanofiber and Nanopwticle Figure 1 shows the SEM of ZnAc/CA composite nanofibers. Composite nanofibers generated from 6 wt% ZnAc solution are generally smooth and round, with fiber diameter in a narrow range of 115-185 nm (Fig. I(a)), while that from 10 wt% ZnAc solution are mixtures of beads and nanofibers with a large size variation (Fig. I(b)). The large nanofibers are 27&380 nm in diameter with a regular round morphology; whereas the size of the whisker-like nanofibers is 99.9%, Toyo Aluminum KK, Tokyo, Japan), AIN diluent powder (type H, >99.9%, Tokuyama K.K., Hino, Tokyo, Japan), NH4C1 ( > 99%, Nacalai Tesque, Inc., Kyoto, Japan), and Y203(99.9%, Shin-Etsu Chemical Co., Tokyo, Japan) were used as the starting materials. Figure 1 shows the morphological characteristics of Al and AIN powders; the average particle size is 23 and 0.5 pm, respectively. The reaction charge was composed of Al, AIN (AI/AlN=40/60 mol%), and promoting additives of NH4CI and YzO3 (5 wt% each). The reaction powders were mixed using mortar and pestle for 10 min and then sieved through a 212-pm sieve to disperse any large agglomerates. Fifty grams of the powder mixture was poured into a porous graphite container ($42 mm x 90 mm H) and packed by tapping to a relative density -60% of the theoretical density. The density was estimated by measuring the weight of the powder mixture in a fixed volume of the container. The graphite container was then placed in the combustion chamber. Two W-Re thermocouples (connected to a data acquisition system) were inserted into the center of the charge (one at the middle and another near the top surface) at a fixed distance of 30 mm and used to record the temperaturetime pattern of the combustion and determine the combustion speed by measuring the time lapsed for the wave passage between the two thermocouples. The chamber was evacuated and subsequently filled with high-purity N2 gas (99.999%) up to 0.25 MPa pressure. The combustion was initiated from the bottom by igniting a 2-g ignition pellet (AI+AIN, 1:l) placed at the bottom of the packed powder by passing an electric current (60 A, 20 V) for 10 s through a carbon ribbon under the pellet. The combustion reaction was completed in about 5 min and the chamber was then cooled to room temperature in -30 min. The reaction product was visually observed. The product phases were identified by Xray powder diffraction (XRD; JEOL, JDX-3530, Tokyo, Japan) using CuKa radiation. The morphology of as-synthesized powder was observed by field emission scanning electron microscopy (FE-SEM; ERA-8800, ELIONIX, Tokyo, Japan). Samples for SEM observation were coated with thin films of sputtered gold to reduce electrical charge-up. 111. Results and Discussion
The content of diluent (60 mol%) in the reaction mixture was chosen according to a previous study for the relationship between Al molar ratio and nitrogen pressure on the yield and properties of AIN p r o d u ~ t . ~It~ is , ~used ' to reduce the reaction temperature and prevent coagulation of melted aluminum
particles. The as-synthesized AIN cake was very fragile with only a white color. The microstructure of as-synthesized AIN powder was observed by FE-SEM using representative samples from three different locations in the product cake: top surface, side surface, and middle center. The grain morphologies of as-synthesized AIN particles are given in Fig. 2. The microstructure consists of two major types: aggregates of irregular particles (-0.5 pm, same as original AIN diluent) and ball-like grains (same size and shape as original Al particles) consist of thin crust ( 5150 nm) covers unique quasi-aligned AIN nanofibers grown in the interior. Their cross-sectional view is similar to oval disk of the sea anemone. Figure 3 shows the XRD pattern of the as-synthesized product. The diffraction lines are assigned to a hexagonal AIN structure similar to the bulk AIN powder reported (JCPDS-file 2 5 1 133). Residual metallic Al has not been detected. From the SEM observations, one can conclude that: ( I ) The AIN diluent did not participate in the formation of AIN particles and played only a passive role in controlling the combustion temperature and dispersing the Al particles. (2) There was no grain growth or sintering for both the formed and original AIN diluent particles. (3) The quasi-aligned AIN nanofibers were formed inside the reacting Al particles. As far as we know, this mode of growth, encapsulation in reacting particles, has never been reported before. The question is how were these balls of quasi-aligned AIN nanofibers formed? The formation of a shell-core system during the course of nitridation of Al metal is known in the direct nitridation (DN) method and reported as a "coreshell" mode1.29-34In this method, the nitridation takes place via three steps: nitridation at the surface of the particles with the formation of a crystalline nitride shell, breakaway or flowout of molten or vaporized Al core, and volume nitridation outside the shell with a remaining hole or an empty core. The final morphology of the AIN product is honeycomb like (a clear SEM photograph of this morphology can be seen in our recent paper by Radwan and Bahgat34).In the combustion synthesis method, both the surface nitridation and breakaway were also observed when moderate combustions were promoted by using small amounts of additives (low),,"I such as C and NH C13s38 and/or ignition under an appropriate nitrogen pressure.3LL The nitride skins were formed at the preheating stage of the combustion and then molten Al flowed out with the formation of an eggshell-type AIN morphology. Neither of the previous observations can account for the present SEM observations, which requires a new growth model without a breakaway. The typical temperaturetime history of the nitridation reaction (Fig. 4) sheds light on the behavior of the combustion. The temperature was measured in the middle center of the charge. The pattern shows a mild combustion with no explosive mode and has a relatively low rate of temperature increase. It took -53s to reach 600°C (below melting of Al), -73s to reach
Fig. 1. Scanning electron microscopy micrographs of starting powders: (a) A1 and (h) aluminum nitride diluent.
296
Progress in Nanotechnology: Processing
Fig. 2. Field emission scanning electron microscopy images of the as-synthesized aluminum nitride product.
-
lOWC, 160 s to T', (1620"C), and lasted -27 s in the afterburning stage. We noticed that in the temperature-time histories of other combustion experiments without NH4C1 and Y203 additions, once the combustion wave begins, the temperature increases rapidly close to its maximum value. Owing to the
low combustion temperature, the grain growth and sintering of
AIN particles were avoided. The speed of the combustion reaction was determined by measuring the time lapsed for the wave passage between two thermocouples inserted into the center of
A1N
20
30
40
50
20 (7
60
70
80
Fig. 3. X-ray diffraction pattern of the as-synthesized aluminum nitride
product. Nanotubes, Nanorods, and Nanowires
Fig. 4. Temperaturetimevariation at the combustion front.
297
Fig.5. Schematic illustration of the growth model of quasi-aligned aluminum nitride nanofibers by the combustion synthesis method
the reaction bed at a (vertical) distance of 30 mm. The combustion had a slow speed (0.26 mmjs). In Fig. 5, we propose a new growth model to explain the formation of quasi-aligned AIN nanofibers by the CS based on a “micro-reactors’’ model. The formation of nanofibers occurred through two stages.
( I ) Formation of Micro-Reactors This stage started in the preliminary stage of combustion and occurred outside the Al particles through two steps. In the first step (below 600”C),NH4CI dissociated into HCI and NH3 vapors (NH4C1 sublimes at 350°C and dissociates at 520°C): NH~CI(S, V)
4
HCl(g)
+ NH,(g)
(2)
The sublimation and decomposition of NH4CI are endothermic and produce several gaseous species. These reactions absorb sufficient heat and disturb the direct nitridation of Al particles with N2 gas, which retards the wave propagation. It also provides enough time for another endothermic reaction between yttria and the surface alumina layer (at T 2 melting point of Al) with the creation of a thin crust on the surface of the Al particles:
+
+
YzO~(S) A1203(S) N2(g) - O ( s )stable thincrust
+
A1 - Y - N
(3)
The energy dispersive X-ray spectroscopy microanalysis showed the presence of yttrium in the composition of the thin crust (Fig. 6). The formation of these new crusts also suppresses the AI-N interaction and slows down the combustion propagation. Because of the low heat evolution and the presence of new protective crusts, there was neither breakaway nor explosion of molten cores.
Ammonium chloride plays a critical role because it produces hydrogen chloride, which can be considered to be a key intermediate product, acting as a “catalyst.” HCI(v) promotes the vaporization of molten aluminum cores into volatile aluminum chloride species and facilitates progress of nitridation through sequence of spontaneous chlorination and nitridation intermediate reactions. The semi-molten AI-Y-N-O crusts seemed to function as catalyzed self-substrates and provided active sites that promoted homogenous nucleation of AIN embryos on the inner surface of the crusts from the vapor phase by a VLS mechanism. The AIN nanofibers might condense from the vapor phase, after a critical (low) degree of supersaturation inside the “micro-reactors’’ is attained, and grow on the preceding embryos by a VS growth mechanism in an epitaxial way according to the classical crystal growth theory.43 No droplets could be observed at the tips of these nanofibers. This results in a unique oriented growth in the interior of the reacting particles normal to the inner crust surface. The first stage of the combustion, formation of micro-reactors, was the essential step for the formation of AIN nanofibers inside reactant particles. The postulation of nucleation of AIN by VLS in molten droplets or layer with a subsequent fiber growth through VS mechanism is in close agreement with the observations of Moya and colleagues.2494345 To the best of our knowledge, this mode of growth inside reactant Al particles has not been observed before and it was not expected in combustion reactions due to the high increase
( 2 ) Nitridation and Growth The crust-core systems function as unique “micro-reactors.’’ The various gaseous species present (HCl(g), NH3(g), N2(g)) diffuse through the crust (through pores or cracks) into the molten AIN cores. Both the nitridation and growth steps then occur inside the developed micro-reactors. The nitridation proceeds via spontaneous chlorination-nitridation sequences similar to that reported in our previous results of direct nitridation of an A1/NH4C1 mixture.42 Gaseous hydrogen chloride is very active and spontaneously reacts with molten Al cores to produce AICI, vapors, which are nitrided by a gas-gas reaction as follows: AI(I) 298
+ 3HCl(g)
+
AICI,(V)+#H2(g)
(4)
Fig.6. Energy dispersive X-ray spectroscopy pattern of the thin nitride crust. Progress in Nanotechnology: Processing
in nitridation temperature and fast speed of combustion. The current combustion condition was successful creating new micro-reactors during the combustion reaction, which promoted this mode of oriented growth inside.
IV. Conclusion Quasi-aligned AIN nanofibers were formed by the combustion synthesis according to a new micro-reactor model. This route might be a possible method for economical growth of AIN nanofibers, which are obtainable only by complicated nitridation reactions at elevated temperatures. AIN with this unique morphology can be used not only for AIN ceramics and composites but also in nanotechnology applications.
References ‘L. M. Sheppard, “Aluminum Nitride: A Versatile but Challenging Material,” Ceram. Bull., 69[11] 1801-12 (1990). *J. H. Harris, “Sintered Aluminum Nitride Ceramics for High-Power Electronic Applications,” JOM, 50 [6] 56-60 (1998). ’B. H. Mussler, “Advanced Materials and Powders-Aluminium Nitride (AIN),” Am. Ceram. Soc. Bull., 79 [6] 45-7 (2000). 4J. A. Haber, P. C. Gibbons, and W. E. Buhro, “Morphological Control of Nanocrystalline Aluminum Nitride: Aluminium Chloride-Assisted Nanowhisker Growth,” J. Am. Chem. Soc., 119, 5455-6 (1997). 5Y. Zhang, J. Liu, R. He. Q . Zhang. X. Zhang. and J. Zhu, “Synthesis of Aluminum Nitride Nanowires from Carbon Nanotubes,” Chem. Maler., 13,389p905 (2001). 6Q.Wu, Z. Hu. X. Wang, Y. Lu. K. Huo, S. Deng, N. Xu, B. Shen, R. Zhang, and Y. Chen. “Extended Vapor-Liquid-Solid Growth and Field Emission P r o p erties of Aluminium Nitride Nanowires,” J. Marer. Chem., 13, 2024-7 (2003). ’M. Radwan and M. Bahgat. “Novel Growth of Aluminium Nitride Nanowires,” J. Nanosci. Nanorechnol., 6 [2] 55861 (2006). ‘4. Wu, Z. Hu, X. Wang, Y. Hu, Y. Tian, and Y. Chen, “A Simple Route to Aligned AIN Nanowires,” Diamond Relat. Muter., 13, 3 8 4 1 (2004). 9H. Chen, Y. Cao, and X. Xiang, “Formation of AIN Nano-Fibers,” J. Cryst. Growrh, 224, 187-9 (2001). “Q. Wu, Z. Hu, X. Wang, Y. Chin, and Y. Lu, “Synthesis and Optical Characterization of Aluminum Nitride Nanobelts,” J. Phys. Chem. B, 107 [36] 97269 (2003). I’M. Yoshioka, N. Takahashi, and T. Nakamura, “Growth of the AIN NanoPillar Crystal Films by Means of a Halide Chemical Vapor Deposition under Atmospheric Pressure,’’ Marer. Chem. Phys., 86, 74-7 (2004). I2T. Xie, Y. Lin, G. Wu, X. Yuan, Z. Jiang. C. Ye, G. Meng, and L. Zhang. “AIN Serrated Nanoribbons Synthesized by Chloride Assisted Vapor-Solid Route,” Inorg. Chem. Commun., 7, 545-7 (2004). ”Q. Wu, Z. Hu, X. Wang, Y. Lu, X. Chen, H. Xu, and Y. Chen, “Synthesis and Characterization of Faceted Hexagonal Aluminum Nitride Nanotubes,” J. Am. Chem. Soc., 125, 101767 (2003). I4L.-W. Yin, Y. Bando, Y.-C. Zhu. D. Golberg. and M.-S. Li, “A Two-Stage Route to Coaxial Cubic-Aluminum-Nitride-Boron-Nitride Composite Nanotubes,” Adv. Mater., 16 [ I I] 929-33 (2004). ”V. N. Tondare. C. Bahsubramanian, S. V. Shende, D. S. Joag, V. P. Godbole, S. V. Bhoraskar, and M. Bhadbhade, “Field Emission from Open Ended Aluminum Nitride Nanotubes,” Appl. Phys. Lett., 80 [25] 4813-5 (2002). I6C. Liu, Z. Hu, Q . Wu, X. Wang, Y. Chen, H. Sang, J. Zhu, S. Deng, and N. Xu, “Vapor-Solid Growth and Characterization of Aluminum Nitride Nanocones,” J. Am. Chem. Soc., 127, 1318-22 (2005). ”C. Liu, Z. Hu, Q . Wu, X. Wang, Y. Chen, W. Lin, H. Sang, S. Deng, and N. Xu, “Synthesis and Field Emission Properties of Aluminum Nitride Nanocones,” Appl. Sur. Sci., 251, 220-4 (2005). ‘*S.-C. Shi, C.-F. Chen, S. Chattopadhyay, Z.-H. Lan, K.-H. Chen, and L.-C. Chen. “Growth of Single-Crystalline Wurtzite Aluminum Nitride Nanotips with a Self-Selective Apex Angle,” Adv. Funcr. Mater., 15 [5] 7 8 1 4 (2005).
