Ceramic Nanomaterials and Nanotechnology
Related titles published by The American Ceramic Society: Progress in Nanotechnology
0 2002, ISBN 1-57498- 169-2 Innovative Processing and Synthesis of Ceramics, Glasses, and Composites VI (Ceramic Transactions Volume I35)
Edited by Narottam f? Bansal and ].I?Singh 0 2002, ISBN 1-57498- I50- I Advances in Ceramic Matrix Composites Vlll (Ceramic Transactions Volume I 39)
Edited by ].I? Singh, Narottam F? Bansal, and Mritunjay Singh 0 2002, ISBN 1-57498- 154-4 Innovative Processing and Synthesis of Ceramics, Glasses, and Composites V (Ceramic Transactions Volume I 2 9) Edited by Narottam F? Bansal and J.f? Singh 0 2002, ISBN 1-57498- 137-4 Advances in Ceramic Matrix Composites VII (Ceramic Transactions Volume 128) Edited by Narottam F? Bansal, J.f? Singh, and H.-T. Lin 0 200 I, ISBN 1-57498- 136-6 Boing-Boing the Bionic Cat and the jewel Thief By Larry L. Hench 0 200 I,ISBN 1-57498-I29-3 The Magic of Ceramics By David W. Richerson 0 2000, ISBN 1-57498-050-5 Boing-Boing the Bionic Cat By Larry L. Hench 0 2000, ISBN 1-57498- 109-9 Ceramic Innovations in the 20th Century Edited by John B.Wachtman Jr. 0 1999, ISBN 1-57498-093-9
For information on ordering titles published by The American Ceramic Society, or to request a publications catalog, please contact our Customer Service Department at 6 14-794-5890 (phone), 6 14-794-5892 (fax),<
[email protected]> (e-mail), or write to Customer Service Department, 735 Ceramic Place, Westerville, OH 4308 I, USA. Visit our on-line book catalog at <www.ceramics.org>.
ceramic Transactions Volume 137
Lerarnic Nanomaterials and Nanotechnology
Proceedings of the Nonostructured Materials and Nonotechnology Symposium held at the 104thAnnual Meeting ofThe American Ceramic Society,April 28-May I, 2003 in S t Louis, Missouri.
Edited 6y Michael Z. Hu Oak Ridge National Laboratory
Mark R. D e Guire Case Western Reserve University
Published by The American Ceramic Society 735 Ceramic Place Westerville, Ohio 4308 I www.ceramics.org
Proceedings ofthe Nanostructured Materials and Nanotechnology symposium held at the 104th Annual Meeting of The American Ceramic Society,April 28-May I , 2003 in S t Louis, Missouri.
Copyright 2003,The American Ceramic Society. All rights reserved. Statements of fact and opinion are the responsibility of the authors alone and do not imply an opinion on the part of the officers, staff,or members ofThe American Ceramic SocietyThe American Ceramic Society assumes no responsibility for the statements and opinions advanced by the contributors to its publications or by the speakers at its programs. Registered names and trademarks, etc., used in this publication, even without specific indication thereof, are not to be considered unprotected by the law.
No part of this book may be reproduced,stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical,photocopying,microfilming, recording,or otherwise, without prior written permission from the publisher: Authorizationt o photocopy for internal or personal use beyond the limits of Sections 107 and I08 of the US. Copyright Law is granted by the American Ceramic Society, ISSN 1042- I I22 provided that the appropriate fee is paid directly to the Copyright Clearance Center; Inc.,222 Rosewood Drive, Danvers, MA 0 I923 USA,www.copyright.com.Prior to photocopying items for educational classroom use, please contact Copyright Clearance Center; Inc. This consent does not extend to copying items for general distribution or for advertising or promotional purposes or to republishing items in whole or in part in any work in any format. Please direct republication or special copying permission requests to the Senior Director; Publications,TheAmerican Ceramic Society, PO Box 6 136,Westerville, Ohio 43086-6 136, USA. Cover photo: High-resolution scanning electron micrograph of single-wall carbon nanotubes dispersed by ethanol is courtesy of Gudlong Zhan,loshua Kuntz,julin Wan, lavier Garay, and A.K. Mukherjee, and appears as figure I b in their paper “Single-Wall Carbon Nanotubes reinforced Alumina Nanocomposites Consolidated by Spark-Plasma-Sintering,” which begins on page I 6 I .
For information on ordering titles published byThe American Ceramic Society, or to request a publications catalog, please call 6 14-794-5890. Printed in the United States ofAmerica.
4 3 2 1-05 04 03 02 ISSN 1042-1 122 ISBN 1-57498-152-8
Preface..
............................................
vii
Synthesis and Processing of Nanomaterials Nucleation and Growth Mechanism of Silicalite- I Nanocrystal During MolecularlyTemplated Hydrothermal Synthesis. . . . . . . . . . 3 L. Khatri, M.T. Harris, M.Z. Hu, E.A. Payzant,and L.F.Allard,Jr:
Synthesis and Characterization of Mono-sized SiO, Ceramic Particles in Meso-Structure ..............................
23
T.-W. Chen and W,-C. J, Wei
Early Stage Silica Formation in Methanol and Ethanol . . . . . . . . . . 33 D.L. Green, M.T. Harris,J.S. Lin,Y-F. Lam, S. Jayasundara,and M.Z. Hu
Nanoparticles in Chip Miniaturization ......................
59
X. Feng, M.S. H. Chu, H.E. Rast, and B.C. Foster
Synthesis and Characterization of Nanoparticles of Stabilized Zirconia ....................................
75
C.R. Foschini, B.D. Stojanovic, C.O. Paiva-Santos, L. Perazolli, and J. A.Varela
Metal Oxide Particle Synthesis by Electric-Field Induced Water-in-Oil Emulsions ..........................
83
T.L.Terry, R.T. Collins, and M.T Harris
Synthesis and Characterization of &Sic Nanowires and h-BN Sheathed B-Sic Nanoanocable ...................... K. Saulig-Wenger, D. Cornu, F. Chassagneux, G. Ferro, t? Miele, and T. Epicier
.93
Virtual Processing of Advanced Nanomaterials . . . . . . . . . . . . . . I 0 I L.A. Pozhar,V.F.de Almeida, and M.Z. Hu
Nanocomposites Synthesis of Nanostructured WC/Co Powders through an Integrated Mechanical and Thermal Activation Process . . . . . . . . I29 L.L. Shaw, R. Ren, Z. Ban, and Z.Yang
V
Synthesis of Nanocrystalline Ni Coatings Reinforced with Ceramic Nanoparticles ...............................
143
J. He and 1.M. Schoenung
Single-Wall Carbon Nanotubes ReinforcedAlumina Nanocomposites Consolidated by Spark-Plasma-Sintering.. . . . . 161 G.-D. Zhan, J. Kuntz,J. Wan, J. Gamy, and A.K. Mukerjee
Processing and Microstructure of A I,O,/TiO, Nano-Composite from Plasma-SprayedA I,TiO, ........................... I 7I J. Wan, G-D. Zhan, A.K. Mukherjee, B.H. Kear Microstructural Characterization of Silicon Nitride/Boron Nitride Nanocomposites ..................................... I 8I T. Kusunose, H. Kondo,YYamamoto, M.Wada,T.Adachi,T. Sekino, TI Nakayama , and K. Niihara
Characterization of Fiber Coatings and Glass Composite Interfaces I89 by Atomic Force Microscopy (AFM) ...................... I? Fehling, D.Hulsenberg,Th. Mache
A Study on The Processing of Oxide-Based Nanocomposites. . . I97 S. Mullens,J. Cooymans, C. Smolders, and J. Luyten
Nanoscale Phenomena in Glasses, Glass-Ceramics, and Glass4ontaining Composites Interstitial Nanostructures in Engineered Silicates . . . . . . . . . . . . 209 L.I?Davila, S.H.Risbud, and J.F. Shackelford
Recent Advances in LAS-Glass Ceramics. . . . . . . . . . . . . . . . . . .22 I W. Pannhorst
Nanophase Formation in Different Glass-Ceramic Systems. . . . . 235 M. Schweiger;W. Holand, and V. Rheinberger
Uniaxial Plastic Deformation in Zirconia-Based Nanocrystalline Ceramics Containing a Silicate Glass. ..................... 245 R. Chaim, R. Ramamoorthy,A. Goldstein, I. Eldror; and A. Gurman
Transparent Gallate Spine1 Glass-Ceramics . . . . . . . . . . . . . . . . . 265 L.R. Pinckney 6.N. Samson, G.H. Beall,j. Wang, and N.F. Borrelli Electronic Structure of Interfaces in Cu-Glass Nanophase Composites .............................. .277 M.Backhaus-Ricouk,M.-F.Trichet,F. Maurel,A. Dezellus,L. Samet, D. lmhoff Index.............................................. v1
293
Nanotechnology - the creation and utilization of functional materials, devices and systems with dimensions on the order of 0.1 t o 100 nanometers, and exhibiting novel properties and functions - has become a major national, and indeed international,research initiative. Much of the promise of nanotechnology will not be possible without continued advances in the synthesis of nanostructured materials, combined with progress in analytical and physical characterization techniques capable of probing phenomena at this length scale.Applications are foreseen in medicine, electronics, structural materials, catalysis, fluid separations, power generation, environmental management, and materials design. Nanomaterials are the foundation for nanotechnology. Ceramics such as oxides, non-oxides (nitrides, carbides, etc.), and their composites represent a significant category of materials with great impact upon many applications.The synthesis, processing, and characterization of ceramic nanomaterials (or nanoceramics,for short) has thus become fundamentally important subjects for research and development. This is the first book focusing on nanoceramics and related nanomaterials, ranging from precursor nanoparticles and coatings t o nanocomposites. This volume of CeramicTransactions is based on papers presented and submitted t o the Symposium C2 on Nanostructured Materials and Nanotechnology, which was held during the 104th Annual Meeting of The American Ceramic Society (ACerS),April 28-May I , 2002 in St. Louis, Missouri.The symposium consisted of 65 contributions (52 oral presentations and I3 posters) spanning the entire three days of the meeting. Reflecting the truly international character of the symposium and of nanoceramics research, this compilation contains papers with authors from nine countries and five continents. While the coverage of this book is by no means exhaustive, the papers presented in this volume represent major current topics in nanoceramics-related research. The papers here are organized into three chapters reflecting these major topics: Synthesis and Processing of Nanomaterials (in the form of nanoparticles, coatings, nanowires, and fi bers); Nanocomposites (processing, properties, and characterization); and Nanoscale Phenomena in Glasses, GlassCeramics, and Glass-Containing Composites.
It is hoped that this book will not only become a useful reference for scientists and engineers interested in nanostructured ceramics and ceramic-based nanomaterials, but also will serve as a stimulus for interdisciplinary collaboration that
is important for the advancement of nanotechnology. The symposium on Nanostructured Materials and Nanotechnology was sponsored by ACerS and the Basic Science Division of ACerS, with cooperation from the Engineering Ceramics, Electronics, and Glass and Optical Materials Divisions of ACerS.The editors express special thanks t o the co-organizers of this symposium, Professor Masahiro Yoshimura (Tokyo Institute of Technology, Japan) and Wolfram Holand (Ivoclar Vivadent AG, Liechtenstein). Dv:Holand was especially instrumental in initiating, organizing, and chairing the session on nanophase formation/structure in glasses and glass ceramics (most of these presentations appear as papers in the third section of this volume). Prof. Yoshimura’s influence was felt throughout the symposium, particularly in the areas of innovative synthesis and processing of nanomaterials, where his group contributed several outstanding presentations. Finally, the editors thank all of the authors who contributed manuscripts to, or who assisted with reviewing manuscripts for; this volume.
Michael Z. Hu Mark R. De Guire
9..
VIU
Synthesis and Processing of Nanomaterials
This page intentionally left blank
NUCLEATION AND GROWTH MECHANISM OF SILICALITE-1 NANOCRYSTAL DURING MOLECULARLY TEMPLATED JWDROTHERMAL SYNTHESIS Lubna Khatri Department of Chemical Engineering 21 13E Chemical & Nuclear Eng. Build. University of Maryland College Park, MD 20742
Michael T. Harris School of Chemical Engineering 1283 CHME Building Purdue University West Lafayette, IN 47907-1283
Michael Z. Hu*, E. Andrew Payzant, and Lawrence F. Allard Jr. Oak Ridge National Laboratory *Bldg.4500N, MS-6 18 1 Oak Ridge, TN 37931 *Correspondingauthor ABSTRACT Nano-/micro-sized zeolite particles with narrow size distribution are important building-block materials for nanofabrication and many applications. Template-directed self-assembly, nucleation and growth of zeolite crystal particles f?om solutions are not fully understood. This study is aimed at understanding the early-stage nucleation process of silicalite-1 nanocrystals, using both in-situ and ex-situ measurement techniques. In alkaline aqueous solutions of sodium silicate, tetrapropyl ammonium hydroxide (TPA) is used as organic templating molecule to assist silicalite crystallization. TPAsilicalite-1 particles have been synthesized under various hydrothermal conditions with incubation temperatures ranging from 1OOOC- 180°C. The changes in the morphology, size and crystal structure of the particles formed during synthesis are carefblly monitored. An X-ray diffraction (XRD) technique and a high-resolution transmission electron microscope (HRTEM) with a cryogenic holder were successfully used to study the gradual nucleation and growth of zeolite (i.e., TPA occluded silicalite-I) crystals in solutions. Amorphous gel particles (- 50 nm), the first solid phase evolved in the solution, were for the first time imaged by HRTEM. Further hydrothermal
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication,reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
3
processing of the gel particles in solutions led to the formation of nanosized crystal nuclei, which were embedded in the amorphous gel. This observation supports the gel-to-crystal transformation mechanism for silicalite crystal evolution and growth.The Fast Fourier Transforms (FFT) of the TEM images provided complementary information to XRD on the crystallinity of transient forms from gel to fully-grown crystal. An in-situ high temperature X-Ray diffraction (HTXRD) was used for real-time monitoring of crystallization from gel precursors under progressive heating-up conditions (30-12OO0C). The HTXRD spectral patterns proved that zeolite crystal nucleation from gel is more than a thermally induced crystallization phenomenon because the dried amorphous gel failed to crystallize into silicalite during high-temperature processing. However, further silicalite crystal growth was observed in dried gels containing coexisting nanocrystal seeds. INTRODUCTION Zeolites, which are generally microporous crystalline aluminosilicates with framework of corner-sharing TO4 tetrahedra (in which the T-sites are occupied either by silicon or aluminurn), have rather complex but precisely repetitive atomic network with submicroscopic channel or pores typically 3 to 10 A in size112.Their name was coined in 1756 by a Swedish chemist and mineralogist, Axe1 Fredrick Cronstedt (1722-1765), whch was derived fiom the Greek words meaning, “boiling stones.” 334 Zeolites have outstanding characteristics that lead to diverse and widespread uses such as catalysts and molecular sieves in chemical and petrochemical ind~stries,~ ion exchangers6 (for example in purification and treatment of wastewater), absorbents (for example in deter ents), nonconductive supports in microelectronics: energy-storage material , membrane and most reactors where catalysis and separation is completed in one recently in micro-hano-scale fabrications and devices.’1712
f
Although the crystalline zeolite ZSM-5 (aluminosilicate with MFI-type structure) was discovered in 197213 and silicalite (the hydrophobic polymorph of silica with MFI-type structure) was synthesized in early 1978,14 little is known about the nucleation step or the early-stage crystal growth process in their hydrothermal synthesis solutions. Nucleation and growth mechanisms for different zeolite systems ranging fkom liquid-phase ion-transplantationto solid hydrogel reconstruction, have been proposed,’ but the complete conversion process from the synthesis solution to a fblly-grown crystal is yet to be
4
Ceramic Nanomaterials and Nanotechnology
understood. Additionally, there is a lack of crystallization kinetics data, which is required to develop a reliable predictive growth model. The simplest way to crystallize a zeolite (discrete particle) is to produce a highly supersaturated aqueous solution of appropriate composition contaiding an alkali hydroxide at a relative low dissolution temperature. It is known that zeolite crystal formation in solutions typically go through a gel transient state, however, such gel material has not been observed by an electronic microscope imaging technique. Numerous studies have provided evidence that coprecipitated gels uridergo an aging process at low temperature (-25 "C)in which the bulk physical nature (and consequently the intimate atomic linkages) changes, producing the appropriate structural units (or building blocks) that grow further at crystallization temperatures (-50-200 "C).
As silicalite-1 has not yet been obtained without a molecular template,16 the importance of the structure-directing species is evident. Many organic templating (structure-directing) ionic species have been used for the synthesis of purely siliceous MFI-type zeolite, but tetrapropylammonium cation (TPAS) is the most suitable one for silicalite-l crystallization, l 8 and is, therefore, used in the synthesis solution studied here. Silicalite-1 unit cell (uc) parameters for the studies were taken from the original reference listed in the International Zeolite Association @A) Tab~lation'~. During the synthesis, the organic structure-directing species are trapped (or occluded) in the framework through intermolecular van der Waals forces and can only be removed by thermal decomposition, but no covalent bonds A study by small angle neutron scattering ( S A N S ) suggested that the TPA' ion is incorporated into the zeolite framework during aging and early nuc1eation2l as it is too large to enter the channels afterwards. Investigation of the mobility of occluded TPA' inside a purely siliceous MFI framework using a combination of Magic Angle Spinning (MAS) and Cross Polarization (CP) NMR revealed an inter-cation dynamic equilibrium22 to satisfj the lowest energy state required for the crystallization process. Others have shown that hydrophobic TPA' ions are occluded into the channel interactions of silicalite, with one cation per channel intersection, through N M R techniques.23 Ultra-small-angle X-ray scattering (USAXS) studies have shown that amorphous colloidal aggregates (up to 9 nm in size) are formed during aging of the clear synthesis solution before crystallization begins? Combined in situ small-angle X-ray scattering (SAXS) and wide-angle X-ray scattering
Ceramic Nanomaterials and Nanotechnology
5
( W A X S ) suggest a solution-mediated growth mechanism where primary particles are less than 2.5 but the complete process from amorphous particle to the fully grown crystals is still not filly understood. According to Watson et al., the nucleation process is represented by the formatioq of particles Rg= 81 A which have nearly the same density as that of the fully crystalized silicalite particle containing the tetrapropylammonium te1nplate.2~ S A X S f S W A X S data show that alkalinity of the synthesis solution affects the formation of different sized amorphous gel particles during silicalite-1 crystallization process.25 Further study suggested that the primary unit and aggregates, formed during early nucleation stage, are consumed later to form crystals.26 The purpose of this research is to obtain fbndamental understanding of zeolite crystal nucleation and growth processes (in particular, at the early stage) during hydrothemal solution synthesis and thermal processing of gel precursor materials. Such understanding on zeolite crystallization has significant impact upon large-scale synthesis of high-quality zeolite particles, obtaining large single crystals of zeolite, growth of zeolite films, and fabrication of zeolite-top-layer inorganic membranes using gel precursors. As a better understanding of the early-stage nucleation mechanism helps to identifL the role of the organic templating molecules in self-assembling synthesis of delicate microstructures/textures in single crystal materials, new methodologies in molecular-directed synthesis of exotic materials can be explored. EXPERIMENTAL Nanocrystals of TPA-silicalite-1 were hydrothennally synthesized using the composition mention in the Exxon following the procedure outlined in Vroon et al? The silicalite synthesis solution was prepared in a capped teflon container by dissolving weighed amount of sodium hydroxide pellets (99.99%, Aldrich) in tetrapropylammonium hydroxide (TPAOH) solution (1M in water, Aldrich). The mixture was then heated and kept at 8OoC while weighed amount of fumed silica (99.98%, Aldrich) was dissolved under vigorous stirring to obtain a clear solution. Filtered (0.2 pm filter), deionized water was then stirred in, so that the final molar ratio of the synthesis solution was 10 Si02 : 3 TPAOH : 1.05 NaOH : 140 H20 (20 gm SiO2 : 100 ml(1M solution in water) TPAOH : 1.4 gm NaOH : 3.2 gm H20).
The clear synthesis solution were cooled down and kept at room temperature for 3 hrs, which counts as the aging time. The aged clear solutions
6
Ceramic Nanomaterials and Nanotechnology
were then transferred to a Teflon-lined autoclave vessel and heated in a convection oven to the specified synthesis temperature (180°C and 100°C; Error $2 "C) and the exact synthesis times were recorded. The samples used for the in-situ high temperature X-ray diffraction (HTXRD) study were not washed. Also, unwashed samples were used for the early stage sample synthesized at 100°C for 2 hrs, to preserve the fragile amorphous gel particles. For the rest of the samples, the amorphous hydrogel andor solid crystals were retrieved by centrifugation (20,000 rpm, 20 min) and washed by redispersion in filtered deionized water. The centrifugation-redispersion procedure was repeated until pH of the dispersion is around 8. The size, morphology and microstructure of the synthesized zeolite particles were investigated using a scanning electron microscope (SEM) (Jeol JSM-T220A) and a high resolution transmission electron microscope (HRTEM) (Jeol IOOCX1I), depending on the particle size of the samples. Silicalites have a 3D framework structure that can easily be decomposeddestroyed by electron beam during HRTEM analysis. As silicalites have a lower density (typically 1-2 gm/cm3) than silica SiO2 (2.3-2.6 g/cm3), structure fragments resulting fiom broken bonds can quickly diffuse away, making bond reformation much more difficult than in ~ 0 2 . ~ ' This creates the difficulty for high-resolution nanoscale imaging of a silicalite crystal by HRTEM. Therefore, a liquid nitrogen cooled cryogenic sample holder was used to enhance the stability of nanoscale crystal structure under electron beam. While focusing and stigmatism correction were performed on neighboring area away fkom the nanocrystal of interest, an image was captured by a Gatan 794 multiscan camera immediately after the beam is moved on the silicalite nanocrystal of interest.
The effective hydrodynamic diameter of zeolite particles was measured by a custom designed dynamic light scattering (DLS) spectrophotometer (details given in Hu et al.30), which is accurate for measuring submicrometer-size particles, typically ranging from 5nm to 1 p. The amorphous gel-type samples are air-dried overnight at near room temperature whereas the grown crystal samples are dried at 6OoC for 1 h to get powder samples for room temperature X-ray diffraction (XRD) analysis. The crystallite size was estimated fiom the broadening of the diffiaction peak [loll by using Sherrer equation from the diffraction peak breadths (full width at half maximum)31.
Ceramic Nanomaterials and Nanotechnology
7
An in situ high temperature X-ray diffraction (HTXRD) instrument (details given in Hu et al.32)at the O W L HTML user facility was utilized for detailed study of the heating conversion process of gel or gel-containing precursor materials to crystals. Two samples were used for this study, the first is amorphous sample (Sl) and the second is partially crystalline (S2) in nature. S1 is formed by hydrothermal treatment of synthesis solution at 100°C for Zhr, v d the drylng the resulting suspension, while 52 is incubated for 5.5hr at 100°C. The gel precursor material was placed on the platinum heating strip, and the temperature was then increased programmatically fiom room temperature (30°C) to 1200°C and then cooled down to room temperature. The heating and cooling rate was 10 degreehin. The spectra at various constant temperatures were scanned in the following order: 3OoC, 50°C, increasing temperature every 50°C till 5OO0C, then increasing the temperature every 100°C till 1200°C, and then at 30°C. Each scan was taken fkom 6 to 50 degree of 2-theta at a rate of 1 minute/degree. Data acquisition was performed using DMS-NT soffware (Scintag Inc., Cupertino, CA) and data analysis was undertaken using Jade software (Materials Data Inc., Livermore, CA).
RESULTS AND DISCUSSION Discrete zeolite (silicalite-1) crystal particles were first synthesized at 180°C. Particles appeared to be monodispersed and spherical (aspect ratio close to 1/1, Figure 1). At this temperature, particle growth kinetics are fast and submicron sized crystals were already well grown at reaction time as short as 1 hr. Particle size was found to increase with reaction time. Therefore, lower hydrothermal processing temperature (i.e., 100°C) was chosen to be able to produce nanosized crystal particles and to monitor the transition processes fiom initial aged clear solution to a fully-grown crystal particle. At this lower temperature, reaction kinetics and nucleation and growth dynamics are slow enough for one to collect a solid sample with a transient state fiom solution to a fully-grown crystal. 1) Cryo-HRTEM study of transient solid phases during hydrothermal conversion of solution species to zeolite crystals The complete zeolite growth process fkom clear solution, via amorphous gel particle formation, including several steps like early-stage zeolite nanocrystal nuclei formation and then the growth of these nuclei to filly-grown crystals were captured using HRTEM with the cryogenic sample holder. The field emission mode was used for our study as it gives a high amount of structural information with minimum interference with the zeolite sample. The HRTRM
8
Ceramic Nanomaterials and Nanotechnology
Figwe 1: SEM images of monodispersed silicalite crystalline particles synthesized at 180°C for various hydrothermal reaction times.
images for samples incubated at 100°C for 2hr, 3hr, 4.5 hr and 6hr are shown in Figure 2, 3, 4, and 5 , respectively. The corresponding electron diffraction pattern of these images, fast Fourier transform (FFT), are given as figuie 2A, 3A, 4A and 5A. The FFT of the image for the zeolite sample incubated at 100°C for 2hr (figure 2A) confirms that the sample is amorphous, where as the FFT of the image for the zeolite sample incubated at 100°C for 6hr (figure 5A) confirms that the sample is crystalline. The first observed sample was obtained by incubating the aged, clear synthesis solution in the sealed Teflon lined stainless steel vessel for 2hrs. The
Ceramic Nanomaterials and Nanotechnology
9
Figure 2: HRTEM image of unwashed zeolite sample heated at TSwth=lOO°C and incubated for 2hrs
Figure 2A: Fast Fourier Transform (FFT) diffi-actionof the HRTEM image of unwashed zeolite sample heated at T,,th=lOOOC and incubated for 2hrs
HRTEM image of the unwashed sample (Figure 2) shows spherical particles of approximately equal size of a diameter of 45 to 55 nm. The FFT of this image (Figure 2A) does not show any lattice reflections, therefore, suggesting that amorphous particles are present. Also, the XRD of the dried samples from this solution did not give any reflection peak (Figure 8). This confms that the particles are indeed amorphous and no crystal structure is yet evolved.
10
Ceramic Nanomaterials and Nanotechnology
Figure 3: HRTEM image of washed zeolite sample heated at T,,lh=lOOOC and incubated for 3hrs
Figure 3A: Fast Fowier Transform (FFT) diffraction of the HRTEM image of washed zeolite sample heated at T,,~=lOO°C and incubated for 3 hrs
A s the incubation time is increased to 3 hrs, the sample changes from clear solution to slightly milkish in color. The HRTEM image of this sample (Figure 3) shows the presence of nanocrystals (average size -1 Snm). FFT of this image (Figure 3A) indicates the presence of some crystalline materids. It is important to note that the early zeolite crystal nuclei co-existed with extensive amount of gels. The XRD spectra for the dried sample (Figure 8) confirmed that a small percentage of silicalite crystal structure is present in the sample,
Ceramic Nanomaterials and Nanotechnology
11
Figure 4: HRTEM image of washed zeolite sample heated at Tsyntt,=lOOoCand incubated for 4.5hrs
Figure 4A: Fast Founer Transform (FFT) diffraction of the HRTEM image of washed zeolite sample heated at T,t~=lOOoC and incubated for 4.5 hrs The HRTEM image of the sample incubated for 4.5 hrs (Figure 4) shows a partially grown single crystal of around 60 nm. The crystal lattice structure corresponds to [loll plane of silicalite (Figure 4A). The gel is still present around this crystal structure, suggesting that there is considerable amount of amorphous silicalite species to help the crystal grow firther. It appears that the silicalite crystal grows by direct consumption or reconstruction from the surrounding gels, supporting the gel-tranformation mechanism for silicalite growth. Such mechanism of crystallization by direct gel transformation has in solutions. also been recently reported for zeolite A nanocrystal
12
Ceramic Nanomaterials and Nanotechnology
Figure 5 : HRTEM image of washed zeolite sample heated at TSpt~=10O0C and incubated for 6 hrs
Figure 5A: Fast Fourier Transfonn (FFT) diffraction of the HRTEM image of washed zeolite sample heated at Ts,t~=lOOOCand incubated for 6 hrs However, for silicalite crystallization, we observed neither the ideal core nanocrytallization phenomena inside a discrete gel particle nor the one-to-one correlation between a gel particle and a silicalite crystal. The HRTEM image (Figure 5) for sample incubated for 6 hrs shows a completely grown crystal of about 80 nm and no transient gel materials were observed around the crystal particle. The crystal shows a reflection of the [1001 plane of silicalite from analysis of the FFT of this image (Figure 5A). A closer view of the crystal shows the intricate crystal lattice structure on the surface of Ceramic Nanomaterials and Nanotechnology
13
Figure 6: Enlarged HRTEM image of a fully grown crystal showing the [1001plane
Figure 7: Computer simulated 3-D image of [loo] plane for silicalite-1 showing the characteristic 10 ring channels (using Mac Tempas) the crystal particle (Figure 6). This is also matched by a computer-simulated image of the [1001plane of silicalite (Figure 7). Framework structure of silicalite-1 zeolite comprises of two different channel systems, each defined by 10-membered rings. Straight channels with an elliptical cross section of approximately 5.7-5.2 A are parallel to the crystallographic axis b, and sinusoidal channels with nearly circular cross
14
Ceramic Nanomaterials and Nanotechnology
section of 5.4 A run along the crystallographic axis a. The resulting intersections are elongated cavities up to 9.0 A in diameter. A single crystallographic cell, shown as a black box in the center of the computersimulated image (Figure 7), contains 96 Silicon, 192 Oxygen. 2) XRD studies on hydrothennal gel-to-crystal transformation in solutions In addition to the above HRTEM images, the evolution from early-stage gel to fully-grown crystals during hydrothennal treatment of the synthesis solution is also monitored by room temperature XRD study of washed, dried solid samples collected at different reaction times (Figure 8). Using the XRD spectrum for each sample, the size of the crystallite was calculated using the Sherrer’s approximation for the [loll peak (Table 1). When compared with the DLS effective hydrodynamic diameter measurements, the earlier gel samples are somewhat agglomerated because the particle size measured by DLS is larger than the crystallite size determined by XRD, which is agreeable to those determined by Cryo-TEM analysis. Note that the gel phase around the crystallite cannot be detected by the XRD size analysis method, but is clearly detected through imaging and DLS determination of particle diameter.
Figure 8: Room Temperature X-Ray Diffkaction patterns of the silicalite nanocrystal evolution and its growth during hydrothennal synthesis
Ceramic Nanomaterials and Nanotechnology
15
Table I. Comparison of zeolite particle size using various characterization tools
1 Time 2hr 3hr 3.5hr 4.5hr 5hr 6hr IOOhr
DLS
80
TEM 55
XRD Amorphous
I00 46
20
-
25 35
62 83
60
42
102 I10
80
54.5 59
-
64
-
3) HTXRD real-time studies of gel-to-crystal conversion process by heat treatment ill crystallize from some gel Above results clearly show that silicalite w precursors in solutions under hydrothermal processing conhtions. It is now necessary to verify the possibility of the crystallization of gel or early-stage gel-containing solid precursors materials crystallization by thermal treatment.
A HTXRD was used to allow the real-time monitoring of crystallization from the precursor materials. The first precursor material (Sl) put on the Platinum-Molybdenum heating strip was an unwashed amorphous gel. The amorphous nature is clearly indicated by the characteristic hump of the initial HTXRD spectrum (Figure 9). TPA molecules, although being mixed and trapped inside the dried gel (proved by EDX elemental analysis), do not guide the nucleationlgrowthof desirable silicalite inorganic crystal phase. During the thermal treatment by increasing temperature, the hump spectra transform gradually into sharp peaks, which signifies the occurrence of crystallization. However, the crystals evolved from the dry gel are not silicalite but Crystoballite High. This result indicates that the silicalite crystallization process via TPA templates is not simply a heat induced crystallization event but requires some water and also possibly the availability of ionic species and their mobility in the gel.
16
Ceramic Nanomaterials and Nanotechnology
Figure 9 : HTXRD spectra during heat-induced crystallization from amorphous gels, showing no zeolite formation and frnal crystal is CrystoballiteHigh.
Increasing temperature :
Figure 10 : HTXRD spectyra during heat-induced crystallization from amorphous gel containing nanocrystals of silicalite, indicating that firther silicalite growth can occur and at higher temperatures silicalite disappears and CrystoballiteHigh appears.
Ceramic Nanomaterials and Nanotechnology
17
The effect of increasing temperature on the second gel precursor material (S2) was then analyzed by HTXRD. The silicalite crystalline nature of this precursor material can be seen in the initial diffraction spectrum, (Figure 10). As the precursor material is heated till 300 O C , X-ray diffraction peak height intensity increases, corresponding to the growth of silicalite crystals (Figure 10). This indicates the seeded growth OCCUTS prior to the removal of templating TPA molecules fiom silicalite pore channels occurring around 400-500 "C. Xray diffraction peaks are shifted at the onset of TPA removal temperature. With further increase of heating temperature to around 700 "C, the intricate zeolite cages rupture, as the signature peaks for silicalite-1 zeolite starts decreasing and disappearing. At 8OO0C and higher temperatures, the appearance of other peaks at Ztheta 21" signifies the formation of Crystoballite High crystal structure.
-
CONCLUSION The early-stage nucleation and growth during silicalite-1 crystal particle synthesis have been studied by a high-resolution TEM and XRD techniques. The TEM images have captured the important transition phases in the formation of the silicalite crystal, starting from the early-stage amorphous gel particles, through the formation of nanosized zeolite crystallites, to fully-grown crystals. Our observations favor the gel-reconstruction mechanism for the formation of the zeolite nanocrystalline material. The gel particle are formed much earlier in the zeolite hydrothermal synthesis process, as evident through the TEM imaging. Afterwards these gel particles collapse and the zeolite nanocrystals appear, which then grows continuously at the expense of the hydrogel. The amorphous hydrogel is completely consumed during the zeolite synthesis process, as proven by the TEM and SEM images of the grown zeolite crystal samples. In-situ high temperature XRD studies indicate that crystallization f?om amorphous gels is more than just a thermal event because heating of a dry gel material can not produce silicaIite crystals but Crystallite High. Crystalline silicalite fkamework formation requires availability of ionic species, in addition to heating. However, W h e r crystal growth was observed when nanocrystalcontaining gel materials were heated. The silicalite nanocrystals in the starting gel materials may have fiznctioned as seeds for W h e r crystallization of silicalite flom surrounding gels. At temperature above 700"c, nanostructure of silicalite crystals collapse and meanwhile a non-zeolitic crystal form (Crystoballite High) evolves above 800 'C.
18
Ceramic Nanomaterials and Nanotechnology
In summary, amorphous gel plays an important part in the formation of the initial zeolite (silicalite-1) crystal nuclei. In synthesis solution, nanosized silicalite crystal evolve by direct reconstruction or transformation from surrounding gels. Also, the presence of occluded water/solution in the gel is important for the TPA-templated growth of silicalite-1 nanoparticles. REFERENCES ‘D.W. Breck, “Zeolite Molecular Sieves: structure, chemistry and use”, John Wiley & Sons, New York, 1974. 2J.V. Smith, “Origin and Structure of Zeolites”; pp. 3-79 in Zeolite Chemistry and CataZysis. Edited by J.A. Rabo. American Chemical Society Monograph 171, Washington D.C., 1976. 3M.E. Davis, “Zeolites and Molecular Sieves: Not just ordinary catalysts,” Industrial and Engineering ChemistryResearch, 30, pp. 1675-1683, 1991. 4P. Demontis and G.B. Suffiiitti, “Structure and Dynamics of Zeolites Investigated by Molecular Dynamics,” Chemical Reviews, 97, pp. 2845-2878, 1997. ’C.D. Chang and A.J. Silvestri, “The Conversion of Methanol and Other 0Compounds ti Hydrocarbons over Zeolite Catalysts,” Journal of Catalysis, 47, pp. 249-259, 1977. 9.Chu and F.G.Dwyer, “Inorganic Cation Exchange Properties of Zeolite ZSM-5”; pp. 59-78 in Intrazeolite Chemistry. Edited by G.D. Stucky and F.G. Dwyer. American Chemical Society Symposium Series 218, Washington D.C., 1982. ’G.A. Ozin, A. Kuperman and A. Stein, ‘‘Advanced Zeolite Materials Science,” Angewandte Chemie International Edition, 28 [3], pp. 359-376, 1989. 8 J. Weitkamp, M. Fritz and S. Emst, “Zeolite as Media for Hydrogen Storage,” International Journal of Hydrogen Energy, 20 [121, pp. 967-970, 1995. ’A.S.T. Chiang and K. Chao, “Membranes and Fims of Zeolite and Zeolitelike Materials,” Journul of Physics and Chemistry of Solid, 62 [9-101, pp. 1899-1910,2001. ’‘A. Tavolaro and E. Drioli, “Zeolite Membranes,” Advanced Materials, 11 [12], pp, 975-996, 1999. J.L.H. Chau and K.L. Yeung, “Zeolite Microtunnels and Microchunnels,” Chemical Communications,pp. 960-961,2002.
Ceramic Nanomaterials and Nanotechnology
19
12
K.Ha, Y. Lee, D. Jung, J.H. Lee and K.B. Yoon, “Micropatterning of Oriented Zeolite Monolayers on Glass by Covalent Linkage,” Advanced Materials, 12 [21], pp. 1614-1616,2000. 13R.J. Argauer and G.R. Landolt, “Crystalline Zeolite ZSN-5 and Method of Preparing the same,” Mobil Oil Corporation Patent US3702886,1972. I4E.M. Flanigen, J.M. Bennett, R.W. Grose, J.P. Cohen, R.L. Patton, R.M. Kirchner and J.V. Smith, “Silicalite, a New Hydrophobic Crystalline Silica Molecular Sieve,” Nature, 271,pp. 512-516, 1973. I5P.E.A. de Moor, T.P.M. Beelem and R.A. van Santen, “Influence of Aging and Dilution on the Crystallization of Silicalite-1,” Journal of AppZied Crystallography, 30, pp. 675-679,1997. ‘k.Szostak, “Handbook of Molecular Sieves,” Van Nostran Reinhold, New York, 1992 I7A.V.Goretsky, L.W. Beck, S.1. Zones and M.E. Davis, “Influence of the Hydrophobic Character of Structure-Directing Agents for the Synthesis of Pure-silica Zeolites,” Microporous and Mesoporous Materials, 28, pp. 3 87393,1999. 18D.W. Lewis, C.M. Freeman and C.R.A. Catlow, “Predicting the Templating Ability of Organic Additives for the Syntesisi of Microporous Materials,” Journal of Physical Chemistry, 99, pp. 11194-11202,1995. “H. van Koningsveld, H. van Bekkum and J.C. Jansen, “On the Location and Disorder o f the Tetrapropylammonium (TPA) ion in Zeolite ZSM-5 with Improved Framework Accuracy,” Acta Crystallographica Section B: Structural Science, 43, pp. 127-132,1937. 2oS.I. Zones and M.E. Davis, “Zeolite Materials: Recent Discoveries and Future Prospects,” Current Opinion in Solid State and Material Science, 1, pp. 107-117, 1996. 21 J. Dougherty, L.E. Iton and W. White, “Room Temperature Aging of a ZSM-5 Preparation Detected by Small Angle X-ray and Neutron Scattering and N.M.R. Spectroscopy,”Zeolites, 15, pp. 640-649, 1995. 22R.L.Gougeon, L. Delmotte, P. Reinheimer, B. Meurer and J.M. Chkzeau, “High-resolution Solid-state Nuclear Magnetic Resonance Study of the Tetrapropylammonium Template in a Purely Siliceous MFI-type Zeolite,” Ma etic Resonance in Chemistry, 36, pp. 415-421, 1998. ‘A.”. Bell, ‘“MX Applied to Zeolite Ssynthesis,” Colloids and Sugaces A:Ph sicochemical and Engineering Aspects, 158, pp. 221-234, 1999. 2$.N. Watson, L.E. Iton and J.W. White, “h situ Observation of the Growth of Silicalite Nuclei by Small-angle X-ray and Neutron Scattering,” Chemical CommunicQtions,pp. 2767-2768, 1996.
20
Ceramic Nanomaterials and Nanotechnology
25P.E.A.de Moor, T.P.M. Beelem, B.U. Komanschek and R.A. van Santen, “Nanometer Scale Precursors in the Crystallization of Si-TPA-MFI,” Microporous and Mesoporous Materials, 21,pp. 263-269, 1998. 26P.E.A. de Moor, T.P.M. Beelem and R.A. van Santen, “In situ Observation of Nucleation and Crystal Growth in Zeolite Synthesis: A SmallAngle X-ray Scattering Investigation on Si-TPA-MFI,” Journal of Physical Chemistry B, 103, pp. 1639-1650,1999 27J.P. Verduijn, “Nanometer-sized Molecular Sieve Crystals or Agglomerates and Processes for their Production,” Exxon Chemical Patent PCT/EP92/02386,1992. 28 Z.A.E.P. Vroon, K. Keizer, A.J. Burggraaf, H. Venveij, “Preparation and Characterization of Thin Zeolite MFI Membranes on Porous Supports” Journal of Membrane Science, 144, pp. 65-76, 1998. 29M,Pan, “High Resolution Electron Microscopy of Zeolites,” Micron, 27, pp. 219-238, 1996. 30M.Z.Hu, M.T. Harris and C.H. Byers, “Nucleation and Growth for Synthesis of Nanometric Zirconia Particles by Forced Hydrolysis,” Journal of Colloid and Interface Science, 198, pp. 87-99, 1998. 31E.W.Nuffield, “‘X-Ray Diffraction Methods,” John Wiley & Sons, New York, 1974. 32M,Z. Hu, R.D. Hunt, E.A. Payzant and C.R. Hubbard, “Nanocrystallization and Phase Transformation in Monodispersed Ultrafine Zirconia Particles from Various Homogeneous Precipitation Methods,” Journal of American Ceramic. Society, 82, pp. 23 13-2320, 1999. 33S.Mintova, N.H. Olson, V. Valtchev and T. Bein, “Mechanism of Zeolite A Nanocrystal Growth from Colloids at Room Temperature,” Science, 283, pp. 958-960,1999.
Ceramic Nanomaterials and Nanotechnology
21
This page intentionally left blank
SYNTHESIS AND CHARACTRIZATION OF MONO-SIZED SO2 CERAMIC PARTICLES IN MESO-STRUCTURE Thg-Wei Chen and Wen-Cheng J. Wei* Department of Materials Science and Engineering National Taiwan University, Taipei, Taiwan 106, R. 0. C . ABSTRACT Synthesis of submicron monodisperse silica particles followed the method known as Stober method, using ammonia to hydrolyze tetraethyl orthosilicate (TEOS) in base-catalyzed medium. The nearly monosized spherical particles were assembled into 3-D structure by controlling their surface potential. The structure was fixed with mild-calcination or co-electroplated Ni. The particle interior and the stacking characters of the structures were quantified by TEM and SEM in according to various defects found in the microscopic images.
Keywords: Sol-gel method, silica, assembly, mono-size 1. INTRODUCTION Synthesis of submicron monodisperse oxide particles starting from TEOS (tetraethyl orthosilicate) has been studied extensively in the past four decades, and recently gained the attention of ceramists for the preparation of photonic band gap (PBG) crystals' in the past few years. Essentially, the method reported by Stober et aZ2 is the most important work on the discussing of the synthesis of submicron monodispersive silica particles. It used various concentrations of TEOS, alcohols as reaction solvents, and ammonia solution to hydrolysis as a basic-catalyzed medium. Silica particles were formed in diameter from 0.05 to 2 pm334.However, the method can be modified for better uniformity in size and can be easier controlled to synthesize various mono-sized SiOz particles'. After synthesis of monodispersed Si02 particles, SiO2 opal' can be produced by the sedimentation of the spherical particles. Many researchers tried to arrangement of the monosized ceramic particles on self-assembled monolayers by * Correspondent author, Roosevelt Rd. Sec. 4, Taipei, Taiwan, 106 R.O.C. e-mail:
[email protected]
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
23
using various templete~~”. Most of reported packing structures are FCC structure. Sometimes, it would be found that the structure is a mixture of HCP and FCC.” Because imperfect arrangement of the spheres can produce serious disorders and other defects in the structure, the use of monodispersive ceramic spheres is one of the important issues related to the optical emission of SiO2 opal. Therefore, in this study we tended to quantify the defects, to characterize the order-disorder packing, and to control the density of the defects. 2. EXPERIMENTAL PROCEDURE 2.1 Materials There are three solutions used in the sol-gel method of Stiiber’s work for preparing SiO? powder. In this study, ethanol was used as a solvent (95% pure, Taiwan Tobacco & Wine Monopoly Bureau, Taipei, Taiwan, R.O.C., or extra pure reagent, 99.5%, Shimakyu Pure Chemicals, Japan). The TEOS (tetra-ethyl orthosilicate, MERCK-Schuchardt, Germany) is the source of SiOz, and in purified grade, and ammonia solution (28% NH3, Nacalai Tesque, Inc., Kyoto, Japan). 2.2 Preparation of Si02 Particles Amorphous monodispersive spherical silica particles were synthesized in the laboratory by using a sol-gel method described in previous5, and the procedures are briefly described as follows. The TEOS was first dissolved in ethanol at constant temperame. Then ammonia solution was added into the solution drop-by-drop as the catalyst. After 2 hr of reaction, silica powder was precipitated out from the solution. A dilution and washing of SiOz particles in ethanol was taken in order to stop the reaction. Repeat the procedures 3 times, then dilute in ethanol.
2.3 Arrangement of Si02 Particles Pure SiOz suspension was settled on alumina crucible and other substrates by gravitational sedimentation at room temperature for 2-3 days, and then dried in oven at 105°C for 24 hr. Some samples were calcined at 500°C for 4 hr and 950°C for 3 hr in order to burn out the organic residue and for bonding strength between particles. After coating Au by sputtering on sample swface, the observation of Si02 packing microstructures were done by SEM (XL30, Philips Co., Holland, and E O L JSM-100 scanning microscope). 2.4 Fixation of Si02 Particles by Ni Composite electroplating was proven to be a good way to fix silica spheres, and then the microstructure of cross section would be easy to observe. The composite electroplating fixation technique of SiOz/Ni are briefly described as
24
Ceramic Nanomaterials and Nanotechnology
follow. The electrolyte was 500 ml nickel sulphamate mixed with 500 ml D.1 Water, 3 g NiC12, and 30 g H3B03. The solution is controlled at pH=5, the temperature at 50°C. SO2 powder about 6.56 g (0.37 ~01%)was mixed into the electrolytic solution and dispersed by stirring magnet. CO per plate was used for plating substrate, and electroplating in 2, 6 , 8 ASD (NdmP) for 2 hr. Finally, silica particles would be composite- electroplated with Ni on Cu plate. 2.5 Characterization of Si02 Structure Ni with SiOz particles was cut in thin section for the following TEM thin foil sample. After grinding and ion milling, the samples were analyzed by TEM (JEOL lOOCXTI and Hitachi Model HF-2000 FEEM, Japan) and X-ray energy dispersive spectroscopy (EDS). The counting of coordination number (CN) of each Si02 sphere was done on the SEM micrograph. It is assumed that one sphere optimally connect t 6 other spheres as in close-packing. The coronation number Ni=6, and so on. Then we can define the Degree of order D is Therefore, D=l .OO-0.00.
i=o
/
i=O
3. RESULTS AND DISCUSSION By using sol-gel method, we could synthesize amorphous silica powder from solution and showed an average yield about 28 % 30 %. The particles size was measured from more than 200 particles of SiOz on TEM micrographs (e.g. Fig. 1 (a)), which had been calibrated in accuracy better than 0.4%. Homogenous and dispersive SiOz particles were observed by TEM in Fig. l(a), which were synthesized by a formation TEOS: NH40H: C2H50H = 14: 17:200 ml at 25'C for 2 h, and then diluted and washed in ethanol. The size of the particles was measured in an accuracy of f 0.01 clm, and quantified in Fig. 1 (b), which showed an average particle diameter of 299 nm and a standard variation of 7.9 nm, That represents 2.6%variation in size distribution. The figure also includes one set of the data reported by St6ber2 in 1968, which particle average diameter is 240 nm and the standard deviation is 1.04. The amorphous SiOz particfes can be fixed by calcination at 500°C on the temperatures up to 105OOC. Then dramatic shrinkage was detected above 1050°C'0. The sample in Fig. 2 was prepared by gravitational sedimentation and calcined at 950°C for 3 h. The microscopic picture of SEM shows a typical powder packing structure. The picture shows that an artificial opal was made of 400-nm spheres. Some regions present face-centered cubic (FCC) packing and fractured along { 111 ) and { 1 l O } cleavage crystalline planes. The sample shows similar surface character as the one calcined at 500°C for 4 h (Fig. 3), but there are
-
Ceramic Nanomaterials and Nanotechnology
25
apparent sintering points formed between particles if calcined at 950°C for 3 h. Sintering can greatly improve the strength of packing structure. Figure 3 depicts that SiU2 particles line up in 2-D arrays with short range of order. There are cubic packing and close packing domains mixed on the surface plane with domain boundaries separating one to each other. The domains in fact have different orientation caused by the packing orientation. The defects are similar to the faults, boundaries, and vacancies found in of atomic structures, which are pointed out in Fig. 3. The concentration of defects can be counted in order to compare the intactness of particle arrangements. When silica spheres are homogeneous dispersed in a suspension, similar sedimentation structures in some degree of order are reported in literature""'. Figure 4 shows the cases of the particle packing in various degrees of ordered structure. The particles in Fig. 4 (a) were prepared from the solution containing TEOS:C~HSOH:NH~OH =12: E 2 0 0 ml, and synthesized at 5°C. The average particle size is 433 nm. Then dried in A1203 crucible and left in the oven at 105'C overnight. The particle packing in Fig. 4 (a) has been calcined at 500'C for 4 hr after drying. Figure 4 (b), (c), (d) show the particle packing prepared from a fomiation with higher TEOS (TEOS:C2HsOH:NH40H =15: 15:200 ml), and synthesized at 25°C of which the average particle size is 445 nm. Then the particles sedimented on a glass plate for 3 days, and dried in a 105°C oven for 24 hr. By observation of SEM, Fig. 4 (b) shows nearly perfect 2-D hexagonal closed packing. Figure 4 (c) shows 2-D cubic packing, which is identified as (100) of FCC. The unit cell of the FCC is plotted on the diagram. Figure 4 (d) shows the mixture of 2-D cubic and hexagonal packings. Long crack and vacancies are noted as well. The degree of order ranges from 1.0 (perfect hexagonal closed packing) to 0 (completely dispersed isolated particle). Note that the observation by SEM is limited to the surface layer of the particle packing. Since the packing density of surface layer can be counted fkom the SEM pictures, the more the defects, the less the packing density. Because the regularity of particle packing is one of the important points for spectra diffraction, the arrangement of mono-sized particles is the controlling factor. Base on previous observation, cubic packing plane { 100) and hexagonal packing plane { 11 1 of FCC structure are two most familiar planes appeared on surface layer. If there is only one packing pattern among one domain, which is free from defects, the structure has continuous, long-rage packing in order. The structure (e.g. Fig. 4 (b)) can be called a perfect order structure. Figure 5 shows the cases of crack in structure. Due to the shrinkage of the packing during drylng and calcination, cracks are usually generated and become the boundaries of the domains. Fig. 5 (a) and 5 (b) show the samples prepared by same procedures as Fig. 4 (b). It is noted that the cracks are mostly across the interior of big domains. The domain boundaries are usually free from the
26
Ceramic Nanomaterials and Nanotechnology
formation of cracking. Figure 6 shows the SEM images of the cross section of Si02 spheres surrounded with co-deposited Ni. Aggregation of Si02 particles was observed in Fig. 6 (a), but some on higher magnification are in good order (Fig. 6 @)), which are surrounded with Ni and imaged by BSE (Back Scattered Electron) mode. The figure shows that some empty sites are originally occupied by the particles, but are palled-out from packing structure due to weaker bonding. One TEM bright field (BF) image of ordered Si02 spheres surrounded with co-deposited Ni is shown in Fig. 7. The inserted diffraction pattern reveals polycrystalfine character of Ni bonding. The interior of the Si02 particles is amorphous determined by selective area diffiaction (SAD) pattern. However, the milled surface of the SiOz is possibly covered with very thin Ni layer if the ion milling is in-appropriately operated. Figure 8 is TEM BF of one thin Si02 particle and the microbeam EDS results along the indicated line perpendicular across the interface of SiOz/Ni. The elemental analysis of Ni reported in wt% gotten fiom a spoty region less than 10 nm indicated that there was no Ni element diffused into SiOz particles and nearly no Ni contamination on the cross section of Si02 particles. This implies that the Si02 sphere has a dense surface, which cannot allow the penetration of Ni-plating solution. Figure 9 is the TEM BF image of $ 3 0 2 particle packing layers surrounded with Ni. Because 2-3 layers of packing could be observed, 3-D packing can be identified according to the packing style between layers. Fig. 9 (b) schematically shows that the marked Si02 packing in a form of FCC { 1001, and Fig.9 (c) shows the hexagonal closed-packing, which may be the FCC { 11 1] plane. The region is an assembly mixed wi%ththe particles of different packing orientations. The interchange of the particle assembly may be as thin as two particle layers. The region is 111 of particle packing information in 3-D condition. 4. CONCLUSION
Sedimentation of Si02 particles assisted by electroplating helps the assembly of SiOz. Nearly ordered Si02 opal structure was prepared, but point and planar defects are left in the interior and surface of the packing. The cross-section of Si02 sphere appears in amorphous state. Because the interior is free fkom contamination of Ni, the skin of the SiO2 before Ni-codeposition must be dense. The image of 3D structure revealed the assembly of the particles can be FCC type, but in packing orientations. The particle assembly may change to the other orientation as thin as two particle layers. REFRENCE ‘P. Ni, P. Dong, B. Cheng, X. Li, and D. Zhang, “Synthetic Si02 Opals”, Adv.
Ceramic Nanomaterials and Nanotechnology
27
Matex, 13 [6] 437-441 (2001). 2W. Stober and A. Fir&+“Controlled Growth of Monodisperse Silica Spheres in the Micron Size Range, J. Colloid Interface Sci., 26,62-69 (1 968). 3M. T. Harris, R. R. Brunson, and C . H. Byers, ‘The Base-Catalyzed Hydrolysis and Condensation Reactions of Dilute and Concentrated TEOS Solution,” J. Colloid Interface Sci., 121,397-403 (1 990). %. T. Harris, 0. A, Basaran, and C. H. Byers, ‘‘The Precipitation Dynamics of Silica Particles,” pp. 1 19-127 in Ceramic powder science 111. Edited by G. L. Messing, S . Hirano, and H. Hausner (Westerville, Ohio, American Ceramics Society, c 1990) T. W. Chen and W. J. Wei, “Ripening of SiOz Particles in Stober’s Process,”(in preparation). ‘Y. Musuda, W. S. Seo, K. Koumoto, “Two-Diniensional Arrangement of Fine Silica Spheres on Self-Assembled Monolayers,” Thin Solid Films, 382, 183-189 (2001). ’Y. Masuda, W. S. Seo and K. Koumoto, “Arrangement of Nanosized Ceramic Particles on Self-Assembled Monolayers,” Jpn. J. Appl. Phys. 39, 4596-4600 (2000). 8Y. A. Vlasov, X. Z . Bo, J. C. Sturm & D. J. Norrism, “On-Chip Natural Assembly of Silicon Photonic Bandgap Crystals,” Letters to Nature, 414 El51 289-293 (2001). ’T. Cassabneau and E Caruso, “Semiconducting Polymer Inverse Opals Prepared by Electropolymerization,” Adv. Matex, 14 111 34-38 (2002). 10 H. Miguze, A. Blanco, C. Lopez, E Mesegure, H. M. Yates, M. E. Pemble, F. L. Tejeira, F. J. G. Vidal, and J. S. Dehesa, “Face Centered Cubic Photonic Bandgap Materials Based on Opal-Semiconductor Composites,” J. Lightwave Tech., 17 1111 1975-1981 (1999). 86.0 800
80.0 70.0 80.0 40.0 m.0
ao.0 20.0 100
SO 2.0 0.6 1.0 0.1
Fig. 1 200 210 220 230 240 250 280 270 280 290 SO0 310 320 Partlcle dlarneter (nm) (a)TEM picture and (b)statistic data of SiOz H40H: C2H50H=14:17:200 ml at 25OC, Op = particles synthesized by volume ratio of TEOS: NJ 299 nm and m =18, compared with St6ber 119681 Dp = 240 nm and m=lO.
28
Ceramic Nanomaterials and Nanotechnology
Fig. 2 SEM micrographs of naturally seditnented SiOz particle packing, calcined at 950°C for 3 h. An artificial opal was made of 400 nm spheres.
Fig. 3 Defects were identified on the surface of Si02 particle packing after calcination at 500°C for 4 h. The dekcts are defies as stacking faults, dotnain boundaries, and vacancies similar to those in atomic structure.
Fig.4 SEM micrographs of various SiOz packings which show the degree of order (a) D=0.7, (b) D 1 .U (close packing), (c) D-0.66 (cubic packing), (d) D=0.73 (mixing conditions).
Ceramic Nanomaterials and Nanotechnology
29
Fig. 5 SEM micrographs of various SOzpackings (a) 2-D plane with vacancies, domain interfaces and crack defects, (b) mixed defects of cubic and close packings.
Fig. 6 SiOzparticles fixed by electro-plated Ni revealed by SEM. (a) An aggregation of SOz particles surrounded with Ni and imaged by BSE mode, (b) vacant spaces between SiO, particles are filled with Ni, and some particles are missing due to weaker bonding.
Fig. 7 TEM BF micrograph and inserted diffraction pattern revealing polycrystalline character of Ni bonding.
30
Ceramic Nanomaterials and Nanotechnology
110
(b)
'00 90
a0 70
40
30 20 10
0 -150
-100
-50
D
50
100
150
200
250
Interface distance (nrn)
Fig. 8 (a) TEM of thin section of SiO2 particles and (b) microbeamEDS results along the indicated line of SiOz/Ni interface.
Fig. 9 TEM BF image of 3-D Si02 particles packing layers surrounded with Ni,(a) original image. fb) layers of FCC (100) packing, (c) layers of hexagonal closed-packing-planepacking.
Ceramic Nanornaterials and Nanotechnology
31
This page intentionally left blank
EARLY STAGE SILICA FORMATION IN METHANOL
ETHANOL
D.L. Green University of Minnesata (Twin Cities) Department of Chemical Engineering and Materials Science 151 Amundson Hall, Box 198 421 Washington Avenue SE Minneapolis, MN
M.T. Harris Pwdue Unversity School of Chemical Engineering 1283 CHME Building West Layfayette, IN 47907-1283
S. Jayasundara George Mason University Center for Bioresource Development MS 3A3 Fairfax, VA 22030
J.S. Lin Solid State Division Oak Ridge National Laboratory Oak Ridge, TN 3783 1
M.Z. Hu Chemical Technology Division Oak Ridge National Laboratory Oak Ridge, TN 3783 1
Yui-Fai Lam NMR Lab, Department of Chemistry University of Maryland College Park,MD 20742
ABSTRACT 29Si-NMR is used to monitor the soluble reaction intermediates of tetraethylorthosilicate (TEOS) - its hydrolyzed monomers - that lead to the formation of the first primary particles, while small-angle x-ray scattering (SAXS) and dynamic light scattering (DLS) are used to monitor the conversion of the soluble Si into silica nanoparticles. Reactions are conducted at a [TEOS] = 0.5 M, low concentrations of ammonia ([NH3] = 0.01 - 0.1 M), and [HzO] = 1.1 4.4 M to slow the reaction kinetics, to resolve the initial size of the first nuclei, and to follow their structural evolution. We found that [NH3] and [HzO] control
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
33
the balance between hydrolysis of TEOS and the condensation of its hydrolyzed monomers. Transesterification between methanol and TEOS did occur; however, it is negligible compared to the production of hydrolyzed intermediates. A supersaturation of the hydrolyzed monomers controls the formation of nuclei, which are initially fiactal and open in structure. Interestingly, the nuclei are twice as large in ethanol (Rg = 8 nm) as those in methanol (Itg = 4 nm). The particle number concentration and the volume fi-action of the silica particles are calculated independently from the S A X S , DLS, and 29Si-NMRresults. Finally, the rate of nucleation is obtained from the particle number concentrations. ZNTRODUCTION The industrial use of silica is widespread. The further processing of monodisperse, spherical silica nanoparticles permits their use in many fields including ceramics, chromatography, catalysis, and colloids [11. Additionally, precursor silica particles have been used in stabilizers, coatings, glazes, emulsifiers, strengtheners, and binders [2]. Despite the plethora of uses, past research of Stober particles has been primarily concerned with empirically predicting the final size based on the initial reactants [3,4]. However, more recently, two models: (1) monomer addition [5,6]; and (2) controlled aggregation [7,8], have been created to elucidate the chemical andor physical growth mechanisms of silica. These models divide the formation of silica into two events: nucleation and growth. Nucleation, the formation of the first nuclei or primary particles, occurs at the early stages. Subsequently, the particles grow to achieve a final size distribution. The two models use different approaches to describe particle growth. Matsoukas and Gulari in the monomer addition model maintain that after an initial burst of nucleation, growth occurs through the addition of hydrolyzed monomers to the particle surface. In contrast, Bogush and Zukoski in the controlled aggregation model claim that nucleation occurs throughout the reaction. These primary particles or nuclei aggregate with themselves or larger aggregates to eventually produce a narrow size distribution. The work of McCormick and coworkers [9] supports controlled aggregation and indicates that the presence of the singly hydrolyzed TEOS monorner [(OH)Si(OC&)3] is necessary for particle formation. Hanis, Brunson, and Byers [lO] and Van Blaaderen et al. E1 11 believed both mechanisms are responsible for growth by suggesting that controlled aggregation occurs for much of the reaction, and then monomer addition caused the smoothing of the colloid surface. Specifically, the findings of Van Blaaderen et al. [ 11J are based on experiments that measured the effect of growth by changing the ionic strength of the solution through salt addition at the beginning of the reaction, and
34
Ceramic Nanomaterials and Nanotechnology
salt addition after the reaction had formed particles. When salt is added with the initial reactants, controlled aggregation was thought to be responsible early on due to the salt causing a large increase in particle size compared to the same conditions without salt. Salt addition does not significantly affect growth after the larger particles are formed. This indicates that monomer addition is responsible for an increase in particle size toward the later stages. Overall, the work of Van Blaaderen et al. [ll] suggests that the final particle size is strongly dependent upon the colloidal stability of the primary and intermediate particles. The controlled aggregation and monomer addition models are based upon dynamic and static light scattering (DLS and SLS) and transmission electron microscopy (TEM) measurements of the particles relative to time. DLS and SLS lack the sensitivity to detect the initial scattering of the primary particles during the early stages. TEM also has limitations. The preparation of TEM samples can obscure particle sizes through drying and TEM images do not indicate whether particles are solvated [121. Because of these inadequacies, there is still uncertainty about the about the size and the structure of the initial nuclei that should be incorporated in the monomer addition and controlled aggregation models. For example, Matsouskas and Gulari [6] in the monomer addition model define a primary particle as the combination of two hydrolyzed TEOS monomers. This suggests that the diameter of a primary particle should be within the molecular domain - on the order of several Angstroms since the radius of a TEOS molecule is 3 A 1131. The controlled aggregation model of Bogush and Zukoski [ 5 ] predicts primary particle radii between 1 - 2 nm without actually detecting these structures. While the techniques mentioned above could not observe nucleation, studies do exist that investigate the early stages of formation. Bailey and Mecartney [12] used cryo-TEM, which incorporates the fast fi-eezing of a sample with liquid ethane to avoid the effects of drylng encountered using regular TEM. Harris’ group, Boukari et al. [ 14-16], employed small-angk x-ray scattering (SAXS), an in-situ method that can probe molecular (0.I - 0.2 nm) to colloidal (1 - 20 nm) length scales. Both groups of researchers observed an induction period where no particles were detected. Following this induction period, the researchers observed a low-density primary particle with radii from 4 nm to 10 nm. Subsequently, these particles densified and grew as hydrolysis and condensation proceeded. Overall, the work of Bailey and Mecartney [ 121 and Boukari et al. [14-16] show that the monomer addition and controlled aggregation models do not adequately account for the size and structure of the initial nanostructures at the early stages of particle formation and growth. The induction period before particle formation represents the build-up of singly hydro1yzed monomers that form the initial low-density structures, possibly throvgh the self-assembly of these monomer molecules. Increased hydrolysis and
Ceramic Nanomaterials and Nanotechnology
35
condensation would occur within this network, ultimately yielding compact and spherical silica particles. With the above considerations, it is clear that a better understanding is needed of the mechanisms that control the formation of primary silica nanoparticles during the early stages and their subsequent growth. Therefore, the specific objectives of this research are as follows: (1) quantify the production of singlyhydrolyzed TEOS monomers and to establish the concentration of the singlyhydrolyzed rnonomer at which nucleation occurs; (2) examine how the primary particle size and structure are affected by variations in the solution chemistry [i.e., ammonia and water concentrations, and the nature of the solvent (methanol and ethanol)]; and (3) study the effect of ionic strength o f the solution on the aggregation dynamics at the early stages of particle growth. To accurately resolve the first colloids, reactions were carried out at low m3Jand [H20] as listed in Table I. These initial reactant conditions provide an extra benefit of allowing us to decouple the fonnation of nuclei fiom their growth. To explain the results obtained in this study, a short theoretical treatment of the information collected by DLS, SAXS, and 29Si-NMR is discussed in the following three steps. First, the general scattering theory and the determination of particle size by DLS is covered. Second, the size and structure of the evolving nanoparticles as determined from SAXS is reviewed. Lastly, the determination of the volume fi-action of Si02 by 29Si-NMR,SAXS, and DLS is explained. The calculation of the time-dependent particle number concentration, and thus the nucleation rate, is readily obtained from the SAXS and DLS measurements.