Nanotubes, Nanorods, and Nanowires
I9S.-C. Shi. S. Chattopadhyay, C.-F. Chen. K.-H. Chen, and L.-C. Chen. “Structural Evolution of AIN Nano-Structures: Nanotips and Nanorods.” Chem. Phy.7. Lett., 418. 152-7 (2006). ”A. G. Merzhanov, “History and Recent Developments in SHS.” Ceram. Inr., 21, 371-9 (1995). *‘K. Tanihata and Y. Miyamoto, “Reaction Analysis on the Combustion Synthesis of Aluminum Nitride,” Inr. J. SHS, 7 121 209-17 (1998). ”V. V. Zakorzhevskii and I. P. Borovinskaya, “Regularities of Self-Propagating High-Temperature Synthesis of AIN at Low Nitrogen Pressures,” Inr. J. SHS., 7 [2] 199-208 (1998). 21 V. V. Zakorzhevskii, I. P. Borovinskaya, and N. V. Sachkova. “Combustion Synthesis of Aluminum Nitride.” Inorg. Mater., 38 [ I I] 1 1 3 1 4 (2002). 24J.S. Moya, J. E. Iglesias, J. Limpo, J. A. Escrioa. N. S. Makhonin. and M. A. Rodriguez, “Single Crystal AIN Fibers Obtained by Self-propagating HighTemperature Synthesis (SHS),” Acta Mater., 45 [8] 3089-94 (1997). 2SG.Jiang, H. Zhuang, J. Zhang, M. Ruan, W. Li, F. Wu, and 8 . Zhang. “Morphologies and Growth Mechanisms of Aluminum Nitride Whiskers by SHS Method-Part I.” J. Muter. Sci., 35, 57-62 (2000). 26 G. Jiang, H. Zhuang, J. Zhang, M. Ruan, W. Li, F. Wu, and B. Zhang, “Morphologies and Growth Mechanisms of Aluminum Nitride Whiskers by SHS Method-Part 2,” J. Mater. Sci., 35, 63-9 (2000). 27T. Sakurai, Y. Miyamoto, and 0. Yamada, “Combustion Synthesis of Fine and High-Purity AIN Powder and Its Reaction Control,” J. Soc. M a t . Sci. Jpn, 54 [6j 574-9 (2005) (in Japanese). 28T. Sakurai. 0. Yamada, and Y. Miyamoto. “Combustion Synthesis of Fine AIN Powder and Its Reaction Control,” Muter. Sci. Eng. A , 415. 40-4 (2006). 291. Kimura, K. lchiya, M. Ishii, N. Hotta, and T. Kitamura, “Synthesis of Fine AIN Powder by a Floating Nitridation Technique using an N2/NHJ Gas Mixture,” J. Mater. Sci. Lett., 8. 3 0 3 4 (1989). ’OH. Scholz and P. Greil, “Nitridation Reactions of Molten AI-(Mg, Si) Alloys,” J. Marer. Sci., 26, 669-77 (1991). ”K. Komeya, N. Matsukaze, and T. Meguro. “Synthesis of AIN by Direct Nitridation of Al Alloys,” J. Ceram. Soc. Jpn, 101 [I21 1319-23 (1993). ’*A.-J. Chang, S.-W. Rhee, and S. Baik, “Kinetics and Mechanisms for Nitridation of Floating Aluminum Powder,” J. Am. Ceram. Soc.. 78 [I] 3 3 4 0 (1995). ”T. Fujii, K. Yoshida, K. Suzuki. and S. Ito. “Direct Nitriding of Large Grains of Aluminum with 2 mm Size,” SolidStare Ionics, 141-142, 593-8 (2001). 14M. Radwan and M. Bahgat, “A Modified Direct Nitridation Method for Formation of Nano-AIN Whiskers,” J. Mater. Process. Techno/., 181, 99-105 9 7 ) . G. J. Jiang, H. R. Zhuang, W. L. Li, F. Y. Wu, B. L. Zhang, and X. R. Fu, “Mechanisms of the Combustion Synthesis of Aluminum Nitride in High Pressure Nitrogen Atmosphere (2);’ J. Marer. Synth. Process., 7 [I] 1-6 (1999). ‘k.-N. Lin and S.-L. Chung, “Combustion Synthesis of Aluminum Nitride Powder using Additives,” J. Marer. Re.7.. 16 [XI 2200-8 (2001). 37R.-C.Juang, C:J. Lee, and C . C . Chen, “Combustion Synthesis of Hexagonal Aluminum Nitride Powders under Low Nitrogen Pressure,” Mater. Sci. Eng. A. 357, 219-27 (2003). 38 C.-N. Lin, C.-Y. Hsieh, S.-L. Chung, J. Cheng, and D. K. Agrawal, “Combustion Synthesis of AIN Powder and Its Sintering Properties,” I n / . J. SHS, 13 [2] 93-106 (2004). I9S. M. Bradshaw and J. L. Spicer, “Combustion Synthesis of Aluminum Nitride Particles and Whiskers,” J. Am. Ceram. Soc., 82 [9] 2293-300 (1999). *J. Shin, D.-H. Ahn, M.-S. Shin, and Y.-S. Kim, “Self-Propagating HighTemperature Synthesis of Aluminum Nitride under Lower Nitrogen Pressures,’’ J . Am. Ceram. Soc., 83 [5] 1021-8 (2000). 4’G.J. Jiang, H. R. Zhuang, W. L. Li, F. W. Wu. B. L. Zhang, and X. R. Fu, “Mechanisms of the Combustion Synthesis of Aluminum Nitride in High Pressure Nitrogen Atmosphere (2),” J. Mater. Synth. Process., 9 [I] 49-56 (2001). 42M. Radwan, M. Bahgat, and A. A. El-Geassy, “Formation of Aluminium Nitride Whiskers by Direct Nitridation,” J. Eur. Ceram. Soc., 26 [I31 248S8 (2Ow. 4’W. B. Campbell, “Growth of Whiskers by Vapor-Phase Reactions”; pp. IS46 in Whisker Technology, Chapter 2, Edited by A. P. Levitt. John Wiley & Sons Inc., New York, 1970. “P. G. Caceres and H. K. Schmid, “Morphology and Crystallography of Aluminum Nitride Whiskers,” J. Am. Ceram. Soc., 77 [4] 977-83 (1994). 4SR. Fu, H. Zhou, L. Chen, and Y. Wu, “Morphologies and Growth Mechanisms of Aluminum Nitride Whiskers Synthesized by Carbothermal Reduction.” Muter. Sci. Eng. A , 266,4451 (1999). 0
299
This Page Intentionally Left Blank
Synthesis and Optical Properties of Mullite Nanowires Han-Kyu Seong, Ungkil Kim, Myoung-Ha Kim, and Heon-Jin Choit Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, Korea
Youngho Lee and Won-Seon Seo Reliability Assessment and Materials Evaluation Center, Korea Institute of Ceramic Engineering and Technology, Seoul 153-801, Korea
-
Single-crystalline mullite (3AI2O3 2Si02) nanowires have been synthesized on silicon substrates by forcing aluminum and chromium chloride powders to react under an ammonia gas flow. The diameter and length of the nanowires have uniform diameters of < 100 nm and several micrometers, respectively. High-resolution transmissionelectron microscope and selected-area electron diffraction analyses indicated that the nanowires were almost structural defect free and had a single-crystalline phase with a (0001 ) growth direction. The photoluminescence spectra showed that the mullite nanowires reached an emission peak at the center wavelength of 442 nm originating from the AI-0 bonds in preference to the A 1 4 and Si-0 bonds in the mullite bonding structure. Comparison of the photoluminescence between as-synthesized, oxygen-annealed, and plasma-etched nanowires indicates that the nanowires have few defects (i.e., oxygen vacancies).
1.
Introduction
D
shape, and size have been known to play a significant role in determining the properties of In this regard, one-dimensional nanostructures such as nanotubes and nanowires have been the focus of considerable research into their use to improve the physical and mechanical proper tie^.^^ Recently, carbon nanotubes (CNTs) have been reported as potential elements for reinforcement in ceramic matrix composites (CMCs) owing to their unique mechanical properties, such as excellent tensile strength (>200 GPa) and Young's modulus (1 TPa).6 Meanwhile, mullite whiskers have been investigated as reinforcements for high-temperature CMCS.~-" However, the whiskers are limited by whisker damages and matrix cracking during the course of processing and densification of composites. Such problems may be solved by using mullite nanowires that could have excellent mechanical properties due to a perfect single-crystalline nature free from defects. Their nano-size could be also useful for processing and densification of composites without reinforcement damage and matrix cracking. In this study, the synthesis and optical properties of single-crystalline mullite (3A1203 . 2Si02) nanowires using the chloride vapor transport (CVT) process are reported. IMENSIONALITY,
11. Experimental Procedure
The growth of the single-crystalline mullite nanowires was performed in a horizontal hot-walled CVT system (Fig. I). Si/Si02 substrates oxidized by the wet oxidation method were used. The solid Al (purity 99.5%) and CrCI2 (purity 99.9%) powders placed in quartz susceptors, respectively, were inserted into the center of a quartz tube at 2 in. intervals. The oxidized substrates, deposited with a 0.2-nm layer of Ni by sputtering, were placed in the quartz tube at a distance of 2 in. from the Al powders. The temperature of the furnace was increased at a heating rate of 5O"C/min from room temperature to the reaction temperature of 1000°C under a flow of ammonia gas at a rate of 20 cm3/min, maintained for 10 min under a constant flow of ammonia, and then cooled down to room temperature. The quartz tube was then degassed and purged with argon gas. The morphologies and crystal structures of the nanowires grown on the substrates were characterized using scanning electron microscopy (SEM) and X-ray powder diffraction. Further structure and stoichiometry analyses of the nanowires were performed using high-resolution transmission electron microscopy (HRTEM), selected area electron diffraction (SAED), and energy-dispersive X-ray spectroscopy (EDS). The photoluminescence (PL) spectra were measured at room temperature with an excitation wavelength of 325 nm (He-Cd C W laser). 111. Results and Discussion
Figure 2 shows the SEM images of the typical mullite nanowires synthesized on the substrates using a CVT process. The nanowires with a high aspect ratio were distributed over the entire area of the substrates (Fig. 2(a)) and had uniform diameters of < 100 nm and lengths of several micrometers (Fig. 2(b)). The overall crystalline nature of these nanowires was confirmed as a high-quality mullite structure using an X-ray diffraction pattern. The low-magnification TEM image in Fig. 3(a) shows that most nanowires have uniform diameters of < 100 nm and lengths of several micrometers. Although Ni was used as the catalyst and it
X. Miao--contributing editor
Manuscript No. 22766. Received February 3, 2007; approved February 28, 2007. This research supported in part by a grant from the Korea Research Foundation (MOEHRD, KRF-2005442- wO203) and the Second Stage of Brain Korea 21 Project in 2006. H. J. C. thanks the RAME Center, KlCET for the use of their facilities. 'To whom correspondence should be addressed. e-mail:
[email protected] Nanotubes, Nanorods, and Nanowires
Fig. 1. Schematic illustration showing the setup for the growth of mullite nanowires.
301
Fig.2. Scanning electron microscopy images of mullite nanowires grown on substrates by annealing at 1000°C for 10 min. (a) Low and (b) high magnification.
Fig. 3. (a) A low-magnification transmission electron spectroscopy (TEM) image of mullite nanowires. (b) Energy-dispersiveX-ray spectroscopy (EDS) spectrum of a selected individual nanowire in the image (a). (c) High-resolution TEM image of mullite nanowire. The inset shows the selected area electron diffraction pattern of the wire, recorded along the [110] zone axis.
302
Progress in Nanotechnology: Processing
clearly contributed to the growth of the nanowires, no Ni globules or elements in the nanowires were identified in the SEM observations and the EDS analysis. The failure to observe any Ni globules or elements suggests that the catalysts were etched out by the chloride vapors during the growth of nanowires. As shown in Fig. 3(b), the EDS analysis of a selected individual nanowire shows that it is composed of Al (80.7 wtX), Si (19.3 wt%), and 0 elements, which correspond to the stoichiometric composition of 3A1203.2Si02. The HRTEM observation (Fig. 3(c)) also shows that the mullite nanowires are smooth and almost structural defect free. The SAED analysis, recorded along the [ 1101 zone axis, further demonstrated that the nanowires are a single-crystalline form of the orthorhombic Pbum structure with a (0001) growth direction (inset of Fig. 3(c)). The HRTEM lattice image shows that the fringe spacing of the mullite nanowire corresponding to the interplanar distance of (0001) planes is 2.95 A. This lattice expansion on the [OOOI] plane compared with the bulk mullite may be ascribed to the quantum confinement effects and/or stresses generated on the curved surfaces of the nanowires, which is expected to be significant as the size approaches the nanometer scale.” In the present process, the major vapor species in the CVT reactor are Sic&, AIC13, NH3, H20, and HCI. Therefore, the overall reactions for the mullite nanowire growth could be as follows:
+
+
2SiC14(v) 6AICIj(v) 13HzO(v) = 3A1203 .2SiO~(s)+ 26HCl(v)
(1)
The SiCI4,AICI,, and H 2 0 vapors can be formed through the sub-reactions as follows:
+
+
~NH~(v) 2Al(s) 9cd&(s) = 2AICl,(v) IZHCI(v) 9Cr(s) 1 +2N2(v)
+
+
+
+
3sio2(s) 6CrC12(s) 4NH3(v) = 3SiC14(v) 6H2O(v) 6Cr(s) 1 +2N2(v)
+
+
Photoluminescence spectra of the mullite nanowires as-synthesized, annealed a t 0 2 for 2 h, and plasma etched f r o m CF,+O2 mixed gas at room temperature, respectively.
vacancies, the band gap of the silica becomes narrow dramatically.13 Similarly, an increase of the oxygen vacancies in the nanowires may induce a narrow band gap. As the mullite nanowires synthesized in this study are single crystalline with few defects, it could be used as novel reinforcements for hightemperature CMCs.
IV. Conclusions
(2)
(3)
As the reactants in Eq. ( 3 ) , SiCI4 and H20 vapors as the silicon and oxygen sources originate from the amorphous silica thin film on the oxidized substrate formed during wet oxidation. The CrC12 powder remains in the Cr powders on the quartz susceptor of the CVT reactor after the reaction and thus acts as a chloride source for the overall CVT process. Figure 4 shows a typical PL spectrum from the mullite nanowires in the as-synthesizedstate, annealed at 600°C in 0 2 for 2 h, and plasma etched by a CF4+7.8% O2 mixing gas at room temperature. The excitation wavelength was 325 nm at room temperature. All spectra show a strong PL peak at the center wavelength of 442 nm (2.8 eV in photon energy) that can be attributed to the quantum confinement effects and/or the AIL@ Si bonds of the mullite bonding structure. The intensity of the PL peak of annealed nanowires is similar to that of the as-synthesized one; however, it considerably increases for the plasmaetched nanowires. This result implies that the radiative centers leading to the measured PL are generated from the AIL0 bonds in preference to the AIM-Si bonds in the mullite bonding structure. Peng et a/.” observed that AI2O3 nanowires and nanobelts have various emission intensities in different thermal annealing atmospheres such as O2 and H2 owing to the optical transitions in oxygen-related defects, F+ (oxygen vacancy with one electron) center. In the mullite nanowires, however, the PL peak intensity and the position for the annealed nanowires were barely changed. This may be due to the fact that the mullite nanowires synthesized in this study had a perfect single-crystalline nature with few defects (i.e., oxygen vacancies). On the other hand, the increase of the PL intensity as well as red shifting of the peak position in the plasma-etched nanowires can be principally attributed to the increase of the oxygen vacancy density, resulting from the destruction of the Si-0 bonds in the nanowires. It has been reported that with the increase of the oxygen Nanotubes, Nanorods, and Nanowires
Fig. 4.
The mullite nanowires with high aspect ratios were synthesized on the substrates using CVT process. The nanowires had uniform diameters of < 100 nm and a length of a few micrometers. The TEM characterization and PL measurements from the nanowires with as-synthesized, annealed in oxygen, and plasma etched by a CF4+O2 mixing gas indicate the single-crystalline and defect-free nature of nanowires. The mullite nanowires could offer good opportunities for highly toughened ceramic, metal, and polymer composites. References ‘J. Ning, J. Zhang, Y. Pan, and J. Guo, “Fabrication and Mechanical Properties of Si02 Matrix Composites Reinforced by Carbon Nanotube,” Mazer. Sci. Eng., A357, 3 9 2 4 (2003). ’H. J. Choi. H. K. Seong, J. Chang, K. 1. Lee, Y. J. Park, J. J. Kim. S. K. Lee. R. He, T. Kuykendall. and P. Yang, “Single-Crystalline Diluted Magnetic Semiconductor GaN:Mn Nanowires,” A h . Muter., 17, 1 3 5 1 4 (2005). ’H. K. Seong, Y. Lee. J. Y. Kim, Y. K. Byeun, K. S. Hdn, J. G . Park. and H. J. Choi, “Single Crystalline AIGaN:Mn Nanotutes and Their Magnetism,’’ Adv. Mater., 18, 3019-23 (2006).
4A. Peigney, Ch. Laurent, E. Flahaut. and A. Rousset, “Carbon Nanotubes in Novel Ceramic Matrix Nanocomposites,” Ceram. IIII., 26, 6 7 7 4 3 (2000). ’M. B. Nardelli, B. 1. Yakohson, and J. Bernholc. “Brittle and Ductile Behavior in Carbon Nanotubes,” Phys. Rev. Leu., 81, 4 6 5 6 9 (1998). 6B. 1. Yakobson and P. Avouris, “Mechanical Properties of Carbon Nanotubes,” Appl. Phys., 80,287-327 (2001). ’T.I. Mah and K. S. Mazdiydsni, “Mechanical Properties of Mullite,” J. Am. Ceram. Sac., 66.699-703 (1983). ‘ S . Kanzaki, H. Tabata, T. Kumazawa. and S. Ohtd. “Sinteringand Mechanical Properties of Stoichiometric Mullite.” J. Am. Ceram. Suc., 68 [ I ] C - 6 c - 7 (1985). 9K. Okada and N. Otuska, “Synthesis of Mullite Whiskers and Their Application in Composites,” J. Am. Ceram. SOC.,74 [lo] 2412-8 (1991). ‘OH. J. Choi and J. G. Lee, “Synthesis of Mullite Whiskers.” J. Am. Ceram. Suc., 85 [2] 481-3 (2002).
“H. J. Choi, H. K. Seong, J. C. Lee, and Y. M. Sung, “Growth and Modulation of Silicon Carbide Nanowires,” J. Crystal Growth, 269, 472-8 (2004). ‘*X. S. Peng, L. D. Zhang, G. W. Meng, X. F. Wang, Y. W. Wang, C. 2.Wang, and G. S. Wu. “Photoluminescence and Infrared Properties of a-Al,O, Nanowires and Nanobelts,” J. Phys. Chem. B., 106. I 1163-7 (2002). ”H. R. Philipp, “Optical Properties of Non-Crystalline Si, SiO. SiO, and SiOz.” 0 J . Phys. Chem. SolidF, 32, 1 9 3 H 5 (1971).