THEORY General Scattering of Particles The DLS and SAXS techniques detect the scattering intensity I@) of dilute sols relative to the scattering vector Q = (4z/il)sin(@/2)where 8 is the scattering angle and h is the wavelength of the light beam. For the He-Ne laser used in the DLS apparatus, h = 632.8 nm, and for the S A X S setup, h = 1.54 A. The general relationship for I(Q) is:
where N is the number of particles per unit volume in the sample and F(Q)is the structure factor which depends on the shape, and spatial distribution of particles in the sample [171. For homogenous spheres, F(Q)is known and Eq. 1 yields [171 I(Q) = KNM2[3sin(QR) - Q R cos(QR)
QR’
36
I’
Ceramic Nanomaterials and Nanotechnology
Here, K is an apparatus-dependent optical constant that is independent of the scattering produced by the sample [18], R is the particle radius, and A4 is the mass of the particle equal to (4/3)psi02nR3 where psi02 is the mass density in g/cm3 of SiOz. In dilute sols, R is equivalent to the hydrodynamic radius Rh. The bracketed term in Eq. 2 is essentially unity for small colloids (R < 15 nm) when h >> R. In dilute soh, R is equivalent to the hydrodynamic radius Rh,DLs, which is indirectly calculated from DLS by time-correlated intensity measurements of particles undergoing Brownian motion in solution. The DLS technique more directly determines the particle self-diffkion coefficient D that is related to RR,JLS through the Stokes-Einstein equation.
The solvent viscosity q and the temperature ?are input parameters that have to be known beforehand; and k is the Boltzmann constant. In contrast to DLS which detects differences in refractive indices, SAXS probes differences in electron density in which the scattering-intensity profile, I(Q) vs Q, of a sample yields the two following scattering regimes of interest to this study: (1) Guinier and (2) Porod. The Guinier region occurs at low Q (Rg<< Q-*)in which the limiting value of F(Q) simplifies to a Gaussian function that yields the particle size through the radius of gyration Rg [171.
where I0 is the intensity at P O . 10is directly proportional to N and Rg [ 171. The Porod regime, distinguished by scattering connected to the structure of a particle, is encountered at high Q (Rg >> 8'). Here, F'(Q) renders a power-law relationship between I Q ) and Q [ 17,191.
The exponent a, which takes non-integer values between 1 and 4 in a threedimensional space, relates the nanostructure to a fiactal object [20]. For complex and random systems, the concept of a fiactal has been applied to interpret some of the measured scattering profiles [21-241. In the fiactal description, the fi-actal
Ceramic Nanomaterials and Nanotechnology
37
dimension D quantifies how a property of the fiactal object such as the mass M in mass-fiactals or the surface area A in surface-fiactals change with R. For mass fractals, which can be pictured as open, polymeric, low density structures, the mass M scales as [22-241:
where Dmis the mass fi-actal dimension that takes non-integer numbers equivalent to a for 1 < a < 3. If D, = 3, the Euclidean (non-fkactal) description of the mass is recovered. Surface fractals, characterized by dense cores with rough surface structures with an area A , are described by [22-241:
where D,is the surface fractal dimension that covering non-integer number fiom 2 c D,< 3. In contrast, a = 6 - D,for surface fractals, and thus, a ranges between 3 < a < 4. The well-known Porod's Law for dense, smooth, and spherical interfaces is obtained when a = 4. Silica Volume Fraction and Number Concentration The conversion of soluble Si - TEOS and its intermediates - into SiO2 nanoparticles can be tracked experimentally by %i-NMR, SAXS, and DLS through the determination of the volume fraction, Q, of the SiOz particles. This material balance fiom the start of the reaction at t = 0 through the SAXS detection of the initial nanostructures allows one to determine if the induction period does represent the build-up of hydrolyzed monomers. Subsequently, correlating the species detected in solution to the evolution of nanostructure provides greater evidence that continued reactions between soluble monorners and the particles are responsible for their densification. Liquid-phase 29Si-NMRmonitors the conversion of soluble Si monomers into insoluble SiOz through a mole (mass) balance on Si. Si, where S ~ Tis the initial molar concentration of TEOS used in the reaction; is the molar concentration of the unhydrolyzed and hydrolyzed
Si(t)soluble
38
Ceramic Nanomaterials and Nanotechnology
monomers; and Si(t)insoluble is the molar concentration of Si02. The volume fraction of Si02 particles measured by 29Si-NMRis:
where Msio2 is the molecular weight of silica, and psjo2is the density of the 5 0 2 particles. Th9 volume fraction calculated by S U S , +SUS, is determined by modeling silica as a porous solid comprised of units with uniform electron density, piio2, suspended in a solvent with electron density piolvenr. For a porous solid that occupies a fraction of the sample volume, $SAXS is related to a normalization constant of the two-point density-density correlation function of the scattering system [ 171 - otherwise known as the Porod invariant, Qp [25].
Here, Ape is the difference in scattered electron density or conkast variation between the primary particle and the solvent displaced by the formation of the separate phase. The quantity Ape is calculated from the following relationship:
where NA is Avogadro's number; b the scattering length of an electron equal to 0.28 x 10'l2 cm [26]; Bolvent is the mass density in g/cm3 of the solvent; and Msolvent is the solvent molecular weight. The parameters nsio2 and n ~ are the ~ number of electrons in the solid and liquid phases. The Porod Invariant Qp(t) in Eq. 10 is determined by numerically integrating the SAXS scattering profiles and addmg the small contribution outside its experimental limits by extrapolating Eq. 4 to Q = 0 and extending Eq. 5 to Q = 00 [25]. Here, the high-Q limit can be approximated by extending Q to the size of diatomic nitrogen NZ at Q = 2n/4A, and then analytically determining the remainder for Porod's law at a = 4. This approximation at infinite Q is experimentally valid since Nz molecules in Brunauer-Emmett-Teller(BET) [27] surface area measurements "see" the sufface as a smooth, flat interface [28].
Ceramic Nanomaterials and Nanotechnology
39
l
~
The volume fraction from DLS, ~ D U , is calculated by Eq. 12 after determining &(t) from Eq. 3 and N ( t ) ~ from u Eq. 2
where V~,DLS is the volume for a spherical particle. The number concentration of silica from SAXS is determined by N ( t ) s m = + ( t ) s ~ s / Y p , For ~ ~ spheres, . Vp= 4/3nRim where R,,, is related to Rgin Eq. 4 through R , =fiR,. Finally, the law of mass conservation for 29Si-NMR,SAXS, and DLS dictates that
which indicates that R(t) = RDLS= R m s and N(t) = N(t)sMs = N(t)Ds. The nucleation rate gn(t)is determined from the slope of the N(t) versus time curve. EXPERIMENTS Quantification of Singly-HydrolyzedMonomer, HO(SiOC2H5)3 Kinetic 29Si-NMR experiments were performed to measure the disappearance of unhydrolyzed TEOS and the appearance of the singlyhydrolyzed monomer. The Bruker AM400 NMEt s ectrophotometer was used to detect these concentrations. The results fiom the 2 9Si-NMR data were compared to the results of the small angle scattering (SAXS) experiments to test whether a consistent concentration of singly-hydrolyzed monomer corresponds to the detection of the primary particles. SAXS experiments are discussed below. Determination of Effect of Solution Chemistry on Primary Particle Size and Structure The SAXS apparatus [29,30] and the DLS apparatus [31] were employed to detect the size and structure of the primary particle as a function of the initial ammonia, water, and salt (LiCl) concentrations and as a h c t i o n of the solvent (methanol and ethanol). The SAXS instrument was used to determine the size and structure of particles with a radius of gyration less than 30 nm. The DLS equipment was employed to measure the size of nanostructures with a radius of gyration as large as a several hundred nanometers. The scattering curve fiom the solvent and freshly prepared solutions of TEOS, water and alcohol (without ammonia) were subtracted from the SAXS scattering
40
Ceramic Nanomaterials and Nanotechnology
curve of the reacting solution. The TEOS monomer was not a sufficient scatterer in the fleshly prepared solution of TEOS, water and alcohol to yield scattering curves that were significantly different from the scattering profile of the pure solvent. Thus, the scattering cwve from the solvent was subtracted from scattering curve of the reacting solution. Effect of Ionic Strength on Aggregation Dynamics LiCl in concentrations ranging fiom 104 M to 5 ~ 1 0 M - ~ was added to a reacting solution consisting of initial concentrations of N H 3 , H20 and TEOS that fall within the range for Stober silica synthesis. Methanol and ethmol were used as the solvents. A dynamic light scattering (DLS) apparatus detailed Yudin et al. (1997) was used to measure the particle size evolution of silica nanoparticles.
RESULTS AND DISCUSSION 29 Si-NMR was used to detect the liquid phase reactants and products generated by the Stober reaction in methanol and ethanol. The NMR spectral information was used to determine the effect of [NH3],[HzO], and the solvent on the hydrolysis of TEOS and to resolve which soluble silica species are present during the induction period before primary particle formation. Fig. 1 shows a comparison of the time dependent changes in the 29Si-NMRspectra of a reaction mixture containing 0.1 M N H 3 , 2.2 M H20, and 0.5 M TEOS in ethanol versus methanol (samples El and M5). In both solvents only two 29Sifrequencies were observed, a primary TEOS peak and an intermediate peak. The primary TEOS or e," peak shifts appeared at -81.9 pprn in ethanol and at -81.3 pprn in methanol, when referenced to TMS at 0 ppm. The intermediate product, the singly hydrolyzed monomer of TEOS or Qi ,which was previously identified by Brinker and Scherer [32] in ethanol systems was detected at -78.9 ppm (Fig la). In Fig. lb, an intermediate product was detected at -80.4 ppm in the NH3 / TEOS / methanol system. It has been shown that this species is also Qi [33]. The results from the 29Si-NMRexperiments are summarized in Fig. 2. The plots show the transient behavior of the (3: and Qi during the based-catalyzed hydrolysis of TEOS. The most important aspect of this plot is the concentration of the hydrolyzed monomer, Qi at the point of nucleation. This precursor specie is responsible for nanophase (nuclei) formation. The [Qi]at the point of nucleation is higher in methanol (Fig. 2a) than in ethanol (Fig. 2b); thus, it should be expected from a thermodynamic argument that the size of the primary particles formed in methanol should be smaller than those formed in ethanol [34]. Thus, the size of the primary particles is governed by the supersaturation of Qi .
Ceramic Nanomaterials and Nanotechnology
41
The kinetic data shown in Fig. 12 indicate that the following reactions occurred in solutions [NH3] = 0.1 My[H20] = 2.2 My and in methanol (sample M5) and ethanol (sample El):
Qi
+H,O
&
SiO, +3ROH
where kh1 is the hydrolysis rate constant for TEOS, and ki2 relates to an overall hydrolysidcondensationrate constant. The rate constants for this reaction scheme are given in Table II. The hydrolysis of TEOS is faster in methanol than in ethanol, which has also been found by other researchers [10,35]. However, this research is the first to compare the condensation rates of Qi between both solvents, whch indicates that Qt condenses at essentially the same rate in ethanol as in methanol. A typical time sequence of the S A X S data are is shown in Fig. 3. Figs. 4a and 4b show the “average” slope, a, in the power law region and the radius of gyration Rg of silica growth relative to the ammonia and water concentration in methanol by S A X S . Increasing the or [H20] decreases the induction time; however, elevating [NH3] has a more pronounced effect on the initialization of silica growth. The lower cutoffs of the fiaction dimension (a 2 2.0) for the primary particle and size (Rg2 4.0) show that primary particles are open, ramified structures. There is no sign that the primary particles aggregate at the low water and ammonia concentrations when no additional salt is added to the reacting solution. The increase in a as the Rg remains constant indicates that densification occurs within the nuclei to produce ultimately compact, non-fiactal particles. The effect of the solvent on the rate of change of a and on the primary particle size can be seen by comparing reactions M5 for methanol and El for ethanol. These results are shown in Figs 5a-b, where in Fig. 5b, the first two Rg for El were obtained from the SAXS, while DLS was used to acquire the remaining size (RA)measurements. Figs. Sa-b indicate that primary particles in ethanol are approximately twice as large (&,&hano[ m 8 nm, Rg,mthanol = 4 nm) and form about 30 minutes after those in methanol. Similarly, the initial nuclei in both solvents are mass fiactal, possessing the same a = 2. The rate of increase of a is slower in ethanol where it reaches a value of 3.5 after 720 minutes as compared to 4.0 for methanol. Therefore, the densification of silica in methanol outpaces that in ethanol. Finally, low [HzO] and [NH3] in ethanol results in the nucleation of
m3]
42
Ceramic Nanomaterials and Nanotechnology
particles that retain their size as they densify. This characteristic is common to both solvents. The particle volume fiactions determined from the SAXS scattering profiles are shown in Fig. 6. Fig. 6a shows that the rate of increase of 4 increases with increasing [NH3] and [H20]. Furthermore, 9 calculated fiom SAXS and DLS for sample M5 in Fig. 6b agree with those determined fiom liquid-phase 29Si-NMR. Thus, these three techniques link the conversion of soluble silica (TEOS and its intermediates) into solid Si02. This indicates that (3: is responsible for the formation of nuclei and their densification through monomer addition. In ethanol (sample El), however, Fig. 6b shows that $I from 29Si-NMRwas nearly 0.001 higher than the corresponding measurements by DLS and S A X S towards t = 700 min. It is possible that the remaining mass denotes the presence of a highly condensed soluble species, such as 44, which would be undetectable under the methodology used to collect the 29Si-NMR data. Highly condensed silica species and low molecular weight soluble monomers have been detected by Klemperer and Ramamurthi [36,37] using gas chromatography. Figs. 7a and b show the effect of "€331, [H20], and solvent on the number concentration, N, of the silica particles. In Fig. 7a, the increasing N indicates that nucleation is continuous or that primary particles form by continuous nucleation. The production or nucleation rate of the silica particles is increased by raising the concentrations of NH3 and H20. Since the particles grown in methanol are essentially half the size of those in ethanol, the particle number concentration in sample M5 (Fig. 7 4 exceeds that produced in sample E l (Fig. 7b). The initial nucleation rate gn(t) for samples Ml - M5 and sample El are shown in Table IV, which also gives the polynomial nucleation rate fits to the N fiom Fig. 7. Higher-order polynomial fits were attempted, but a second order fit to the N(t)sms data resulting in a first order gn(t) captured the behavior of silica. These nucleation rate profiles are shown in Fig. 8. Table IV and Fig. 8 show that across samples M1 - M5, an order of magnitude increase in [NH3] fkom 0.01 0.1 M has a pronounced effect on the initial gn(t) and the nucleation rate profiles. Overall, Fig. 8 indicates that the nucleation rates decrease over the reaction period. However, Table IV indicates that the initial gn(t)is constant for samples M1 - M3, where [NH3] = 0.01 M and [H20] increases from 1.1 - 4.4 M. Comparison of gn(t)between samples M5 and E l shows the effect of the solvent. The initial nucleation rate is 16 times higher in methanol than in ethanol for reactant conditions of [NH3] = 0.1 M I' [H20] = 2.2 M / [TEOS] = 0.5 M. Table rV summarizes the effect of the salt on the first detectable size of the primary particles formed during the base-catalyzed hydrolysis of TEOS in methanol and ethanol by SAXS. The data indicate that although the initial particle size is affect by the solvent, the initial particle size is not affected by the ionic strength of the solution.
Ceramic Nanomaterials and Nanotechnology
43
To investigate the influence of the electrolyte concentration on the growing and reacting nanoparticles, the effect of the LiCl addition on sample M5 and on sample El was studied. Table IV summarizes the effect of the salt on the first detectable size of the primary particles formed dwing the base-catalyzed hydrolysis of TEOS by S A X S . The data indicate that although the initial particle size is affect by the solvent, the initial particle size is not affected by the ionic strength of the solution. Figs. 9a-b combine the of the primary particle determined fiom SAXS in Table IV with DLS measurements that monitor the subsequentparticle growth. Interestingly, Figs. 9a-b displays the greater ability of SAXS-to detect the primary particle over DLS. The extrapolation of the DLS growth curves to earlier times appears to converge around the initial detection of the primary particle by SAXS. This suggests that the SAXS detection of the primary particle was close to the time of the “actual” initial nucleation event. Overall, the particle growth data reveal an increase the rate of growth with an increase in the salt concentration. Thus, higher ionic strength enhances the aggregation of the initial primary particles.
CONCLUSIONS 29Si-NMR coupled with SAXS, and DLS are powerful techniques for measuring the initial stages of nanophase formation. m 3 ] and [H20] control the hydrolysis of TEOS and the condensation of the first hydrolyzed monorner to form the fnst primary particles. The difference in the size of the primary particles formed in methanol and ethanol is caused by a difference in the supersaturation of the first hydrolyzed monomer. This hydrolyzed specie controls the formation of fiactal nuclei through continuous nucleation. ‘Furthermore, the initial nanostrucutures densify through monomer addition of the hydrolyzed monomers. The ionic strength of the solutions does not affect the size of the initial nanophase; however, it does affect the aggregation dynamics of the initial nanoparticles formed during the base-catalyzed hydrolysis of TEOS. ACKNOWLEDGEMENTS This research was supported by the National Science Foundation @MR 9700860) and the Sloan Foundation. The author would like to thank Dr. J.S. Lin and Dr. Hacene Boukari for helpful discussions about SAXS.
44
Ceramic Nanomaterials and Nanotechnology
pprn
-77
-78
-79
-80
-81
-82
Figure 1. 29Si-NMR spectra of a 0.1 M N H 3 , 2.2 M HzO, 0.5 M TEOS mixture. a) Ethanol (Sample El); and b) Methanol (Sample MS). NMR spectra are proportional to peak area under initial scan.
Ceramic Nanomaterials and Nanotechnology
45
Figure 2. Soluble silica concentrations of Q," , and Qi from 29Si-NMR:a) 0.1 M NH3 / 2.2 M HzO (Sample El) in ethanol; and b) 0.1 M / 2.2 M H2O (Sample M5) in methanol ([Q:]o = 0.5 M). Solid lines are model fits to the data.
46
Ceramic Nanomaterials and Nanotechnology
Figure 3. The measured scattering profiles I(Q) vs. Q for 0.1 M NH3 / 2.2 M H 2 0 / 0.5 M TEOS in methanol (sample M5). Each curve is labeled with the time in minutes following the mixing of the initial reactants. The solid lines are the fitted power-law functions.
Ceramic Nanomaterials and Nanotechnology
47
and EH201 on the formation of Stober silica in Figure 4. The effect of methanol: a) a;and b) Rg as a function of time.
48
Ceramic Nanomaterials and Nanotechnology
Figure 5 . The effect of solvent on the formation of Stober silica in methanol (M5) and ethanol (El): a) a - SAXS; and b) Rg and Rh as a function of time. M5 sizes are from SAXS, whle the first two sizes for El are from SAXS and remaining sizes are from DLS.
Ceramic Nanomaterials and Nanotechnology
49
Figure 6. The effect of [NH3], [HzO], and solvent on the determination of the volume fraction $ by S U S , DLS, and 29Si-NMR:a) M1 - M5 (SAXS only); and b) M5 and El.
50
Ceramic Nanomaterials and Nanotechnology
Figure 7. The effect of [NH3], [H20], and solvent on the number concentration [NI of silica as determined fiom SAXS and DLS: a) M1 - M5; and b) El. Solid lines are the polynomial fits to the data.
Ceramic Nanomaterials and Nanotechnology
51
1 E13
-
n Y C
U]
1E12
Time (min)
Figure 8. The effect of “€331, [H20], and solvent on the nucleation rate profiles for polynomial fits in Table 111 for sample M1- M5 and El.
52
Ceramic Nanomaterials and Nanotechnology
30 No sait 0001 M O.a)EM
25
ODEM
20 h
E C
- 12
15-
v
-8 - 4 SAXS RimaryPartide
0
,
0
I
,
100
,
K
200
l
I
.
300
400
'
500
I
'
I
'
I
'
800
700
600
O
900
Time (min)
J
25
-
- 20
20
-
-
16
n
E C
v
15-
-12
10-
-8
E
ac' - 4 5 1
o
0
l
,
l 100
,
l
200
,
l
300
,
l 400
,
l 500
,
l 600
,
l 700
,
l 800
,
I0 900
Time (min)
Figure 9. The effect of [LiCl] on the particle size R, and Rh of silica. a) methanol (sample M5);b) ethanol (sample El) ([LiCl] = 0 - 0.005 M).
Ceramic Nanomaterials and Nanotechnology
53
Table I. Initial Stober reactant concentrations studied by 29Si-NMR,SAXS, DLS. Solvent
Sample
SAXS Configuration
Methanol
M1
2m
0.01
1.1
0.5
M2
2m
0.01
2.2
0.5
M3
2m
0.01
4.4
0.5
M4
2m
0.05
1.1
0.5
M5
2m
0.1
2.2
0.5
El
2m, 5m
0.1
2.2
0.5
Ethanol
fNH31
(M) [H201 (M)
[TEos] (M)
Table 11. Rate constants for k h ] and kh2,. 1
1,
1,
dl
I
I
0.10 / 2.2 / 0.5
ktnanor
.08 3 , Y l f u.03
Sample E l
88.3 k 0.4”
I
1
78.0 k 1.0”
Table 111. The silica nucleation rate gn(t)from SAXS and DLS in methanol and ethanol. Sam- gn(t’fnuc)SAXS x l 0 l 2 gn(t=tnuc)DLS x 10’’ gn(t) - Polynomial Fit [l/cm3min]
54
ple
[l/cm3min]
M1
3.6 f 0.2
M2
3.9 rt 0.1
M3
3.8 f 0.1
M4
9.68 f 0.03
M5
19.4 f 0.7
El
1.2 f 0.1
1/cm3min]
---
(-1 .O f 3.0)e+08t + (5.0 & 3.0 )e+12 (-9.4 -t- 0.3)e-t08t + (4.6 f 0.2)e+12 (-5.2 f 0.2)e+09t + (7.7 fO.l)e+12
---
(-9.6 k 0.3)e+09t + (1.3 f O.l)e+13 (-2.6 5 0.2)e+10f + (3.0 5 OS)e+13
1.47 f 0.01
(-1.9 k Oml)e+09t+(1.9 f 0.4)e+12
Ceramic Nanomaterials and Nanotechnology
Table IV.The effect of [LiCl] on the size (R3 of the primary particle in methanol and ethanol (SAXS - m 3 ]= 0.1 M; [HzO] = 2.2 M; [TEOS] = 0.5 M; [LiCl] = 0 - 0.001 M).
Solvent and Size
No Salt
Intermediate
High - 0.001 M
Methanol - Rg (nm)
4.8 k 0.3
4.0k 0.2
4.8 f 0.4
Etha01 - R g (m)
8.1 k 0.6
8.6 f 0.8
8.6 k 0.6
a - [LiCl] in methanol = 0.0075 M; [LiCl] in ethanol = 0.0005 M
Ceramic Nanomaterials and Nanotechnology
55
REFERENCES Iler, R.K., f i e Chemistv of Silica, John Wiley & Sons, 1979. Payne, C., “Applications of Colloidal Silica: Past, Present, and Future,” in H.E. Bergna Ed., The Colloid Chemistty of Silica, American Chemical Society, p. 58 1 (1 994). Stober, W., Fink, A., Bohn, E., “Controlled Growth of Monodisperse Silica Spheres in the Micron Size Range,” J. Colloid Interface Sci., 26,62 (1968). Bogush, G.H., Tracy, M.A., Zukoski, C.F., “Preparation of Monodisperse Silica Particles: Control of Size and Mass Fraction,” J. Non-Crystalline Solids, 104,95 (1988). Matsoukas, T. and Gulari, E., “Dynamics of Growth of Silica Particles fi-om Ammonia -Catalyzed Hydrolysis of Tetra-ethyl-orthosilicate,” J. Colloid. Interface Sci., 124,252 (1988). Matsoukas, T. and Gulari, E., “Monomer-Addition Growth with Slow Initiation Step: A Growth Model for Silica Particles from Alkoxides”, J. Colloid Interjiace Sci., 132, 13 (1989). Bogush, G.H. and Zukoski, IV, C.F., “Studies of the Kinetics of the Precipitation of Uniform Silica Particles through the Hydrolysis and Condensation of Silicon Alkoxides”, J. Colloid Interface Sci.,142, 1 (1991). Bogush, G.H. and Zukoski, C.F. N ,“Uniform Silica Pdrticle Precipitation: An Aggregative Growth Model,” J. Colloid Interface Sci., 142, 19 (199 1). Lee, K., Look, J., Harris, M.T., McConnick, A.V., “Assessing Extreme Models of the Stober Synthesis Using Transients under a Range of Initial Composition”,J. Colloid InteP-faceSci., 194,78 (1 997). Harris, M.T., Brunson, R.R., Byers, C.H., “The Base-Catalyzed Hydrolysis and Condensation Reactions of Dilute and Concentrated TEOS Solutions,” J. Non-Crystalline Solids, 121,397 (1990). van Blaaderen, A. and Kentgens, A.P.M, “Particles Morphology and Chemical Microstructure of Colloidal Silica Spheres Made from Alko~ysilanes,’~ J. Non-Crystalline Solids, 149, 161 (1992). Bailey and Mecartney, “Formation of Colloidal Silica Particles from Alkoxides,” Colloids and Surfaces, 63, 15 1 (1 992). Himmel, B. and Gerber, Th., “X-ray Diffraction Investigation of Tetraethoxysilane and Ethanol in Liquid and Vapor Phases,” J. Non-Cryst. Solids. 159,235 (1993). Boukari, H., Lin, J.S., Hamis, M.T., “Small-Angle Scattering Study of the Formation of Colloidal Silica Particles fiom Alkoxides: Primary Particles or Not?”, J. Colloid Interjiace Sci.,194,3 1 1 (1997). Boukari, H., Lin, J.S., Harris, M.T., “Probing the dynamics of the Silica Nanostructure for Growth by SAXS, Chem. Mat. , 9,2376 (1997).
56
Ceramic Nanomaterials and Nanotechnology
Boukari, H., Long, G.G., Harris, M.T., “Polydispersityduring theFormation and Growth of the Stober Silica Particles from Small-Angle X-ray Scattering Measurements,”J. Colloid Inte$ace Sci. 229,129 (2000). Guinier, A. and Foumet, G., “Small-Angle Scattering of X-Rays”, Wiley, 1955. Weiner, B., “Particle Size Using Photon Correlation Spectroscopy,” in “Modern Methods of Particle Size Analysis” (G. Barth, Ed.), John Wiley and Sons, 1984. Bale, H.D. and Schimdt, P.W., “Small-Angle X-Ray Scattering Investigation of Submicroscopic Porosity with Fractal Properties,” Phys. Rev. Lett. 53, 596 (1984). Texiera, J., “Small-Angle Scattering by Fractal Systems,” J. Appl. C’st. 21, 781 (1988). Schaefer, D.W. and Keefer, K.D., “Fractal Geometry of Silica Condensation Polymers,” Phys. Rev. Lett. 53, 1383 (1984). Schmidt, P.W., in “The Fractal Approach to Heterogeneous Chemistry: Surface, Colloids, Polymers” @. Avnir, Ed.), Wiley, England, 1989. Schmidt, P:W., “Small-Angle Scattering Studies of Disordered, Porous, and Fractal Systems” J. Appl. Cryst. 24,414 (1991). Schmidt, P.W., “Some Fundamental Concepts and Techniques Useful in Small-Angle Scattering Studies of Disordered Solids,” in “Modern Aspects of Small-Angle X-Ray Scattering” (H. Brumberger, Ed.), Kluwer, Netherlands, 1995. Schaefer, D.W., Brow, R.K., Oliver, B.J., Rieker, T., Beaucage, G., Hrubesh, L., Lin, J.S., “Characterization of Porosity in Ceramic Materials by Small-Angle Scattering: Vycor Glass and Silica Aerogel” in ‘Modern Brumberger, I. Ed.), p. 299, Aspects of Small-Angle X-Ray Scattering” (€ Kluwer, 1995. Stuhnnann, H.B.,“Contrast Variation,” in “Modern Aspects of Small-Angle X-Ray Scattering” (H. Brumberger, Ed.), p. 22 1, Kluwer, Netherlands, 1995. Hiemenz, P.C. and Rajagopalan, R., “Principles of Colloid and SunFace Chemistry, 3rdEdition”, Marcel Dekker, 1997. Beaucage, G., Personal Communication,2001. Hendricks, R.W., “The ORNL 10-Meter Small-Angle X-Ray Scattering Camera,”J. Appl. Cryst., 11, 15-30 (1978). Wignall, G.D., Lin, J.S., Spooner, S., “Reduction of Parasitic Scattering in Small-Angle X-Ray Scattering by a Three-Pinhole Collimating System,” J. Appl. Cvst., 23,241-245 (1990). Yudin, I.K., Nikolaenko, G.L., Kosov, V.I., Agayan, V.A., Anisimov, M.A., Sengers. J.V., “A Compact Photon-Correlation Spectrophotometer for Research and Education,” Int. J. of Themzophysics, 18, 1237 (1997).
Ceramic Nanomaterials and Nanotechnology
57
[32] Brinker, C.J.and Scherer, G.W., “Sol-Gel Science”, Academic Press, 1990. [33] Green, D.L., Jayasundera, S., Lam, Y.-F., Harris, M.T., “Chemical Reaction Kinetics Leading to the First Stober Silica Nanoparticles - NMR and SAXS Investigation,” J. Non-Crystalline Solids (submitted 2002). [34] Green, D.L., Lin, J.S., Lam, Y.-F., Hu, M.2.-C., Schaefer, D.W., Harris, M.T., “Size, Volume Fraction and Nucleation of Stober Silica Nanoparticles,” J. Colloid Intevface Science (submitted 2002). [35] Sadasivan, S., Dubey, A.K., Li, Y , Rasmussen, D.H., “Alcoholic Solvent Effect on Silica Synthesis - NMR and DLS Results,” J: Sol-Gel Sci. Tech. 12,5 (1998). [36] Klemperer, W.G. and Rmamurthi, S.D., “Molecular Growth Pathways in Silica Sol-Gel Polymerization,” Mat. Res. Soc. S’mp. Proc. 121, 1 (1988). [37] Klemperer, W.G. and Rammurthi, S.D., “A Flory-Stockmayer Analysis of Silica Sol-Gel Polymerization,” J. Non-Cyst. Solids. 121, 16 (1990).
58
Ceramic Nanomaterials and Nanotechnology
Nanoparticles in Chip Miniaturization Xangdong Feng*, Mike S.H. Chu, Heber E. Rast, and Brian C. Foster Ferro Corporation, 7500 E. Pleasant Valley, Cleveland, OH 4413 1 ABSTRACT Reductions in semiconductor operating voltages coupled with a demand to produce higher volumetric efficiencies in multilayer ceramic capacitor (MLCC) devices have led to a significant reduction in the internal dielectric thickness of these components and dramatic increases in layer counts. This trend has resulted in the rapid demand in nanosized dielectric powder such as BaTi03, electrode metal powders, and chemical mechanical planarization slurries. A brief review of the nanoparticle opportunities and synthetic methods for manufacturing nano BaTi03 is discussed in terms of advantages and challenges.
NANOPARTICLES IN DIELECTRICS The main drivers for technology innovation in the electronic industry have been miniaturization, higher frequency, lower operating voltages and reduction of component part count. In a multiband cell phone, there are typically 150-300 passive components with 50-70% capacitors, 20-40%resistors and 10% inductors. The worldwide consumption of multilayer ceramic capacitor (MLCC, Fig. 1.) has increased steadily as shown in Fig.2
'.
* Corresponding author,
[email protected]
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
59
Fig.1. A schematic drawing of Multilayer Ceramic Capacitor
Fig.2. Worldwide Consumption of MLCC
to
60
Ceramic Nanomaterials and Nanotechnology
Standard ceramic capacitors have continued to evolve in two primary directions: smaller-siked components and larger capacitance values2. The use of lower operating voltages in handheld devices and in microprocessors has allowed dielectric layer thickness in multilayer ceramic capacitor (MLCC) devices to be reduced, and, consequently, higher layer counts become possible. It is predicted that the dielectric thickness of the MLCC device will reach less than 800 nm by the year 2003 and to less than 600nm by 2005. On the other hand, the layer count in the MLCC is increased dramatically and it is predicted to be over 1000 layers in MLCC beyond year 2002 as shown in Fig. 3.
Fig.3. Dielectric Layer Thickness and Count Trend
In order to keep reliability level, there should be at least a few sintered grains between two electrode layers. Therefore, dielectric powder such as BaTi03 should be less than 1/5 of layer thickness (assuming no grain growth after sintering and five grains are needed). For a 0.6-micron layer, dielectric powder size should be less than 120 nm, i.e., in the nanosize range of about 1OOnm. This presents a great challenge for the industry to make such BaTi03 at commercial
Ceramic Nanomaterials and Nanotechnology
61
scale with reasonable cost and it also presents an exciting opportunity for nanoparticle synthesis technologies. NANOPARTICLESIN CHEMICAL MECHANICAL PLANAIUZATION Continued miniaturization of the device dimensions and the related need to interconnect an increasing number of devices on a chip have led to building multilevel interconnections on plmarized levels. Photolithography would be hard pressed to meet such miniaturization demand of uncompromised precision. Chemical Mechanical Planarization (CMP) is the process of smoothing and planning aided by the chemical and mechanical forces. CMP is playing an important role in chip miniaturization. With its ability to achieve global planarization, its universality (materials insensitivity), its ability to multimaterials surfaces, and its relatively cost-effectiveness, CMP is the ideal planarizing medium for the interlayered dielectqcs and metal films used in silicon integrated circuit fabrication. Fig.4 shows the impact of CMP on a multilevel integrated circuit. Fig. 4. Cross-Sections of Integrated Circuitsbefore (left) and after CMP (right)
The CMP process is accomplished by the utilization of CMP slurry, which usually consists of nanoparticles silica or cerium oxides in a formulated aqueous solution. The CMP process is illustrated in Fig. 5 .
62
Ceramic Nanomaterials and Nanotechnology
Fig.5 . An Illustration of Chemical Mechanical Planarization Process
As shown in Fig.5, CMP process involves placing a wafer face-down on a carrier. The rotating carrier is pressed against a rotating polishing pad and abrasive CMP slurry is dropped onto the pad. The combination of downward pressurekhemical actiodrotation smoothes and planarizes the surfaces.
CMP slurries present another large and emerging opportunity for nanoparticles. Its market size is expected to go rapidly from the current $400 million per year to about $770 million dollars in the year 2006 at an annual growth rate of over 20% as shown in Fig. 6 .
Ceramic Nanomaterials and Nanotechnology
63
Fig.6. CMP Sluny Market Projection
NANOPARTICLE SYNTHESIS METHODS Nanoparticles have large surface areas and high reactivities due to the unsaturated bonds on their pristine surfaces. Nanoparticles tend to react among themselves to form necking to agglomerate into large secondary particles. This imposes extreme challenges for the industry to manufacturing nanoparticles in large quantities. To select the proper manufacturing methods for a specific application it requires a thorough understanding of the fundamental aspects of the nanoparticle formation of the methods. There are numerous methods for making nanoparticles and these methods can be classified 3:
64
-
By synthesis strategy4’ o Topdown o Bottomup
-
By nature of the process: o Physical o Chemical
Ceramic Nanomaterials and Nanotechnology
o o
-
By Energy sources o o o o o o o o o o o o o
-
Biological Combinations Laser Plasma Joule heating Sputtering ElectronBeam Microwave Hydrothermal Freeze Drying High-energy ball mill Combustion Flame Supercritical Chemical reaction
By media in which nanoparticles are formedG9 o Gas phase synthesis o Solid phase synthesis o Liquid phase synthesis
In t h s review, the emphasis is on BaTi03 synthesis. There have been a number of excellent reviews on BaTi03 synthesis *-13. However, most of these reviews were limited in scopes and were not updated to reflect the considerations for making nano BaTi03 particles. GAS PHASE SYNTHESIS
In general, gas phase syntheses refer to those methods where gaseous molecules and atoms are directly converted into nanoparticles through gas-phasereaction 14* , condensation, and/or decomposition, nucleation, growth, and possibly agglomeration. However, in the synthesis of BaTi03 there have been no reported methods involving only pure gases molecules. The gas phase syntheses discussed here are only pseudo gas phase syntheses since they usually involve small liquid droplets dispersed in gases during synthesis. These pseudo gas phase syntheses can be further classified into the following two broad types: aerosol pyrolysis and flame aerosol pyrolysis.
Ceramic Nanomaterials and Nanotechnology
65
Aerosol Pyrolysis
An aerosol pyrolysis process is also called spray pyrolysis or vapor pyrolysis. It consists of three parts: aerosol generating, particle formation inside a furnace, and particle collection as shown in Fig.7. A homogeneous liquid solution (organic, aqueous, or a mixture) containing the reactants are converted into aerosol using air, nitrogen gas or mixture of gases. The sizes of the droplets formed in the aerosol depending on the types of aerosol generators used, carrying gas volume, and the nature of the liquid 18. An ultrasonic generator usually gives very fine droplets that may be suitable for making nano particles ". The aerosol is then carried by the gas into a preheated fixmace, where the liquid droplets are going through solvent evaporation, solute precipitation, solute decomposition, and oxide sintering to final particles. Eventually the particles are carried out by the gases into particle collection section, where the particle may be quenched with cold gases and collected by filters or bags (Fig.7). Fig.7. Schematic representation of an aerosol pyrolysis process
66
Ceramic Nanomaterials and Nanotechnology
In order to generate uniform and small droplets of aerosol, the first step of the process is to prepare a uniform of solution containing barium and titanium. This solution can be made fiom simple inorganic salts such as T i c 4 and BaC12 19, or Tic4 and Ba(N03)2 **in a mixture of water and ethanol. The starting materials can also be a mixture of Ti alkoxides and barium salts such as Ba(N03)2 in a 22. The sources of titanium and barium have also mixture of water and alcohols211 been tried with both organometallic compounds such as titanium alkoxides with barium acetate 22 or titanium lactate and barium acetate 23. The different starting materials have shown effects on the BaTi03 formation mechanism and the characteristics of BaTi03 formed. Tick has a strong tendency to disassociate and hydrolyze according to the following reaction in solutions as a bulky solution or as aerosol droplets'': T i Q + H20 = Ti02' + 2H++ 4C1' = TiOC12 + 2HC1
(1)
When the droplets are carried into the h a c e , the solvent begins to evaporate and further hydrolysis happens according to: Ti02++ (x+l)H20= 2H++Ti02.xH20
(2)
TiOC12 + (x+l)HzO = Ti02.xH2O + 2HC1
(3)
Or
Further heating: TiO, .xH,O
-380uc
> TiO, (anatase) + x H,O
(4)
If BaC12 involved: BaC12(s) + Ti02(anatase) + H20(g) = BaTi03 (s) + 2HC1
(5)
In the case of Ba(NO3)2 used, there may be also Reaction (5) involved due to the formation of BaC12: Ba(N03)2 + 2HCl= BaC12 + 2HN03 The following reaction was also reported:
Bu(NO,),
+ TiO, (anatase)
540-6500c
Ceramic Nanomaterials and Nanotechnology
> BaTiO, + 2N0,
+ -10, 2
67
An example of BaTi03 particles prepared fiom aerosol process is shown in Fig. S(a), where particles of 50 - 600nm were obtained. These particles are tetragonal in crystal structure with a BdTi ratio of 0.999924.A narrower size distribution of the BaTi03 particles from this process requires quicker quenching Fig. 8. BaTi03 prepared fiom aerosol pyrolysis (a) and flame pyrolysis (b). .j i cl 1.. LP p 3 ,
, I
f W\
of the pyrolyzed gas phase and the generation of truly nanosized BaTi03 may require such a large dilution of the particle stream with the carrying gas, which makes this process extremely inefficient. The scale-up to production of tonnage quality may also be difficult. Flame-aerosol pyr01ysis Flame synthesis is a process where aerosol of barium and titanium containing compounds are bumed in a flame 16. An example of such BaTi03 are shown in Fig.8@).The BdTi ratio is near unity with cubic structure. Particle size is between 50nm to 1l O O n m , which is very wide. It would be a great challenge to produce unagglomerated BaTi03 with narrow size distribution.
68
Ceramic Nanomaterials and Nanotechnology
SOLID PHASE SYNTHESIS Solid phase synthesis of barium titanate can be further classified into solid state and mechanochemical process. Solid state synthesis
This is the most traditional approach for making BaTi03. It involves milling BaC03 with Ti02 to achieve comminution and mixing. This mixture is then calcined at 900-22OOC to generate BaTi03: BaC03 + E 0 2 = BaTi03 + C02. Ths process is simple, economical, but usually produces large and agglomerated particles. Contaminations fiom milling media are also a concern. Intensive milling of BaC03 and Ti02 has reduced calcination temperature to 750C in order to reduce BaTi03 sizes25.Replacement of Ti02 with TiO(OH)2 produced 200 nm BaTi03 through calcination at 850C26. Efforts have also been spent on using ultrafine raw materials of Ti02 and BaC03 andor on more sophisticated milling technology of the raw materials in order to reduce calcinations temperature and achieve smaller particles. Mechanochemical synthesis In this process BaC03 and solid Ti02 are milled together with the presence of large amount of sacrificial salts such as NaCl to form nanoparticles. NaC1 was to physically prevent the resulting BaTi03 fi-om agglomeration. A final washing of NaCl with large amount of water is needed to recover BaTi03. BaTi03 of 60 nm was reported to be made after heating the milled mixture at 700C27. However, the ability to achieve phase pure materials with the proper Ba/Ti ratio, to prevent contamination from milling media and NaCl, and to reduce the amount of waste water fiom extensive water-washing are still issues. LIQUID PHASE SYNTHESIS
Liquid phase syntheses are the most common and diverse methods for BaTi03 synthesis. The critical particle formation step occurs in the liquid phase. For supercritical conditions, the particle formation phase is at both liquid and gas
Ceramic Nanomaterials and Nanotechnology
69
phase, since there is no distinction between gas and liquid under those conditions. The most common solvent is water, but more and more organic solvents are also used in an effort to reduce agglomeration of nanoparticles. Common to all the liquid phase synthesis is that they begin with solution preparation. A true solution means homogeneity at atomic or molecular scale. The central goal for the diverse solution synthetic methods is to preserve as much of this homogeneity as possible in the particle formation process.
The numerous liquid phase synthesis methods can be classified into six basic groups, according to the difference in their chemistry. They are: 1. Double compounding Oxalate Citrate Catecholate Isoproxide
2. Sol-gel Gel Precipitates Complexgel Sol-emulsion 3. Stabilization Freezedrying Microemulsion Spontaneous-combustion Stable alkoxides
4. Low temperature direct formation Alkoxides (SP, and water vapor) Inorganic salts Crystalline Ti02 5 . Hydrothermal
-
In-situ transformation Dissolution and crystallization
6. Other methods Supercritical Dynamic cavitation
70
Ceramic Nanomaterials and Nanotechnology
A review paper on the liquid phase BaTi03 synthesis is being preparedz8. A brief summary on these liquid phase syntheses is provided here. Among the double compounding synthesis methods, oxalate is the only commercial method that has produced large quantity of BaTi03 particles at sizes of 500 nm or smaller with exceptional dielectric properties. Its advantages include economical raw materials, simple process, lower calcination temperature, and smaller particle size than solid state methods. There have been extensive methods developed to improve the oxalate methods aiming at smaller particle sizes. Da et. al.29 have made 200-400 nm BaTi03 by lowering the solution pH to 2.5-3.0 during oxalate process. Microemulsion was also used to control the size of barium titanium oxalate to achieve particles around 100 m3'.Smaller particle sizes in oxalate process by manipulatin calcination processes were also explored by calcination in vacuum3', control ratj2, and by using inhibitor^^^. However, it would be extremely challenging to use oxalate methods to make nano-sized BaTi03 at large scale. Sol-gel methods have been shown to be able to make small size BaTi03 particles with good sintering and electrical properties. However, various precursors often show different decomposition behavi02~. The formation of intermediate compounds may destroy the initial homogeneity. Higher temperature and prolonged calcinations are usually needed to restore homogeneity, which leads to larger particles. Stabilization synthesis of BaTi03 consists of two steps: stabilizing a molecular mixture of Ba and Ti by complexing, polymerization, or high soluble nitrates; preserving the molecular homogeneity through fieez drying35, and microemul~ion~~~ 38. The stabilization methods are spontaneous cornbu~tion~~, usually complicated and require high temperature calcinations, which promotes agglomeration. Hydrothemal synthesis is another advanced commercial method for ultrafine BaTi03 synthesis. While a variety of chemistries can be used, the common feature of hydrothermal processing is that reaction temperatures of 100400 "C are used to crystallize barium titax~ate~'~~. This involves the use of an autoclave for batch processing or a tubular reactor for continuous processing. Two companies, Sakai Chemical in Japan, and Cabot Corporation in the USA, have had manufacturing capability for hydrothennal barium titanate and some formulated dielectrics for about 10 years. Sakai Chemical is reported to have a
Ceramic Nanomaterials and Nanotechnology
71
capacity of about 250 metric tons per month and is a major supplier to several large companies such as Kyocera and Samsung. Hydrothermal synthesis has the advantages of direct formation of BaTi03 without calcinations, easy control of particle size from 30 to 30Onm without much agglomeration, and versatile chemistry and raw materials. The challenges are the control of BdTi ratio, reaction completion, and elimination of hydroxyl defects in the BaTi03 crystals. Low temperature direct synthesis is usually carried out at temperature below lOOC at ambient pressure in air or C02-free atmosphere434*.The reaction time is between a few minutes to a few days. The solution pH is usually 13 or higher. Very fine BaTi03 particles of as small as 10 nm have been made using this method. The challenge for this method is the elimination of alkali contamination, control of BdTi ratio, improvement in sintering and dielectric properties.
SUMMARY
Chip miniaturization has provided great opportunities for nanoparticles. The demand for nanoparticles of dielectrics, electrodes, and CMP is increasing exponentially each year. Because of extremely high activities of nanoparticles, it has imposed great challenges for the manufactures to produce unagglomerated nanoparticles. Among the many synthetic methods for BaTi03, some methods are much more suitable for the manufacturing nano BaTi03 than others to achieve derive the desired industrial properties such as particle size distribution and stoichiometry. The selection of the proper methods requires a solid understanding of the fundamentals of each synthesis.
REFERENCES 1. 2.
3.
72
Paumanok Publications, “Ceramic Capacitors World Markets, Technologies & Opportunities:A 1998-2003 Technical-Economic Analysis” Tech. Report No. ISBN# U-929717-64-3 (1998). T. G. Reynolds III, Am. Cerarn. Soc. Bull. 80,29 (2001). X. Feng, R. Harris, paper presented at the Fine, Ultrafiie and Nano Particles ‘01, Chicago, IL, October 14-17,2001,
Ceramic Nanornaterials and Nanotechnology
4.
5.
6. 7. 8. 9. 10. 11. 12. 13.
14. 15. 16. 17. 18. 19.
20. 21. 22.
23. 24. 25.
26. 27.
M. C. Roco, in Nanostmctured Materials G. M . Chow, N. J. Noskova, E&. Kluwer Academic Publishers, (1998), vol. 3, pp. 71. E. L. Hu, D. T. Shaw, in Nanostructure Science and Technology R. W. Siegel, E. Hu, M. C. ROCO,Eds, Kluwer Academic Publishers, (1999) pp. 16. F. E. h i s , H. Fissan, KONA 17, 130 (1999). D. W. Johnson, Jr., in Advances in Ceramics: Ceramic Powder Science. The American Ceramic Society, Inc., (1987), vol. 21, pp. 3. V. R.W. Amarakoon, B. C . LaCourse, 0. Hernandez, J. G. Fagan, J. Pietras, J. L. Peterson, in Ceramics Transactions. (1995), vol. 49, pp. 15. K. S. Mazdiyasni, in Ceramic Transactions. (1990), vol. 15, pp. 23. P. P. Phule, S. H. Risbud, J. Mater. Sci. 25, 1169 (1990). A. D. Hilton, R. Frost, Key Engineering Materials 66 & 67, 145 (1992). A. K. Maurice, R. C. Buchanan, Ferroelectrics 74,61 (1987). J. M. Herbert, Proceedings of the British Ceramic Societ). 18, 157 (1970). S. Pratsinis, Prog. Energy Combust. Sci. 24, 197 (1998). H. D. Jang, J. Jeong, J. Aerosol Sci. Techn. 23,553 (1995). T. T. Kodas, M. J. Hampden-Smith, Aerosol Processing of Materials Wiley-VCH, New York, (1999). K. Kimura, in Fine Particles: Synthesis, Characterization, and Mechanisms of Growth. (2000) pp. 513. H. Muhlenweg, A. Schild, A. Gutsch, C. Klasen, S. Pratsinis, Journal of Metastable and Nanocrystalline Materials 8,941 (2000). 0. B. Milosevic, M. K. Mirkovic, D. P. Uskokovic, J. Am. Ceram. Soc. 79,1720 (1996). H.-J. Cho, J.-H. Lee, S.-J. Park,Ceramic Transactions22,51 (1991). K. Nonaka, S. Hayashi, K. Okada, N.Otsuka, T. Yano, J. Mater. Res. 6,1750 (1991). T. Ogihara, H. Aikiyo, N. Ogata, N. Mizutani, Advanced Powder Technology 10,37 (1999). J. Ortega, The University of New Mexico (1992). X. Feng. (2000). V. Berbenni, A. M a d , G. Bruni, Thermochimica Acta 374, 151 (2001). K. R. Han, H. J. Koo. Korea Institute of Science and Technology, International, (2001). P. McCorrnick, paper presented at the Fine, Ultrafine and Nano Powders 1999.
Ceramic Nanomaterials and Nanotechnology
73
28. 29. 30. 31. 32. 33. 34. 35.
36. 37. 38. 39. 40. 41.
42. 43. 44.
45.
46. 47. 48.
74
X. Feng, M. S. H. Chu, E. Rast, B. Foster, Manuscript in preparation (2002). G. Da, G. Yanfang, Z. Baicun, H. Liming, Journal ofEast China University of Science and Technology 22, 536 (1996). J. Wang, J. Fang, S.-C. Ng, L. M. Gan, C. H. Chew, X. Wang, Z. Shen, J. Am. Ceram. Soc. 82,873 (1999). M. D. Alcala, F. J. Gotor, L. A. Perez-Maqueda, C. Real, M. J. Dianez, J. M. Criado, Journal of nermal Analysis and Calorimetry 56, 1447 (1999). A. V. Ragulya, Nanostructured Materials 10,349 (1 998). Y . Enomoto, A. Yamaji, Am. Ceram. Soc. Bull. 60,566 (1981). H . P. Beck, W. Eiser, R. Haberkorn, Journal of the Ceramic Society of Japan, 687 (2001). L. I. Martynenko, 0. A. Shiyakhtin, D. 0. Charkin, Inorganic Materials 33,489 (1997). P. Pramanik, Bull. Mater. Sci. 19,951 (1996). H . Hemg, R. Hernpelmann, Nanostructured Materials 9,241 (1997). W. Xiangde, Z. Xiaoming, W. Yinhua, Z. Bin, Z. Xiujuan, Technology of Water Treatment 24 (1998). S. Erwei, X. Changtai, Z. Weixhou, H. Sukun, Y . Zhiwen, Journal of the Chinese Ceramic Society 24,45 (1996). K. W. Seo, J. K. Oh, Korean J. Chem. Eng. 17,47 (2000). J. Moon, J. A. Kerchner, H. Krarup, J. H. Adair, J. Mater. Res. 14,425 (1999). J . 0. Eckert Jr., C. C. Huang-Houston, B. L. Gersten, M. M. Lencka, R. E. Riman, J. Am. Ceram. Soc. 79,2929 (1996). H . Matsuda, T. Mizushima, M. Kuwabara, Journal of the Ceramic Society of Japan 107,290 (1999). M . Leoni, V. Buscaglia, C. Aliprandi, Journal of the European Ceramic Society 14, 85 (1995). M. Leoni, M. Viviani, A. Testino, P. Nanni, C. Mingazzini, British Ceramic Proceedings 61,47 (1999). M . Leoni, V. Buscaglia, P. Nanni, M. Viviani, Advances in Dielectric Ceramic Materials 88, 159 (1998). S. Wada, H. Chikamori, T. Noma, T. Suzuki, Journal of Materials Science 35,4857 (2000). H . S. Potdar, P. Singh, S. B. Deshpande, P. D. Godbole, S. K. Date, Materials Letters 10, 112 (1990).
Ceramic Nanomaterials and Nanotechnology
SYNTHESIS AND CHARACTERIZATIONOF NANOPARTICLES OF STABILIZED ZIRCONIA
C. R. Foschini Massachusetts Institute of Technology,MIT 77. Massachusetts Avenue, 13-1406 Cambridge,MA USA 02139
B, D.Stojanovic; C.O.Paiva-Santos L. Perazolli and J. A. Varela Instituto de Quimica, W E S P Aruraquara, SP Brasil 14801
ABSTRACT Reactive powders have been synthesized by using several chemical processes. In this work the method used consisted of the complexation of zirconium metal from zirconium hydroxide through a solution of 8hydroxiquinoline. The crystallization kinetics of zirconia was followed by X-ray diffkaction, scanning electron microscopy and surface area measured by nitrogen adsorptiorddesorption technique. The results indicated that zirconia with surface area as high as 100m2/g can be obtained fiom this method after calcining at 600°C. The tetragonaVcubic phase was stabilized at room temperature where the particle size was lower than 20nm. Keywords: Nanopowder; Stabilized Zirconia; Rietveld Method INTRODUCTION Zirconia is an important material for high temperature applications'. There are four well-defined polymorphic forms: monoclinic, tetragonal, orthorhombic and cubic phases2. At room temperature, only the monoclinic form is stable, although both the tetragonal and orthorhombic phases can be quenched to the ambient conditions. Addition of appropriate dopants, however, is known to stabilize the tetragonal and cubic phases3*'. The crystal structure of and mechanisms of the transformations between the monoclinic, tetragonal and cubic phases are of considerable technical interest, since they can be manipulated to provide optimized ph sical and chemical properties of the materials fabricated from stabilized zirconiaX 8 . The so-called partially stabilized zirqonia (PSZ), which are typically two phases (cubic and tetragonal) or single phase (tetragonal), are of importance for mechanical and structural applications. The fully stabilized zirconia (FSZ), normally single phase cubic, are of interest for heating elements, oxygen sensors, To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
75
fuel cells, electrolytes, coatings, structural and wear components, and toughening agents in many ceramic compositesg~". The interest on the synthesis of submicrometricspowders is increasing due to the necessity of products with high performance in the ceramic industry'*. Theses powders are characterized by high purity (>99,9%), controlled and reproducible chemical composition (including dopant addition), chemical homogeneity on atomic level and controlled particle size and shape. Colloidal particles can be obtained by several methods, in liquid or vapor phasesI3-l6, but the techniques involving chemical solutions are more convenient for preparing fine powders having high purity. On this method, the solid phase can be obtained from a solution that contains the desired cations by precipitation, evaporation or solvent extraction. The segregation is reduced by combining the ions in a gel phase or by means of quick solvent extraction (milliseconds). The solid phase is usually a salt that can be calcined at low temperatures. This calcined powder is porous and brittle, so it can be easily ground to submicrometric size. In this work an alternative chemical route used to obtain ZrO2 powders, with a high surface area was investigated.
EXPERIMENTAL PROCEDURE An aqueous solution of zirconium ions was prepared by dissolution of Zr(OH)4 in acid medium. To this solution 8-hydroxiquinoline (Q) 12% in acetic acid was added, followed by neutralization with concentrated ammonium hydroxide solution, to give the precipitate. The precipitate was collected by filtration on Biichner funnel, washed with hot water to remove possible excess of 8-hydroxiquinoline, air dried at room temperature during 24 hours and stored in dessicator. The resulting powders were calcined at different temperatures from 300°C to 1200°C. X-ray difiaction (XRD) (Model D5000, Siemens), with Cu Ka radiation and a graphite monochromator was used for the phase formation study. For refinements purposes it was added 30 weight % of SiOz internal standard. The pseudo-voigt h c t i o n was used to refine the Bragg's peaks. The background was adjusted trough a degree 5 polynomial function. It was used the DBWS-9807a program which is an upgrade version of DBWS-9411, described by Young, Sakthivel, Moss and Paiva-Santos". Specific surface areas for the zirconia was followed powders were measured by the BET multipoint method (ASAP 2000, Micromeritics), using N2 as the adsorptioddesorption gas. RESULTS AND DISCUSSION Fig.1 presents the X-ray diffraction patterns of powders calcined in the range of 300 to 600°C. The X-ray diffraction (XRD) results showed that zirconia
76
Ceramic Nanomaterials and Nanotechnology
is formed after nucleation from the amorphous precursor with no intermediate phases, at very low temperatures. When the dried powder is calcined at 3OO0C, occurs the crystallization of ZrO2 in the tetragonalhbic phases. Crystallization to metastable tetragonakubic zirconia occurs between 300"-500"C and the transformation to stable monoclinic phase starts at about 600°C and seems to be related with the crystallite growth with temperature increasing. 250,
I
(5.888
l
2-thrta rc.1
l
1
1
1
75.w
Figure 1: X-ray diffraction of zirconia powders calcined in the range of 300 to 600°C for 1 h. The analyzed powders follow two distinct stages: in a first step the powders are formed of amorphous zirconia and organic materials at phase of decomposition, mainly carbon, in a next step OCCUTS the crystallization of zirconia and total elimination of organic. These crystals have a high surface area, with extremely reactivity, which makes it suitable for catalysis. The particles were in the range of 8 to 1 2 m as measured by X-ray line broadening analysis, between 300" to 600°C. According to Garvie et a16 it occurs an arrangement of stabilized zirconium at tetragonal structure when the crystallites was lower than few microns. The increasing in temperature and consequently increasing the crystallite sizes, promotes a destabilization of zirconium what changes from tetragonal to a monoclinic phase in agreement with was shown in Fig. 1, above 600°C. As it was expected the increasing in the temperature of calcination decreases the surface area. However it could be observed fkom Table I that in the temperature range of interest, up to 6OO0C, it is possible to obtain a high reactive zirconia powder stabilized in the tetragonallcubic phase. The hysteresis curves, shown in Figure 2, for the adsorptiorddesorption isotherms, obtained from powders calcined at 250°C showed a type 111 shape", which can be attributed to
Ceramic Nanomaterials and Nanotechnology
77
the high organic matter content. In this type of behavior the adsorbed gas molecules have higher affinity among them than with the solid surface, hindering the surface area analysis in this samples. Surface Area (rn'ig) 21.93 127.32 108.96 107.22 89.19 11.21
Temperature ("C)
250 360 430 520 600 1020 260 240
-
220
n ' m
200
g
180
'0
160
v
a,
8
2
140 120
100
a,
80
>o
40
5
60 20 0
0,O
0,2
0,4
0.6
0.8
1.0
Relative Pressure (P/P,)
Figure 2: Nitrogen adsorptioddesorption of calcined zirconia powders in the range of 250 and 1020°C for 1 h. The powders calcined at 6OO0C,were analysed by Rietveld Method. It was verified the presence of nanocrystallites of cubic zirconia (a = OSlOO(1) nm) with spatial group Fm3rn. Fig. 3 presents the Rietveld plot where it is shown the X-ray patterns of observed, calculated and the difference between the 21-02 and Si02 internal standard, besides the Bragg's peak for each phase.