303
This Page Intentionally Left Blank
(Nao.8Ko.2)o.5Bio.5Ti03 Nanowires: Low-Temperature Sol-GelHydrothermal Synthesis and Densification Yu-Dong Hou,+ Lei Hou, Ting-Ting Zhang, Man-Kang Zhu, Hao Wang, and Hui Yan Department of Materials Science and Engineering, Beijing University of Technology, Beijing 100022, China
The sol-gel-hydrothermal processing of ( N ~ O . ~ K ~ . ~ ) ~ . ~ Bthe ~ . number ~ T ~ Oof~possible spontaneous polarization directions for (NKBT) nanowires as well as their densification behavior were the compositions near the MPB due to the coexistence of rhoinvestigated. The morphology and structure analyses indicated mbohedral and tetragonal phases. Therefore, NBT-based solid solutions that have an MPB structure and can be poled easily that the sol-gel-hydrothermal route led to the formation of phase-pure perovskite NKBT nanowires with diameters of 50have been studied r e c e n t l ~ . ~Among -~ the NBT-based binary lead-free piezoelectric systems, (Nao,sI&2)o.5Bio,5Ti03(NKBT) 80 nm and lengths of 1.5-2 pm, and the processing temperature systems have shown good piezoelectric performance and have was as low as l W C , but the conventional sol-gel route tended been the subject of the most extensive investigations because the to lead to the formation of NKBT agglomerated porous struccomposition was close to the MPB and the ceramics can be tured nanopowders, and the processing temperature was higher than 650°C. It is believed that the gel precursor and hydrotherpoled easily in a relatively low dc field ( 4 5 kvjmm). NKBT powders are traditionally prepared by a solid-state reaction.64 mal environment play an important role in the formation of the nanowires at a low temperature. Owing to the better packing In case of solid-state reactions, the starting materials are oxides o r carbonates of Bi, Na, K, and Ti. The particle size of these efficiency and therefore a good sinterability of the freestanding nanowhiskers, the pressed pellets made by NKBT nanowires starting materials is in the micrometer or submicrometer range. showed >98% theoretical density at 1100°C for 2 h. The solA perovskite phase-forming temperature of 800°C and above is gel-hydrothermal-derived ceramics have typical characteristics needed so that the components of the mixture have sufficient thermal energy to overcome the atomic/ionic diffusion barriers of relaxor ferroelectrics, and the piezoelectric properties were for the reaction. In most cases, the resulting powders are not better than the ceramics prepared by the conventional sol-gel nanocrystalline and undergo drastic agglomeration and yield an and solid-state reaction. inhomogeneous particle size as a result of the high-temperature treatment; therefore, they are unsuitable for enhancing the dielectric and piezoelectric properties of ceramics for high-perforI. Introduction mance uses. In comparison with a conventional solid-state reaction, the sol-gel process has some advantages, including exT present, lead zirconate titanate (PZT)-based ceramics are cellent compositional control, and homogeneity at the molecular the most widely used in electronic devices due to their high level due to the mixing of liquid precursors. However, sol-gel piezoelectric performance. However, the pollutant of toxic lead was regarded as a solid rather than solution process because solduring the fabrication and waste of products cause a crucial gel-derived precipitates are amorphous in nature and calcinaenvironmental problem. Therefore, there is an increasing intertions in air are inevitable for the formation of the crystalline est in developing lead-free piezoelectric ceramics to replace material. The sol-gel-hydrothermal processing represents an alPZT-based ceramics and to minimize lead pollution. It is well ternative to the calcinations for the crystallization of an objecknown that covalency between unoccupied states of the Pb6d in tive compound under mild temperatures. As a novel method to the perovskite structure, and O p states favor ferroelectric prepare oxide powders, the sol-gel-hydrothermal technique has ground states.' Compared with Pb2+, Bi3+ ions were in an the double advantage of both sol-gel and hydrothermal syntheisoelectronic state and also showed a long pair effect, which ses and has become attractive in the last decade due to its high encouraged studies of Nao.sBio.5Ti03(NBT) as an alternative degree of crystallinity, well-controlled morphology, high purity, to PZT ceramics. and narrow particle size distribution of the prepared powSodium bismuth titanate, NBT, which was found by Smolenders.'.'' In our previous work, K,,5Bi,,5Ti03 (KBT) nanowires skii et al.: is a kind of perovskite-type relaxor ferroelectric with with good sinterability were successfully synthesized by the sola Curie temperature Tc= 320°C. At room temperature, it has a gel-hydrothermal technique, and the dielectric properties of the rhombohedra1 structure (a = 0.389 nm, rx = 89.6"), and shows a derived KBT ceramics were superior to that prepared by all the relatively large remanent polarization (Pr= 38 pC/cm2). Howother methods reported Thus, the direct generever, it is difficult to pole NBT due to the high coercive field ation of NKBT nanowires with good sinterability at a low tem(Ec = 7.3 kvjmm), making it difficult to obtain the desirable piperature is of considerable interest. ezoelectric properties. In addition, unlike PZT ceramics, NBT In the present work, the sol-gel-hydrothermal process was has no morphotropic phase boundary (MPB), which plays a very presented as a new route to produce NKBT nanowires at a important role in PZT ceramics. The electromechanical propertemperature below 200"C, which is comparatively lower than ties show a maximum over a compositional range around that synthesized by the normal sol-gel route, which requires a the MPB of PZT, which can be attributed to an increase in temperature at least 650°C. The nanowires were characterized by X-ray diffraction (XRD), FT-IR, Raman, and transmission electron microscope (TEM) analysis. The densification behdvJ. Ninwontrihuting editor ior, the final microstructure of the sintered material, and the electric properties of the densified disks were investigated in detail. To the best of our knowledge, this is the first time that the Manuscript No. 22459. Received November 8, 2006; approved January 29, 2007. synthesis of NKBT nanowires has been reported, and the elecThis work was supported by the National Natural Science Foundation of China (Grant tric properties of sol-gel-hydrothermal-derived ceramics were No. 60601020) and the Natural Science Foundation of Beijing (Grant No. 4072006). 'Author to whom correspondence should be addressed. e-mail ydhou@bjut,edu.cn superior to that prepared by other methods.
A
Nanotubes, Nanorods, and Nanowires
305
11. Experimental Procedure
The raw materials used in the present work were analyticalgrade bismuth nitrate pentahydrate (Bi(N03)2.5H20), sodium nitrate (NaN03), potassium nitrate (KNO,), tetrabutyl titanate (Ti(OC4H9)4)racetic acid (CH3COOH), and ethanol (CH,CH20H). Firstly, bismuth nitrate pentahydrate was dissolved in acetic acid; potassium nitrate and sodium nitrate were dissolved in C02-free distilled water, respectively. The mixture was then introduced into a prepared solution of a stoichiometric amount of tetrabutyl titanate in ethanol. After stirring vigorously for 2 h, a thin yellow homogeneous sol was formed. Then, the sol was heated at 80°C for 12 h to prepare a dry gel. The obtained gel precursor was added to a NaOH solution to form a suspension. The initial concentration of the NaOH solution was 10M. The as-prepared mixture was poured into a Teflon vessel, and then subjected to hydrothermal treatment at an appropriate temperature under auto-generated pressure for 48 h. After cooling, the product was filtered, washed with distilled water, and dried at ambient temperature. To compare the results with the conventional sol-gel process, a part of the gel was calcined at different temperatures from 300" to 800°C for 2 h in air. The crystal phase of the powders was determined using X-ray diffractometry (XRD; Model D8 Advance, Bruker AXS, Karlsruhe, Germany) in the 8-28 mode with graphitemonochromatized CuKu radiation (h = 0.154178 nm). The powder morphology was observed on a transmission electron microscope (TEM; Model JEM-2000 F, JEOL, Tokyo, Japan) and scanning electron microscope (SEM; Model S-3500N, Hitachi, Tokyo, Japan). Fourier transform infrared absorption spectra of the powders were obtained using an FT-IR apparatus (Model NEXUS670, Nicolet, Madison, WI). Raman scattering spectra of powders were recorded at room temperature from a Raman spectrometer (Model T64000, Jobin-Yvon, Paris, France) under backscattering geometry. Excitation was taken as the 488 nm line of an Art laser with a 50 mW output power. For the sintering experiments, the obtained powders were pressed into pellets with a diameter of 12 mm under an isostatic pressure of 150 MPa. Conventional sintering was performed at 50°C temperature intervals between 1000" and 1200°C for 2 h in a sealed alumina crucible. The bulk densities of the sintered pellets were measured by the Archimedes method. The microstructure of the sintered pellets was observed using SEM; (Model S-3500N, Hitachi, Tokyo, Japan) on the fracture side and free-top surfaces of the pellets. When the fracture surface did not reveal clear grain boundaries, the polished and thermally etched surface was observed. To measure the electrical properties, silver paste was coated on both sides of the sintered pellets and fired at 560°C for 30 min to form electrodes. The dielectric property and its dependence on temperature were measured using a precision LCR meter (Agilent 4284A, Agilent Technologies Inc., Palo Alto, CA) with an automated temperature controller. Before testing the piezoelectric properties, the specimens were poled in a silicone oil bath at 120°C by applying a dc field of 5 kV/mm for 30 min and aged for 24 h. The piezoelectric coefficient (&) was measured using a quasi-static piezoelectric 4, meter (Model ZJ3D, Institute of Acoustics, Chinese Academy of Sciences, Beijing, China). The electromechanical coupling factor (kp) and the mechanical quality factor (Q,,,)were estimated by the resonance and anti-resonance technique using an impedance analyzer (Agilent 4294A, Agilent Technologies Inc.).
,
20
.
,
30
'
,
'
,
40
-
,
50
60
-
I
-
,
80
70
2 Theta (Degree) Fig. 1. X-ray
diffraction (XRD) patterns of powders hydrothermally treated at different temperatures: , perovskite. Inset: Fine scanning XRD patterns in 28=37"-48" for N%5Bio5TiO, (NBT), (Nao *KO2)o ,Bi, 5Ti03 NKBT, and KO5Bio 5Ti0, (KBT) powders hydrothermally treated at 160°C. to 48" for NBT, NKBT, and KBT powders hydrothermally treated at 160°C. It can clearly be seen that (1 11) and (200) peak positions of NKBT shift to high degrees compared with that of pure KBT, which can be attributed to the larger radius of K+ (1.33 A) than that of Na+ (1.02 A) in the A site of the AB03 s t r ~ c t u r eTo . ~ compare the results with the conventional sol-gel process, a part of the gel was calcined at different temperatures from 300" to 800°C for 2 h in air, and the XRD results are shown in Fig. 2. As can be seen in the figure, an amorphous phase was formed at a temperature below 400°C. When the temperature was increased to 450"C, some diffraction peaks corresponding to the Bi2Ti207pyrochlore phase appeared. However, for the powders heated at S O T , an obvious change was observed in the XRD patterns. The characteristic peaks of the NKBT perovskite phase appeared. After further increasing the temperature to 650°C and above, only a pure NKBT phase could be observed and there was no evidence of a second phase. The above results revealed that a temperature as high as 650°C is needed for the transformation from the pyrochlore to perovskite phase in a conventional sol-gel process, which is about 500°C higher than that required for the sol-gel-hydrothermal technique. Figures 3(a) and (b) show the images of NKBT samples synthesized by the sol-gel-hydrothermal and conventional sol-gel
111. Results and Discussion
Figure 1 shows the XRD patterns of powders hydrothermally treated at different temperatures. As can be seen from the XRD pattern for the sample synthesized at IWC, peaks corresponding to the perovskite phase had begun to appear, but the peaks were ill defined, which was indicative of the low crystallinity of this phase. Well-crystallized phases of NKBT were obtained for the samples hydrothermally treated at 160°C and above. The inset in Fig. I shows the fine scanning XRD patterns in 28 = 37"
306
7
20
30
, 1
40
-
1
*
50
I
60
-
I
70
.
I
80
2 Theta (Degree) Fig. 2. X-ray diffraction patterns of the dried gel heated at different
temperatures
m , perovskite, 0 , pyrochlore
Progress in Nanotechnology: Processing
2000
1500
1000
Wavenumber (cm-'>
500
Fig.4. Infrared spectra of the dried gel, and (Nao.&&sBio (NKBT) nanowires hydrothermally treated at 160°C.
Fig. 3. (a) Transmission electron spectroscopy (TEM) photograph of (Nan & 2)0 5BioSTi0, (NKBT) powders synthesized by the sol-gelhydrothermal method at 160°C. Inset: A typical NKBT nanowire and its SAED pattern. (b) Scanning electron microscope photograph of NKBT powders synthesized by the conventional sol-gel method at 650°C. Inset: High-magnification TEM image of powders.
method, respectively. From Fig. 3(a), it can clearly be observed that the sample synthesized at 160°C by the sol-gel-hydrotherma1 process demonstrates the morphology of nano-sized wirelike fibers, which are monodispersed and are not fused to one another. Furthermore, each nanowire is uniform in width along its entire length, with diameters of 5&80 nm and lengths of 1.5-2 pm. The inset in Fig. 3(a) shows the SAED pattern recorded from an isolated nanowire. The diffraction spots have been well indexed to the planes of NKBT, confirming the formation of a Nanotubes, Nanorods, and Nanowires
STiO3
single-crystal NKBT nanowire. This sol-gel-hydrothermal method for preparing NKBT nanowires was highly reproducible, and in all cases, similar morphologies were obtained. Compared with nanowires synthesized by the sol-gel-hydrothermal process, the powders synthesized at 650°C by the conventional sol-gel method consist of spherical particles with an average particle size of about I pm, as can be seen in Fig. 3(b). However, the high-magnification TEM micrograph in the inset of Fig. 3(b) highlights the particulate morphology of the crystallized NKBT particles. It can be seen that large aggregated particles consisting of many smaller primary particles of about 2 M nm are formed. This particulate morphology can be attributed to the porous nature of the as-prepared xerogel precursor, which minimized the interparticle contact by the open structure, thereby resulting in the formation of porous agglomerated powders in calcination. In order to further analyze the structure changes during the synthetic process, the FT-IR spectra of the gel and the NKBT nanowires hydrothermally treated at 160°C were obtained, respectively, and the results are shown in Fig. 4. The bands at approximately 1630 cm-l on the IR spectrum of the gel can be attributed to the bending mode of H-O-H. The peak at 1541 cm-' can be attributed to COO vibration. The sharp and intensive peak at 1390 cm-' is due to the presence of nitrate, while the weak peaks at around 1037, 920, and 830 cm-l can be attributed to CO vibration. The bands in the low-wavenumber region (45&650 cm-I) can be attributed to Ti-0 bond vibrations. After hydrothermal treatment at 160"C, nitrate and carbon peaks were reduced significantly, but a large band appeared around 600 cm-l, which can be attributed to the characteristic vibration of T i 4 octahedron and indicates the formation of the perovskite phase.13 Raman scattering is known to be an appropriate technique for the investigation of the short-range order and phase structure in perovskites. For comparison, pure KBT has been synthesized at 160°C by the sol-gel-hydrothermal method and the room temperature Raman scattering experiments of both KBT and NKBT have been investigated and the results are shown in Fig. 5. For pure KBT, there are four obvious peaks appearing, in an ascending order, at around 273, 330, 529, and 632 cm-' in the Raman shift ran e of 200-700 cm-', which is similar to already reported data.g4 For NKBT, it can be seen that the dominant peak at 273 cm-' roughly maintains its frequency position. This observation, together with the fact that a strong peak at essentially the same position has been observed in the Raman 307
*
1
-
rci
1050°C I
.
30 200
300
400
500
600
~
1
40
~
50
60
2 Theta (Degree)
1
1000"c ~
1
80
70
Fig. 7. X-ray diffraction patterns of the sol-gel-hydrothermal-derived (Nao.xKo.2)o.sBio.sTi03 ceramics sintered at different temperatures: , perovskite; 0 , pyrochlore.
700
Raman shift (cm-'1 Fig. 5. Raman spectra of G.5BiO.5Ti07(KBT) and (N+.~&.2)o,sBio.~TiO~ (NKBT) nanowires hydrothermally treated at 160°C.
spectra of BaTi03 and PbTi03, is an indication that this band must be dominated by T i 4 vibrations. However, compared with pure KBT, the Raman spectra of NKBT show quite a broad feature. The radii of Na+ and K+ areoquite different, r(Na+) = 1.02 A compared with r(K+) = 1.33 A. The decrease of the ionic radii, when going from K+ to Na+, leads naturally to a distortion of the structural framework. On the other hand, the incorporation of NBT and KBT leads to a cation disorder (Bi, K, Na) on the 12-fold coordinated site, which results in the broad feature of NKBT Raman spectra due to the overlapping of Raman modes. In the present work, NKBT nanowires, which were not obtained by other methods, were prepared by a simple sol-gelhydrothermal method. Owing to an extremely small size and anisotropy, the control of nucleation and growth of one-dimensional nanostructure materials is a big challenge. In our work, it is believed that the gel precursor and hydrothermal environment play a key role in the formation of the nanowires. It is known that the growth of nuclei in the calcinations of the gels was controlled by the short-range diffusion of ions in a limited space; therefore, it is difficult to control the morphology of the final products. In the hydrothermal process, the initial condition of
94 98
96
92 90 -
84 82 -
88 86
so!
--Ic Sol-gel method
-aI
1000
Sol-gel-hydrothermal method
.
I
1050
.
I
1100
.
I
1150
.
1
1200
I
Temperature ("C) Fig. 6. Relationship between the relative density and the sintering temperature of the (Nao.xKoz)o.sBio.sTi03ceramics prepared by the sol-gel and sol-gel-hydrothermal methods.
308
the reactant exerts an impact on the crystal nucleation and growth, which is responsible for the morphology of the product. It is supposed that chained nanoclusters with a well-defined morphology are formed during the gelation of an NKBT sol by controlling the hydrolysis of the sol. These chained nanoclusters could serve as the nuclei of NKBT during hydrothermal treatment and gradually grow freely in an aqueous solution to form the nanowires. One-dimensional wire-like nanostructures have received considerable attention in recent years for providing a good system to study the dependence of properties, such as nonlinear optical phenomena, thermal transport, electrical, and mechanical properties on dimensionality and size r e d ~ c t i o n . ' ~It' ~is important to note that nanowires also show superior sinterability, which aided in obtaining a well-densified single-phase ceramic.' However, until now, work related to the ceramics derived from nanowires has rarely been reported. Thus, it is of great significance to investigate the sintering as well as electrical properties of the pressed pellets made by NKBT nanowires. Figure 6 gives the relative density as a function of sintering temperature for the NKBT ceramics prepared by sol-gel and sol-gel-hydrothermal methods, respectively. It can be seen that the pressed pellets made by NKBT nanowires showed >98% theoretical density at 1100°C for 2 h. However, the sol-gel-derived NKBT ceramics showed only 93% theoretical density at the same temperature. It is well-known that the relative density of a sintered ceramic is directly related to the green density of the pressed pellets, which in turn is highly de endent on the morphology of the precursor oxide powders.2032PAscan be seen in Fig. 3, the powder synthesized by the sol-gel-hydrothermal method yields free-standing nanowires of NKBT. The free-standing nanowires have a better packing efficiency and therefore a high green density of about 60% of the theoretical density. In comparison, the powder synthesized by the sol-gel method yielded a porous-structured nanopowder of NKBT, which does not have a high packing efficiency and is responsible for the low green density of about 50% of the theoretical density. Figure 7 shows the XRD patterns of the sol-gel-hydrothermal-derived NKBT ceramics sintered at different temperatures. At 1000"-1 IOWC, all peaks can be indexed to the pure NKBT perovskite phase. However, when the sintering temperature reached up to 115O"C, peaks corresponding to the Bi2Ti~07 pyrochlore phase were observed and peak intensities of Bi2Ti207 became stronger with a further increase in sintering temperature. The appearance of the Bi2TiZO7phase indicated that NKBT was decomposed due to the high volatilization of the potassium element at a high temperature, and the latter could be responsible
'
Progress in Nanotechnology: Processing
~
1
Fig. 9. Temperature dependence of dielectric constant at various frequencies for the sol-gel-hydrothermal-derived (Nao.8Ko.2)o.sBio.STi03 (NKBT) ceramics sintered at 1100°C.