78
Ceramic Nanomaterials and Nanotechnology
.
x)o-
-0bs
-Calc
-0obs-calc I Bragg Pos
B
g5 5
T-B
2500-
f
.. I
0-
I 1
.I
I
I Ill I
I I I I I I I 1 I I 1 1 I l l I I I I I I I I 11111 I l l 1
20
40
80
80
.
1
loo
I
Zro,
IIIII Qwr!~ '
120
Figure 3: Rietveld plot for Zr02 powder with Si02.
Figure 4: FWHM of Zr02 and %OZ.
On Fig. 4, its is shown the FWHM of Zr02 and standard quartz. It could be observed that the zirconia pattern presents extremely large peaks, indicating very small crystallites. The maximum width for the standard quartz was 0.233' at 28 = 120" and for zirconia it was 3.37" at 118" (28).
CONCLUSION The chemical route using the complexation of zirconium metal from zirconium hydroxide through a solution of 8-hydroxiquinoline was suitable in obtaining pure stabilized zirconia. This method was efficient to produce pure and stabilized (tetragonalkubic) zirconia powders with a very low degree of agglomeration. Through this synthesis method, aAer calcination of powders at very low temperatures, it was possible to obtain stabilized zirconia with a high specific surface area (500°C > 100m2/g) and a small crystallite size ( 8 - 12 nm) as measured by X-ray line broadening analysis. ACKNOWLEDGEMENTS The authors are grateful to CAPES, CNPq and FAPESP for financial support of this work. REFERENCES 'Heuer, A.H. and Hobbs, L.W. in Advances in Ceramics, 3, Science and Technology of Zirconia, American Ceramic Society, 198 1, Columbus, OH. 2Smith, D.K. and Newkirk, H.W., The Crystal Structure of Baddeyite (Monoclinic Zr02) and Its Relation to the Polymorphism of ZrO2. Acta Cytallogr., 1965,18,983-991.
Ceramic Nanomaterials and Nanotechnology
79
3Li, P., Chen, I-W. and Penner-Hahn, J.E., X-ray Absorption Studies of Zirconia Polymorphs. I1 Effect of Y2O3 Dopant on on Zirconia Structure. Phys. Rev. B., 1993,48(14), 10074-10081. 4Foschini, C.R., Souza, D.P.F., Paulin, P.I., Varela, J.A., AC impedance study of Ni, Fe, Cu, Mn doped ceria stabilized zirconia ceramics. J. Eur. Ceram. Soc., 2001,29(9) 1143-1150. 5 Subbarao, E.C., Maiti, H.S. and Srivastava, K.K., Martensic Transformation in Zirconia. Phys. Status Solid A, 1974,21,9-29. 6 Garvie, R.C., Stabilization of the Tetragonal Structure in Zirconia Microcrystaliites. J. Phys. Chem., 1978,82,218-224. 7 Howard, C.J., Hill, R.J. and Reinhert, B.E., Structures of the ZrO2 Polyrnolphs at room Temperature by High-Resolution Neutron Powder Diflraction. Acta Crystallogr., 1988,844, 1 16-120. 'Garvie, R.C., Hannink, R.H.J. and Pascoe, R.T., Ceramic Steel. Nature (London), 1975,258,703-705. 'Claussen, N., Riihle, M. and Heuer, A.H., Science and Tecnhology of Zirconia 11. Advances in Ceramics. Vol. 12, The Am. Ceram. Soc., 1984, Westerville, OH. l0S6miya, S., Yamarnoto, N. and Yanagina, H., Science and Technology of Zirconia 111, Advances in Ceramics. Vols 24A and 24B, The Am. Ceram. Soc., 1988, Westerville, OH. "Badwal, S.P.S., Bannister, M.J. and Hannink, R.H.J., Science and Technology of Zirconia V. Austalia Ceramic Society, 1992, Melbourne. '2Kosmac, T., Drofenik, M., Malic, B., Besenicar, S. and Kosec, M., The Influence of Dispersed Zr02 Particles on the Properties of some Electronic Ceramic. Advanced Ceramics IT. Ed. S.Somiya, Elsevier Applied Science, 1988, 29-44. 13Hardy,A B., Gowda, G., Mcmahon, T.J., Riman, R.E., Rhine, W.E. and Bowen, H.K., Preparation of Oxide Powders. Ultrastructure Processing of Advance Ceramics, Ed. John D.Mackenzie and Donald R. Ulrich, John Wiley & Sons, Inc, 1988,407-428. 14Alvarez,M.R., Landa, A.R., Otero-Diaz, L.C. and Torralvo, M.J., Structural and textural study on Zr02-Y203 powders. J. Eur. Ceram. Soc., 1998,18(9) 12011210. "Chaim, R., Basat, G. and Kats-Demyanets, A., Effect of oxide additives on grain growth during sintering of nanocrystalline zirconia alloys. Mat. Lett., 1998, 35 (3-4), 245-250. '6Harris, M.T., Sisson, W.G., Scott, T.C., Basaran, 0 A, Byers, C.H., Ren, W. and Meek, T.T., Multiphase Electrodispersion Precipitation of Zirconia Powder. Better Ceramics through Chemistry VI. Ed. Anthony K. Cheetham, C. Jefiey
80
Ceramic Nanomaterials and Nanotechnology
Brinker, Martha L. Mecartney and ClkmeatlS. Sanchez. Materials Research Sym osium Proceeding Vol. 346, Materials Research Society, 1994, 171- 176. '%BWS-9807 RELEASE 03Feb99: AN UPGRADE OF "DBWS-9411, "An upgrade of the DBWS programs for Rietveld' Refinement with PC and mainframe corn uters", J. Appl. Crust., 1995, 28, 366. "Brunauer, S., Emmett, P.H. and Teller, E., J.Am. Chem. Soc., 1938,60,309.
Ceramic Nanomaterials and Nanotechnology
81
This page intentionally left blank
METAL OXIDE PARTICLE SYNTHESIS BY ELECTRIC-FIELD INDUCED WATER-IN-OIL EMULSIONS Treniece L, Terry, Robert T. Collins and Michael T. Harris School of Chemical Engineering Purdue University West Lafayette, IN 47907-1283
ABSTRACT Multiphase electrodispersion is being used to synthesize ultrafine metal oxide powders for the development of advanced ceramic materials. Electric fields are used to atomize aqueous salt solutions emanating from a nozzle into an organic continuous phase. The condensatiodprecipitation reaction takes place in the dispersed droplets, which serve as microreactors. Particle diameters of 0.1- to 10-pm are formed by electrodispersion precipitation when 2-ethyl- 1-hexan01 is used as the continuous phase. Most (90 wt%) of the particles are in the 0.5- to 5micometer size regime. Larger particles (2 to 30 pm)are formed when vegetable oil is used as the continuous phase. Porous shells and microspheres are readily produced by electrodispersion precipitation. Metal oxide systems that have been produced include zirconia-alumina, titania-nickel oxide, zinc oxidefcopper oxide, cerium oxide and hafinium oxide.
INTRODUCTION Studies have shown that electric fields may be employed to enhance transport processes in a variety of multiphase systems. The use of high-intensitypulsed electric fields can control droplet size in a dispersed liquid system. This technique was utilized in the development of the electric dispersion reactor (EDR), a device that employs electric fields to produce aqueous microreactors in an organic liquid (€ianisl, Harris2 and Harris’). The EDR combines the features of emulsion theory and homogeneous precipitation by dispersing a conducting aqueous drop into a nonconducting organic liquid. Each of these droplets become a localized mircroreactor where reactants in the organic phase diffuse into the aqueous droplet causing precipitation and gelation to occur, while water and remaining reaction products diffuse into the organic phase. The initial liquid droplet controls the size of the final particles formed. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
83
Two methods of chemical synthesis have been explored in the EDR: external gelation and internal gelation (Harris4 and Harris3). With the external gelation method, submicrometer to micrometer-sized droplets containing soluble metal salts are dispersed into a continuum that contains a precipitation agent such as ammonia. A schematic hagram of this technique is shown in Fig. 1. In this figure, metal salts in the aqueous droplet react as the ammonia transfers fi-om the organic continuous phase into the droplet. Precipitation and gelation occur forming a porous hydrous metal oxide gel-sphere. Another form of external gelation explored places the metal containing species in the continuum and the precipitation agent in the disperse phase. In internal gelation, all reactants are placed in a chdled aqueous feed solution or broth. Dispensing the droplets from a nozzle into a warmer continuous phase induces precipitation and gelation. Chilled concentrated aqueous solutions of metal salts, hexamethylenetetramine, and urea are common chemicals used in the internal gelation process. The external gelation technique allows the formation of oxide powders over a wider range of conditions than the internal gelation because of the added flexibility of locating reactants in both liquid phases (Harris4). SPECIES TRANSFERRING
ORGANIC CONTINUOUS PHASE
AQUEOUS DROPLET
2-ETHYL- 1-HEXANOL NH3, ETHANOL
Figure 1. External gelation technique for metal salt solutions in microdroplet reactor The current research compares metal oxide particle synthesis in a vegetable oil continuous phase to those previously made in 2-ethyl-1-hexanol. The effects of aqueous flowrate, temperature and reactant composition on particle size and particle morphology in the liquid-liquid system are characterized using scanning electron microscopy.
84
Ceramic Nanomaterials and Nanotechnology
EXPERIMENTAL Apparatus and Method A schematic diagram of the laboratory-scale EDR unit is given in Fig. 2. The aqueous solution is electricallygrounded and is introduced through a 2.97mm o.d./l.67mm i.d. glass nozzle that is concentrically located in a 6.35mm 0.d. glass tube. The organic (insulating) continuous phase in pumped in the annulus region between the two glass tubes. The grounded nozzle is located approximately 1 cm above a concentric ring electrode that is located outside the 6.35mm 0.d. glass tube. This electrode is connected to a high voltage power supply system which includes a Hewlett-Packard 6654A programmable unit, a custom-built and proprietary high voltage-switching unit, and sweep function generator (BK Precision - Model 3030). The output from the high voltage-switching unit provides the input to a 12V ignition coil (Mallory 29625) which supplies the high voltage to the electrode. The high voltage is pulsed at frequencies of approximately 500 - 5000 Hz and with amplitudes of approximately 30-40 kV. Once the aqueous phase issues fiom the nozzle, it is immediately dispersed into a large number of micron-sized droplets and is accelerated downward by the electric field and the flow of the continuous phase. In the illustration of Fig. 3, the organic phase is introduced through an annulus surrounding the grounded nozzle to provide cocunrent flow of the two liquid phases. This ensures quick removal of the aqueous phase fiom fiuther influence of the electric field and thus reduces droplet coalescence. Since the characteristic time for diffusion is a few hundredths of a second, mass transfer and chemical reactions proceed rapidly in the micro-droplet.
Experimental Procedure Hydrous metal oxide particles are synthesized in the EDR by the external gelation method. Standardized solutions of 1.63 M Tic14 in HCl, 1.013 M ZrO(NO3)2 and 1.675 M (NH&Ce(N03)3, 1.01 M Al(NO3)3, and 0.993 M H E 4 in HCl were used in these experiments. Solutions of IM NiC1, 1M CuCl and IM ZnC12 were also used. The stock solutions were mixed to obtain the desired concentration of the metal salts in the aqueous phase. Finally, 1:2:3 Y:Ba:Cu mixed oxide particles were formed from an aqueous salt solution consisting of 0.25M Y(NO3)3, 0.5M Ba(N03)2, and 0.75M CuC12. The aqueous solution is pumped into the EDR as the conductive disperse phase and the organic phase containing the precipitating agent (0.012 to 0.12M N H 3 ) is pumped into the system as the nonconducting continuous phase. The continuous phase flowrate is varied fiom 50 to 100 mUmin. The volumetric ratio of the disperse phase to continuous phase flow rate ranges from 0.004 to 0.04. Higher ratios will result in extensive coalescence in the electrode region. The peak voltage is >30kV and the pulsing fiequency is fiom 500 to 5000 Hz.
Ceramic Nanomaterials and Nanotechnology
85
Figure 2. Laboratory-scaleElectric Dispersion Reactor
Tu SalWfsm€Yva$ $pm
Figure 3. Schematic of electrodispersion precipitation process.
86
Ceramic Nanomaterials and Nanotechnology
To remove the residual organic liquid, the particles are collected and washed with ethanol and subsequently with water and centrifuged at speeds ranging fkom 2000 to 3500rpm. The particles are allowed to air dry and are then imaged using a scanning electron microscope. Nitrogen adsorption is used to obtain the specific surface area of the dried particles. The distribution of the metal oxides in the two component metal oxide particles is determined by elemental dot mapping using an energy dispersive X-ray spectrometer (HNU Systems, Inc., Model 5000) that is attached to a scanning electron microscope (JOEL JSM-5300). RESULTS AND DISCUSSIONS Figure 4 shows scanning electron micrographs of metal oxide particles formed from the external gelation technique where the metal salts were dissolved in the aqueous phase and the ammonia was dissolved in the continuous phase 2ethyl-1-hexanol. The total concentration of metal salts in the aqueous phase is 1 moVL. This figure shows that single-component, two-component and threecomponent metal oxide particles are easily formed by the electrodispersion precipitation process. Furthermore, any n-component system that can be fomed fiom water-soluble metal salts can be produced by electrodispersion precipitation. A dot mapping of Cu and Zn in a ZnO/CuO particle are shown in Figure 5 . The copper is evenly in the region of the particles that is penetrated by the electron beam. The Zn dot mapping is more difficult to see; however, it too shows a uniform distribution of Zn in the probed region of the particle. Figures 6 shows the dot mapping of Y, Ba and Cu in the particles produced from the (1:2:3) Y, Ba and Cu precursor salt solution. The photographs clearly show a uniform distribution of the metals in the probed region of the particle. The uniform distribution of the mixed metal oxides in probed region of the particles suggests that precipitation is fast enough such that diffusional segregation of the metal oxide components is low. A typical differential volume particle size distribution is shown in Figure 7. The particle size distribution is measured by a Coulter LS130 light scattering particle sizer. The particle sizes range fiom 0.2 to 10 pm. The scanning electron micrographs show that the particles produced with 2-ethyl-1-hexan01 as the continuous phase are 10 pm or less, which is in agreement with the light scattering results. Nitrogen adsorption measurements reveal that the particles are also porous with a BET surface area of approxGately 100 m2/g.
Ceramic Nanomaterials and Nanotechnology
87
Figure 4. Hydrous metal oxide particles produced by electrodispersion precipitation/external gelation of metal salt solutions dispersed in 2-ethyl-lhexanol.
88
Ceramic Nanomaterials and Nanotechnology
Figure 5. SEM of and EDX-dot mapping of ZnO/CuO particles.
Figure 6. SEM and EDX-dot mapping of YBa2Cu3OX(OH),particles.
Ceramic Nanomaterials and Nanotechnology
89
Figure 7. Typical particle size distribution obtained from Coulter LS light scattering particle size. (The particle size distribution in this plot is for the hafnia particles.)
Figure 8. Typical porous ceramic shells that are formed at high concentrations of precipitating agent (ammonia) in the continuous phase.
90
Ceramic Nanomaterials and Nanotechnology
Particles with dimples or holes (Figure 8) are attributed to very rapid precipitation-gelationnear the droplet/oil interface where porous ceramic shells are formed. Such conditions occur when the concentration of the precipitation agent (ammonia) is high in the continuous phase. The high ammonia concentration at the interface results in very low concentrations of the soluble metal salts at the interface; thus, the metal ions in the disperse phase diffuse rapidly to the interface and precipitate. The dimples and holes form when shells collapse during the drylng process. Figure 9 shows the zirconia particles that were produced by spraying a 1M-zirconyl chloride aqueous solution in vegetable oil containing 0.026M ammonia. Figure 10 illustrates the particle size distribution for these particles. Metal oxide particles produced by electrodispersion precipitation in vegetable oil were larger (typical particle sizes: 2 to 100 pm) than those produced with 2-ethyl1-hexanol as the continuous phase. A possible explanation for the difference is that larger droplets are formed due to the higher viscosity of the vegetable oil. Thus, larger particles are formed.
Figure 9. Zirconia particle produced by electrodispersion precipitation of a 1M-zirconyl chloride solution in vegetable oil. CONCLUSIONS Electrodispersion precipitation has been used to produce single and two component metal oxide particles in the size range of 0.2 to 30 pm. Smaller particles were produced when the lower viscosity 2-ethyl-1-hexanol was used a continuous phase. Future studies will involve studying the effect of temperature and a more extensive study of the effect of viscosity on the particle size and morphology.
Ceramic Nanomaterials and Nanotechnology
91
0.45 0.4
0.35 >r 0.3 0
sa 0.25 E
0.2
LL
0.15 0.1
0.05 0 0
20
40
60
80
100
120
Particle Diameter (pm)
Figure 10. Particle size distribution for zirconia particle produced by electrodispersioa precipitation of a 1M-zirconyl chloride solution in vegetable oil. ACKNOWLEDGEMENTS This research was supported by the National Science Foundation (DMR9700860) and GEM.
LITERATURE CITED I M. T. Hanis", T. C. Scott, 0. A. Basaran, and C. H. Byers, "Morphology Control in Precursor Cerarnic Powder Production by the Electrical Dispersion Reactor," Mat. Res. Soc. Symp. Proc., 180,853-856 (1990). 2M. T. Hank*, T. C. Scott, 0. A. Basaran,and C. H. Byers, "Formation of YBa-Cu (1-2-3) Hydrous Oxide Precursor Powders in the EIectric Dispersion Reactor," AIChE/Symposium Series, Superconducting Engineering, 88, 44-46 (1992). 'M. T.Harris", W. G. Sisson, and 0. A. Basaran,"Computation, Visualization, and Chemistry of Electric Field-Enhanced Production of Ceramic Precursor Powders," Mat. Res. Soc. Symp. Proc., 271,945-950 (1992). 4 M.T. Harris, "Ultrafine Ceramic Precursor Powders by Homogeneous Precipitation and Electrodispersion", Ph.D. Thesis, University of Tennessee, Knoxville, 1992. 'M. T. Harris, Scott, T.C. and. Byers, C. H, "The Synthesis of Metal Hydrous Oxide Particles by Multiphase Electrodispersion," Mat. Sci. Eng. ,A168, 125-129 (1993).
92
Ceramic Nanomaterials and Nanotechnology
SYNTHESIS AND CHARACTERIZATION OF p-SIC NANOWIFES AND hBN SHEATHED p-Sic NANOCABLES. Karine Saulig-Wenger, David Cornu, Fernand Chassagneux, Gabriel Ferro and Philippe Miele Laboratoire Multimatdriaux et Interfaces, UMR CNRS 56 15 Universitd Claude Bernard - Lyon 1 43 Bd du 11 Novembre 1918 F-69622 Villeurbanne Cedex, France Thierry Epicier GEMPPM UMR CNRS 5510 INSA Lyon 20 Avenue Albert Einstein F-6962 1 Villeurbanne Cedex, France ABSTRACT Under nitrogen, direct thermal treatment at 1200°C of silicon powder placed in a graphite crucible yields cubic silicon carbide @-Sic) nanowires mixed with micrometric silicon particles. When the pyrolysis is conducted under argon and in presence of boron nitride powder, h-BN sheathed p-Sic nanowires are obtained. The structure of both nanomaterials have been investigated by EDX, HRTEM and EELS. INTRODUCTION Though carbon nanotubes have been the most extensively studied nanoobjects,' the interest in preparing nanostructwes such as nanotubes, nanowires or nanocables of various chemical composition (WS:, GaN3, Tic4,...) has been clearly emphasized for applications in nanoelectronics or as reinforcement agents in nanocomposites. Particularly, cubic silicon carbide (pand hexagonal boron nitride (Iz-BN)~occupy a unique place among high technological ceramics due to their special chemical and electronic properties. Several routes have been reported for the preparation of p-Sic nanowires (NWs) or SiO2 sheathed p-Sic NWs. The carbothermal reduction of (a) SiO(g) with
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
93
carbon nanotubes as templates4 or (b) carbon nanoparticles embedded in silica xeroge17 yield S i c NWs or SiO2 sheathed S i c nanocables. Moreover S i c NWs have been prepared using chemical vapor deposition (CVD) techniques (c) directly with metal nanoparticles which act as catalysts for the vapor-liquid-solid (VLS) growth process' or (d) by using hot filament assisted In addition, Tang et al. have reported that the reaction of CC14(g) with SiCb(g) or Si powder in presence of Na yield Sic NWs." The main drawback of the synthetic methods described above is their high cost due either to the experimental technique andor to the starting materials involved. EXPERIMENTAL 1) Sic nanowires In a typical experiment, silicon powder (Aldrich 99.999%, 60 mesh) was placed in an alumina boat and then inserted in a graphite crucible. The thermal treatment was performed in an inductive vertical furnace with a cooled wall silica reactor. The reaction chamber, the graphite crucible and the alumina boat were degassed in vacuo before filling with pure nitrogen or argon up to the atmospheric pressure. The graphite crucible was rapidly heated up to 1200°C and held at this temperature for 1 hour. After cooling, a dark powder was scraped fiom the alumina boat and analyzed by means of XRD, scanning electron rnicroscopy (SEM, Model N"S800, Hitachi), high-resolution transmission electron microscopy (HRTEM, field emission gun microscope JEOL 2010F), energy dispersive X-ray spectroscopy (EDX, Link-Isis OXFORD analyzer) and electron energy-loss spectroscopy (EELS, Digi-PEELS GATAN). 2) BN coated Sic nanowires Turbostratic BN powder was prepared by pyrolysis of BN preceramic polymer under nitrogen at 1000"C.6 A mixture of boron nitride and silicon particles was suspended in dry diethyl ether and sonicated for dispersion. After drying in vacuo, the solid mixture was placed in the alumina boat and treated thermally under argon following the same process as described above. RESULTS AND DISCUSSION 1) Sic nanowires When the pyrolysis is conducted under argon, SEM investigations showed no nanowires. In contrast, SEM images of the crude product obtained under nitrogen revealed the presence of numerous nanowires mixed with micrometric silicon particles (Fig.1). The nanowires were highly curved and had diameters falling in the range of 5-60 nm with length from 10 to 100 p.
94
Ceramic Nanomaterials and Nanotechnology
Figure 1 : SEM image of S i c nanowires. EDX and EELS analyses indicated that the nanowires are composed of silicon and carbon with a chemical composition consistent with Sic. As an example, EELS analysis performed with 0.5 to 2 nm electron probes (Fig. 2c) displayed two main features at -100 eV and -284 eV corresponding to C-K and Si-L edges, respectively. The feature at -532 eV corresponding to 0-K edge was not observed, which clearly points out the absence of oxygen in the observed nanostructure.
Figure 2 : HRTEM of a Sic nanowire (a) with the Selected Area Electron Diffraction (SAED) (b). Typical EELS spectrum of a SIC nanowire (c)
Ceramic Nanomaterials and Nanotechnology
95
Selected area electron diffkaction (SAED) (Fig. 2b) demonstrates that the nanowires are composed of cubic silicon carbide (p-Sic). According to electron diffraction and HRTEM (Fig. 2a), Sic NWs grew preferentially along the [l 111 direction and possess a high density of stacking faults. Without using carbon nanotubes as templates, three main growth mechanism have to be considered, namely a screw dislocation, a VLS or a Vapor-Solid (VS) growth process.l 1 The main feature of a screw dislocation growth nucleation is the formation of spiral nanowires. The latter were not observed in our samples thus the formation of NWs should proceed via either a VLS or a VS growth mechanism. The tips of the NWs were examined by HRTEM showing they mostly exhibit no globule at their ends whereas few have a nanosized silicon particle. This result strongly suggests that the N W s grew via a VS nucleation mechanism. In addition, this implies the presence of carbon in the vapor phase and therefore a transport of carbon fkom the solid source (crucible). Moreover the N W s did not grow under argon. Based on these results, we assume that nitrogen plays a role in the carbon transport mechanism consistently with previously reported works.l 2 Further investigations are in progress in order to obtain experimental evidence of the transport mechanism of carbon and of the growth mechanism of the NWs. 2) BN coated Sic nanowires When the thermal treatment of silicon powder is performed under argon and in presence of boron nitride, SEM analyses show the presence of curved nanowires (Fig. 3).
Figure 3 : SEM image of h-BN coated Sic nanowires
96
Ceramic Nanomaterials and Nanotechnology
HRTEM images (Fig. 4b) show that N W s are similar to those depicted above but also indicate that they are coated with graphitic materials as evidenced by FFT of the coating region which exhibits the (0002) reflections typical of layered compounds (Fig. 4a). Noteworthy is also the observation in HRTEM images (Fig. 4b) of lattice fringes resolved at -0.33 nm on (0002) planes and -0.21 nn-i on planes normal to (0002) planes. EELS analyses performed with 0.5 to 2 nm electron probes indicate that the core of the N W s is composed of Sic whereas the sheaths contains boron and nitrogen. As an example, the spectrum depicted in figure 4c displays two main features at -188 eV and -400 eV corresponding to those reported for B-K and N-K edges in hexagonal boron nitride, re~pective1y.l~ Two minor features at -284 eV and -532 eV gave evidence of small carbon and oxygen impurities (C-K and 0-K edges, respectively). Since a faint shoulder-like feature is observed at -100 eV (Si-L edge), the carbon should come fiom the Sic NW, owing to the electron beam spreading within the sample. The presence of oxygen may be due to the superficial oxidation of the NW. These results clearly indicate that the thermal treatment of a mixture of silicon powder and boron nitride powder yields nanocables composed of h-BN sheathed p-Sic nanowires.
Figure 4 : FFT of the cercled region (a) and HRTEM image of a p -Sic nanowire with a BN coating (b). Typical EELS spectrum of the BN outer layer (c) Coatings of different thickness have been observed by HRTEM (Fig. 5). Mainly the outer layers were of 2-4 nm thickness. They are composed of h-BN (0002) planes parallel to the growth axis of the NWs. When the thickness of the coating increased, its structure was clearly modified and resembles the bamboolike structure reported for carbon nanotubes. As illustrated by Fig. 5, thick layers were composed of two thin layers of h-BN separated by an amorphous region, the two thin layers being periodically joined by crystalline h-BN bridges. Line scan
Ceramic Nanomaterials and Nanotechnology
97
EELS are in progress in order to determine the chemical composition of the amorphous external region.
Figure 5 : HRTEM images of different BN coatings. Except for a few silicon nanoparticles, no globule was detected at the tips of the nanocables and thus we assume a VS growth mechanism for these nanocables similar to that described above for Sic NWs synthesis. It implies also that the carbon is transported in the vapor phase by the nitrogen of BN. Based on these assumptions, we can propose two different routes for the formation of the BN coatings (a) the Sic NWs are sufficiently boron-doped to permit the formation of a BN coating in a similar way as the mechanism described by M. D. Sacks for the synthesis of BN coated Sic fibers14 (b) the BN films grew directly fiom a boron and nitrogen rich vapor phase but the Sic N W s grew earlier due to faster kinetics. Further experimental work is in progress in order to obtain experimental evidence of the growth nucleation process of these nanocables. ACKNOWLEDGEMENT We gratefilly thank the CLYME (Consortium Lyonnais de Microscopie Electronique) for the access to the TEM-FEG microscope. REFERENCES
98
Ceramic Nanomaterials and Nanotechnology
'E.T. Thostenson, Z. Ren, T.-W. Chou, "Advances in the science and technology of carbon nanotubes and their composites : a review", Composites Science and Technology, 61 1899-1912 (2001). *R. Tenne, L. Margulis, M. Genut, G. Hodes, "Polyhedral and Cylindral structures of tungsten disulphide", Nature, 360 444-446 (1992). 3W. Han, S. Fan, Q. Li, Y. Hu, "Synthesis of Gallium Nitride Nanorods Through a Carbon Nanotube-Confined Reaction", Science, 277 1287-1289 (1997). 4 H. Dai, E.-W. Wong, Y.Z. Lu, S. Fan, C.M. Lieber, "Synthesis and characterizationof carbide nanorods", Nature, 375 769-772 (1995). 5P.G.Neudeck, "Progress towards high temperature, high power Sic devices", Institute of Physics Conferences Series I41 : Compound Semiconductors,1-6 (1994). 6R.T. Paine, C.K. Narula, "Synthetic Routes to Boron Nitride", Chemical Reviews, 90 [11 73-91 (1990). 7G.W. Meng, Z. Cui, L.D. Zhang, F. Phillipp, "Growth and characterization of nanostructured QSiC via carbothermal reduction of Si02 xerogels containing carbon nanoparticles", Journal of Crystal Growth, 209 801-806 (2000). *Y. Zhang, N. Wang, R. He, X. Chen, J. Zhu, "Synthesis of Sic nanorods using floating catalyst", Solid State Communications, 118 595-598 (2001). 'K.W. Wong, X.T. Zhou, F.C.K. Au, H.L. Lai, C.S. Lee, S.T. Lee, "Fieldemission characteristics of Sic nanowires prepared by chemical-vapor deposition", Applied Physics Letters, 75 [ 191 29 18-2920 (1999). 10 Q. Lu, J. Hu, K. Tang, Y. Qian, G. Zhou, X. Liu, J. Zhu, "Growth of SIC nanorods at low temperature", Applied Physics Letters, 75 [4] 507-509 (1999). 11 H.Y. Peng, X.T. Zhou, H.L. Lai, N. Wang, S.T. Lee, "Microstructure observations of silicon carbide nanorods", Journal of Material Research, 15 [9] 2020-2026 (2000). 12 C. Popov, M.F. Plass, R. Kassing, W. Kulisch, "Plasma chemical vapor deposition of thin carbon nitride films utilizing transport reactions", Thin Solid Films, 355-356 406-411 (1999). 13D.Goldberg, Y. Bando, M. Eremets, K. Takemura, K. Kurashima, H. Yusa, "Nanotubes in boron nitride laser heated at high pressure", Applied Physics Letters, 69 [ 141 2045-2047 (1996). I4M.D. Sacks, "Silicon Carbide fibers with boron nitride coatings", U.S. Pat. No. 6.040.008, Mar. 21,2000.