Fig.8. Scanning electron microscope (SEM) micrographs of the freetop surface of (Nao.8Ko.2)o.5Bio.5TiOz (NKBT) ceramics prepared by the sol-gel method at (a) 1100°Cand the sol-gel-hydrothermal method at (b) I 100°C and (c) 1200°C. Inset: SEM micrographs of the polished and thermally etched cross section of NKBT ceramics.
for the decrease of the relative density at a high temperature above 1150°C. Strong differences in the microstructure of the NKBT ceramics prepared by sol-gel and sol-gel-hydrothermal methods, respectively, are evidenced by the SEM photographs, as shown in Figs. 8(aHc). It can be seen from Fig. 8(a) that the visible intergranular porosity, irregular polyhedral shape of the grains, and inhomogeneous grain size distribution clearly evidence the lower degree of densification of the sol-gel-derived specimen sintered at 1100°C. Comparatively, the sol-gel-hydrothermalderived specimen sintered at 1100°C showed a dense and homogenous microstructure, as can be seen in Fig. 8(b). This agrees well with the relative density of 98% measured for the material. In addition, it should be noted that the sol-gel-hydrothermal-derived specimen sintered at 1200°C is less homogenous and the surface micrograph shows some evidences of melting grain, as can be seen in Fig. 8(c). A similar feature of the formation of amor hous phase has also been observed in a PZTbased ceramic.2PLeaching the obtained specimens with dilute nitric acid completely removes the amorphous phase. The origin Nanotubes, Nanorods, and Nanowires
of the amorphous phase can be attributed to vaporizing of the potassium component during the sintering process, and a further study has been carried out to investigate the detailed mechanism. Owing to the high density and homogenous microstructure, the sol-gel-hydrothermal-derived NKBT ceramics sintered at I100"C had been selected to investigate the dielectric and piezoelectric properties. Figure 9 shows the temperature dependence of dielectric constant at various frequencies for the sol-gelhydrothermal-derived NKBT ceramics sintered at 1100°C. A frequency dependence and a hump were observed between room temperature and 180°C ( T , , corresponding to a change from a ferroelectric phase to an anti-ferroelectric phase),23and a diffusion of the dielectric constant at temperatures higher than 320°C (T,, corresponding to a transition from an anti-ferroelectric phase to a paraelectric phase) was also found.8 As N a f, K+, and Bi3+ are randomly distributed in the 12-fold coordination crystallographic sites, the diffuse behavior of NKBT ceramics could reasonably be attributed to the compositional fluctuation and structural disorder in the arrangements of these cations at the A site. On the basis of the above results, it is obvious that for a temperature higher than T,, the variation of the dielectric constant does not follow the classical Curie-Weiss law. Whatever the measurement frequency, the dielectric constant varies according to the Uchino and Nomura function,24
-11 -12 -13 -14
-15 -16
-17 0
1
2
3
4
In (T-TJ Fig. 10. In (l/sr-l/&max) as a function of In (T-Tmax)at 10 kHz for the sol-gel-hydrothermal-derived (N~.8Ko.2)o.sBio.sTi03 (NKBT) ceramics sintered at 1 100°C.
309
Table I. Comparison of the Dielectric and Piezoelectric Properties of NKBT Ceramics Prepared by Different Methods Method
Sol-gel-hydrothermal method Sol-gel method Solid-state reaction Solid-state reaction Solid-state reaction
E,
( I kHz)
-
tan S ( I kHz)
Q3”
41(PC/N)
Td
(“c)
Reference
1220
0.022
0.35
165
156
180
This work
1100
0.030
0.30 0.27
120 109
125
160
-
-
-
-0.050 0.033
-
-
__
0.28
105
120
150 130
This work Sasaki et aL6 Jidng et 01.’ Li et
1030 1200 760
I/IZ-I/E,,,~~ = (T-TmaJ7/C,with y close to 1.51 as clearly shown in Fig. 10 for 10 kHz. In this relation, y acts as a diffusion coefficient and ranges between 1 (normal ferroelectric) and 2 (a complete diffuse phase transition (DPT) ferroelectric). This behavior would be significant of a relaxation process due to an Asite cation mixture. The dielectric and piezoelectric properties of the NKBT ceramics obtained by our sol-gel-hydrothermal methods are compared in Table I with those of ceramics prepared by other methods, including sol-gel and solid-state reactions. It can be found that excellent electrical properties (d33= 156 pC/N, k, = 0.35, Qm = 165, E, = 1220, and tan 6 = 0.022 at room temperature) are obtained for the sol-gel-hydrothermal-derived NKBT ceramics, which were superior to those of previously reported ceramics obtained by other m e t h o d ~ . 6 >Besides ~ . ~ ~ this, it should be noted that the sol-gel-hydrothermal-derived NKBT ceramics retained a high Td value of 180°C. It is known that Td is the depolarization temperature, which is an important factor for BNT-based piezoelectric ceramics in view of their practical use.26The enhanced electric properties in the sol-gel-hydrothermal-derived ceramics can be attributed to the good sinterability of nanowhiskers, which resulted in an increase in density and a more homogeneous microstructure.
IV. Conclusion Free-standing nanowires of NKBT have been synthesized successfully by using the sol-gel-hydrothermal method. In the solgel-hydrothermal process, the hydrothermal conditions create a gentle environment to promote the formation of crystalline NKBT nanowires at a very low processing temperature of 160”C, which reduced the synthesized temperature for the NKBT perovskite phase in the conventional sol-gel process by about 500°C. Owing to the good sinterability of nanowires prepared by the sol-gel-hydrothermal route, the NKBT ceramics with a relative density of 98% are easy to prepare at 1100°C for 2 h and the dielectric and piezoelectric properties were superior to that prepared by the conventional sol-gel route and solid-state reactions. The sol-gel-hydrothermal route, without the presence of catalysts and requiring no expensive equipment, will ensure higher purity in the products and greatly reduce the production cost, and thus offer a novel and simple synthetic route for onedimensional nanoscale materials and high-quality ceramics.
References ‘R. E. Cohen, “Origin of Ferroelectricity in Perovskite Oxides,” Nafure, 358, 136-8 (1992). ’G. A. Smolenskii, V. A. lsupov, A. 1. Agranovskaya, and N. N. Krainik, “New Ferroelectrics of Complex Composition,” Sov. Phys. Solid Stare, 2, 265 1-4 (1961). ‘H. Ishii. H. Nagata, and T. Takenaka, “Morphotropic Phase Boundary and Electrical Properties of Bisumuth Sodium Titanate-Potassium Niobate Solid-Solution Ceramics,” Jpn. J. Appl., 40 [Part I] 566C-3 (2001).
310
k”
4B. I. Chu, D. R. Chen, G. R. Li. and Q. R. Yin, “Electrical Properties of Na,, 2Bi112Ti03-BaTi03Ceramics.” J. Eur. Ceram. Soc., 22, 21 15-21 (2002). 5S. C. Zhao, G. R. Li, A. L. Ding, T. B. Wang, and Q. R. Yin, “Ferroelectric and Piezoelectric Properties of (Na, K)o 5BiosTi03 Lead Free Ceramics,” J. Phys. D, 39, 2277-81 (2006). ‘A. Sasaki. T. Chiba. Y. Mamiva. and E. Otsuki, “Dielectric and Piezoelectric Properties of (BiosNao.s)TiO~-(Bi~.S~.s)TiO, Systems,” Jpn. J . Appl. Phys. Part I , 38,5564-7 (1999). ’Y. M. Li, W. Chen, J. Zhou. Q. Xu,H. I. Sun, and M. S. Liao, “Dielectric and Ferroelectric Properties of Lead-Free Nao.5Bio5Ti03-K0.5Bi0.5Ti03Ferroelectric Ceramics,” Ceram. fnr., 31, 13942 (2005). *X.P. Jiang, L. Z. Li, M. Zeng, and H. L. W. Chan, “Dielectric Properties of Mn-Doped (Nao8Ko,~)0.5Bi0.sTiO~ Ceramics.” Marer. Lerf.. 60, 178690 (2006). 9Z. J. Li, B. Hou, Y. Xu, D. Wu, Y. H.Sun, W. Hu, and F. Deng, “Comparative Study of Sol-Gel-Hydrothermal and Sol-Gel Synthesis of Titania- Silica Composite Nanoparticles,” J. Solid Stare Chem., 178, I 3 9 W 5 (2005). ‘OH. Wang, L. Wang, J. B. Liu, B. Wang, and H. Yan, “Preparation of Phl_,LayTil-r,403(x = 0.28) Powders by a Sol-Gel-Hydrothermal Method,” Mater. Sci. Eng. B, 99, 495-8 (2003). “L. Hou, Y. D. Hou, X. M. Song, M. K. Zhu, H.Wang, and H. Yan, “SolGel-Hydrothermal Synthesis and Sintering of KO5Bio5TiOl Nanowires,” Marer. Res. Bull., 41, l 3 3 M (2006). ”Y. D. Hou, M. K. Zhu, L. Hou. J. B. Liu, J. L. Tang. H. Wang, and H. Yan, “Synthesis and Characterization of Lead-Free KOsBio5TiOl Ferroelectrics by Sol-Gel Technique.” J. Cryst. Growth, 273, 500-3 (2005). I3C.Y. Kim, T. Sekino. and K. Niihara. “Synthesis of Bismuth Sodium Titanate Nanosized Powders by Solution/SolCel Process,” J . Am. Ceram. Soc., 86. 1464-7 (2003). I4J. Kreisel, A. M. Glazer, G. Jones, P. A. Thomas, L. Abello, and G . Lucazeau, “An X-Ray Diffraction and Raman Spectroscopy Investigation of A-Site Substituted Perovskite Compounds: The (Nal-,K.Jo 5Bio.sTi03( 0 5 x 5 I) Solid Solution,” J . Phys.: Condens. Marrer. 12, 3267-80 (2000). ”H. J. Dai, E. W. Wong, Y. Z. Lu, S. S. Fan, and C. M. Lieber, “Synthesis and Characterization of Carbide Nanorods,” Narure. 375, 769-72 (1995). I6L. A. Bumm. J. J. Arnold, M. T. Cygan. T. D. Dunbar, T. P. Burgin. L. Jones 11, D. L. Allara, J. M. Tour, and P. S. Weiss, “Are Single Molecular Wires Conducting,” Science, 271, 1705-7 (1996). ”J. D. Holmes, K. P. Johnston, R. C. Doty, and B. A. Korgel, “Control of Thickness and Orientation of Solution-Grown Silicon Nanowires,” Science, 287, 1471-3 (2000). “J. J. Urban, J. E. Spanier, L. Ouydng. W. S. Yun, and H. Park, “Single Crystalline Barium Titanate Nanowires,” Adv. Mater., 15, 423-6 (2003). ”G. Feng. W. D. Nix, Y. Yoon, and C. J. Lee, “A Study of the Mechanical Properties of Nanowires Using Nanoindentation.” J . Appl. P h j x , 99,074304-1-10 (2006). 2oY, D. Hou, L. Hou. M. K. Zhu, and H. Yan, “Synthesis of (&,sBi,5)04. B% 6Ti03 Nanowires and Ceramics by Sol-Gel-Hydrothermal Method, Appl. Ph s Lerr.. 89, 243114-1-3 (2006). ‘A. Banerjee and S. Bose, “Free-Standing Lead Zirconate Titanate Nanoparticles: Low-Temperature Synthesis and Densification,” Chem. Marer., 16, 561&5 (2004). 22Y.D. Hou, M. K. Zhu, H. Wang, B. Wang, C. S. Tian. and H. Yan, “Effects of Atmospheric Powder on Microstructure and Piezoelectric Properties of PMZNPZT Quaternary Ceramics,” J. Eur. Ceram. Soc., 24. 3731-7 (2004). ”Y. Hiruma, H. Nagata, and T. Takenakd, “Phase Transition Temperatures and Piezoelectric Properties of (Bil~2Nal,2)Ti03-(Bili2Kli2)TiO~-BaTi03 LeadFree Piezoelectric Ceramics,” Jpn. J . Appl. Phys.. 45 [Part I] 7409-12 (2006). 24K.Uchino and S. Nomura, “Critical Exponents of the Dielectric Constants in Diffused-Phase-Transition Crystals,” Ferrwlecrr. Lerf. Sect., 44,55-61 (1982). ”Y. M. Li, W. Chen, Q. Xu, J. Zhou, and M. S. Liao, “Dielectric and Piezoelectric Properties of (Nao 8&,.2)o sBioSTi03Ceramics,” J. Inorg. Marer., 19, 81722 (2004). 26X, X. Wang, X. G. Tang, and H. L. W. Chan, “Electromechanical and FerLead-Free Piroelectric Properties of (Bil,2Nal,2)Ti03-(Bil~2Kl,2)Ti03-BaTi03 ezoelectric Ceramics.” Appl. Phys. Letr., 85, 91-3 (2004). n
Progress in Nanotechnology: Processing
Synthesis and Characterization of Ce, --xCdx02-6Nanorods Jong So0 Leet and Sangtae Kim* Department of Chemical Engineering and Materials Science, University of California, Davis, California 956 16
Single crystalline Cel_,Gd,02-g nanorods with x = 0.05 and 0.1 as well as x = 0 were synthesized by a sol-gel process without using any template and/or applying high pressure. The nanorods grow preferentially along the [ l l o ] direction regardless of the dopant concentration. The diameter and the length of the synthesized nanorods range from 10 to 50 nm and from 200 to 700 nm, respectively. The length of the nanorods decreases with increasing dopant concentration. The role of the surfactant as a structuredirecting agent is critical to the growth of Ce, -,GdX02-g nanorods.
I. Introduction
C
ERIUM dioxide (Ce02) is a fluorite-structured nonstoichiomeric oxide with a greater structural tolerance to reduction. Its electrical property as well as oxygen storage capacity have been the center of attention for decades because of the potential for various applications such as exhaust catalysts, oxygen sensors, and, in particular, a solid electrolyte (SE) for solid oxide fuel cells (SOFCS).'-~In recent years, the oxygen ion conductivit of nano-structured Ce02 doped with acceptors such as GdY+ has been of particular interest as the nano-structured CeOz may serve as an SE for low-temperature S O F C S . This ~~ hypothesis is based on the assumption that the large surface/ interface area in the nanomaterials may serve as fast conduction pathways leading to high conductivity. Hence, the preparations as well as the oxygen ionic conductivity of Gd-doped Ce02 in the forms of nanocrystalline (crystallite size < 50 nm), thin films (two dimensional (2 D)), and ceramics (three dimensional (3 D)) have been extensively e ~ p l o r e d . ~Unlike other oxide conductors,' however, little attention has been paid to the synthesis of one dimensional (1 D) single crystalline nano-structured (e.g., nanorods) CeO2 and thus to their electrical properties, although they can serve as ideal model systems to study the size effects on their physical properties. In particular, preparation of 1 D Gddoped Ce02 has not been reported to date, while few results have described the synthesis of 1 D nominally pure Ce02 that are practically less relevant as far as the SOFC applications are concerned.*-I2 In this communication, we report a simple large-scale synthesis of single crystalline Cel-,Gd,02-6 (CGO) nanorods with x = 0.05 (CGO-5) and 0.1 (CGO-10) as well as x = 0 (CEO), based on sol-gel routes without using any template and/or applying high pressure typically needed for the synthesis of the 1 D Ce02 reported previously.8-12 J.-H. Lee--contnhuting editor
Manuscript No. 22076. Received August 12, 2006; approved October 16, 2006. J.S.L. acknowledges a partial support from the Korea Research Foundation (KRF2005-214-DOO305) for this work. 'Author to whom correspondence should be addressed. e-mail
[email protected] 'Member, American Ceramic Society.
Nanotubes, Nanorods, and Nanowires
11. Experimental Procedure
For the synthesis of 1-D CGO of interest, 30 mL of aqueous sulfuric acid solution (pH 2) containing 1.52 g of CeC13.7H20 (Aldrich, St. Louis, MO; 99.9%), together with appropriate amounts of Gd(N03).6H20 (Aldrich, 99.99%) and 5 g of cetyltrimethyl ammonium bromide (C19H42N. Br, CTAB, Aldrich, 99%), was prepared and stirred for l h at 60°C. Seventeen milliliters of aqueous ammonia (Aldrich, 28-30 wt%) was then added to the solution in a dropwise manner. The addition results in a change in the pH value of the solution from about 2 to about 11, leading to precipitation. The color of the solution changes from colorless to dark purple accordingly. After stirring for another hour, the glass beaker containing the mixture was sealed with an aluminum foil to prevent possible contaminations and was placed in a dry oven at 80°C for 36 h. The product was then separated from the solution using a centrifuge and washed multiple times with pure ethanol and deionized water. The collected product was dried at 60°C for 12 h in a dry oven. Finally, for crystallization, the product was calcinated at 550°C in N2 for 6 h and then in air for 2 h. The structure analysis of the as-calcinated product was performed using an X-ray diffraction (XRD) spectrometer (Scintag XDS-2000, Cupertino, CA) with CuKa radiation, a transmission electron microscope (TEM, Philips CM- 12, Hillsboro, OR), a high-resolution TEM (HRTEM, Tecnai F-20, Hillsboro, OR), and an energydispersive X-ray spectrometer (EDX, Gatan, Pleasanton, CA). 111.
Results and Discussion
Figure 1 shows the XRD patterns of the synthesized CGO-10 (a) and CGO-5 (b) samples. Also included in Fig. 1 is the synthesized CEO (c) sample for comparison. All the peaks revealed are indexed to confirm a fluorite structure. Neither secondary nor impurity phases are detected in the samples as can be seen in Fig. 1. Also shown in Fig. 1 is a peak (marked with a solid square) of a high-quality standard quartz (JCPDS, No. 33-1 161) used as a reference for the angular correction, which remains unshifted in all the samples. The XRD peaks of the CGO samples are found to be slightly shifted to the lower diffraction angle with increasing Gd content while those of the CEO sample remained nearly unshifted compared with those of the bulk Ce02 with a lattice constant of a = 0.541 nm (JCPDS, No. 43-1002). The shifts of the (220) and (311) peaks of the CGO-5 and CGO-10 are A(20) = 0.02, 0.04" and 0.03, 0.18", respectively, indicating that the lattice parameter of the CGO is increased with increasing Gd content relative to that of CEO. Such an increase in the lattice parameter can beattributed to the slightly larger radius of the dopan: Gd3+(1.05 A) compared with that of the host ion Ce4' (0.97 A) in the 1atti~e.I~ This result strongly supports the fact that well-crystallized CGO with the dopant uniformly distributed into the lattice has been synthesized. Figure 2 shows a TEM image (a) and its associated EDX spectrum (b), and an HRTEM image (c) of the CGO-5 sample. Figure 2(a) clearly demonstrates that the synthesized CGO-5 is in the form of nanorods with a diameter (d)and length (I) of 311
Fig. 3. Transmission electron microscope (TEM) (a) and its associated energy dispersive X-ray spectrometer spectrum (b), and high-resolution TEM image (c) of a 10 mol% Gd-doped Ce02 nanorods. Fig. 1. X-ray diffraction patterns of as-synthesized 10 mol% (a), 5 mol% (b) Gd-doped Ce02. and nominally pure C e 0 2 (c) samples.
about 1&50 nm and about 100-400 nm, respectively. A small amount of the nanopowder of CGO-5 (d- 10 nm) was found to coexist with the nanorods (see Fig. 2(a)). The EDX spectrum of Fig. 2(b) reveals the peaks associated only with Ce, Gd, and 0 (the Cu-related peaks come from the Cu grid). The relative atomic percent of Gd in the nanorod estimated based on the EDX data is about 6.8%, leading to the fact that the exact chemical composition of CGO-5 nanorods is Ceo.932Gdo.06802-s. The selected area diffraction (SEAD) patterns of one of the nanorods shown in the bottom inset of Fig. 2(a) indicate that the nanorod is a single crystalline phase with growth along the [110] direction. The HRTEM image (Fig. 2(c)) confirms that the CGO-5 nanorod is a structurally uniform single crystal with a regular periodicity of the lattice. The interplannar distance between the adjacent lattice fringes calculated from Fig. 2(c) is about 0.31 nm (see the inset), corresponding to that of the (1 11) lattice plane of the bulk Ce02. These results support the fact that the CGO-5 nanorod grows preferentially along the [I 101 direction. Figure 3 shows the TEM image (a), EDX spectrum (b), and HRTEM image (c) of the synthesized CGO-I0 sample. Similar to CGO-5, CGO-I0 was found to form nanorods coexisting with its nanopower (d< 10 nm, see in Fig. 3(a)). The volume fraction of the nanopowder is, however, found to be slightly higher than that in the CGO-5 sample. The CGO-10 nanorods are about lO(r200 nm long and are about 30-50 nm wide, indicating that those nanorods are shorter and thicker compared
with the CGO-5 nanorods shown in Fig. 2(a). Both the CGO-5 and the CGO-10 nanorods did not grow further even at a longer reaction time. This may imply that the length of the CGO nanorods and the dopant concentration are interrelated. Indeed, CEO nanorods reveal the smallest aspect ratio among the nanorods of interest in this study. The SEAD patterns of a selected single nanorod (see the bottom insets of Fig. 3(a)) of CGO-10, together with the EDX spectrum (Fig. 3(b)), indicate that the synthesized nanorod is a structurally uniform single-crystal CGO with growth along the [I 101 direction, consistent with the CGO-5 nanorods. This is also confirmed by Fig. 3(c) from which the interplannar distance between the adjacent lattice fringes is calculated to be about 0.19 nm, corresponding to that of the (220) lattice plane of the bulk CeO2. The exact chemical composition of the CGO-10 nanorods calculated from the EDX data was c % ssGdo 120z-s. For comparison, CEO nanorods (see the EDX spectrum shown in Fig. 4(b)) were also synthesized and their morphologies are shown in Fig. 4. Note that, unlike the CGO samples discussed above, the as-synthesized CEO sample consists of nearly 100% single crystalline nanorods that appear to be thinner (ca.l(r30 nm) and longer (ca. 200-700 nm) than the CGO nanorods. The CEO nanorods also grow along the [I 101 direction, which confirms the results reported previously.’ The interplannar distance between the adjacent lattice fringes calculated from Fig. 4(c) is about 0.27 nm, corresponding to that of the (200) lattice plane of the bulk Ce02. The cetyltrimethyl ammonium bromide (CTAB, see “Section II”), which serves as a structure-directing species and/or a reactant, has been used for the synthesis of various nano-
Fig. 2. Transmission electron microscope (TEM) (a) and its associated energy dispersive X-ray spectrometer spectrum (b). and high-resolution TEM image (c) of 5 mol% Gd-doped Ce02 nanorods.
Fig.4. Transmission electron microscope (TEM) (a) and its associated energy dispersive X-ray spectrometer spectrum (b), and high-resolution TEM image (c) of pure Ce02 nanorods.
312
Progress in Nanotechnology: Processing
structure^.^^'^^'"'^ It was indeed found that only nanopowders with d 1300°C). With Z r 0 2 addition, the decomposition temperature is further lowered (in the range 1000- 1050"C).'0 Efforts to achieve densification at a lower sintering temperature involved hot pressing/hot isostatic pressing, spark plasma sintering,' and use of additives (calcium fluoride (CaF,)),' etc. All the above composites in the HAZrO2 system involved a high volume fraction of Z r 0 2 addition. On the other hand, the available literature on the processing, densification behavior, and mechanical properties of HA-Zr02 composites containing a low volume fraction of ZrO2 ( I 20 ~ 0 1 % )is comparatively
'
''
359
limited.8,'2,'3 Kim et aL8 studied the effect of CaF2 addition on the reaction sintering and mechanical properties of HA-Zr02 composites containing 20 vol% Z r 0 2 . Addition of CaF2 promoted densification and retarded the HA to T C P decomposition. The densification temperature also decreased with CaF2 addition. The highest flexural strength was 125 MPa and fracture toughness was 2 MPa/m'/2, which were higher than pure HA. Pyda et all2 studied the effect of chemical composition and morphology of zirconia particles on the properties of HA-Zr02 particulate composites containing 11.9 vol% Ca or Y-stabilized ZrOz. The composites were hot pressed in the temperature range 11501250°C in an argon atmosphere. It was observed that isometric ZrO2 crystallites incorporation to the HA matrix increased both the strength (18OMPa) and the toughness (1.1 MPa/m'/2), which were higher than that of pure HA. Ahn et studied the effect of nano zirconia reinforcement on the strength of hotpressed HA-Zr02 composites containing 1.5-8 wt% Z r 0 2 . They observed the highest strength and hardness at 3 wt% Z r 0 2 , which subsequently decreased at higher Z r 0 2 loading. They also reported that a lower volume fraction of Z r 0 2 addition will help to retain both HA and t-Zr02 phases, which may be due to the combined effect of matrix constraint and uniform dispersion of fine zirconia particles. Thus, it is expected that HAZrO2 composites containing a lower volume fraction of t-Zr02 may yield a transformation-toughening effect that can lead to improved mechanical properties of HAzirconia composites.7,8,12-17 The aim of the present investigation is to observe the effect of low volume percent TZP addition on the density, strength, and microstructure of HA-Zr02 composites. An attempt has also been made to analyze the observed results with respect to phase analysis and microstructure of the samples for different HA-Zr02 composites. The exact effect of Z r 0 2 addition (both at low and high Z r 0 2 loading) cannot be understood unless a systematic investigation in this system is carried out.
Experimental The HA-zirconia composite preparation consisted of two parts: (a) Preparation of nano-TZP (2.5 mol% Y2O3) powder using the precipitation method.
360
(b) Preparation of an HA-zirconia composites (containing 2, 5, 7.5, and 10wtYo TZP) by the reverse strike precipitation method, denoted as HZ2, HZ5, HZ7, and HZ1 0, respectively. The precursors used for TZP (2.5 mol% Y2O3) were ZrOC12 . 8 H 2 0 (Loba Chemicals, India) and Y2O3 (Indian Rare Earths Limited, India). The concentration of ZrOCl2 solution was 0.75 mol/L in which the requisite amount of Y2O3 powder was dissolved. Precipitation was carried out using an N H 4 0 H (1:l) (Oster Chemicals, India) solution at p H 10. The precipitates were washed with water to remove chloride ions, followed by drying and calcination at 850°C/2 h to obtain TZP powder. The outline of the powder preparation steps is shown in Fig. la. The HA-Zr02 composite powders were prepared from an equimolar aqueous solution (1.63 mol/L) of both Ca(NO&. 2 H 2 0 (Oster Chemicals, India) and (NH4)2HP04 (Nice Chemicals, India). These two solutions were mixed together, along with the addition of a few drops on H N 0 3 to obtain a colorless solution. In the usual method of HA-Zr02 composite powder preparation, N H 4 0 H is added to the above-mixed solution. However, in the present investigation, the mixed solution was added dropwise to a beaker containing calcined TZP powder dispersed in N H 4 0 H , while the beaker was being vigorously stirred. This allowed better and uniform reaction of Ca(N03)2and (NH,J2HP04 with NH40H and for the present investigation, this precipitation process has been termed as the "reverse strike precipitation" method. The outline of the precipitation process is shown in Fig. 1b. The as-precipitated amorphous powders having varying proportions of TZP were subjected to hot water and propanol washing to prevent particle agglomeration on drying of these powders. It has been reported that the presence of hydroxyl and other ions (e.g., CI-) affects the crystallization and densification behavior of many ceramic powders. l 3 The washed powders were oven dried for 24h. The FTIR spectra of the as-dried precipitated powder of HA and HA-TZP were observed in a Perkin Elmer FTIR spectrophotometer (Spectrum RX- 1). The spectrum was taken in the KBr pellet mode in the wave number range 4400-400 cm-'. A small part of the dried powder was subjected to thermal analysis in the DSC/TG mode (Netzsch 402 C) till 1200°C at a heating rate of 1O"C/ min in air. The dried powders were calcined in different temperature ranges from 650°C to 850°C/2h in air. The phase analysis of the calcined powders and sintered
Progress in Nanotechnology: Processing
length = 20 mm) as well as in diametral compression of the cylindrical disk (12.5 mm diameter) at a crosshead speed of 0.2 mm/min. The microstructures of the sintered and polished pellets were studied by SEM USM6480LV). In all the above characterizations of the HATZP composite, HA was used as the control sample.
I
Addiumof*xccrr Ni,oH l o imnasothopH (9-10) I
I
Results and Discussion
I +
ov *l~iw FW
Water Warhod
24b
Glcination
At 8500Cnhr
1.63 M of
I
1.63 M of
Ca(NQ)2.4&0
1
Solution (A)
rnh~04 Solution (B)
2.5 mow Y-TZP powder dispasedin (1:l) MI3solution kept in
Was-
of the
FTIR The FTIR spectra of pure HA (Fig. 2) show all the peaks corresponding to (OH)- and stretching, and no extra peak was observed. The broad peak between 3750 and 2900 cm-' corresponds to the hydrogen-bonded 0-H group stretch of HA and water. Thus, the processed HA was stochiometric. The three strong peaks at around 1060, 570, and 606cm-' could be attributed to the P-0 vibration modes. The band at 570 and 606cm-' corresponds to the bending mode of and the strong band at 1060 and 963 cm-' is due to the vibration of the group. The absence of broadening of the peak at 1060cm-' further confirms that no decomposition of HA to P-TCP occurs at 850°C. The decrease in broadness of the peak between 3750 and 2900cm-' with an increase in calcination temperature indicates the removal of absorbed H20.
Fig. I . (a) Flow diagram f i r the synthesis of 2.5Y-TZP (6) flow diagram f i r the synthesis of hydroyapatite-(2, 5, 7.5, and 10 wt%).
pellets was studied by XRD (Philips PW1830 Holland, the Netherlands). The relative amount of the different phases present in the sintered pellet was calculated from the relative XRD peak intensities of the different phases. The particle morphology was studied by TEM (Philips CM 200). The densification behavior was studied in a dilatometer (Netzsch DL 402 C). The powders were uniaxially compressed at 280 MPa using 3 wt% PVA as the binder. The density of the sintered samples was measured by the Archimedes principle using kerosene as the immersion liquid. The strength of the sintered composites was measured using three-point bending (span
Nanocomposites and Nanostructures
I
I
4000
I
3500
I
3000
I
2500
I
2000
IV
1500
I
1000
500
Wave number (cm-')
Fig. 2. FTIR spectra ofhydroyapatite as a firnction of calcination temperature.
361
DSC/TG The DSClTG curve of the precipitated powders of HA is shown in Fig. 3a. The curve shows a total weight loss of 5.78%, which rakes place in two stages. It has a broad but sharp endothermic peak in the temperature range 3O-10O0C, with a peak at 70°C. This endothermic peak is associated with a weight loss of 3.55%, which could be related to the loss of adsorbed water from HA. Following this peak, there is a broad and diffused exothermic peak in the temperature range 250400°C. This exothermic peak also shows a gradual weight loss of about 2.23%. X-ray diffraction of the precipitated powder heated at 400°C shows peaks corresponding to HA only. Thus, this diffuse exothermic
0.6
c
-0.8"
2w I
'
'
400
'
'
600
"
800
"
loo0
'
1200 I
'
Temperature ("C)
I
-0.41'
I
'
200
.
'
400
'
'
600
.
800
"
'
1000
.
12w)
Temperature ("C)
Fig. 3. DSC/TG curve of (a) hydroxyapatite powder, (b) HZ2 powder.
362
peak is due to the crystallization of HA from the precipitated amorphous powder. Figure 3b shows the DSC/TG curve of the precipitated HZ2 powder. The nature of the curve and the total weight loss (4.74%) is comparable with that of pure HA (5.78%). This curve also shows one broad endothermic peak in the temperature range 30-1 OO'C, with a peak at about 70"C, along with a weight loss of 3.21% corresponding to the loss of adsorbed water. The weight loss is lower than pure HA as the amounr of HA is less by 2%. Similarly, the second phase of weight loss from 100°C to 400°C is also lower in HZ2 in comparison with pure HA. However, the exothermic peak of HZ2 is at a lower temperature and it is sharper as compared with pure HA. This implies that the addition of TZP reduces the crystallization temperature and also improves the crystallization behavior of HA. However, no further change in either the crystallization behavior or the crystallization temperature could be observed in HA with a still higher number of TZP (viz. HZ5, HZ7, and HZ10) samples. Further, the DSC/TG curves of both HA and HATZP powders show a broad exothermic peak in the temperature range 800-1000°C. This exothermic peak is not associated with any weight loss. Therefore, the peak should correspond to phase changedordering in HA and/or ZrOz. In order to validate the DSC curve, the positions of the first three peaks of HA (or triplets) (d = 2.81, 2.72, and 2.78A0, respectively) were monitored as a function of calcination temperature. The results are shown in (Fig. 4b), where the left-hand pattern (Pattern A) shows the complete XRD profile of HA and HA-TZP composites, and the right-hand pattern (Pattern B) shows the "zoomed-in" pattern of the same in the 28 range 30-35". It is clear from the patterns that at 650"C, the triplets are not well separated and only two peaks (d = 2.8 1, 2.72A") are distinguishable. With an increase in the calcination temperature to 850"C, besides the above two peaks a third peak (as split peak) appears at d = 2.78. Finally, at 1050"C, all the three peaks could be clearly distinguished. It could also be seen from the zoomed-in pattern that the peak position shifts toward a lower d value and the peak intensity and sharpness increase. Thus, we can infer that at 65OoC, H A is poorly crystallized (disorder state) and it becomes perfectly crystalline at 1050°C. This phase change from poorly crystallized HA to perfectly crystallized HA is indicated as up a broad exothermic peak in the temperature 800-1000°C. A similar observation was made
Progress in Nanotechnology: Processing
HA is stable up to 850°C. The t-ZrO2/c-ZrO2 peaks could be detected from H Z 2 in Zr02-containing samples (HZ2 to HZ10). The intensity of Z r 0 2 peak increases at a higher Z r 0 2 content. No other phases like TCP, CaZr03, or m-ZrO2 could be detected. As discussed in the DSC/TG section, with an increase in the calcination temperature, the "d" value of HA peaks shifted to a lower value (Fig. 4b). From the XRD pattern, it is clear that in the calcined powder, only HA and t-Zr02 are present. No other phases like TCP, CaZr03, or m-Zr02 could be detected. The intensity of t-Zr02 increases with an increase in wt% of zirconia addition.
Particle Morphology of Calcined Powder Figures 5a, b shows the TEM photograph of calcined HA and HZ2 powder, respectively. The small and spherical particles indicate Z r 0 2 , while the lighter ones indicate HA. It could be seen that HA particles are both spherical and cube shaped in the size range 50-100 nm. Mostly, the Z r 0 2 particles are seen on the surface of HA particles. Some clusters of Z r 0 2 are also seen in the top part of the Fig. 5b.