Ceramic Nanomaterials and Nanotechnology
99
This page intentionally left blank
VIRTU& PROCESSING OF ADVANCED NANOMATERIALS Liudmila A. Pozhar Department of Chemical Engineering University of Tennessee 4 19 Dougherty Hall Knoxville, TN 37996-2200
V a h r F. de Almeida Oak Ridge National Laboratory P. 0. Box 2008 Oak Ridge, TN 37931-6181
Michael Z. Hu Oak Ridge National Laboratory P. 0. Box 2008 OakRidge, TN 37931-6181
ABSTRACT A findamental, non-equilibrium statistical mechanical approach, and equilibrium and non-equilibrium molecular dynamic (EMD and NEMD) simulations have been used to elucidate relations between nanosystem structure, composition and topology, and the transport coefficients. These relations are further utilized to realize a self-consistent, virtual (i.e., theory-based, computational)processing of nanomaterials of sub- 10 nm linear dimensions with designed properties. The developed virtual processing approach is discussed in conjunction with. experimental synthesis of Ilanornaterials, such as nanoporous zeolites. INTRODUCTION Assembling of molecules, monomers, and polymers into clusters a d o r ordered structures, and cluster aggregation are phenomena intrinsic to any natural or industrial materials synthesis. Quantitative description of these phenomena offers immense opportunities to fabricate new materials, in particular nanomaterials, with designed properties. Numerous theoretical and experimental studies performed for over 70 years have led to several approaches that constitute the foundation of the modern technologies of materials synthesis. Various extensions and generalizations of the classical LXshitz-Slezov' approach and other theoretical methods2supply evolution equations for cluster size distributions (CSDs) using thermodynamic considerations and global conservation laws. This is appended by a number of heuristic considerations for nucleation and growth
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication,reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
101
rates, and other parameters in the evolution equations for the CSDs. Among such heuristic considerations is an assumption that homogeneoudheterogeneous nucleation happens at equilibrium, and fiuther growth of the nuclei can be considered as a quasi-equilibrium process that takes place in a weakly homogeneous system. This consideration is used to justify fiuther representation of the appropriate nucleation process rates in terms of the transport coefficients (such as the diffusion coefficient) specific to such weakly homogeneous systems near equilibrium,This “macroscopic”point of view may be reasonable in the case of the later stages of the growth when the nuclei include a great number of atodmolecules (monomers) and are large enough to undergo Ostwald ripening, but it fails entirely at the initial stage of nucleation, when the nuclei are composed of a few atodmolecules, and as they grow while still remaining of the order of several mnometers in linear dimensions. A recent distributionkinetics approach to cluster kinetics and dynamics due to McCoy and his colleagues3 (which, in fact, is a phenomenobgical theory of the CSD time evolution) is capable of reasonable predictions of CSDs, provided the nucleation, growth, dissolution, aggregation and other process rates are known. Tbis approach was successllly applied to self-assembly, dendrimer growth, nucleation, polymerization, depolymerhtion, catalyzed polymer degradation, thermo-oxidativepolymer reactions, pyrolysis, and thermogravimetric processes. With realistic initial conditions, operating parameters, and rate coefficients, McCoy’s approach describes typical crystal growth and re-crystallization processes, including Ostwald and random ripening. Further development of theoretical foundations of the nucleation and growth processes, including materials synthesis, relies on solution of the process rate problem. An important step forward to resolution of this problem can be taken upon recognition that emergence of a nanocluster composed of a few atoms, molecules, or monomers (or just an addition of a few atoms, molecules, or monomers to an already existing nanonucleus, in the case of heterogeneous nucleation) creates strong locd inhomogeneity in a nanoscale dimension region around the nanocluster. While the growth process may still proceed at the local quasiequilibrium conditions, these conditions now are very different fiom those specific to a homogeneous fluid at global equilibrium, as the region surrounding the nanonucleus is occupied by a strongly inhomogeneous fluid. In particular, the local values of the transport coefficients may differ by up to several orders of magnitude fiom those typical h r weakly inhomogeneous systems. This has been long recognized in statistical mechanics of inhomogeneous systems and confirmed by numerous experimental and simulations data (see Ref. [4] and references therein). Theoretical predictions for the tramport Coefficients of inhomogeneous fluids are briefly discussed in the following section of this paper. This discussion is followed by a brief review of nanoscale mechanisms of
102
Ceramic Nanomaterials and Nanotechnology
nucleation and growth processes in electrolyte solutions used in synthesis of zeolite A materials. The theoretical transport coefficients are &her used to analyze transport phenomena taking place during such a synthesis, and possibilities of virtual (i.e., theory-based, computation- and simulation-added) synthesis of advanced materials with designed properties. TRANSPORT PROPERTIES OF NANOSYSTEMS Fluids at interfaces and in nanoscale confinements are called inhomogeneous, because their thermodpamic properties depend upon a position in the fluid. For example, density of such a fluid is not constant, and its transport coefficients are non-local and dependent of a position within the fluid. Another class of inhomogeneous systems is exemplified by nanoscale solids (such as small quantum dots) whose electron and excitation sub-systems can also be viewed and described as highly inhomogeneous fluids. During the recent decade, statistical mechanical approaches to transport processes in nanosystems experienced significant progress fbeled by a technological demand for novel materials of advanced properties (such as nanomaterials for electronics, nanocatalysts, nanoadsorbents, etc.). In addition to predictions of the "mechanical" transport coefficients, such as the viscosity, diffusion and thermal conductivity, that are important for new fabrication process design, novel statistical mechanical approaches suggest means to define charge (electron, hole, phonon, etc.) transport properties and to describe various relaxation phenomena in nanosolids relevant to electronic properties of such nanomaterials. During the 90s PO* and G ~ b b i n s(PG) ~ * ~ developed a hdamental, selfconsistent and tractable approach to transport processes in inhomogeneous systems (that range fiom those composed of several particles to bulk ones) at any physically reasonable density and its gradients. This very general, and at the same time practical, non-equilibrium statistical mechanical technique supplieslexplicit formulae to calculate the transport coefficientsof an inhomogeneous fluid mixture (regardless of the nature of the fluid) in terms of equilibrium or any steady state structure factors. These structure factors are particle-particle correlation functions and densities, such as fluid-fluid and fluid-wall atodmolecde pair correlation hctions and the corresponding densities in the case of classical fluids. Such structure hctors can be obtained solving integral equations of equilibrium statistical mechanics, fiom molecular simulations, and/or fiom experimental data. PG-Thmretical Transport Properties: Viscosity. During recent years, the PG-approach has been successfufly applied to a number of model nanosystems6and supplied reliable predictions of their transport coefficients. In these studies the structure factors (i.e., the density and correlation functions) of the fluids and interfacial solids were simulated using EMD
Ceramic Nanomaterials and Nanotechnology
103
simulation techniques. The data so obtained were further used as input data for simplified PG-theoretical formulae to calcdate the PG-theoretical values of the localized transport coefficients. Note, that “turbulent” contriiutions to the transport coefficients in the case of nanofluids can be neglected, because the atomic dimensions of the nanosystems do not permit development of turbulent eddies. Similar to their general counterparts, the simplified formulae are valid for any pairwise additive, central intermolecular interaction potentials that can be decomposed into the sum
of a short range repulsive (hard-core) contrr’bution,
[with G denoting the effective diameter of the intermolecular interaction, or the hard-core diameter of an atodmolecule], and an attractive, continuous long-range (soft) interaction, & (r, ) , where rij = ri - rj, and rh rj denote position vectors assigned to the centers of mass of interacting molecules i andj, respectively. The only restrictive requirement to the soft intermolecular interaction potential is that it should converge to zero fister than 112 at large separations r between the interacting molecules, r+oo [r is the absolute value of the vector rij]. The decomposition (1) of the actual intermolecular interaction can always be realized by means of the Barker-Henderson (BH) or Weeks-Chandler-Andersen (WCA) methods, that also supply the corresponding effective diameters of the hard-core intermolecular interactions, GBH and OWCA respectively (see the corresponding discussions and references in Refs. 141 and [S]). In the simplest particular case of a nanofluid confined in a narrow slit pore with a strong inhomogeneityonly in one direction (the z-direction across the pore, that is orthogonal to the wall planes drawn through the centers of mass of wall surface atodmolecules facing the pore void) a general expression (3.34) in Ref. [7] €or the PG-theoretical viscosity of strongly inhomogeneous fluids reduces to a very simple formula,
104
Ceramic Nanomaterials and Nanotechnology
where qb =(5/16cr2)(m/zj?B)1'2, P, = l / ( k , T ) , kB is the Boltzmann constant, T denotes temperature, m is the mass of a fluid molecule, G denotes the hard-core diameter of the fluid molecules specific to the hard-core fluid-fluid intermolecular interactions, n*(z) = n(z)c? is the reduced equilibrium number density of the nanofluid, and other quantities are as follows. [Details of the derivation of Eq. (3) in its dimensional form and discussions of other assumptions incorporated into this derivation, in addition to the neglect of inhomogeneity in the x and y directions along the pore, can be found in Ref. [7]]. The dimensionless quantity
z,*(2)= 2 R [ V * (2)3- (1 /3)v,* (2)+ &,*
(43-'
(4)
is proportional to the viscous relaxation time and incorporates two essentially 'fluid' contributions, v"(z) and a-q *(z), R
v * ( z ) = i d 8 sin 9 n * ( z - o c o s 9 ) g ~ ( z , z - ~ c o s 9 ) , 0 A
v * (z)= ~ ~d9sin9[n*(z-acos9)-n(z)~~~(~,z-acos9), 0
and the contribution e*(z) due to fluid-wall intermolecular interactions,
where g, (z,z- ocos 9) is the contact value of the equilibrium, fluid-fluid pair correlation function; 8 is the angle between the vector connecting the centers of mass of the interacting molecules, rg, and the positive z-direction; nw*(z)=nw(z)c3 is the reduced equilibrium number density of the wall molecules; g f i ( z , z - o f icos@ is the contact value of the equilibrium, fluid-wall pair is the hard-core diameter specific to the hard-core correlation function; and ofi~ part of the fluid-wall intermolecular interactions. The quantity p**(z) is defined in terms of g, (2,z- cr cos 9)and the reduced number density,
Ceramic Nanomaterials and Nanotechnology
105
Below we omit the star "*", assuming that all quantities are reduced as described in Table I. As one can see fiom general expressions of Refs. [5] and [7], and fiom the above equations (3) - (8), the local values of the PG-theoretical Viscosity at any position w i t h the nanofluid are defmed by the structure and composition of the surroundingfluid and its confinement [reflected in equations (3) - (8) by the integration over the 8 -angle of the structure factors that depend not only on the position zwhere the viscosity is calculated, but also on the neighboring
Number density, n Velocity, U Viscosity, Temperature, ~ B T The, t
6 d (d&)lD
Figure 1. A model Lennard-Jones nanofluid in a slit nanopore of 3.20 in width composed of immobile wall atoms: the 3D image of the nanofluid number density, nd,where o is the effective diameter of nanofluid atoms. The average nanofluid density .c n$> = 0.603, temperature kBT/&= 0.958; E is the energy parameter o f the Lennard-Jones potential. The origin of the coordinate is in the geometric center of the pore with z-axis orthogonal to the wall planes and horizontal x-axis. The walls are of fcc lattice (3-dimensional) with the [loo] planes fkcing the pore gap.
106
Ceramic Nanomaterials and Nanotechnology
The structure factors in equations (3) - (8) (i.e., the local density n(z), and the pair correlation functions and &) can be calculated using EMD simulation techniques. Such calculations have been performed6 for twenty slit pore nanosystems exemplified by the system depicted in Figure 1. In these calculations the hard-core diameters and potential well depths corresponding to the fluid-fluid and fluid-wall interatomic interactions were the same, 0 - 4 f i y and E=+, and the interatomic interaction potentials were the Lennard-Jones (LJ)and WCA ones defied by expressions
respectively. The width of the slits ran fiom 3.20 to 70,and the pore length was fiom about 110 to 130. As can be seen in Figure 1, there are voids in the slit occupied by the fluid atoms. This means that the probability to find a fluid atom in such positions is extremely low due to exchded volume effects (these regions border with the wall atoms sticking out into the pore space) that are the major cause of inhomogeneity of the above systems. The theoretical localized viscosity and density profiles across the pores reflect these strong inhomogeneity effects and lead to the across-the-pore average theoretical viscosity values that are about an order of magnitude larger than the viscosity of the corresponding “bulk” fluids composed of the same atoms (this value is 0.35 in the units of Table 1). Typical profiles of these quantities for a LJ system with the pore width of 4.10 are pictured in Figure 2. NEMD simulations of the Poiseuille flow in the same slit pores were performed to obtain the corresponding velocity profiles. While the NEMD method on its own can not provide transport coeflicient data, the NEMD velocity profiles can be used to evaluate the poreaverage values of the coefficients. Because such evaluations involve assumptions on the explicit correlation between the stress and Strain rate tensors [the very correlation that has to be derived (and was derived) in the fiamework of a statistical mechanical theory, such as the PGone], their accuracy is better for low density and moderately dense fluids where the NEMD fluid velocity profile is closer to the classical parabolic one. In this case the stress-strain rate correlation is roughly close to that specific to the
Ceramic Nanomaterials and Nanotechnology
107
Newtonian fluid in a gap, fbr which the viscosity and the average velocity are related via an analytical expression (see the detailed discussion in Ref. [6a]). Therefore, this latter expression can be used to evaluate the NEMD average pore fluid viscosity.
Top: the PG-theoretical viscosity for the WCA fluid confined in Figure 2. the pore of H 4 . 1 0 in width at the average density
= 0.603 and temperature k~TkO.958.Curve: equation (3) with the EMD-simulated PCFs; straight line: the corresponding pore-average PG-theoretical viscosity. Bottom: the corresponding fluid number density (curve), average number density (lower line) and temperature (upper line). The wall atom layer separation b k 0 . 6 4 7 and the distance between wall atoms in the directionsalong the pore a l ~ 2 . 3 51.
108
Ceramic Nanomaterials and Nanotechnology
Figure 3. Top: the NEMD velocity profile specific to the WCA fluid at the average density =0.442 and bT/c=0.729 confined in the pore of H=4.10 in width, and its “bulk” parabolic counterpart (parabola). The flow is caused by a small “gravity” force acting on the centers of mass of each of the atom in the ydirection along the pore. Curve: the “smoothed” velocity, vy, in the direction of the flow (y-direction), its fluctuations are also shown; fluctuations along z-axis of the graph: the velocity components in the x-direction, v, (larger fluctuations about the zero-value straight line), and the z-direction, v, (small fluctuations near that line), orthogonal to the flow direction. Bottom: the NEMD density (lower curve) and temperature (upper curve). The wall atom layer separation b/0=0.647and the distance between wail atoms in the directions along the pore dk2.351.
Ceramic Nanomaterials and Nanotechnology
109
For the studied pore fluids the relative error of such an evaluation is fiom 20% (for the widest pore gap of 7.00) to over 50% (for the narrowest pore gap of 3.20). Having this in mind, the PG-theoretical pore-average viscosity and the NEMD one are in excellent agreement for all of the studied systems ranging fiorn 2% (for the WCA system featuring the widest slit) to 51% (for the LJ system featuring the narrowest slit). These data also correlate extremely well with known experimental findings (see Refs. [4b] and 151 for fkther references to experimental publications) that confitn that the viscosity of confined fluids can exceed that of the corresponding “bulk” ones by an order of magnitude. Unfortunately, at present detailed comparison with experimental data is not possible, as experimentally studied systems are very different fiorn those studied by theoretical and simulation means. Typical NEMD velocity and density profiles are shown in Figure 3. As mentioned above, the expression (3) neglects effects caused by inhomogeneity of the structure fixtors in the directions along the pore (in this expression the pair correlation fimction contact values and the density depend only on the coordinate z running across the pore). In practice, this can be an oversimplification. If such is the case, one has to use general PG-theoretical expressions from Refs. IS] and [7]. Preliminary calculations’ of the PG-viscosity for the LJ nanosystem in the slit pore of 3.20 in width using less restrictive simplifications of general PG-formulae that take into account inhomogeneity in all spatial directions show that the theoretical transport coefficient values m y change significantly with regard to those calculated using the above simplified expressions that neglect inhomogeneity effects in the directions along the pore. However, such “inhomogeneity-tuned” formulae are more complicated than the one cited above and therefore, may be less convenient in engineering applications. In particular, the pair correlation function contact values assume their complete functional dependence upon positions of the particles (with the condition that the distance between the centers of the particle is equal to their diameter at contact; there are five of such independent variables). Projections of the EMD pair conrelation fimction contact values for this case are shown in Figures 4 and 5. Preliminary results for the “bulk” viscosity of this system are shown in Figure 6. PG-Theoretical Transport Properties: Diffusion. Contemplating the importance of accounting for all inhomogeneity effects, a simplified expression for the PG-theoretical self-diffusion coefficient was obtained in the zero-order approximation (with regard to pressure contributions) in terms of the complete structure factors depending on the entire position vector q within the pore,
110
Ceramic Nanomaterials and Nanotechnology
where
The theoretical d i h i o n coefficient (measured in the physical units, such as m5/sec) of the form of Eq. (11) occurs naturally (both in the PG-theory, and in classical approaches, which are particular cases of the PG-theory) in the constitutive equations in the process of their derivation fkom the kinetic ones. The above zero-order theoreticaI self-diffusion coefficient Ddq) is related to that measured experimentally, Dp(q), as follows,
Introducing the notations: ? I ,
1
and
one can rewrite the equation (13) in the familiar form
that resembles the heuristic expression for the “dusty gas” self-diffusion coefficient. The advantage of the theoretical expression (17) with regard to the corresponding heuristic one is that the nature of the each contribution in the right hand side is elucidated. In the above expressions cr (and its vector), d q ) , d q , q vo), and
Ceramic Nanomaterials and Nanotechnology
111
The EMD fluid-fluid pair correlation hnction contact values Figure4. (FFPCFCVs), g&,O,z,O,n), in the ( X J ) plane for the slit pore of width H=3.20 and length Lz=11.15240,where o is the hard-core diameter of the fluid atoms. The axes x and z are directed across and along the pore, respectively. The values of the other three coordinates specific to this projection are: y”0, 0 4 , and
112
Ceramic Nanomaterials and Nanotechnology
Figure 5. The E m , fluid-wall pair correlation h c t i o n contact values (FWPCFCVs), gfw(x,1.4s,z,d4,d2), in the ( X J ) plane for the slit pore of width H=3.20 and length Lz=11.15240,where o is the hard-core diameter of the wall and the fluid atoms. The axes x and z are directed across and along the pore, respectively. The values of the other three coordinates specific to this projection are: y=1.40,93/4,and
Ceramic Nanomaterials and Nanotechnology
113
Figure 6. Projection of the localized, PG-theoretical “bulk” viscosity coefficient in the (XJ) plane for the slit pore of width H=3.2cr and length L,=l 1.1524q where CT is the hard-core diameter of the wall and the fluid atom. The axes x and z are directed across and along the pore, respectively (see comments on the coordinate system in the caption of Figure 4). The y-axis (not shown) is oriented along the pore and is orthogonal to the other two axes of the right-hand system of coordinates. The calculations used the EMD air correlation function contact values as input into the PG-theoretical formulae . The data are not smoothed.
r
114
Ceramic Nanomaterials and Nanotechnology
Figure 7. Model system studied in this work by the PG-theoretical and MD simulations means: a side view. The axial half length, L, of the composite pore was Ls+2.5d2, where Lr3.542 is the length of the square cross-section channels of the width d, = 342. The maximum dimension of the central unit (“octahedron”) is = 842. The pore system walls are composed of two dense fcc layers of immobile atoms (ail dimensions are in the units of the wall atoms’ hard-core diameter 0,; the figure is drawn to scale). The black and white colors of the fluid molecules are introduced for convenience. geometry and dimemjons). The LJ interatomic interaction potential parameters for the wall atoms were those specific to oxygen, for fluid atoms to methane, and hid-wall interactions were described by the standard arithmetic average ( 0 ) and geometric average (E) rules. The ratio owp was 0.783. Figure 8 shows contributions of each of the terms in the right hand side of equation (17) to the diffusion coefficient of the nanofluid in a square channel of the pore system. Recently, the authors of this paper studied self-diffusion of the LJ fluid in one of the slit nanopores, H=3.2o, discussed above in conjunction with the viscosity case (see Figure 1). The PG-theoretical self-diffision coefficient was calculated using the above expression (17) and the EMD simulated pair correlation h c t i o n contact values (see examples in Figures 4 and 5). Prelimimry results for this coefficient are shown in Figure 9. Comparison of these results with the data for zeolite-like pore system suggests that the diffusion coefficient of nanosystems is sensitive to the wall-to-fluid atomic diameter ratio (in Ref, [6b3 thisratio is 0.783, while for the above slit pore system it is eqwl to unity): its values change by the factor of 2 or more with a small (about 0.3) change in the ratio (see also Ref. [6b] for more details). The discussed examples codkm that the transport coefficients of nanofluids differ significantly (up to an order of magnitude, even for the simplest model nanofluids) from those specific to the corresponding bulk fluids. Thus, the rates of the processes in nanofluids dehed by diffusion, for example,
Ceramic Nanomaterials and Nanotechnology
115
will decrease by an order of magnitude as compared to similar processes in bulk fluids composed of the same atodmolecules. Both reaction-limited and difhion-bited stages of the synthesis process depend upon d a s i o n of various components of solutions participating in the corresponding processes that occur at the sub-nanoscale (near or inside of gel particles, nuclei, etc.). Therefore, they are defined by the transport coefficients specific to nanofluid difhsion, as those discussed above, and will be inhibited (or even enhanced), depending upon the structure factors, diameter ratios, and thermodynamic parameters of a particular system. Reliable prediction of the dfision coefficient values can significantly optimize old andor suggest new technologies of nanomaterials synthesis. As a particularly important example, in the following section the major experimentally investigated stages of aqueous synthesis of zeolite A materials are reviewed md discussed in conjunction with the utilization of the above analysis of the transport coefficients of nanofluids to outline a route to development of a methodology of synthesis of such materials.
EXPERIMENTAL PICTmJRE OF AQUEOUS SYNTHESIS OF ZEOLITE A MATERIALS. During several decades, experimental studies of ternplated aqueous synthesis of zeolite A materials concentrated on identification and investigation of several subsequent stages of the synthesis process and their role in shaping the outcome of the process. Recent experiments by Mintovaget al. point out four such stages: (1) nucleation of small (presumably, gelled) "droplets" of up to 8 nm in h e a r dimensions fiom the initial "clear" electrolyte solution before addition of a directing agent (tetrapropylamonium hydroxide, or TPA, in the case of Ref. [9J); within minutes after addition of TPA and stirring, coalescence of the (2) initial small gel particles to form gelled particles of about an order of magnitude larger in linear dimensions than the small particles of stage (1); after lengthy process of "aging", nucleation of crystalline zeolite nuclei (3) inside of the large gel particles (of about 80 nm in diameter) followed by growth of the nuclei until all the material inside of every gelled particle is consumed; (4) agglomeration of the (now crystalline) particles, re-crystallization and other Ostwald-ripening -like processes in the agglomerate. Although the experiments of Ref. [9] concern zeolite A synthesis enabled by the use of TMA as a directing agent, a very similar picture'oy1' is observed in the cases when the directing agents were TPA and other organic molecules. Some variations of results of the experiments concerning size of the particles obtained at the stages (1) and (2), and some disagreement on details of the mechanisms involved in each process above can be caused by the fact that the investigated
116
Ceramic Nanomaterials and Nanotechnology
systems were sometimes rather different fiom the system in Ref. [9]. The studies completed at 0RNLl2 recently supply reliable resolution of the synthesis process starting fiom the stage (2), and also confirm the above picture (see Figure 10 for an example of zeolite materials studied at O N ) . Elaborate and state-of-the-art experimental techniques, such as HRTEM, SAXS/WAXS, HRD, SEM, neutron scattering, etc., have been employed by various experimental groups to detail the synthesis. For example, in the study of Ref. [I 13 thirteen solutions of varying concentrationof components (divided in 7 classes) used to synthesize nanoporous silicate materials were investigated in detail. While stage (1) was not captured in that study, stage (2) was investigated in great detail. The obtained results gave rise to the first ever, atltf I.lecessarily very approximate, crystallization field diagram specific to the TPA-basedsynthesis starting fkom stage 2. For all the studied initial solutions, the development of the remarkably stable colloidal system of gelled particles, or entire gellation of initial solution, requires a dynamical description rooted in statistical mechanics of ionic fluids and colloids. Such a description necessarily has to take into account the nanostage of the nucleation and growth of the gel particles that takes place at condition of high spatial inhomogeneity of the solutions. Further, the coalescence of the initial gel particles after addition of the directing agent at the stage 2 constitutes a highly non-equilibrium and fist process in a colloid system, that is, in a system composed of charged particles whose dimensions vary fiom atomic (ions) to nanodimensions. Further development of the crystalline zeolite nuclei inside of these large particles relies both on ion supply fiom the solution, and on the restructurization of the gel particle material. The fist process is entirely defined by diffusion of the ions in the highly inhomogeneous region that surrounds each of the gel particles, and the second process involves formation of a crystalline solid lattice that is highly anisotropic. It has been demonstrated by the authors of this paper m previous sections and publications theoretically, and in existing literature experimentally, that nanofluid transport properties can differ orders of magnitude fiom their bulk counterparts. For example, the PG-diffusion coefficients for slit and zeolite-like pores are at least an order of magnitude different fium the corresponding “bulk” fluid ones. As applied to the zeoilite-A synthesis, this supplies at least part of explanation of why the “aging” process of stage 3 (during which the subnanocrystals nucleate and grow inside large gel particles - the process controlled to a signif”cant degree by supply of ions fiom the solution via diffusion through gel material) takes a long time: the corresponding diffusion coefficients are of at least an order of magnitude less that those in the bulk. Further considerationincludes the nanofluid flows: the viscosity of such flows are at least an order of magnitude larger than that of the corresponding bulk fluids. Therefore, any collective motion in nanofluids is inhibited, which again contribute to the long time of any process development. While the simulation models (and
Ceramic Nanomaterials and Nanotechnology
117
0.0
-6
6
4
3
0.07
0.06 0.05
0.02
I
,
1
1
1
0.01
0.00
z/(2''20w) Figure 8. The zero-order approximation of the PG-theoretical self-difision coefficient, [denoted as D, of Eq. (13) in this paper] of the LJ fluid in a square channel of the zeolite-like pore of Re€ [Sb]. The ratio c ~ of ~ the p wall to fluid hard core diameters was 0.783. DOis a dimensional constant, D0=(3/2) o , 4 (n/@), where p=l/k~T,and k~T=1.5.See Ref. [6bJ for fbrther details.
118
Ceramic Nanomaterials and Nanotechnology
Figure 9. Prehmary results: the PG-theoretical diffirsion coefficient of the LJ nanofluid confimed in the slit nanopore of 3.20 in width, H, and 11.15240 in length, L:cross-sectiony=O. The axisx is orthogonal to the planes drawn through the centers of the wall atoms facing the pore void, and y (not shown) and z-axes are along the pore; o&=l. The EMD pair correlation function contact values were used, and the data was not smoothed. therefore, input data) used in the above theoretical studies do not closely reflect details of each of stages (1) to (4) of the above synthesis, they provide qualitative evaluation of possible changes in transport coefficients (and thus process rates) that happen at the moscale. The experimentalstudies confirm that particle dimensions, process parameters and conditions (for example, the presence or absence of seeds), duration of each stage, etc., m y change observably, sometimes significantly, from one system to another, but the picture of the process of templated synthesis of zeolite A
Ceramic Nanomaterials and Nanotechnology
119
materials fiom clear aqueous solutions does not change: it follows the above pattern roughly outhed by the four stages captured by experimental studies. This is a very important finding that indicates that the theoretical description of the synthesis is feasible and, once achieved, can be very general to cover a great variety of natural and industrial systems. The theoretical methods and results discussed in the previous sections demonstrate how this can be achieved. Such theoretical analysis is needed to predict structure, chemical, physical and electronic properties of the synthesized zeolites, so that experimental work become directed by knowledge and logical considerations, rather than laboratory experience alone, as it happens at present.
VIRTUAL PROCESSING OF ADVANCED NANOMATERIALS Modern statistical mechanical methods and in particular - the PG-theoretical approach discussed above, together with EMD and NEMD methods - suggest a new strategy for development of advanced nanomaterials. Among other advantages, the PG-approach suggests unambiguous and explicit expressions for the transport coefficients and relaxation times in terms of the equilibrium structure hctors (the density and correlation functions) of the systems. This permits prediction of a change in the nature and outcome of a transport process with a change in the surk/fluid structure and suggests a reliable way of
Figure 10. Left: the real-space HRTEM image and a difiaction pattern (top right) for the [ W O ] face of a silicate-1 material synthesized at OFWL. Right (an enlarged part of the real-space HRTEM image on the left): the sub-nm, highly ordered channels are clearly visible and form a precisely defined pattern at the [1001 face.
120
Ceramic Nanomaterials and Nanotechnology
materials processing control. In particular, this approach can be applied to processing of materials at the nanoscale to ensure their desired spatial structure and properties via manipulations the system structure, composition, and the process parameters. The PG-theoretical formulae include this information explicitly. Therefore, one can use the theoretical formulae (such as those for the transport coefficients that govern processing rates of any tramprt processes as well as the process outcome) to identifj. such sets of characteristic systedparameter quantities that would ensure desirable process rates, the fabrication outcome, etc. The means to hcilitate this strategy involve development of theory-based, simulation-added, computationally undemanding, and affordable software for virtual design13of advanced nanomaterials.
VIRTUAL FABRICATION CYCLE Equilibrium Statistical Mechanics
Data on Chemical
I
I
1
EMD: Structure and Statistical ProDerties Density, CorrelationFunctions, etc.
I I
1
1
IlEvaluation
Figure 11.
Scheme of the virtual nanomterial fabrication cycle.
The equilibrium structure fhctors of the processed nanosystems (required as input data by the PG-approach) may be obtained from equilibrium statistical mechanical considerations or simulated using the EMD technique. This latter technique has been used in this work to account €or the surfhce structure effects in detail, and the nanofluid and nano-mfhx structure factors are analyzed and used to calculate localized values of the PG-theoretical transport coeficients. The localized PG-transport coefficients so obtained have been kther used to derive
Ceramic Nanomaterials and Nanotechnology
121
simplified correlations for and to calculate the corresponding system-average values of the system transport coefficients. Figure 11 reflects the major stages of
the virtual fkbrication of nanomaterials. The virtual fhbrication approach to nanomterials development permits theory-based, software-added configuring of the necessary structure, composition and properties of the carrier matrix, pore surfkce on the one band, and the composition,properties and processing parameters concerning the processed fluid mixture on the other hand, to meet techological requirements applied to the fabricated nanomaterial. The data so obtained can be fiuther used to develop such mterhls experimentally. SeveraI objectives can be met using the above strategy. First, virtual kbrication ofnanomaterials is an obvious knowledge-driven, logical alternative to heuristic experimental routes existing at present: it provides means for manipulation of processed systems at atomic scale to obtain nanomaterials of desired properties. Second, the virtual fabrication is a very economical way to novel nanomaterial development, as the only expenses are those to develop s o h e that can be used in every particular technological context; all expenses related to unsuccessful experimental attempts at development of a particular nanomaterial will be spared. Third, virtual fabrication permits considering both the designed rsanomaterial functionality and its integration into the entire scheme of hardware/device/process in which the nanomaterial is supposed to function. Further development of the virtual nanomaterial fabrication strategy will ultimately lead to virtual development of devices, electronic circuitry, chemical reactors, etc.
CONCLUSIONS Virtual fabrication of simple nanomaterials (such as nano-semkonductors) using relatively simple processing, such as chemical vapor deposition, is entirely feasible at present, as the corresponding theoretical and simulation methods exist and continue to develop. Much more in-depth, fbndamentaltheoretical studies and simulations are needed to ensure an advantage of virtual fhbrication of materials at atomic/molecular scale via complicated chemical processing f?om complicated compounds. Experimental studies alone (while extensive and informative) due to their nature, cannot provide a systematic and logical approach to fabrication of advanced nanornaterials. The above discussed theoretical and simulation developments are capable of provision of a fundamental background for development of a unified technology of the synthesis with predictable and optimal outcome, in contrast to the present situation when a technology has to be developed specifically for every particular nanomaterial case, and no prediction of the properties of the fabricated nanomaterials is available before laboratory fabrication attempts.
122
Ceramic Nanomaterials and Nanotechnology
Acknowledgments:This work was sponsored by the project No. KC0203020 fiom the Division of Materials Sciences, Ofice of Science, the U. S. Department of Energy, through the contract No. DE-ACOS-OOOR22725. V.F. de Almeida thanks the Division of Chemical Sciences for partial support of his work through the project No. KC0302020. REFERENCES ‘See, for example, V.V. Slezov and J.W.P. Schmelzer, ‘%omments on Nucleation Theory,” Journal of Physical Chemistry of Solids, 59 E93 1507-19 (1998); V.V. Slezov, J. Schmelzer, and Ya.Y. Tkatch, ‘Number of Clusters Formed in Nucleation-GrowthProcesses,” Journal of Chemical Physics, 105 [1 81 8340-51 (1996); D.W. Heermann, L. Yixue, and K Binder, “Scaling Solutions and Finite-Size Effects in the Lifshitz-Slyozov Theory,” Physics A, 230 [l] 13248 (1996); C. Sagui and M. Grant, “Theory of Nucleation and Growth During Phase Separation,” P?iysical Review E, 59 [4] 4175-87 (1999); S.A. Kukushkin and AV. Osipov, “Kinetics of Thin Film Nucleation From Multi-Component Vapor,” Journal of Physical Chemistry of Soli& 56 [6] 831-8 (1995), etc.; see, also, F.F. Abraham, Homogeneous Nucleation Theory, Academic Press, New York, 1974. Such as G. Ozkan and P. Ortoleva, “A Mesoscopic Model of Nucleation and Ostwald FtipenhgBtepping: Application to the Silica Polyrnorph Systems,” Journal of Chemical Physics, 112 [23] 10510-25 (2000); H. Vehkamakl and I.J. Ford, “Critical Chster Size and Droplet Nucleation Rate From Growth and Decay Simulations of Lennard-Jones Clusters,” Journal of Chemical Physics, 112 [9] 4193-4202(2000); B. Senger, P. S c w D.S. Corti, R. Bowles, J.-C. Voegel, and H. Reiss, “A Molecular Theory of the Homogeneous Nucleation Rate. I. Formulation and Fundamental Issues,” Journal of Chemical Physics, 110 [13J 6421-37 (1999), etc.; see also, 0. Sahnel and J. Garside, Pl‘ecipitatian - Basic Principles and Indushial Applications, Butterworth-Hememann,Oxford, 1992. 3B.J. McCoy, “ Distribution Kinetics Modeling of Nucleation, Growth, and Aggregation Processes,” in UEF-CRE IIN: Special Issues (in press, 2002); B.J. McCoy, “A New Population BaIance Model for Crystal Size Distributions: Reversible, Size-Dependent Growth and Dissolution,’’ Journal of Colloid und Inteface Science, 240, 139-49 (2001); W.J. Sterling and B.J. McCoy, ‘‘Distribution Kinetics of Thermolytic Macromolecular Reactions,” AIChE J o m Z , 47 [lO] 2289-303 (2001); B.J. McCoy, “Hyperbranched Polymers and Aggregates: Distribution Kinetics of Dendrimer Growth,” Journal of Colloid and Interface Scence, 216, 235-41 (1999); B.J. McCoy, ‘‘Vapor Nucleation and Droplet Growth: Cluster Distribution Kinetics for Open and Closed Systems,” Journal of Colloid and Interface Science, 228,64-72, (2000); B.J. McCoy and G.
Ceramic Nanomaterials and Nanotechnology
123
Madras, “Evohrtion to Similarity Solutions for Fragmentation and Aggregation,’’ Journal of Colloid andhte~aceScience, 201,200-9 (1998), etc. 4(a) See, for example, LA. Poihar, Transport Theory of Inhomogeneous Fluids, World Scientific, New Jersey, 1994; L.A. Pozhar and ICE. Gubbins, ‘TranSp01-t Theory of Dense Inhomogeneous Fluids,” Journal of Chemical Physics, 99,8970-96 (1993); C. Rhykerd, 2. Tan,L.A. PO&, and ICE. Gubbins, The Properties of Simple Fluids in Carbon Micropores, Jozxmcrl of Chemical Society. Faraday Transactions, 87,2011-6 (1991); S.S. Sarman, D.J. Evm, and P.T.Cummings, Recent developments in Non - Newtonian Molecular Dynamics, Physics Reports, 305, 1-92 (1998), etc.; (b) J.N. Israelachvili, P.M. McGUiggan, and AM. Homola, “Dynamic Properties of Molecularly Thin Films,” Science, 420, 189-91 (1988); S. Granick, “Motions and Relaxations of Confined Fluids,” Science, 253,1374-9 (199 1). ’L.A. Pozhar arrd ICE. Gubbins, “Transport Properties of Inhomogeneous Fluid Mixtures,” International Journal of Thermophysics,20,805-12 (1999); L.A. Pozhar and KE. Gubbins, ccQuasihydrodynamics of Nanofluid Mixtures,” Physical Review E, 56, 5367-96 (1997); L.A. Pozhar and ICE. Gubbins, “Transport Theory of Dense Inhomogeneous Fluids,” Journal of Chemical Physics, 99, 8970-96 (1993); L.A. Pozhar and ICE. Gubbins, “Dense Inhomogeneous Fluids: Functional Perturbation Theory, the Generalized Langevin Equation, and Kinetic Theory,” Journal of Chemical Physics, 94, 136784 (1991); L.A. Pozhar, “A Master Equation for Dynamical Systems with Thermal Disturbances,” UKrainian Physical JOWXQ~, 34,779-88 (1989). 6(a) LA. Pozhar, E.V. Kontar, and M. Z.-C. Hu, “Poiseuille Flow and Viscosity of Nanofluids Confined in Slit Nanopores, J. Nanosci. Nanotech,‘‘ Journal of Nanoscience and Nanotechnology, 2, 209-227 (2002); LA. Pozhar, “Poiseuille Flow of Nanoflujds Confined in Slit Nanopores,” Discrete CTPUJ Continuous Dynumical Systems, added volume: Llynamical Systems and Dzferential Equations, Eds. J. Du and S. Hu, 3 19 (2001); (b) J.M.D. MacElroy, L.A. Pozhar and S.-H. Suh, “Self-Difision in a Fluid COWithin a Model Nanopore Structure,” Colloids and Swftaces A, 187188,493-507 (2001);
(c)L.A. Pozhar, “Structure and Dynamics of Nanofluids: Theory and Simulations to Calculate Viscosity,” Physical Review E, 61, 1432-46 (2000); E. Akhmatskaya, B.D. Todd, P.J. Daivis, D.L. Evans, ICE. Gubbins, L.A. Pozhas, “A Study of Viscosity Inhomogeneity in Porous Media,” JowPiaZ of Chemical Physics, 106,4684-95 (1997). 7L.A. Pozhar and K.E. Gubbins, ‘‘TransportTheory of Dense Inhomogeneous Fluids,” Journal of Chemical Physics, 99,8970-96 (1993).
124
Ceramic Nanomaterials and Nanotechnology
8L.A. Pozhar et al., “Statistical Mechanical Fundamentals of Synthesis of Zeolite A Materials From Aqueous Solutions” (unpublished results). 9S. Mintova, N.H. Olson, V. Valtchev, and T. Bein, “Mechanism of Zeolite A Nanocrystal Growth from Colloids at Room Temperature,” Science, 283, 958-60 (1999). ”See, for example, P.-P.E.A. de Moor,T.P.M. Beelen, and RA. van Santen, ”In Situ Observation of Nucleation and Crystal Growth in Zeolite Synthesis. A Small-Angle X-Ray Scattering Investigation on Si-“A-MFI,” JournaZ of PhysicaZ Chemistry B, 103, 1639-50 (1999); P.-P.E.A. de Moor, T.P.M. Beelen, B.U. Komanschek, 0. Diat, and R.A. van Santen, “In Situ Investigation of SiTPA-MFI Crystallization Using (Ultra-) SmalI- and Wide-Angle x-Ray Scattering,” Journal of Physical Chemistry B, 101,11077-86 (1997); P.-P.E.A. de Moor, T.P.M. Beelen, and R.A. van Santen, “SAXS/WAXS Study on the Formation of Precursors and Crystallizationof Silicates,” Microporous Materials, 9,117-30 (1997); P.-P.E.A. de Moor,T.P.M.Beelen, B.U. Komanschek, and RA. van Santen, ‘Wanorneter Scale Precursors in the Crystallization of Si-TPA-MFI,” Microporous and Mesoporous Materials, 21,263-9 (1998). “J.N. Watson, L.E. Iton, R.I. Keir, J.C. Thornas, T.L. Dowling, and 3.W. White, “TPA-SilicaliteCrystallization fiom Homogeneous Solution: Kinetics and Mechanism of Nucieation and Growth,” JuurmZ of Physical Chemistry B, 101, 10094-10104 (1 997). ”L. Khatri, M. 2.-C. Hu, C.J. Rawn, E.A. Payzant, M.T. Harris, J.-S. Lh, and L.F. Allard, Jr., “Molecularly Ternplated Nucleation and Growth of Silicate-1 Nanocrystals Dyring Hydrothml Synthesis,” in press (2002); M.2.-C. Hu, D.W. DePaoli, and D.T. Bostick, “Dynamic Particle Growth Testing: Phase I Studies,” Scientific Report ORNL/TM-ZOO1/100, Oak Ridge National Laboratory, 2001; L. Kbatri, unpublished results. I3L.A. Pozhar, ‘‘Vktual Nanohbrication of Electronic Materials,” Proceedings of the 2001 International Conference 08 ComputationalNunoscience, March I9 21,2001, Hilton OceanfEont Resort, South Carolina, USA, 188-91 (2001).
Ceramic Nanomaterials and Nanotechnology
125
This page intentionally left blank
Nanocomposites
This page intentionally left blank
SYNTHESIS OF NANOSTRUCTURED WC/Co POWDERS THROUGH AN INTEGRATED MECHANICAL AND THERMAL ACTIVATION
PROCESS Leon L. Shawl, Ruirnin. Ren'", Zhigang Ban', and Zhenguo Ymg2 Department of Metallurgy and Materials Science University of Connecticut, Storrs, CT 06269 Pacific Northeast National Laboratory, Richland, WA 99352
'
ABSTRACT In this study we describe a new process that allows for production of lowcost nanostructured WC/Co composite powders. The process is termed as the integrated mechanical and thermal activation (IMTA) process because of the integration of mechanical activation (via high-energy ball milling) and thermal activation (via thermal treatment) in the process. Using tungsten oxide (WO& cobalt oxide (COO)and graphite as the starting material, we have demonstrated that mechanical activation before carbothermic reduction and carburization reactions can greatly enhance the formation of WC. The product from this new process is nanostructured powder rather than coarse-grained counterparts. The high-energy milling allows the desired reaction to take place at low temperatures, thus preserving the nanostructure, while the thermal treatment is necessary to obtain the reaction between the constituents. The fundamentals associated with the IMTA process are discussed. INTRODUCTION WC/Co is one of hard cennets used widely in industry as tips for cutting tools and wear-resistant parts. Their reasonable resistance to oxidation and corrosion at high temperatures also makes them desirable as a protective coating for devices that require resistance to both corrosion and wear at elevated temperatures. The WC/Co nanocomposite materials have great potentials in further improving properties over the conventional coarse-grained counterparts [1-4]. The traditional method of making WC/Co cemented carbides is by carburizing of W powder, followed by crushing, grinding, and blending, and the Visiting Scientistfrom the Department of Materials Science and Engineering,Dalian Railway Institute,Dalian (116028). P. R. China. a
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
129
end-product is micrometer-sized powder [5,6]. The nanocrystalline WC can be synthesized in situ by ball milling elemental W and C powders at room temperature [7]or WO3, C and Mg powders at room temperature for more than 20 hours [8]. In the latter process, WO3 is reduced by elemental Mg to form MgO and W, and then W reacts with C to form WC. The product MgO needs to be leached away before the powder can be used. In recent years, another new process, called Spray Conversion Processing, has been invented to synthesize nanophase WC/Co composite powder [9-131. This new processing method consists of three sequential steps: (1) preparation and mixing of aqueous solutions of the precursor compounds to fix the composition of the starting solution; (2) spray drying of the starting solution to form a chemically homogeneous precursor powder; and (3) thermochemical conversion of the precursor powder to the desired nanostructured end-product powders. Recently, we have reported a different process that allows for production of low-cost nanostructured pure carbides (e.g. S i c and Tic) and nitrides (e.g. Si3N4, CrN and TiN) [14-181. The new process combines mechanical and thermal activation to enhance the formation of carbides and nitrides, and thus is termed as Integrated Mechanical and Thermal Activation ( M A ) process. The basic form of the M A process is to mechanically activate reactants at room temperature through high-energy ball milling, which is followed by heating the milled reactants to high temperatures to complete the synthetic reaction. For example, we have demonstrated that using TiO2 and graphite as the starting materials, nanostructured Tic powder can be synthesized via high-energy milling the reactants at ambient tern erature for 2 hours, followed by annealing this milled powder mixture at 1400! C in argon for 1 hour [15]. In comparison with the current industrial carbotherrnic reduction for synthesizing coarse-grained Tic powder, the IMTA process has reduced the temperature of carbothermic reduction [Eq.(l)]by -5OOOC and at the same time shortened the reduction time from 10-20 hours to 1-2 hours. Thus, the IMTA process not only results in the formation of nanostructured Tic, but also substantially reduces the reaction temperature and time [15]. TiOz + 3C = Tic + 2 CO
(1)
Although IMTA has been successfully demonstrated in the synthesis of nanostructured carbides and nitrides, its applicability to the synthesis of nanostructured materials with metallic components has not been studied to date. Thus, in this study we have investigated the feasibility of using the IMTA process to synthesize nanostructured materials with metallic component(s) such as WCKo cermets. In the following, details for synthesizing nanostructured WCKo composite powders through the IMTA process from tungsten oxide (WO3)and element cobalt (CO)or cobalt oxide (COO) are described. It will be shown that
130
Ceramic Nanomaterials and Nanotechnology
low-cost nanostructured WCKo powder can be produced via the IMTA process. The fundamentals associated with this new process will also be discussed.
EXPERIMENTAL The raw material for synthesizing nanostructured WClCowas a mixture of tungsten trioxide (WO3) powder of purity 99.8%(10 - 20 micrometers), graphite powder of purity 99.0% (-100 mesh) and cobalt oxide (COO)powder of purity 95% (-300 mesh) or pure cobalt powder of purity 99.0% (-250 mesh). The highenergy ball milling was conducted using a modified Szegvari attritor with tungsten carbide (WC)balls (4.76 mm in diameter). The modified Szegvari attritor has been shown to be effective in preventing the formation of the dead zone and producing uniform milling products within the powder charge [19]. A ball-to-powder weight ratio (i.e., the charge ratio) of 60:l and a milling speed of 600 RPM were employed in all the experiments. The charged canister was evacuated up to 10'2 ton, flushed with argon, followed by evacuation and finally back filled with argon of purity 99.95%at a pressure of 1.5 atm before the onset of milling. During milling the canister was cooled using circulation water with a flow rate of about 770 ml/min throughout the process and the temperature of the canister was monitored with an E-type thermocouple attached to the bottom of the canister. The powders milled for different times were subsequently heated in an argon atmosphere at a desired temperature ranging from 800 - 1000°C. The holding time for high temperature reactions depended on the temperature used. A heating rate of either 50O0C/h or 36OoC/hwas employed for all the experiments. Phase identification of the milled and heated powder mixtures was carried out using X-ray diffraction (XRD) methods with Cu Ka radiation (NorelcoPhilips Diffractometer). The average size of crystallites was determined based on XRD peak broadening using the Scherrer formula without consideration of internal strain [20]. Although the internal strain within powder particles was neglected in estimating crystallite sizes via XRD analyses, cross-examination of the crystallite size was conducted using an analytical transmission electron microscope ("EM,Philips EM420). Determination of particle morphology and crystal structures were also performed using "EM coupled with selected area electron diffraction (SAED). An environmental scanning electron microscope (Philips ESEM 2020) was also utilized to characterize the morphology and size of the powders. Specific surface area (SSA) analysis was carried out using nitrogen adsorption based on the Braunauer-Emmett-Teller (BET) theory (Quantachrome NOVA 1000). To investigate the effect of milling on carbothermic reduction, thermogravimetry analysis (TGA) was performed using a thermogravimetric analyzer (PERKIN-ELMER TGA7) under the protection of a high- urity argon flux ramping from room temperature to 1000°C at a heating rate of 10gC h i n .
Ceramic Nanomaterials and Nanotechnology
131
RESULTS AND DISCUSSION SEM images of a WO3-CO-graphite powder mixture before and after milling are shown in Fig. 1. It can be seen that the sizes of WO3, graphite and CO powders are all reduced to the submicrometer range in the first 3 hours of milling, beyond which little change in the particle size occurs even after 24 hours of milling (not shown in Fig. 1). XRD spectra of the powder mixtures shown in Fig. 1 are presented in Fig. 2. It can be seen that the integrated area of graphite peaks was substantially decreased after 3 hours of milling. This is in part due to a stronger absorption of X-ray radiation by tungsten than carbon, and in part due to its easy delamination along its basal plane and amorphization during milling [21]. A related study 1191 has shown that amorphization of graphite in the Si-plus-graphite powder mixture occurs with only 6-hours of milling under a similar milling condition as the present study. Figure 2 also shows that XRD reflections of WO3 exhibit broadening, suggesting small sizes of crystallites andor the presence of the internal strain. It is also noted that Fig. 2 does not exhibit X R D reflections of the CO metal that has been added to the powder mixture in order to prepare a WC - 10 wt.% CO composite. The absence of the COreflections is due to a stronger absorption of xray radiation by tungsten than by cobalt. The crystallite size of WO3 in the powder mixture shown in Fig. 1 and 2 was estimated as a function of milling time according to broadening of the (200) peak of WO3 and is presented in Fig. 3. Note that the crystallite size of WO3 has reached nanometer scales with only 3 hours of milling. The specific surface area (SSA) of the powder mixture shown in Fig. 1 - 3 is presented in Fig. 4 as a function of milling time. It can be seen that the SSA of the powder mixture increases sharply at the early stage of milling, peaks in the range of 6 - 12 hours of milling, and then decreases gradually. A similar phenomenon has also been observed in SiO2-plus-graphite and TiO2-plus-graphite powder mixtures [22,23]. The subsequent reduction in the SSA of the powder mixture after 6-hours of milling has been attributed to the change of carbon morphology with milling time ~31.
A comparison of the XRD spectra of powder mixtures after heating at 1000°C for 2 hours with and without prior milling is shown in Fig. 5. The two powder mixtures shown have the same starting composition for making cermets with a composition of WC - 18 wt.% Co. However, the tungsten oxide in the powder mixture with prior milling has been completely reduced and carburized to form WC, whereas the powder: mixture without milling still contains a substantial amount of tungsten oxide. This result clearly demonstrates that high-energy milling before carbothermic reduction and carburization can enhance
132
Ceramic Nanomaterials and Nanotechnology
carbothermic reduction and formation of WC. Furthermore, based on the X-ray peak broadening, the crystallite size of the WC formed from the curve with 12hours of milling in Fig. 5 is estimated to be 32 nm. SEM and "EM images of the WCICo product corresponding to the curve with 12-hours of milling shown in Fig. 5 are shown in Fig. 6 and 7,respectively. It is found that the particle size of the WCICo powder formed is in the range of 0.3 to 0.5 micrometers (Fig. 6), whereas the size of W C crystallites ranges from 30 to about 100 nm (Fig. 7), which is slightly larger than the average crystallite size (32 nm) estimated according to the X-ray peak broadening. Thus, it is concluded that the WCICo powder produced through the IMTA process is submicrometer-sized (0.3 to 0.5 pm particles) with nanostructures (30 to 100 nm grains). Figure 8 shows TGA curves of WO3-plus-graphite powder mixtures without and with milling for different times. Several features are noted from these curves. First, there is little weight loss during heating to 1000°C for the powder mixture without milling. Weight loss is an indicator for the degree of carbothermic reduction of WO3 and the formation of WC. Assuming that the overall reactions for the formation of coarse-grained WCICo at relatively hi h i? temperatures [S] are applicable here, the formation of WCICo during the 1000 C thermal treatment can then be written as:
WO3 + 4 c = wc + 3 CO coo + c = CO + CO Thus, the present result implies that it is necessary to heat the powder mixture without milling to above 1000°C for carbothennic reduction of WO3 and thus weight loss to occur [see Eq. (2)]. Second, the high-energy milling has reduced the carbothermic reduction temperature significantly. Finally, the reduction in the onset temperature for the reaction increases with increasing the milling time. For example, the onset temperature for the reaction is about 94OoC and 86OoC for the powder mixture milled for 3 and 24 hours, respectively. The results above clearly indicate that high-energy milling of WO3, COO (or CO)and graphite mixtures has led to such changes as (i) fine particles (Fig. l), (ii) increased specific surface areas (Fig. 4), (iii) small crystalhtes pig. 3), (iv) a large amount of structural defects andlor internal strains in crystals (Fig. 2), and (v) formation of amorphous carbon (Fig. 2). These structural changes induced by high-energy milling have greatly enhanced W C and CO formation (Fig. 5 and 8). On one hand, without high-energy milling a reaction temperature higher than 1000°C would be required (Fig. 5 and 8), which could lead to a loss of the nanostructure by grain growth, as occurred in conventional methods for making coarse-grained WCICo materials [5,6]. On the other hand, without thermal
Ceramic Nanomaterials and Nanotechnology
133
treatment the low-cost starting material could not be used because WC and CO cannot be formed by prolonged milling of WO3, COOand C mixtures at ambient temperature. To form WCKo at ambient temperature, either high-cost elemental powders 171 or reactive compounds combined with subsequent chemical leaching [8] has to be used. Thus, integration of mechanical activation and thermal activation is necessary in order to form nanostructured WC/Co from low-cost materials at low reaction temperatures. In order to understand the reaction sequence during the IMTA process, WO3 + graphite and WO3 + graphite + COpowder mixtures have been milled for 12 hours and then heated at 1000°C in argon for different times. The reaction products are subsequently examined using XRD. The results from these experiments are shown in Fig. 9 and 10. It is found that for the W03-plus-graphite system the reaction sequence at 1000°C is WO3 + W18049 +WO2 +W 3 W2C + WC. The reduction sequence found (from WO3 to W) is consistent with the result reported by other investigators working on reduction of coarse-grained WO3 [24], while the carburization sequence (from W to WC) is in agreement with the W-Cequilibrium phase diagram that shows the presence of W2C between W and WC phase zones [25]. This result indicates that high-energy milling prior to the high temperature reaction, although enhancing the formation of WC, does not alter the reduction sequence established for coarse-grainedWO3. With the addition of CO (or COO)to the WO3-plus-graphite system, the reduction sequence of WO3 remains almost the same; however, the carburization sequence becomes much more complicated with the appearance of several ternary carbides such as cOgw&, cow$ (previously known as co3Wgc4) and co3w&. Although the addition of CO (or COO) has complicated the carburization sequence, the presence of CO has substantially enhanced the formation of WC. For example, the WO3-CO-C system after heating at 10oO°C for 30 minutes is primarily composed of WC (Fig. lO), whereas the WO3-C system with the same processing condition consists of only W2C, and for this system 120 minutes of heating is required in order to substantially reduce WzC (Fig. 9). The enhanced formation of WC in the CO-containing system is most likely to be related to the catalytic behavior of Co. It is we11 known that COis one of the best catalysts for the decomposition of CO, which in turn provides active carbon to reduce various oxides and to carburize elemental W and many intermediate compounds (e.g. CObw&, cOw3c,cO3w3c and w 2 c ) to f0m wc.
CONCLUDING REMARKS The feasibility of using an integrated mechanical and thermal activation (IMTA) process to prepare nanostructured WCKo has been demonstrated in this study. The WC/Co powder produced using the IMTA process is submicrometersized (0.3 to 0.5 pm particles) with nanostructures (30 to 100 nm grains). The
134
Ceramic Nanomaterials and Nanotechnology
integration of mechanical activation (via high-energy ball milling) and thermal activation (via thermal treatment) is the key to the success of the IMTA process. Without mechanical activation the reaction temperature would have to be high and the nanostructure could be lost. Without thermal activation the low-cost starting material could not be used; in stead, high cost elemental or reactive compound powders combined with prolonged milling times would be required if nanostructured WC/Co powder is to be formed at ambient temperature, Because of the integration of mechanical activation and thermal activation in the IMTA process, the nanostructured WC/Co can be prepared from low cost materials at low temperatures with short processing times. Thus, the IMTA process offers a promising cost-effective approach for large-scale fabrication of nanostructured WCKo materials.
Acknowledgements - The financial supports from the National Science Foundation under grant No: DMR-9710265 as well as the University of Connecticut Research Foundation are greatly appreciated.
REFERENCES 1. K.Jia, and T. E. Fischer, "Sliding wear of Conventional and Nanostructured Cemented Carbides," Wear, 203-204,3 10-318 (1997). 2. K. Jia, and T. E. Fischer, "Abrasion Resistance of Nanostructured and Conventional Cemented Carbides," Wear, 200,206-214 (1996). 3. B. K. Kim,G. H. Ha, G. G. Lee and D.W. Lee, "Structure and Properties of Nanophase WC/Co/VC/TaC Hardmetals," Nanostruct. Mater., 9, 233-236 (1997). 4. B. H. Kear and L. E. McCandlish, "Chemical Processing and Properties of Nanostructured WC-COMaterials," Nanostruct. Mater., 3, 19-30 (1993). 5. E. K. Storms, The Refractory Carbides, Academic Press, New York, pp. 143154 (1967). 6. D. H. Jack, "Cemented Carbide as an Engineering Material," in Engineering Applications of Ceramic Materials: Source Book, M. M. Schwartz, Eds., American Society for Metals, Materials Park,OH, pp. 147-153 (1985). 7. S. Mi and T. H. Courtny, "Synthesis of WC and WC-CO Cermets by Mechanical Alloying and Subsequent Hot Isostatic Pressing," Scripta Mater., 38 [l] 171-176 (1997). 8. M. Sherif El-Eskandarany, M. Omori, M. Ishikuro, T. J. Konno, K. Takada, K. Sumiyama, T. Hirai, and K. Suzuki, "Synthesis of Full-Density Nanocrystalline Tungsten Carbide by Reduction of Tungsten Oxide at Room Temperature," Metall. Mater. Trans., 27A [121 4210-4213 (1996).
Ceramic Nanomaterials and Nanotechnology
135
9. L. E. McCandlish, B. H. Kear, and B. K. Kim, “Carbothermic Reaction Process €or Making Nanophase WC-CO Powders,” U.S. Patent, 5.65 1,808 (1997). 10.L. E. McCandlish, B. H. Kear and B. K. Kim, ”Chemical Processing of Nanophase WC-CO Composite Powders,” Mater. Sci. Tech., 6, 953-957 (1990). 11. L. Gao and B. H. Kear, “Low Temperature Carburization of High Surface Area Tungsten Powders,’’Nanostruct. Mater., 5,555-569 (1995). 12.L. Gao and B. H. Kear, “Synthesis of Nanophase WC Powder by a Displacement Reaction Process,” Nanostruct. Mater., 9,205-208 (1997). 13. P. Seegopaul, L. E. McCandlish and F. M. Shinneman, “Production Capability and Powder Processing Methods for Nanostructured WC-COpowder,” Int. J. of Refractory Metals &Hard Materials, 15, 133-138 (1997). 14. L. Shaw, R.-M. Ren, and 2.-G. Yang, “High-Energy-Milling Enhanced Synthesis of Sinterable Carbides from Their Oxides,” U.S. Patent # 6,214,309,April 2001. 15. R.-M.Ren, 2.-G. Yang and L. Shaw, “Synthesis of Nanostructured Tic via Carbothermic Reduction Enhanced by Mechanical Activation,” Scripta Mater., 38 [5] 735-741 (1998). 16.R. Ren, Z. Yang and L. Shaw, “A Novel Process for Synthesizing Nanostructured Carbides: Mechanically Activated Synthesis,” Ceram. Eng. Sci. Proc., 19 [4] 461-468 (1998). 17. R.-M. Ren, Z.-G. Yang and L. Shaw, “Synthesis of Nanostructured Chromium Nitride through Mechanical Activation Process,” Nanostruct. Mater., 11 [11 25-35 (1999). 18. R.-M.Ren, Z.-G. Yang and L. Shaw, “Nanostructured TiN Powder Prepared via an Integrated Mechanical and Thermal Activation Process,” Mater. Sci. Eng., A286,65-71 (2000). lg.Z.-G. Yang and L. Shaw, “Synthesis of Nanocrystalline Sic at Ambient Temperature through High Energy Reaction Milling,” Nanostruct. Mater., 7 [8] 873-886 (1996). 20. H. P. Klug and L. E. Alexander, X-Rav Diffraction Procedures for Polycrvstalline and Amorphous Materials, John Wiley & Sons, Inc., London, pp. 491-494 (1954). 21. J. Tang, W. Zhao, L. Li, A. U. Falster, W. B. Simmons, Jr., W. L. Zhou, Y. Ikuhara and J. H. Zhang, “Amorphization of Graphite Induced by Mechanical Milling and Subsequent Crystallization of the Amorphous Carbon upon Heat Treating”, J. Mater. Res., 11 [3]733-738 (1996). 22. R.-M. Ren, Z.-G. Yang and L. Shaw, “Polymorphic Transformation and Powder Characteristics of TiOz during High Energy Milling,” J. Mater. Sci., 35,6015-6026 (2000).
136
Ceramic Nanomaterials and Nanotechnology
23. R.-M. Ren, Z.-G. Yang and L. Shaw, “Synthesis of Nanostructured Silicon Carbide through Integrated Mechanical and Thermal Activation Process,” J. Am. Ceram. Soc., in press. 24. J. Haber, J. Stoch and L. Ungier, “Electron Spectroscopic Studies of the Reduction of WOs,” J. Solid State Chem., 19, 113-1 15 (1976). 25. R. V. Sara, “Phase Equilibria in the Tungsten-Carbon System”, J. Am. Ceram. SOC., 48 [S] 251-257 (1965).
Figure 1. SEM images of a WO3-CO-C powder mixture with (a) no milling, (b) milling €or 3 hours, and (c) milling for 12 hours. The molar ratio of W03:C:Co in the powder mixture is 1:4.5:0.34that gives rise to a WC - 10 wt.% COcennet after synthesis. 0, G and W in (a) represent CO,graphite and WO3, respectively.
Ceramic Nanomaterials and Nanotechnology
137
Figure 2. XRD patterns of the WO3-Co-C powder mixture shown in Fig. 1.
60 $50
8-40
3
30 20 10
10
1
Figure 3. The crystallite size of WO3 in the WO3-Co-C mixture as a function of milling time.
138
0
Figure 4. The specific surface area (SSA) of the W03-Co-C mixture as a function of milling time.
Ceramic Nanomaterials and Nanotechnology
0 : WO2 0 ;Graphite
*:WC .:CO
With 1Zhours milling
20
40
so 20
80
70
80
Figure 5. XRD patterns of original and milled WO,-Coo-graphite powder mixture annealed at 1000°C in argon for 2 hours.
Figure 6. SEM image of as-synthesized WCKo powders corresponding to the curve with 12-hours of milling shown in Fig. 5.
Figure 7. TEM image of as-synthesized WUCo powders shown in Fig. 6.
Ceramic Nanomaterials and Nanotechnology
139
70 - 65 --
60
7
750
I
I
I
1
:
800
850
900
-
950
1000
Temperature ( C )
Figure 8. Thermogravimetic analysis of W03-C powder mixtures with and without milling. The continuous heating is conducted in an argon atmosphere with a heating rate of 1o0C/min.