Densifcation Behavior and Phase Analjsis of Sintered Compacts Fig.4. (a) XRD analysis of bydroxyapatite (HA), HA-ZrO, compositepowder calcined at 850aC/2b, (b) expanded region of XRD pattern to show the peak sbifing of HA.
from the XRD pattern of calcined HA-TZP powders. Moreover, the presence of Z r 0 2 also contributes to an exothermic peak because t-Zr02 prepared from an amorphous powder undergoes the following phase change: amorphous-cubic-tetragonal. It has been observed in our earlier study18 that the cubic-tetragonal phase change in Z r 0 2 takes place around 850°C. This could also be observed as an exothermic peak in DSC. Thus, in HA-TZP powder, we could observe two distinct peaks: one due to HA phase change and the other due to Z r 0 2 phase change.
Phase Analysis of Calcined Powder Figure 4a shows the XRD pattern of calcined HA and HA-TZP (HZ2, HZ5, HZ7, and HZ10) composite powders calcined at 850°C. The pattern shows that
Nanocomposites and Nanostructures
Figure 6 shows the nonisothermal sintering behavior of HA and HZ2, HZ5, HZ7, and HZlO composites. However, a significantly lower density is observed at 7.5 and 10wtYo TZP additions. The reduction in density at 7.5 and 1Owt% TZP occurred due to the higher sintering temperature of TZP and hence could not be densified at 1250°C. However, for HZ2, it appears that some other mechanism is playing a role in improving the densification. The relative sintered density at 1250°C is given in Table 11. The XRD pattern (Fig. 7) shows phases in the sintered samples of HA and HZ2, HZ5, HZ7, and HZ10. The XRD pattern shows that beyond 2wt% TZP additions, the sintered composites also contain P-TCP and CaZrO3. In HZ2, although the shrinkage starts at nearly the same temperature as that of HA, HZ2 has higher shrinkage in comparison with HA. This implies that the sintered density of HA increases on addition of 2 wt% TZP.
Microstructural Development on Sin tering The HZ2, HZ5, HZ7, and HZlO samples were sintered in air in the temperature range 1150-1250°C
363
Fig. 5. TEM of (a) hydroxyapatite (HA) powder and (6) HZ2 powder calcined at 850°C/2h.
for 2 h. The microstructure development of different HA-TZP composites as a function of varying TZP contents was studied for samples sintered at the highest temperature. Figures 8a-c shows the microstructure of different HA-TZP composites: (a) HZ2, (b) HZ7, and (c) HZ1O. All these samples were sintered at 1250°C for 4 h. For comparison, the microstructure of pure HA is also shown in Fig. 8d, which shows a mixed grain size for HA. The larger grains of HA are in 4-6pm, while
the smaller grains are in the range 1-2pm. Figure 8a shows that H Z 2 samples have a dense microstructure with an HA grain size in the range 2 - 4 p m , although some larger grains of 5-6pm could also be seen. The microstructure also contained void spaces and bright small grains (< 1 pm) distributed in the matrix. It appears that many of these void spaces were created due to the pullout of ZrOz grains during specimen preparation. The microstructure of HZ7 (Fig. 8b) clearly shows
Fig. 6. Nonisothermal densiJcation curve of bydroxyapatite (HA), HA-ZrO, composites.
Fig. 7. XRD analyJis of hydroxyapatite (HA), HA-Zr02 composites sintered at 1250" C/4h.
364
Progress in Nanotechnology: Processing
Fig. 8. SEM of hydroxyapatite (HA), HA-ZrO2 composites sintered at 125O"C/4h (a) HZ2, (b) HZ7, (c) HZlO, and (d) HA.
the effect of TZP addition on the grain size distribution of HA. In this sample, although some large grains (- 5 pm) are present, many small grains (1-2 pm) could also be seen. The ZrOz pullout during polishing is evident and some ZrOz particles are also present as clusters. The microstructure is further refined in HZlO (Fig. 8c). Although this microstructure had a mixture of a large grain size ( - 5 p m ) and very small grain size (0.5 pm), the distribution became narrow. The grain size distribution is shown in Table 11. The bright clusters of ZrOz grains can be seen near the multiple grain junctions. The microstructure (Figs. 8a-c) thus shows that addition of TZP to HA refines the grain size and produces an increased population of finer grains. However, the distribution of TZP is not uniform and it produces grain clusters. The TZP cluster size increases with increasing weight fraction of TZP. The effect of TZP addition on the grain size distribution is shown in Table I.
Table I.
Mechanical Properties Table I1 lists the diametral compression as well as the bending strength of sintered HA and HA-TZP composites. The table shows that following the density trend, the strength of HZ5, HZ7, and HZlO decreases with increasing TZP content. However, the strength increases significantly for H Z 2 composites. This is an interesting observation and needs more careful and detailed study. As already stated, this study only deals with the effect of TZP addition on the density, strength and microstructure. The exact effect of low TZP addition requires detailed TEM analysis, which will be taken up in a future study. Although the microstructure indicates grain refinement with TZP addition, the addition of TZP also increases the porosity due to incomplete densification due to the higher sintering temperature of TZP. Thus, the mechanical properties decrease with increasing TZP content.
Grain Size Distribution of Hydroxyapatite (HA), HA-Zr02 Composites Sintered at 125OoC/4h ~
HA Grain size (pm) Average Mean
Coarse
HZ2 Fine
Coarse
HZ7 Fine
Coarse
HZlO Fine
Coarse
Fine
3.63f1.15 1.13k0.3 3.64f1.76 0.93f0.36 2.7011.09 0.79f0.39 1.57f0.45 0.62k0.24 2.48 k 1.49 2.73 f 1.94 1.50+ 1.17 0.99 f0.57
Nanocomposites and Nanostructures
365
Table 11.
Density and Mechanical Properties of Sintered Compact of Hydroxyapatite (HA) and HA-Zr02 Composite
Relative Diametral Three-point sintered compression bending Composition density (Yo) strength (MPa) strength (MPa)
HA HZ2 HZ5 HZ7 HZlO
98.0 99.6 96.3 92.4 90.0
13.0f 1.03 34.5 k 2.763 11.3f1.04 17.0f 1.362 13.0 L- 0.907
35f 1.2 72 3.6 31 f 1.55 40f2 36L- 1.75
Phases in sintered HA sample (volo/o) HA
t-ZrOz
CZ
TCP
100 96.50 64.9 71.2 79
3.50 11.3 12.7 14.50
-
-
-
17.6 11 6.5
6.2 5.1 Tr.
Conclusions
5.
M. Knepper, 8. Milthrop, and S. Moricca, “Interdiffusion in Short-Fibre Reinforced Hydroxyaparite Ceramics,” J Mater. Sci. Mater. Med., 9 [ 101
HA and its composite with zirconia containing 2, 5, 7.5, and 10wtYo TZP were prepared by the reverse strike precipitation method. A DSC/TG study revealed that addition of TZP to HA reduced the decomposition
6.
Y.-M. Kong, S. Kim, and H.-E. Kim, “Reinforcement of Hydroxyapatite Bioceramic by Addition of Z r 0 2 Coated with A203,”J. Am. Ceram. Soc., 82
7.
2.Shen, E. Adolfsson, M. Nygren, L. Gao, H . Kawaoka, and K. Niihara,
589-596 (1998).
weight loss and promoted low-temperature crystallization of HA. TEM analysis of calcined powders showed spherical particles of TZP dispersed in cuboidal/spherical HA grains. The sintered density of HA increases with 2wt% TZP addition and it reduces with further addition of TZP. XRD revealed that in the sintered samples, only HA and t-Zt02 were present at 2wt% TZP addition. For a higher TZP content (5, 7.5, and 10wt%), both CaZr03 as well as a-TCP could also be seen, along with HA and t-Zr02. The bending strength of HA (35 MPa) increases to 70 MPa at 2 wt% T Z P addition, followed by a decrease at a higher TZP content. SEM microstructures show a grain refinement of HA with TZP addition. The decrease in strength at higher TZP addition could be related to the improper densification of the composite at a higher TZP loading.
References L. L. Hench, “Bioceramics,”J. Am. Ceram. Soc., 81 [7] 1705-1728 (1998). L. L. Hench and J. Wilson, Science, 226 1111 630-636 (1984). G. De With, H. J. A. Van Dijk, N. Hattu, and K. Prijs, “Preparation, Microstructure and Mechanical Properties of Dense Polycrystalline Hydroxyapatire,”J Mater. Sci., 16 [6] 1592-1598 (1981). K. Ioku, S. Somiya, and M. Yoshimura, “Hydroxyapatite Ceramics with Tetragonal Zirconia Particles Dispersion Prepared by HIP Postsinrering,” J. Ceram. Soc. Jpn. Int. Ed., 99 [3] 196203 (1991).
366
1111 2963-2968 (1999).
“Dense Hydroxyapatite-Zirconia Ceramic Composites with High Strength for Biological Application,” Adu. Mater., 13 [3] 214-216 (2001). 8. H.-W. Kim, Y.-H. Koh, B.-H. Yoon, and H.-E. Kim, “Reaction Sintering and Mechanical Properties of Hydroxyapatite-Zirconia Composites with Calcium Fluoride Addition,” J. Am. Ceram. Soc., 85 [6] 1634-1636 (2002).
9. J. Li, L. Hermansson, and R. Soremark, “Synthesis and Sinrering of Hydroxyapatite-Zirconia Composite,” J. Marer. Sci.: M a w . M e d , 4 [ 11 50-54 (1993). 10. R. Ramachandra Rao and T. S. Kannan, “Synthesis and Sintering of Hydroxyapatite-Zirconia Composites,” M a w . Sci. Eng. C, 20 [ 1-21 187-193 (2002). 11. J. Li, H. Liao, and L. Hermansson, “Sintering of Partially Stabilized Zirconia
and Partially Stabilized Zirconia-Hydroxyapatite Composites by Hot Isostatic Pressing and Pressureless Sintering,” Eiomaterialr, 17 1181 17871790 (1996). 12. W . Pyda, A. Slosaryk, Z. Paszkiewicz, A. Rapaa-Kmita, M. Haberko, and A. Pyda, “Effect of Chemical Composition and Morphology of Zirconia
13.
14
15
16. 17.
18.
Particles on Properties of HAP-Zirconia Particulate Composites,” Key Eng Mater., 2 0 6 2 1 3 1567-1570 (2002). E. S. Ahn, N. J. Gleason, and J. Y. Ying, “The Effect of Zirconia Reinforcing Agents o n the Microstructure and Mechanical Properties of Hydroxyaparite-Based Composites,” J. Am. Ceram. Soc., 88 [ 121 3374-3379 (2005). S. H. Kim, H. C. Lee, H. G. Bang, and S. Y. Park, “Effect of MgFz Additive on the Mechanical Properties in HydroxyapatitelZirconia Composites,’’ Mater, Sci. Forum, 510-511 4 7 8 4 8 1 (2006). F. Mezahi, A. Harabi, S. Zouai, S. Achour, and D. B. Assolant, “Effect of Stabilized Zr02, A203and T i 0 2 on Sintering of Hydroxyapatite,” Mater. Sci. Forum. 492-493 241-246 (2005). A. R. Kmita, A. Slosaryk, and Z. Paszkiewicz, “Mechanical Properties of HAp-Zr02 Composites,”J. Eur. Ceram. Soc., 26 1481-1488 (2006). C.-Y. Chiu, H. C. Hsu, and W. H. Tuan, “Effect of Zirconia Addition on the Microstructural Evolution of Porous Hydroxyapatire,” Ceram. Int., 33 715-718 (2007). R. P. Rana, S. K. Pratihar, and S. Bhattachalyya, “Effect of Powder Treatment on the Crystallization Behaviour and Phase Evolution of Alz03-High Z r 0 2 Nanocomposites,” J Mater. Sci., 41 7025-7032 (2006).
Progress in Nanotechnology: Processing
SYNTHESIS AND CHARACTERIZATION OF CHALCOGENIDE NANOCOMPOSITES J. Martin and G. S. Nolas Department of Physics, University of South Florida, Tampa, Florida 33620 ABSTRACT Lead Telluride (PbTe) nanocomposites were prepared by densifying 100 - 150 nm PbTe nanocrystal powders synthesized from a low temperature alkaline aqueous solution reaction in a high yield (> 2 grams per batch). Densification using Spark Plasma Sintering (SPS) successfully integrated nanostructure dispersions within a bulk matrix. We report the synthesis and low temperature transport measurements on novel chalcogenide nanocomposites, including resistivity, Seebeck coefficient, and thermal conductivity. INTRODUCTION Thermoelectric effects couple thermal and electric currents, allowing for the solid-state inter-conversion of heat and electricity. The thermoelectric figure of merit, ZT=POT/K,defines the effectiveness of different thermoelectric materials, where S is the Seebeck coefficient, o is the electrical conductivity, T is the absolute temperature, and K is the total thermal conductivity. Recent progress in a number of higher efficiency thermoelectric materials (room temperature ZT > 1) can be attributed to the nanoscale enhancement of thermoelectric properties. These new materials demonstrate increased Seebeck coefficient and decreased thermal conductivity due to the phenomenological properties of nanometer length scales, introducing both quantum confinement and interfacial scattering effects. One consequence of nanostructure is the increase of interfaces. These interfaces serve to scatter phonons more effectively than electrons and reduce the thermal conductivity. Additionally, the presence of interfacial energy barriers filters the carrier energy traversing the interface, restricting those energies that limit the mean carrier energy.' This increases the Seebeck coefficient, as its value depends on the mean carrier energies relative to those at the Fermi level. Experimental evidence in expitaxial films of n- and p-type lead chalcogenides verifies the enhancement potential of this mechanism.* The primary research into nanostructured enhancement of thermoelectric properties remains limited to thin films, heterostructures, and nanowires. For example, p-type BizTe3/Sb2Te3 10 angstrod50 angstrom supperlattice structures demonstrated a room temperature ZT = 2.4. The thermal conductivity in this system was reduced by a factor of 2 compared to other Bi2Te3 alloys.J Harman and co-workers reported a room temperature ZT = 1.6 in PbTe/PbTeSe quantum-dot superlattices (QDSLS)~that contain PbSe nanodots imbedded in a PbTe matrix. Kong and co-workers reported an enhancement in Si/Ge supper lattice^.^ Recently, Heremans and co-workers reported an increased Seebeck coefficient for PbTe with the inclusion of Pb precipitate nanostructures.6 One method to incorporate nanoscale dimensions into bulk materials is by ball-milling.7,s This procedure rapidly grinds powders to sub-micron dimensions. Ball-milled Si-Ge nanocomposites demonstrated an increased Seebeck coefficient and a reduced thermal conductivity.' Although the electrical conductivity also increased, the overall thermoelectric performance of the material was enhanced. Ball-milled PbTe materials have also demonstrated thermoelectric enhancement.l o However, the syntheses discussed leave suspect a direct correlation between nanostructure and thermoelectric enhancement, due to unaccounted lattice
Nanocomposites and Nanostructures
367
strain affects. This stresses the need to directly synthesize nanocrystals and incorporate them into bulk structures. SYNTHESIS AND CHARACTERIZATION Lead telluride nanocrystals were synthesized from the low temperature reaction of a selenium or tellurium alkaline aqueous solution and a lead acetate trihydrate solution.' Several experiments were performed to optimize product yield by varying both the alkaline and precursor lead acetate trihydrate concentrations. To prepare PbTe nanocrystals, the precursor solutions were prepared separately at 90" C by dissolving elemental Te in a 20 M KOH aqueous solution and by dissolving Pb(CH3C00)2-3H20 in distilled water. After 30 minutes, the lead acetate trihydrate solution was dripped into the rapidly stirring deep purple alkaline solution to immediately form PbTe nanocrystals. After 5 minutes, the reaction mixture was removed from the heat source and dilute HN03 was added to flocculate the nanocrystals. The solution was removed from the grayish-black precipitates that were subsequently washed 4 times with the dilute nitric acid solution, removing lead hydroxide impurities. Excess lead acetate trihydrate in the reaction favors the formation of easily removable impurities. The precipitates were then XRD analysis washed 4 times with distilled water and dried overnight in a fume hood. confirmed nanocrystal phase purity. This procedure reproducibly synthesizes 100 - 150 nm spherical PbTe nanocrystals, confirmed by TEM (Figure I), with a high yield of over 2 grams per batch. High yields of nanocrystals are required to synthesize practical nanocomposite materials. PbTe nanocrystals were subjected to Spark Plasma Sintering (SPS) to achieve 95 YO theoretical density. In the SPS procedure, a pulsed DC current conducts through both the graphite die and the sample under high pressure to minimize nanostructure deterioration. Prior to SPS, five batches of PbTe nanopowders were mixed together in a glass vial for each sample. The final densities are 7.67 g/cm3 (sample PbTel) and 7.75 g/cm3 (sample PbTe2). Densifying solely the nanocrystals results in a uniform dispersion of non-conglomerated nanostructure within a bulk matrix.
'
-
-
Figure 1. TEM image of PbTe nanocrystals after precipitation.
368
Progress in Nanotechnology: Processing
Figure 2. SEM image indicating the nanostructure of PbTel .
SEM (Hitachi S-800) images of a PbTel fracture surface indicate the preservation of nanostructure following the SPS procedure with grain diameters ranging from 100 nm to over one micron (Figure 2). Figure 3 shows the standard X-ray diffraction scans for the two PbTe nanocomposites post SPS. All samples exhibit peaks characteristic of PbTe. The successive spectra are normalized and shifted in intensity for clarity. While the PbTe nanopowder spectra indicates phase purity, the sintered nanocomposites exhibit a PbTe03 impurity phase. Densified nanocomposites of PbTe were cut into 1 x 2 x 5 mm parallelopipeds for transport property measurements. Four-probe resistivity, Seebeck coefficient, and steady-state thermal conductivity were measured on the same sample from 300 K - 12 K. Table 1 summarizes the room temperature data. Both samples are p-type with large room temperature Seebeck coefficients of 330 pV/K (Figure 4). The temperature dependence and the magnitude of the Seebeck share consistency throughout the measured temperature range. Transport measurements also demonstrate a correlation between resistivity and porosity (Figure 5). An increase of only 1.5 % theoretical density between PbTel and PbTe2 decreased the room temperature resistivity by a factor of 2. Since porosity has no affect on the Seebeck coefficient'2, increasingly denser nanocomposites should facilitate optimal thermoelectric performance. The thermal conductivity values also correlate to the difference in sample density. As shown in Table I, PbTel demonstrates a slightly lower thermal conductivity than PbTe2 due to the increased porosity.13 The room temperature ZT for PbTe2 is 0.1 without optimization or doping.
-
-
Table 1. Percent theoretical density, resistivity, Seebeck coefficient, and total thermal conductivitv at 300 K for the PbTel and PbTe2 nanocomDosites. Sample Density (%) S (pV/K) K (W-m-lK-') p (mOhm-cm) PbTe 1 94 328 2.2 24.9 PbTe2 95 324 2.5 12.6
Nanocomposites and Nanostructures
369
Figure 3. XRD spectra for the two PbTe nanocomposites post SPS procedure. Arrows indicate PbTe03 impurity.
Figure 4. Temperature dependence of Seebeck coefficient for the two PbTe nanocomposites. Closed circles identify PbTel and open triangles identify PbTe2.
370
Progress in Nanotechnology: Processing
Figure 5. Temperature dependence of resistivity for the two PbTe nanocomposites. Closed circles identify PbTel and open triangles identify PbTe2. CONCLUSIONS Dense PbTe nanocomposites were prepared by densifying PbTe nanocrystal powders synthesized from high yield (> 2 grams per batch) alkaline aqueous solution reactions. These nanocomposites dimensionally integrate nanostructure within a bulk matrix without conglomeration. Transport measurements indicate a strong correlation between resistivity and porosity, suggesting increasingly denser nanocomposites might hrther optimize the transport properties for thermoelectric properties. To our knowledge, these samples represent the first preparation of dense nanocomposites from solution-phase synthesized nanocrystals, including transport measurements. The authors acknowledge support by GM and DOE under corporate agreement DE-FC26-04NT42278. REFERENCES
B. Y. Moizhes and V. A. Nemchinsky, In Proceeding for the 1 lthInternational Conference on Thermoelectrics, Institute of Electrical and Electronics Engineers, Inc., 1992. Y. I. Ravich, In CRC Handbook of Thermoelectrics, edited by D. M. Rowe, pages 67-73, CRC Press New York, 1995. R. Venkatasubramanian, E. Siivola, T. Colpitts, B. O’Quinn, Nature 413, 597 (2001). 4 T. C. Harman, P. J. Taylor, M. P. Walsh, and B. E. LaForge, Science 297,2229 (2002). 5 T. Kong, S. B. Cronin, M. S. Dresselhaus, Applied Physics Letters 77, 1490 (2000). 6 J.P. Heremans, C.M. Thrush and D.T. Morelli, J. Appl. Phys. 98,063703 (2005). 7 K. Kishimoto and T. Koyanagi, J. Appl. Phys. 92,2544 (2002). 1
Nanocomposites and Nanostructures
371
P.C. Znai, W.Y. Zhao, Y. Li, X.F. Tang, Z.J. Zhang and M. Nino, Appl. Phys. Lett. 89, 0521 1 1 (2006). 9 M. S. Dresselhaus, G. Chen, M. Y. Tang, R. G. Yang, H. Lee, D. Z. Wang, Z. F. Ren, J. P. Fleurial and P. Gogna, Proc. Mater. Res. SOC.886 3 (2006). l o J. P. Heremans, C. M. Thrush, and D. T. Morelli, Thermopower Enhancement in PbTe nanostructures, Physical Review B 70, 115334 (2004). 11 W. Zhang, L. Zhang, Y. Cheng, Z. Hui, X. Zhang, Y. Xie, and Y. Qian, Materials Research Bulletin 35,2009 (2000). 12 L. Yang, J. S. Wu and L. T. Zhang, J. Alloys and Compounds 364,83 (2004). l 3 I.Sumirat, Y. Ando, S. Shimamura, J. Porous Mater. 13,439 (2006).
372
Progress in Nanotechnology: Processing
SELF ASSEMBLED FUNCTIONAL NANOSTRUCTURES AND DEVICES Cengiz S. Ozkan Department of Mechanical Engineering Bourns Hall, A305 University of California, Riverside Riverside, CA 92521 ABSTRACT This paper reports the self assembly of functional nanostructured materials including multi-walled Carbon Nanotube-Quantum Dot (CNT-QD) heterojunctions using the Ethylene Carbodiimide Coupling procedure (EDC). Thiol stabilized ZnS capped CdSe quantum dots containing amine terminal groups (QD-NH2) were conjugated with acid treated Multi-Walled Carbon Nanotubes (MWCNT) ranging from 400 nm to 4pm in length. SEM, TEM, EDS and FTIR were used to characterize the conjugation process. The versatile electrical properties of carbon nanotubes (CNT) make them promising candidates for nano electronic devices, especially tunneling diodes and transistors"*. In previous CNT based nano-electronic devices, control over electrical properties of the devices has been limited. In addition, researchers have relied on overlapping CNTs2 for forming junctions, which introduces local bending that could adversely affect the electrical properties of the tubes3. Alternatively, covalent modifications of carbon nanotubes with metal colloids4 and semiconducting quantum dots ( Q D s ) ~have , ~ been reported for the synthesis of heterojunctions for electronics applications. These approaches were associated with carbon nanotube sidewall functionalization. Here, the synthesis of heterojunctions with controlled conjugation of water stabilized, amine terminating, ZnS coated CdSe QDs (QD-NH2) to acid treated ends of multi walled CNTs (MWCNT) are reported. Scheme 1 illustrates the steps involved in the synthesis of the heterojunctions. Carboxyl groups were introduced to the ends of as-grown MWCNTs (Nanostructured & Amorphous materials, Inc., Los Alamos) by refluxing at 130 "C in concentrated nitric acid for 24 hours. The tips of MWCNTs, which have the highest defect sites, get oxidized first. Oxidized MWCNTs are shorter and have terminating carboxylic groups that impart a hydrophilic nature and facilitate further functionalization. ZnS capped CdSe QDs (Evident Technologies, Inc., New York) were used in the functionalization of the MWCNTs. The use of ZnS coating over the CdSe core improves the quantum yield by passivating7 surface dangling bonds (carrier trap sites) and also, enabling them for use in biosystems'. Scheme 1. Conjugation of ZnS capped CdSe QDs to MWCNTs. Nanocomposites and Nanostructures
373
To prepare water stabilized QDs (QD-NH2), ZnS capped CdSe nanocrystals were suspended in chloroform by sonication for 30 minutes. Equal volumes of 1.0 M 2-aminoethane thiol hydrochloride (AET) were added to this solution. This resulted in a two-phase mixture with the aqueous AET forming an immiscible layer above the organic chloroform-QD suspension. The mixture was stirred vigorously on a magnetic plate for 4 hours. When ZnS capped CdSe QDs were reacted with AET, the mercapto group in the thiol group in AET bonded to the Zn atoms and the amine groups rendered the QDs hydrophilic, in addition to facilitating further functionalization possibilities. The aqueous phase containing QD-NH2 was extracted in phosphate buffer saline (PBS, pH=8.5). Successive washing in PBS produced a good suspension of water stabilized QD-NH2 which were used for the synthesis of MWCNT-QD heterostructures via the two-step coupling procedure using 1-ethy1-3-(3-dimethyl-aminopropyl)carbodiimide HC1 (EDC, Pierce Chemicals, Inc.) in the presence of N-hydroxysuccinimide (sulfo-NHS, Pierce Chemicals, Inc.) The EDC reaction was carried out in PBS for 8 hrs at 50°C under continuous mixing. Characterizations of the heterostructures were conducted using scanning electron microscopy (SEM), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FTIR) and energy dispersive spectroscopy (EDS). Figure 1(A) is an SEM image of QDs conjugated at the end of a MWCNT. No sidewall functionalizations were observed. MWCNTs produce multiple carboxylic groups upon oxidation at their ends and these results in the conjugation of multiple QDs at the ends. Figure 1. (A) SEM image of water soluble QDs before conjugation. (B) SEM image of QDs conjugated at the ends of a MWCNT (4pm) (C) QDs at the ends of a MWCNT (500nm). Previous research6 indicated sidewall QD conjugations for MWCNTs exceeding 200 nm in length. Our conjugation technique is specific that even for MWCNTs as long as 4 pm (Figure 1 (B)), QDs are observed only at the ends. Figure 1(C) is a SEM image indicating QDs conjugated at the ends of a 500 nm long MWCNT. In all the images, we have observed the absence of side wall functionalization. Further TEM (Figure 2 (A) & (B)) and EDS ( Figure 2 (C)) confirms the presence of QDs at the ends of the MWCNTs. The TEM image in Figure 2(A) shows a MWCNT with QDs at its end. Figure 2 (B) is the image of the same QD cluster at a higher magnification. Figure 2(C) shows the EDS data for the QD cluster in Figure 2 (B) using a 1.50 nm probe size. The spectra shows clear peaks corresponding to the elements of Cd, Se, Zn, and S. Notice that, Cu signal is generated from the TEM grid. The potential argument for the clusters to be Fe nano particles (from the MWCNT growth process) instead of QDs can be excluded after observing absence of the characteristic Fe peaks in EDS data. The MWCNT-QD conjugates were also characterized with FTIR, using an AgCl cell in a Bruker Equinox 55 FTIR spectrometer. Figure 3 is the FTIR spectra of oxidized MWCNTs (blue curve) and the MWCNT-QD conjugates (red curve). With plain oxidized MWCNTs, absorption peaks are observed at 1644 cm-1, 1704 cm-1 and 3403 cm-1 (peaks designated A, B, and C), which are characteristic of carboxylic and phenolic groups on acid
374
Progress in Nanotechnology: Processing
treated MWCNTs. For the MWCNT-QD conjugates, new absorption peaks appear at 1653 cm-1, 2977 cm-1 and 3314 cm-1 (D, E, and F), which indicate the C=O stretch in amides, C-H stretch and N-H stretch in amides. C-H and N-H peaks are bigger than the amide C=O peak due to the presence of free QDs in the sample. Nevertheless, a slight blue shift of carboxylic C=O stretch to amide C=O stretch and the appearance of C-H and NH peaks indicate the formation of covalent MWCNT-QD conjugations, via the amide bond formation.
In conclusion, a method to synthesize heterojunctions of individual MWCNTs with QDs, preferentially at the MWCNT ends has been reported. This was confirmed by SEM, TEM and EDS analysis. Due to the mild and Figure 2. (A) SEM image of water soluble QDs well-controlled oxidation of the MWCNTs, before conjugation. (B) SEM image of QDs conjugation occurs only at the ends for conjugated at the ends of a MWCNT (4pm) nanotubes ranging from 400 nm to 4 pm in (C) QDs at the ends of a MWCNT (500nm). length. The controlled conjugation process preserves the electronic properties of the MWCNTs and enables the nanoassembly of heterojunctions. They can be used as building blocks for various nanoscale electronic or optoelectronic devices and three dimensional hierarchical assemblies of multilayered systems.
Figure 3. FTIR spectra of Oxidized MWCNTS (blue) and MWCNT-QD conjugates (red). Absorption Peaks are observed at 1644 cm-', 1704 cm-' and 3403 cm-I in the FTIR spectra for oxidized tubes. New peaks develop at 1653 cm-I, 2977 cm-l and 3314 cm-l in the FTIR spectra of MWCNT-QD conjugates, indicating formations of MWCNT-QD conjugates via amide bond formation.
NanocomDosites and Nanostructures
REFERENCES I (a) Tans, S. J.; Verschueren, A. R. M.; Dekker, C. Nature 1998, 393, 49. (b) Li, J.; Papadopoulos, C.; Xu, J. Nature 1999, 402, 253. (c) Yao, Z.; Postma, H. W. C.; Balents, L.; Dekker, C. Nature 1999, 402(6759), 273. (d) Ahlskog, M.;Tarkiainen, R.; Roschier, L.; Hakonen, P. Appl. Phys. Lett. 2000, 77, 4037. (e) Zhou, C. W.; Kong, J.; Yenilmez, E.; Dai, H. J. Science 2000, 290, 1552. ( f ) Ahlskog, M.; Hakonen, p.; Paalanen, M.; Roschier, L.; Tarkiainen, R. J. Low Temp. Phys. 2001, 124, 335. (g) Rosenblatt, S.; Yaish, Y.; Park, J.;
375
Gore, J.; Sazonova, V.; McEuen, P. L. Nan0 Lett. 2002,2,869. 2 Fuhrer, M. S.; Nygard, J.; Shih, L.; Forero, M.; Yoon, Y. G.; Mazzoni, M. S. C.; Choi, H. J.; Ihm, J.; Louie, S. G.; Zettl, A.; McEuen, P. L. Science 2000,288,494. 3Dai, H. J. Surface Sci. 2002, 500,218. 4(a) Liu, J.; Rinzler, A. G.; Dai, H.; Hafber, J. H.; Bradley, R. K.; Boul, P. J.; Lu, A.; Iverson, T.; Shelimov, K.; Huffman, C. B.; Rodriguez-Macias, F.; Shona, Y.-S.; Lee, T. R.; Colbart, D. T.; Smalley, R. E., Science 1998, 280, 1253. (b) Azamian, B. R.; Coleman, K.S.; Davis, J. J.; Hanson, N.; Green, M.L.H. Chem. Commun. 2000.4,366. 'Banerjee, S.; Wong S. S. Nan0 Lett. 2002,2, 195. 6Haremza, J. M.; Hahn, M. A.; Krauss, T. D. Nan0 Lett. 2002,2, 1253 7Hines, M. A.; Gnyotsionnest, P. J. Phys. Chem. 1996, 100,468. k h a n , W. C. W.; Nie S. M. Science 1998,281,2016.
376
Progress in Nanotechnology: Processing
CARBON NANOTUBE (CNT) AND CARBON FIBER REINFORCED HIGH TOUGHNESS REACTION BONDED COMPOSITES P. G. Karandikar, G. Evans, and M. K. Aghajanian M Cubed Technologies, Inc. 1 Tralee Industrial Park Newark, DE 19711
ABSTRACT Reaction bonded Sic and B4C offer low density (light weight), high hardness, high stiffness, high thermal conductivity, low CTE, excellent ballistic resistance, complex shape producibility and high volume producibility - properties needed in many armor, thermal management, semi conductor capital equipment, and aerospace mirrors and structures markets. However, similar to other ceramics, their toughness and strength are lower compared to incumbent metallic materials. Here we demonstrate the use of carbon fiber and carbon nanotube (CNT) reinforcements for increasing toughness of reaction bonded ceramics. The reaction bonding process depends on the reaction between carbon and molten silicon to achieve infiltration of particulate or fibrous preforms (e.g. Sic, B&, etc.). Thus, if unprotected, carbon fibers and CNTs will convert to Sic in this process. In this work, innovative processing was conducted to successfully incorporate both carbon fibers and carbon nanotubes in reaction bonded materials. Cf/SiC composites were obtained with quasi-isotropic low CTE (< 1 ppm/K) and high fracture toughness (6-10 MPa m112). Fracture toughness of reaction bonded Sic was increased from 4 to 7 MPa m112(a 73% increase) using CNTs. INTRODUCTION Reaction bonded Sic and B4C materials’-’ offer high specific stiffness (stiffnesddensity - E/p) and thermal stability (thermal conductivity / coefficient of thermal expansion - WCTE), and are used for applications such as lithography equipment, optical structures, mirrors, thermal management and aerospace components. Properties of several candidate materials that were historically considered for these applications are compared with those of M Cubed materials in Table 1. In some aerospace applications however, high toughness and strength are also required. Beryllium (Be) offers the best combination of specific stiffness, high strength and high toughness (8-10 MPa m ‘I2), and is therefore extensively used in aerospace structural components. However, Be has many limitations such as high cost, low thermal stability and above all, health hazards associated with Be dust. Reaction bonded Sic and B4C can compete with Be in terms of specific stiffness and have better thermal stability than Be. However, they suffer from the limitation of low (4MPa m ‘ I 2 ) toughness.
’
THE REACTION BONDING PROCESS Reaction bonded Sic was first developed in the 1940’~’-~. Other terms for the process include ‘reaction sintered’ and ‘self bonded”. Figure 1 shows a schematic of this process. In this process, the preform containing the reinforcement and a carbon precursor or binder is “carbonized” in an inert atmosphere above 600°C to convert the precursor to carbon. Next, the preform is placed in contact with Si metal or alloys of Si in an inert or vacuum atmosphere and
Nanocomposites and Nanostructures
377
heated to above the melting point of the alloy. Due to the spontaneous wetting and reaction between carbon and molten Si, the preform is infiltrated completely. The carbon in the preform reacts with the Si forming Sic and in the process bonds the reinforcement together. Some residual Si remains. Table 1.
Summary of properties of selected materials for aerospace applications.
p = Density, a = CTE (-50 to 100°C), k = Thermal Conductivity, E = stiffness, Krc = Fracture Toughness
Figure 1. A schematic of the reaction bonding process
378
Progress in Nanotechnology: Processing
A major advantage of this process is that the volume of the reaction-formed Sic is 2.3 times larger than the volume of the reacted carbon. Thus, by infiltrating Si into preforms that contain high carbon contents, ceramic bodies rich in Sic can be produced. In addition, like water, silicon expands on freezing. As a result, unlike other materials made by casting-type processes, RB composites do not show any shrinkage porosity. The reaction bonding process has several advantages relative to traditional ceramic processes (e.g., hot pressing, sintering). First and foremost, volume change during processing is very low (generally well less than l%), which provides very good dimensional tolerance control and eliminates the need for final machining. In addition, the process requires relatively low process temperatures and no applied pressure, which reduces capital, tooling and operating costs. Moreover, fine high surface area powders capable of being densified are not required, which reduces raw material cost. In the earlier work5 on reaction bonded B4C it was shown that an unwanted reaction between the Si and B4C phases would often occur during the infiltration process, which resulted in cracking of the parts. To prevent excessive reaction from occurring, coarse particles (e.g., 300 microns) were used. This limited the surface area for reaction, but also limited the strength of the resultant material. M Cubed has fbrther optimized the reaction bonding process6-", to produce relatively fine grained Sic and B4C ceramics with favorable mechanical and ballistic properties as well as good machinability. In the specific case of reaction bonded B4C, the unwanted reaction between Si and B& was suppressed by alloying the Si infiltrant with boron (B). The binder formulation was optimized to yield preforms with high strength to allow intricate green machining using CNC machines. Preform bonding technology was developed to allow fabrication of complex shapes, ribbed structures, box structures, cooling channels etc. The shrinkage from the preform stage to final infiltrated stage was reduced to less than 0.5%, allowing net-shape component fabrication with minimal finish machining (that too only for high precision components). This process technology allows fabrication of intricate large structures (500 lb, 1.5m x 0.75 m x 0.3 m), as well as mass production of smaller components (10s of thousand per month). In spite of these refinements, one limitation of these materials remains, namely their lower fracture toughness than Be. As a result, their applications have been limited to stiffness controlled applications and they are not viable for applications needing high strength and toughness. Therefore, the present work pursued two approaches for enhancing the fracture toughness of reaction bonded materials: ( 1) Carbon fiber reinforcement, (2) carbon nanotube (CNT) reinforcement. CARBON FIBER REINFORCED RB SIC
Fiber reinforcement (e.g. Sic, A1203,and carbon) has been shown to increase toughness of ceramics'*. Out of all the fibers, carbon fibers are the most cost competitive due to their wider-scale usage in polymeric composites (e.g. graphite-epoxy). Other fibers (e. g. Sic, A1203) have proven to be cost prohibitive and their use is limited to very high temperature applications (>lOOO"C) where carbon fibers cannot be used due to their oxidation problem. Continuous carbon fiber based composites are typically anisotropic and laminate designs are used for making most of the components. The laminates are typically quasi-isotropic (i.e. the properties are uniform in the plane of the laminate). As described earlier, the reaction bonding process depends on the good wetting and chemical reaction between carbon and molten silicon. This wetting and reaction lead to spontaneous infiltration of preforms containing carbon by molten silicon. For processing carbon fiber reinforced Sic composites by reaction bonding, the key challenge is prevention of attack on
Nanocornposites and Nanostructures
379
the carbon fibers by molten silicon and their conversion to Sic. In the p a ~ t ' ~this - * problem ~ was overcome using carbon, Sic or BN coatings deposited on the fibers in a separate process step by techniques such as chemical vapor deposition (CVD) or chemical vapor infiltration (CVI). Such processes are time consuming and expensive. M Cubed has d e ~ e l o p e d in-situ '~ formed coatings that protect carbon fibers from attack by molten silicon. Figure 2a shows a micrograph of a woven Cf/SiC composite made by reaction bonding and using the in-situ formed protective coating. Energy dispersive analysis of X-rays (EDAX) was conducted on the fibers in this microstructure. Figure 2b shows an EDAX pattern from one such fiber. In this pattern, predominantly carbon is seen with very minor amount of Si. Thus, the fiber was not converted to Si. Typically, the EDAX pattern of Sic predominantly shows Si because in the presence of silicon, carbon X-rays get absorbed and do not reach the detector. Thus, the fact that the EDAX pattern from the fiber shows predominantly carbon is very significant in proving that the fiber was not converted to Sic. A dense carbon coating formed on the fiber surface during the initial processing steps is postulated to form a dense, protective Sic coating on the fiber surface during silicon infiltration, which prevents further attack by molten silicon and complete conversion to silicon carbideI4'l6.
Figure 2. (a) Microstructure of CdSiC composites (b) EDAX analysis from fibers shows predominantly carbon indicating that the fibers are protected from reacting with molten silicon.
To-date several different Cf/SiC composites have been made with different fibers, fiber architectures and in-situ formed coatings. The focus of the present development effort was obtaining composites with quasi isotropic low CTE and high toughness for some specific aerospace applications. Typically 2 x 2 x 0.125 and 6 x 6 x 0.125 inch panels were fabricated. The Cf/SiC composite plates were subjected to extensive microstructural, physical, thermal and mechanical characterization. The densities of the test samples were determined using the Archimedes principle per ASTM C373. For microstructural observations, specimens were sectioned and polished and observed by optical and scanning electron microscopes (SEM). Specimen elastic moduli were measured using strain gages in a tensile test. The flexural strengths were determined using a four-point bend testing apparatus per ASTM C 1161. Fracture toughness measurements were made on selected specimens by the Chevron notch method
380
Progress in Nanotechnology: Processing
(ASTM C1421). Strength and toughness measurements were made on at least 5 specimens of each type. CTE was measured by laser interferometry. Conventional dilatometry techniques are not sufficiently accurate for measuring CTEs of low expansion materials. Thermal conductivities were measured by the axial rod method. The measured properties of these composites are summarized in Table 2. All three composites on which toughness measurements were made show toughness in excess of 6 MPa m”. The data also shows that very low CTE (< 1 ppm/K) Cf/SiC composites were successfully produced. It also shows that the CTE of the composite can be controlled by changing the fiber type (pitch based, PAN based, high modulus, high strength, extent of graphitization etc.) and architecture (unidirectional, cross ply, quasi-isotropic, woven etc.). The ultimate bend strength (UBS) and the modulus (E) of each composite are controlled by the fiber architecture, interface type (fiber pull out) and the extent of fiber-silicon reaction.
Table 2.
Properties of C$SiC composites made by reaction bonding for low CTE applications
IUni-directional
E
2.22
312
9.7
--
-1.02
y: 122 --
*The interface designations refer to various proprietary in-situ coatings. Several components have been fabricated from these composites. Figure 3 shows a photo of a lightweight mirror made out of Cf/SiC. Work continues to further enhance the toughness and strength of these composites.
Figure 3. Photos of the back and the front of a lightweight Cf/SiC mirror.
Nanocomposites and Nanostructures
381
ADVANTAGES OF CNT REINFORCEMENT AND PROCESSING CHALLENGES
Carbon nanotubes (CNTs) were first observed by Iijima in 1991. Since then, a significant effort has been spent in trying to make and characterize n a n ~ t u b e s ' ~The - ~ ~properties . of two typical carbon fibers are compared with the properties of carbon nanotubes in Table 3. Clearly, the properties of CNTs are significantly better than those of carbon fibers. Therefore, CNTs offer a very high reinforcing potential.
Table 3.
Comparison of properties of carbon nanotubes with the properties of carbon fibers.
I
Property
Diameter (micrometer) Density (g/cc) Elastic Modulus (GPa) Ultimate Tensile Strength (GPa) Thermal Conductivity (w/mK) CTE (ppm/K) Electrical Resistivity (micro-ohm-m)
Carbon Fibers Carbon Nanotubes T300 P120 1) (pircn oased) 7 10 0.05 1.76 2.17 -2.0 23 1 827 1000-1400 3.75 2.41 7-10 8 640 >2000 -0.6 -1.45 - 1 (isotropic) 18 2.2 95 (2002). ”M. Mamak, N. Coomhs. and G. Ozin. “Self-Assembling Solid Oxide Fuel Cell Materials: Mesoporous Yttna-Zirconia and Metal-Yttria-Zirconia Solid Solutions,” J . Am. Chem. Soc.. 122, 8932-9 (2000).
Progress in Nanotechnology: Processing
“G. Pang, S. Chen, Y. Zhu, 0. Palchik, Y. Koltypin. A. Zaban. and A. Gedanken, “Preparation and Characterization of Monodispersed YSZ Nanocrystals,” J. Phys. Chem. B, 105. 4647-52 (2001). I7E. L. Crepaldi. G. J. de, A. A. Soler-Mia, A. Bouchara, D. Grosso, D. Durand. and C. Sanchez, “Controlled Formation of Highly Ordered Cubic and Hexagonal Mesoporous Nanocrystalline Yttria-Zirconia and Ceria-Zirconia Thin Films Exhibiting High Thermal Stability,” Angew. Chem. In!. Ed.. 42 [3] 347-51 (2003). ‘“G. Soyez, J. A. Eastman. L. J. Thompson, G.-R. Bai, P. M. Baldo, A. W. McCormick. R. J. DiMelfi, A. A. Elmustafa, M. F. Tambwe, and D. S. Stone, “Grain-Size-Dependent Thermal Conductivity of Nanocrystalline YttriaStabilized Zirconia Films Grown by Metal-Organic Chemical Vapor Deposition,” Appl. Phys. Lett., 77, 1155-7 (2000). ’X. Zhang, X. N. Jiang, and C. Sun, “Micro-Stereolithography of Polymeric and Ceramic Microstructures.” Sens. Actuators A. Phys., 77, 149-56 (1999). 20M.I. Cima, M. Oliveira, H.-R. Wang, E. M. Sachs, and R. Holman, “SlurryBased 3DP and Fine Ceramic Components”; pp. 21623 in Solid Freeform Fabrication Symposium Proceedings, Edited by J. J. Beaman. H. L. Marcus, D. L. Bourell, 1. W. Barlow, and T. Crawford. University of Texas, Austin, TX, 2001. ”8. A. Vojak, S.-J. Park, C. 1. Wagner, J. G . Eden, R. Koripella, J. Burdon, F. Zenhausern, and D. L. Wilcox, “Multistage, Monolithic Ceramic Microdischarge Device Having an Active Length of 0.27 mm,” Appl. Phys. Lett., 78, 1340-2 (2001). *’X. Zhao, J. R. G. Evans. M. J. Edirisinghe, and J.-H. Song, “Direct Ink-Jet Printing of Vertical Walls,” J . Am. Ceram. Soc., 85, 21 13-5 (2002). ”J. E. Smay. G. M. Gratson. R. F. Shepherd, J. 111 Cesarano, and J. A. Lewis, “Directed Colloidal Assembly of 3D Periodic Structures,” Adv. Mater., 14, 1279-83 (2002). “H. A. Lowenstam and S. Weiner, “Mineralization by Organisms and the Evolution of Biomineralization,” p. 191 in Biomineraliiolion and Biological Metal Accumulation, Edited by P. Westbroek and E. W. D. de long. Reidel Publishing Co, Hingham, MA, 1983. ”C. C. Perry, S. Mann. R. J. P. Williams, F. Watt, G. W. Grime, and J. Takacs, “A Scanning Proton Microprobe Study of Macrohairs from the Lemma of the Grass Phuluris Canariensb L,” Proc. R. Sac. 8. Biol. Sci.,222 [I2291 43945 (1984). %. Mann and G. A. Ozin, “Synthesis of Inorganic Materials with Complex Form,” Nature, 382, 3 13-8 ( I 996). ”5. Aizenberg, A A. Tkachenko, S. Weiner, L. Addadi, and G. Hendler, “Calcitic Microlenses as Part of the Photoreceptor System in Brittlestars,” Nature, 412 [6849] 819-22 (2001). 28 J. C. Weaver, L. 1. Pietrasanta, N. Hedin, B. F. Chmelka, P. K. Hansma, and D. E. Morse, “Nanostructural Features of Demosponge Biosilica.” J. Struct. Biol., 144, 271-81 (2003). ’9.C. Sundar, A. D. Yablon, J. L. Grazul, M. Ilan, and J. Aizenberg, “Fiberoptical Features of a Glass Sponge,” Nature, 424 [6951] 899-900 (2003). ’OE. Bauerlein, “Biomineralization of Unicellular Organisms: An Unusual Membrane Biochemistry for the Production of Inorganic Nano- and Microstructures,” Angew. Chem. Int. Ed., 42, 614-41 (2003). ”F. E. Round, R. M. Crawford, and D. G. Mann, The Diatoms: Biology & Morphology ofthe Genera. Cambridge University Press, Cambridge, U.K., 1990. ’%. A. Crawford, M. 3. Higgins, P. Mulvaney, and R. Wetherbee, “Nanostructure of the Diatom Frustule as Revealed by Atomic Force and Scanning Electron Microscopy,” J. Phycol., 37, 543-54 (2001). ”M. Hildebrand and R. Wetherbee, “Components and Control of Silicification in Diatoms”; pp. 11-57 in Progress in Molecular and Subcellular Biology, Vol. 33, Edited by W. E. G. Muller. Springer-Verlag. Berlin, 2003. ’‘V. Martin-Jezequel, M. Hildebrand, and M. A. Brzezinski, “Silicon Metabolism in Diatoms: Implications for Growth,” J . Phycol., 36, 82140 (2000).
Nanocomposites and Nanostructures
’9.Lebeau and J. M. Robert, “Diatom Cultivation and Biotechnologically Relevant Products. Part I: Cultivation at Various Scales,” Appl. Microbiol. Biotechnol., 60,612-23 (2003). 36D. Werner, “Introduction with a Note on Taxonomy”; pp. 1-17 in The Biology oj’Diat0m.v. Edited by D. Werner. University of California Press, Berkeley, CA, 1977. 37 J. Parkinson and R. Gordon. “Beyond Micromachining: The Potential of Diatoms,” Trends Biotechnol., 17, 190-6 (1999). ”J. Zhao, C. S. Caddis, Y. Cai, and K. H. Sandhage, “Free-Standing Microscale Structures of Nanocrystalline Zirconia Nanocrystals with Biologically Re licable 3-D Shapes,” J. Mater. Res.. 20 (21 282-7 (2005). ‘K. H. Sandhage, M. B. Dickerson, P. M. Huseman, M. A. Caranna, J. D. Clifton, T. A. Bull, T. J. Heibel. W. R. Overton. and M. E. Schoenwaelder, “Novel, Bioclastic Route to Self-Assembled, 3-D, Chemically Tailored Meso/ Nanostructures: Shape-Preserving Reactive Conversion of Biosilica (Diatom) Microshells,” Adv. Mater., 14, 429-33 (2002). 40 K. H. Sandhage, M. B. Dickerson, P. M. Huseman, F. M. Zalar. M. R. Rondon, and E. C. Sandhage, “A Novel Hybrid Route to Chemically-Tailored, ThreeDimensional Oxide Nanostructures: The BaSlC (Bioclastic and Shape-Preserving Inorganic Conversion) Process,” Ceram. Eng. Sci. Proc., 23 [4] 6 5 3 4 (2002). 4’R. R. Unocic, F. M. Zalar, P. M. Sarosi, Y. Cai, and K. H. Sandhage. “Anatase Assemblies from Algae: Coupling Biological Self-Assembly of 3 - 0 Nanoparticle Structures with Synthetic Reaction Chemistry,” Chem. Commun..(71 7 9 5 4 (2004). 4zP. Joubert and B. Gaudreau, “Zirconium(1V) Oxyfluorides,” Rev. Chim. Miner., 12 [4] 289-302 (1975). 431. Barin, Thermochemical Data ofPure Substances, Vol. 11, 3rd edition, VCH Verlagsgesellschaft mbH, Weinheim, Germany, 1995. “K. V. Gourishankar, M. Karaminezhad Ranjbar, and G. R. St.Pierre, “Revision of the Enthalpies and Gibbs Energies of Formation of Calcium Oxide and Ma nesium Oxide,” J. Phase Equilibria. 14 [5] 601-1 1 (1993). 4BR. C. Weast, Handbook ofchembtry and Physics, 55th edition, pp. B-1054. CRC Press, Cleveland, OH, 1974. 4%’. A. Tverskov and 1. S. Morozov, “System ZrC14-MgC12-KCI”; p. 340 in Phase Diagramsfor Ceramists 1969 Supplement, Vol. 11. Edited by E. M. Levin. C. R. Robbins, and H. F. McMurdie. The American Ceramic Society, Westerville. OH, 1985. 47B. D. Cullity, Elements of X-Ruy DifJraction, 2nd edition, p. 101. AddisonWesley Publishing Co, Reading, MA, 1978. 48E. Karapetrova. R. Platzer. J . A. Gardner, E. Tome, I. A. Sommers, and W. E. Evenson, “Oxygen Vacancies in Pure Tetragonal Zirconia Powders: Dependence on the Presence of Chlorine During Processing,” J. Am. Ceram. Sac., 84 [I] 65-70 (2001). 49D M. Pasquevich, F. Lovey, and A. Caneiro, “Structural and Microstructural Changes in Zirconia in Dilute Chlorine Atmosphere.” J. Am. Ceram. Sac., 72 [9] 1664-7 (1989). 50 E. V. Armbrust, J. A. Berges, C. Bowler, B. R. Green, D. Martinez, N. H. Putnam, S. Zhou, A. E. Allen, K. E. Apt, M. Bechner, M. A. Brzednski, B. K. Chaal, A. Chiovitti, A. K. Davis, M. S. Demarest, J. C. Detter, T. Glavina, D. Goodstein, M. Z. Hadi, U. Hellsten. M. Hildebrand. B. D. Jenkins, J. Jurka, V. V. Kapitonov, N. Kroger, W. W. Y. Lau, T. W. Lane, F. W. Larimer, J. C. Lippmeier, S. Lucas. M. Medina, A. Montsant. M. Obornik, M. S. Parker, B. Palenik, G. J. Pazour, P. M. Richardson, T. A. Rynearson, M. A. Saito, D. C. Schwartz, K. Thamatrakoln, K. Valentin, A. Vardi. F. P. Wilkerson, and D. S. Rokhsar, “The Genome of the Diatom Thalassiosira Pseudonana: Ecology, Evolution, and Metabolism,” Science, 306,79-86 (2004). ”L. A. Zaslavskaia, J. C. Lippmeier, P. G. Kroth, A. R. Grossman, and K. E. Apt, “Transformation of the Diatom Phneodacty/um Tricornutum (Bacillarrophyceae) with a Variety of Selectable Marker and Reporter Genes,” J . Phycol., 36,379-86 (2000).
479