2.
.n
a
8
a W
Figure 9. XRD patterns of W03-C powder mixtures d l e d for 12 hours and heated at lO0OoCin argon for different times.
140
Ceramic Nanomaterials and Nanotechnology
Figure 10. XRD patterns of WO3-CO-Cpowder mixtures milled for 12 hours and heated at 1000°Cin argon for different times.
Ceramic Nanomaterials and Nanotechnology
141
This page intentionally left blank
SYNTHESIS OF NANOCRYSTALLINE Ni COATINGS REINFORCED WITH CERAMIC NANOPARTICLES Jianhong He and Julie M. Schoenung School of Engineering, University of California Irvine, Irvine, CA 92697-2175
ABSTRACT Nanocrystalline Ni powders and thermally sprayed coatings, containing nanosized AlN particles, were synthesized and characterized. The presence of the AlN particles in the powders drastically decreased the dimension of the Ni agglomerates. AlN particles of approximately 30 nm were dispersed throughout the Ni matrix and enhanced the development of a nanocrystalline structure in the Ni matrix during cryomilling. The Ni coatings containing ultra-fine AlN particles were characterized as having equaixed grains with an average size of 24 nm. The Ni coatings containing AlN particles exhibited improved microhardness and apparent toughness, when compared to corresponding coatings without the AlN particles. The increase in microhardness resulted from both grain refinement and the presence of nano-sized AlN particles, with the latter being the primary factor. INTRODUCTION The significant performance enhancement that results from the presence of small amounts of dispersed fine particles in materials is well recognized, i.e. improving strength through interactions with deformation faults [ 11 and inhibiting grain growth due to Zener pinning [2]. The contributions of nitrides and oxides, formed through reactions between powders and oxygednitrogen fiom the environment during mechanical milling in liquid nitrogen, to the thermal stability of nanocrystalline materials have been realized [3-51. Strengthening has also been found to result from the presence of ultra-fine precipitates in nanostructured coatings [6-71. Mechanical milling in liquid nitrogen has been employed to synthesize nanocrystalline Ni powders that can be used as feedstock powders to fabricate nanocrystalline Ni coatings [S]. The nitrogen atmosphere could result in the formation of hexagonal Ni3N (a=2.670 A, c=4.306 A), but the formation enthalpy of Ni3N at room temperature is quite low (dH298=0.8 KJ/mol [9]), indicating its instability at room temperature. Alternatively, the high hardness and thermal stability of aluminum nitride, AlN, are commonly known [10,11]. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property rJf The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without h e express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
143
Therefore, the aim of the present study was to synthesize nanocrystalline Ni coatings with a low percentage of aluminum nitride (AN)particles and to provide insight into the influence of these AIN particles on the microstructural evolution and mechanical properties of nanocrystalline Ni coatings. In the present study, the mechanical milling in liquid nitrogen process is preceded by the intentional addition of small amounts of AlN particles into the powder system. Nanocrystalline feedstock powders are thus created and are subsequently thermally sprayed into nanocrystalline coatings by using high velocity oxygen fuel (HVOF) technology. It should be noted that mechanical milling is not only a powerful approach to obtain nanocrystalline powders but also an effective technology to introduce dispersed second phase particles. In fact, the process was originally developed for the production of oxide dispersion strengthened superalloys [ 121. EXPERIMENTAL PROCEDURE Commercially available, pure Ni powder (Sulzer Metco Inc. Westbury, NY) with purity 3 99.5 wt. % and a nominal particle size of 45 f 11 ym, and AlN powder (CERAC Inc., Milwaukee, WI) with purity 99% and a nominal particle size of 1.97 ym were selected for the present study. The Ni powders were blended with AlN powders for 0.5 hours in the amounts of 0,0.5 and 2 wt. % AlN (or 0, 1.38 and 5.54 vol. % AlN, using specific gravity values of 3.26 [ 10, 113 and 8.902 [13] g/cm3 for AIN and Ni, respectively). The blended powders were mechanically milled in a modified Union Process 01-ST attritor mill with a grinding tank capacity of 0.0057 m3. Stainless steel balls of 0.635 cm diameter were used with a powder-to-ball mass ratio of 1: 20. The powder charge was 1 kilogram. The mill was operated at 180 rpm for 8 hours and liquid nitrogen was continuously introduced into the attritor tank. The flux of liquid nitrogen was properly controlled using a valve so that the milling temperature, monitored by a thermocouple, was maintained at a relatively constant value of 100 K. On the basis of ASTM E1019 and ASTM E1097 standards, chemical analyses of the as-received and cryomilled powders were conducted by Luvak Inc., a professional chemical analysis company located in Boylston, Massachusetts. The particle size and distribution were determined using a particle analyzer made by Coulter Co. (Miami, FL). Thermal spraying of the powders was carried out by using a Sulzer Metco DJ 2600 HVOF spray system on carbon steel (0.2 wt. % C) substrates that were located at a distance of 0.230 m away from the nozzle of the spray gun. Before spraying, the surface of the substrate was grit blasted with abrasive A 1 2 0 3 particles to create a superficial roughness, which improves the bonding strength between the coating and the substrate. The spray facility and process are described in detail elsewhere [14-151 and the spray parameters are listed in Table I. Four
144
Ceramic Nanomaterials and Nanotechnology
categories of coatings were produced using the following powders. (a) The asreceived Ni powders; (b) The cryomilled (cryo) Ni powders; (c) The cry0 (Ni+0.5 wt. % AIN) powders and (d) The cry0 (Ni+2 wt. % AN) powders. The corresponding coatings are referred to in this manuscript as the conventional Ni coating, the cry0 Ni coating, the cry0 (Ni+0.5 wt. % AIN) coating and the cry0 (Ni+2 wt. % AlN) coating. Five passes of spraying were performed for each coating.
X-ray diffraction (XRD) measurements were performed in the 28 range of 30 to 100 degrees using Cu & (h=0.15406 nm) radiation in a Siemens D5000 diffractometer equipped with a graphite monochromator. A low scanning rate of scans with a step size of 0.01' and a step time of 5 seconds was used to assure the detection of reflections from second phases with low percentages. After the effects of & were corrected, the peak position, full width at half maximum of XRD reflections of Ni, was computed using a software package in the Siemens D5000 diffractometer. In the present work, the hlly annealed Ni powder was used as a standard sample for the determination of instrumental broadening. Scanning electron microscopy (SEM) observations were performed on a Philips XL 30 microscope with a field emission gun. Transmission electron microscopy (TEM) studies were conducted on a Philips CM microscope operated at 200 keV, and the samples were prepared by using a standard grinding-dimpling-ion milling procedure. The microhardness of the cross-section of the coatings was tested and indentation cracking was examined using a Buehler Micromet 2004 microhardness tester. RESULTS AND DISCUSSION Powder Characteristics SEM secondary electron images in Fig. l a through Id show morphological differences among the as-received Ni powder (Fig. la), the cry0 Ni powder (Fig. lb), and the cry0 (Ni+AlN) powders (Fig. l c and d). As a result of cryomilling, spherical Ni powders were transformed into irregular, flake-shaped agglomerates. Compared to the smooth and well-defined agglomerate external surfaces of the cry0 Ni powder (Fig. lb), the agglomerate external surfaces of the cry0 (Ni+0.5 wt. % AlN) powder became rougher due to interaction with the hard particles
Ceramic Nanomaterials and Nanotechnology
145
(Fig. lc). An increase in the amount of hard particles to 2 wt. % led to agglomerates breaking down into small, ragged fiagments. The boundaries to
(a) As-received Ni powder;
(b) Ni powders cryomilled for 8h at lOOK
(c) (Ni+0.5 wt.% AlN) powders cryomilled for 8h at 100K,
(d) (Ni+2 wt.% AN) powders cryomilled for 8h at 100K, arrows indicate uncru shed agglomerates. Fig. 1 SEM secondary electron image indicating morphology of powders.
these ragged fiagments are difficult to distinguish (Fig. Id), which is a common morphological feature of mechanically milled composite powders [14, 16-17]. Observation with high resolution SEM has indicated that such agglomerates still consist of very fine particles [16]. It is noted that in the present cry0 (Ni+2 wt. % AlN) powder, a few uncrushed agglomerates are also seen, as indicated by the arrows in Fig. Id. Fig. 2 shows the distribution of powder particle/agglomerate size for the different powders. The as-received Ni powder had a narrow size range with median 41 pm, consistent with their nominal particle size of 45 k 11 p. Cryomilling increased the median size to 53 pm and broadened the range of the size distribution. The increase in agglomerate size caused by cryomilling has also been observed in other fcc materials [18]. The addition of hard particles led to a
146
Ceramic Nanomaterials and Nanotechnology
decrease in the average agglomerate size and an increase in the range of the size distribution, compared with cry0 Ni powder. Consistent with the SEM morphological images, the median agglomerate size of the cry0 (Ni+0.5 wt. % AlN) powders was measured to be 44 pm, slightly larger than that of the conventional Ni powder but 9 pm smaller than cry0 Ni powder, and the median agglomerate size of cry0 (Ni+2 wt. % AlN) powders drastically decreased to 17 Pm.
-D-
Conventional Ni
-A-
Cryo(Ni+O.S wt.Oh AIN)
-o- Cryo(Ni+2 wt.% AIN)
1
Agglomerate Size (pm)
Fig.2. Distribution of powder particle/agglomerate size (a)- rn -Conventional Ni powder with a median size of 41 pm; (b)- 0- Cryomilled Ni powder with a median size of 53pm; (c) -A-Cry0 (Ni + 0.5 wt. % AlN) powder with a median size of 44 pm; and (d)-o- Cry0 (Ni + 2 wt. % AlN) powder with a median size of 17 pm.
Ceramic Nanomaterials and Nanotechnology
147
During cryomilling, morphological and dimensional changes occurred not only with respect to the agglomerate but also with respect to the second phase, hard particles. After cryomilling the (Ni+AlN) powders, the AlN particles changed from their initial morphology as thin platelets [lO, 111 with an average size of 1.97 pm into very fine round particles with an average size of approximately 30 nm. Figs. 3a and b show a uniform distribution of AlN particles (white particles) in the Ni matrix in the cry0 (Ni+O. 5 wt. % AlN) powders. Milling caused the powder particles to continuously experience overlapping, welding, and fracturing as a result of the collisions between the powder and the milling media. In such a process, the fine, hard AlN particles impacted with the larger Ni agglomerates and were dispersively embedded into the Ni agglomerates.
(a) Distribution of AlN particles;
(b) Detailed view of the region indicated by the arrow in (a) Fig.3 Distribution of AlN particles in the cry0 (Ni+0.5 wt % AlN) powders.
On the basis of reflection line broadening of X-ray difiaction [16, 191, the grain size of the cry0 Ni, cry0 (Ni+0.5 wt. % AlN) and cry0 (Ni+2 wt. % AlN) were calculated to be 105, 65 and 33 nm, respectively. These results indicate that the addition of hard AlN particles accelerated the formation of a nanocrystalline structure. During mechanical milling, powders, which were forced by random cycling impact loading, experienced repeated fkacturing and cold welding [ 121. The fracturing and cold welding led to grain refinement and eventually to the formation of a nanocrystahe structure, because the welded fragments of initial coarse grains should be new grains [ 181. The presence of hard particles in the milling process promoted powders to fracture and grains to be refined partly because the hard particles impacted and partitioned the powder agglomerates into small fragments, as seen in Fig. Id, and partly because the embedded hard particles in the matrix repeatedly deconstructed grains of the matrix. The results of the chemical composition analysis for the powders are listed in Table 11. For comparison purposes, the nominal contents of aluminum and nitrogen for (Ni + AlN) powders, namely the aluminum and nitrogen that result
148
Ceramic Nanomaterials and Nanotechnology
directly from the AlN phase itself, were calculated and are also incorporated in Table 11. The results show that cryomilling leads to noticeable contamination. The increase in iron content resulted from the wear of the stainless steel milling tanks, shafts and balls, which were severely worn by the powders containing hard particles. The increase in aluminum content coming directly from AlN aside, the evident net change in aluminum content is not observed. In general, the contents of nitrogen, oxygen and carbon increased with increasing AlN content in the cryomilled powders. These elements originated from diffusion from the milling environments [ 14-17]. According to Table 11, the net increase in nitrogen content is 0.06 and 0.32 wt. % for the cry0 (Ni+0.5 wt.% AlN) and the cry0 (Ni+2 wt.% AN) powder, respectively. The presence of AlN particles also caused a noticeable increase in oxygen. The increased external surface areas in the cry0 (Ni + AlN) powders, because of the rougher surface and smaller powder agglomerate size (see Fig. l c and Id), is the primary cause for the increase in the oxygen content and the net increase in nitrogen content.
.%&-p
Table I1 Chemical CO nposition of as-received and cry0 Ni, NUAlN powders (wt.%) Code Al Fe N 0 C As-received Ni 0.02 I 0.48 1 0.014 I Crvo Ni 0.034 Cry0 (Ni+0.5 wt.% 0.36/0.33 0.03 1 ANhominal Cry0 (Ni+2 wt.% 0.90/0.68 1.17 0.054 AlNVnominal Microstructure And Mechanical Properties of The Coatings SEM back-scattered electron images in Fig. 4a to 4d show the cross-sectional microstructures of the different coatings. The average coating thickness ranged from 230 to 260 pm. A noticeable difference between the microstructures, observed with SEM, of the conventional Ni, the cry0 Ni and the cry0 (Ni + 0.5 wt. % AN) coatings was not observed. The lamellar structure, which is typically observed in thermally sprayed coatings, and the lamellar structure boundary resulting as the surface of droplets solidified [20], is not clearly visible. However, the cry0 (Ni+2 wt. % AlN) coating, Fig. 4d, exhibits a distinctive coarse lamellar structure at relatively low magnification. It is evident that the coarse lamellar structure corresponds to the spraying pass rather than the surface of droplets solidified. The five passes needed to make the coating are clearly seen, as a result of oxidation that occurred at the inter-pass boundaries.
Ceramic Nanomaterials and Nanotechnology
149
(a) Cross section of the conventional Ni coating;
(b) Cross section of the cry0 Ni coating;
(c) Cross section of the cry0 (d) Cross section of the cry0 (Ni + 0.5 wt.% AN) coating; (Ni + 2 wt.% AN) coating. Fig.4 SEM back scattered electron images of the different coatings produced by using HVOF thermal spray. The microstructures of the conventional Ni and the cry0 Ni coatings were examined using TEM, and the results indicated that the majority of the grains in the cry0 Ni coating were equiaxed with an average grain size of 50 k 23 nm [8]. Influence of the presence of AUV particles on microstructure of the coatings was therefore investigated using TEM in the present study, and the results are shown in Fig. 5. Fig. 5a indicates a typical microstructure of the cry0 (Ni+2 wt. % AN) coating. The majority of grains are also equiaxed, but with a much smaller average size of 24k15 nm. Fine lamellar structures were observed in areas of the sample of approximately 10%, see Fig. 5b. The fine, equiaxed grains were observed in the areas between the lamellar structures. Approximately 20% of the sample area consisted of even smaller equiaxed grains with average size of 16 k 10 nm, in which a large number of twins were present. Mechanical twins are observed in the various cryomilled powders [18], thus the presence of twins in the coating is probably from the preservation of twins in the feedstock powders and the related mechanism is under study.
150
Ceramic Nanomaterials and Nanotechnology
Fig. 5 TEM images of the cry0 (Ni + 2 wt.% AlN) coating. In comparison with the microstructures of the cry0 Ni coating, the following microstructural features were observed as a result of introducing 2 wt. % AN particles into the Ni matrix: (1) a distinctive lamellar structure on SEM crosssectional images, indicating that significant oxidation occurred at the inter-pass boundaries; (2) smaller grain sizes in the coatings, consistent with the smaller grain sizes in the cry0 (Ni+2 wt. % AN) powder and (3) the presence of twins in the cry0 (Ni+2 wt. % AN) coatings. Code
Conventional Ni
Cry0 Ni
Hardness (Kg/mm2> Increase in Hardness (percent)
243
258
Cry0 (Ni+0.5 wt. %Am) 304
N/A
6
25
Cry0 (Ni+2 wt. %rn> 378
59
The microhardness of the different coatings in cross-section was tested with a load of 300g and the results are listed in Table 111. Microhardness increased by only 6 % in the cry0 Ni coating as compared to that of the conventional Ni coating. The addition of 2 wt. % AN particles, however, led to an increase of approximately 60%, which is 10 times as large as the influence of cryomilling alone. Compared with the cryomilled Ni coating, microhardness is increased by
Ceramic Nanomaterials and Nanotechnology
151
18% and 47 % for the cry0 (Ni + 0.5 wt. % AlN) and the cry0 (Ni + 2 wt. % AlN) coatings, respectively. It is universally accepted that both grain refinement (i.e., Hall-Petch equation, H,=Ho+kd-l”, where Ho, k are constants and d is the diameter of the grain) and the presence of fme, hard and dispersed particles lead to an increase in hardness [l]. In the present study, both the grain refinement and the presence of fme, hard and dispersed particles play a role in increasing hardness. On the basis of the Hall-Petch equation, the microhardness of the cry0 (Ni+2 wt. % AN) coating was calculated to be 278 assuming approximate grain sizes of 100, 50 [8] and 24 nm (see Fig. 5a) for the conventional Ni, the cry0 Ni and the cry0 (Ni+2 wt. %AN) coatings, respectively. This predicated value is much lower than the average measured value of 378. In fact, even the calculated value of 278 was an overly high estimated value, because the hardness of nanocrystalline materials has usually been observed to be lower than that predicted using the Hall-Petch equation [21-231. Therefore, the major contribution to the increase in microhardness resulted from the presence of fine, hard and dispersed particles.
(a) Conventional Ni coating at 500g;
(b) Cry0 (Ni + 2 wt. %AN)coating at 500 g.
(d) Cry0 (Ni + 2 wt. %AN)coating at 1000 g. Fig. 6 Indentation cracking. Arrows indicate cracking.
(c) Conventional Ni coating at 1OOOg;
152
Ceramic Nanomaterials and Nanotechnology
The indentation cracking method is often employed to estimate the relative toughness of coatings [15]. In the present study, an indentation cracking examination was conducted, and the results are shown in Fig. 6a to 6d. Under the same load, indentation marks in the cry0 (Ni +2 wt. % AlN) coating are smaller than those in the conventional Ni coating, because the former has a higher hardness than the latter. Under a load of 500g, a crack, located far away from the indentation as shown by the white arrow, was present in the conventional Ni coating. No cracking was observed around the indentation in the cry0 (Ni+2 wt. % AN) coating. Under a load of lOOOg, many cracks near the indentation, even delamination of the coating, were observed in the conventional Ni coating. In contrast, only a short crack was observed in the cry0 (Ni+2 wt. % AlN) coating. It should be noted that indentation cracking results for the cry0 Ni and the cry0 (Ni + 0.5 wt. %) coatings also indicate improved apparent toughness compared with the conventional coating, however, the magnitude is much lower than that of the cry0 (Ni+2 wt. % AlN) coating. Therefore, the present results suggest that the cry0 (Ni+2 wt. % AN) coating possessed higher apparent fracture toughness relative to that of the conventional Ni coating. In a previous study, it was also found that a nanostructured cement coating exhibited higher apparent fracture toughness than its corresponding conventional coating [ 151. Identification of Particle Phase in Coatings X-ray difiaction spectra for the four coatings, as well as for the cry0 (Ni+2wt. % AlN) powders are shown in Fig. 7. In order to analyze the hard particle phases with low volume fraction in the coatings, a very slow scan rate of 0.12 deg/min was employed when the X-ray dif€i-actionexperiments were carried out. Even so, AlN particles were not detected in any of the coatings. Moreover, distinctive reflections from the AN phase was also not observed in the X-ray spectrum for the cryomilled (Ni+2 wt. % AlN) powder, although SEM observations indicate a uniform distribution of AN particles in the cryomilled (Ni+AlN) powders, see Fig. 2. The failure to detect AlN particles by using X-ray diffi-action does not indicate the absence of AlN particles in the coatings. AlN exhibits a melting point approximately 280OoC [24] and is thermally stable up to 220OoC [25], much higher than the temperatures of 1300 to 1900°C that powders experience during thermal spraying [20]. Thus, it is presumed that AlN particles were stable during thermal spraying. However, AlN powder is unstable in moist air and releases ammonia odor [lO]. Thermodynamically, the hydrolysis reaction: 2AlN+3H20=2NH3+Alz03 is possible because the free energy AG of the reaction at 300 K is calculated to be -332.4 kJ,on the basis of the thermodynamic data [9]. In fact, ammonia odor was detected when the cry0 (Ni + AN) powders were sampled. Therefore, it is believed that a small amount of AlN has decomposed, even though A1203 was also not detectable by X-ray diffraction (the angles where
Ceramic Nanomaterials and Nanotechnology
153
the reflections of A1203 would be are also indicated in Fig. 7). Evidently, the concentrations of both AlN and A1203 are too low in the coatings and the powder to be detected by XRD. 800
600
2
I
i
400
0 ........ . ..
200
M
0
30
40
50
60
70
80
90
1 10
Diftaction angle (20) Fig. 7 X-ray diffraction spectra. The symbols and indicate the angle where the reflections of AlN and A1203 phases would be, respectively. (a) cryomilled (Ni+2 wt.% AlN) powder; (b) conventional Ni coating; (c) cryomilled Ni coating; (d) cry0 (Ni+0.5 wt. % AlN) coating; and (e) ) cry0 (Ni+2 wt. % AlN) coating. The presence of NiO phase was confirmed by X-ray difiaction. Only traces of NiO phase were detected in the conventional Ni and the cry0 Ni coatings. In the cry0 (Ni+0.5 wt. % AlN) and the cry0 (Ni+2 wt. % AN) coatings, 1.3 and 4.1 vol. % NiO was detected, respectively. Only a trace of NiO was detected in the cryomilled (Ni+2 wt. % AlN) powders, indicating that the NiO observed in the coatings primarily formed during thermal spraying. The oxidation process occurs rapidly in the presence of oxygen in the surrounding air andor excess oxygen during HVOF spraying; particles exposed to high temperature can oxidize during in-flight and after impingement onto the substrate surface [26]. The formation of an oxide layer comprises the complex processes of adsorption of oxygen
154
Ceramic Nanomaterials and Nanotechnology
molecules on the material surface, dissociation of the oxygen molecules, formation of oxide nuclei and finally the growth of nuclei to form a complete oxide layer [27]. Therefore, the amount of oxide formed during HVOF spraying depends on the oxygen concentration on the particle surface, the free specific surface area of the particles, and the temperature and time that the particles are exposed to oxygen. Because of the increased external surface area and higher oxygen content (Table 11) of the (Ni + AN) powders, as well the fact that small agglomerates experience higher temperatures than do large agglomerates during thermal spraying [20], it is not surprising that the volume fraction of the NiO phase increased with increasing amounts of the AlN particles.
Fig. 8 TEM selected area diffraction pattern for the cry0 (Ni+2wt.% AlN)coating.
Ceramic Nanomaterials and Nanotechnology
155
TEM selected area diffraction pattern of TEM, corresponding to the bright field image shown in Fig. 5a, with its index is shown in Fig. 8. The diffraction patterns from the AlN phase were imaged and are indicated by the arrows. The diffraction patterns from the AlN phase were quite weak and very close to those from the Ni matrix and NiO phase, and the strongest diffraction of { 1 0 0 } ~ was covered in the range of the lamination from the central transmission spot. This led to unsuccessful attempts to take dark field images of the AlN phase with the purpose of exhibiting the dimension and the distribution of particles. using the { 111}NiO reflection, a dark field image was taken and is shown in Fig. 9. The result indicates that the NiO phase Fig. 9 TEM dark field image Of cryomilled in the form of round (Ni + 2 wt*%m)taken by using { 1111NiO, particles, the bigger Ones sized indicating NiO particles. approximately 30 nm and the smaller ones less than 10 nm. CONCLUSIONS In the present study, nanocrystalline Ni powders containing ultra-he AlN particles were synthesized by using cryomilling. The resultant powders were employed as feedstock powders to fabricate hard-particle strengthened nanocrystalline Ni coatings. The results are briefly summarized as follows. 1. The presence of AlN particles in the powders decreased the agglomerate size and increased the surface roughness of the agglomerates. 2. The AlN additive was reduced in size to ultra-fine particles of approximately 30 nm in diameter. These particles were dispersively embedded into the Ni matrix and enhanced the development of a nanocrystalline structure in the Ni matrix during cryomilling. 3. The presence of AlN also led to an increase in the amount of NiO phase that distributed in the coatings in the form of ultra-fine, round particles. 4. Decreased grain size was achieved in the AlN strengthened Ni coating, compared to the cry0 Ni coatings. Smaller initial grains in feedstock powder caused by the addition of AlN and the increase in amount of the dispersed,
156
Ceramic Nanomaterials and Nanotechnology
ultra-fine particles (AlN and NiO) that inhibits grain growth, were attributed to the observed smaller grain size in the AlN strengthened Ni coating. 5. Compared with that of the cryomilled Ni coating, an increase in microhardness by 18% and 47% was obtained for the addition of 0.5 and 2 wt. % AIN, respectively. Indentation cracking results indicated the fine, dispersed AlN particles raised the apparent toughness of the Ni coating. The increase in microhardness resulted from both grain refinement and the presence of ultrafine particles. However, the latter played the primary role in strengthening. ACKNOWLEDGMENTS The authors thank Dr. K. H. Chung and Dr. L. Ajdelsztajn of the University of California, Irvine, for their assistance with the cryomilling and thermal spraying experiments. The authors also gratefully acknowledge financial support provided by the Office of Naval Research under Grants N00014-01-C-0384 and N0001402- 1-0213. REFERENCE 1. A. J. Ardell, Precipitation hardening, Metall. Trans. A , Vol. 16A, No.12, 2131-2165, 1985. 2. D. G. Morris, and M. A. Morris, Microstructure and strength of nanocrystalline copper alloy prepared by mechanical alloying, Acta Metall. Mater., Vol. 39, No. 8, 1763-1770, 1991. 3. R. J. Perez, H. G. Jiang, C. P. Dogan, and E. J. Lavernia, Grain growth of nanocrystalline cryomilled Fe-Al powders, Metall. Mater. Trans. A , Vol. 29A, NO. 10, 2469-2475, 1998. 4. B. Huang, J. Vallone and M. J. Luton, The Effect of nitrogen and oxygen on the synthesis of B2 Nial by cryomilling, Nanostructured Mater., Vo1.5, No.6, 63 1-642, 1995. 5 . J. D. Whittenberger and M. J. Luton, Elevated temperature creep properties of Nial cryomilled with and without Y203, J. Mater. Res., Vol.10, No.5, 11711186,1995. 6. J. He and E. J. Lavernia, Precipitation phenomenon in nanostructured Cr3C2NiCr coatings, Mater. Sci. Eng. A, Vol. A301, No. 1, 69-79, 2001. 7. J. He, M. Ice, J. M. Schoenung, D. H. Shin and E. J. Lavernia, Thermal stability of nanostructured Cr3C2-NiCr coating, J. Thermal Spray Technol., V01.10, NO. 2,293-300, 2001. 8. M. L. Lau, J. He, A. J. Melmed, T. A. Lusby, R. Schweinfest, M. Ruhle and E. J. Lavernia, Microstructural Evolution in Nanocrystahe Ni Coatings, J. Mater. Res.(in press) 2002. 9. E. A. Brands and G. B. Brook, in Smithells Metals Reference Book, 7th edition, Butterworth-Heinemann, Oxford, UK, 1992, p. 8:23-8:26.
Ceramic Nanomaterials and Nanotechnology
157
10. Certificate of Analysis comes with the AN powder, CERAC Inc., Milwaukee, WI, 2001. 11. F. G. Wilson and T. Gladman, Aluminum nitride in steel, Int. Mater. Rev., Vol. 33, NOS, 221-286, 1988. 12. J. S. Benjamin, Dispersion strengthened superalloys by mechanical alloys, Metall. Trans. Vol. 1, No. 10, 2943-295 1, 1970. 13. M. L. Bauccio, in Metals Reference Book, 3rd edition, ASM International, Materials Park, OH, 1999 p. 153. 14. J. He, M. Ice, S. Dallek, and E. J. Lavernia, Synthesis of nanostructured WC12 pct CO coating using mechanical milling and high velocity oxygen fuel thermal spraying, Metall. Mater. Trans. A , Vol.31A, No.2,541-553,2000. 15. J. He, M. Ice, and E. J. Lavernia, Synthesis of nanostructured Cr3C225(Ni20Cr) coating, Metall. Mater. Trans. A, 2000, Vo1.3 1A, No.2, 555-564, 2000. 16. J. He, L. Ajdelsztajn, and E. J. Lavernia, Thermal stability of nanocrystalline WC-CO powder synthesized by using mechanical milling at low temperature, J. Mater. Res., Vol.16, No. 2,478-488, 2001. 17. J. He, M. Ice and E. J. Lavernia, Synthesis and characterization of nanostructured CrsC2-NiCr, NanoStructured Mater., Vol. 10, No. 8, 12711283,1998. 18. J. He and E. J. Lavernia, Development of nanocrystalline structure during cryomilling of Inconel625, J. Mater. Res., Vol.16, No. 9,2724-2732,2001. 19. H. P. Mug and L. E. Alexander, X-ray DifSraction Procedures, John Wiley & Sons, New York, 1974, p. 643. 20. J. He, M. Ice and E. J. Lavernia, Particle melting behavior during highvelocity oxygen fuel thermal spraying, J. Thermal Spray Technol., Vol. 10, No.1, 83-93,2000. 21. X. D. Liu, M. Nagumo, and M. Umemoto, The Hall-Petch relationship in nanocrystalline materials, Mater. Trans., JIM, Vol. 38, No. 12, 1033-1039, 1997. 22. C. Suryanarayana, D. Mukhopadhyay, S. N. Patankar, and F. H. Froes, Grain size effects in nanocrystalline materials, J. Mater. Res., Vo1.7, No. 8, 21142117,1992. 23. D. A. Konstantinidis and E. C. Mantis, On the “anomalous” hardness of nanocrystahe materials, NanoStructured Mater.,Vol. 10, No.7, 1111-1118, 1998. 24. H. A. Wriedt, in Binary Alloy Phase Diagrams, 2ndedition, T. B. Massalski, ed., ASM International, Materials Park, OH, 1990, p. 176. 25. F. Benesovsky, in Encyclopedia of Chemical Technology, 3rd edition, M. Grayson, ed., John Wiley & Sons, New York, 1978, vol. 15, p. 876.
158
Ceramic Nanomaterials and Nanotechnology
26. V. V. Sobolev, J. M. Guilemany, J. Nutting and J. R. Miquel, Development of substrate coating adhesion in thermal spraying, Int. Mat. Rev., Vol. 42, No.3, 117-136, 1997. 27. P. Scharwaechter, M. Wimmer, R. Wurschum, D. Plachke and H. D. Carstanjen, Interaction of oxygen with nanocrystaUine metals, NanoStructured Mat., Vol. 11, No. 1, 37-42, 1999.
Ceramic Nanomaterials and Nanotechnology
159
This page intentionally left blank
SINGLE-WALL CARBON NANOTUBES REINFORCED ALUMINA NANOCOMPOSITES CONSOLIDATEDBY SPARK-PLASMA-SINTERG
Guo-Dong Zhan, Joshua Kuntz, Julin Wan, Javier Garay, and A. K. Mukheqjee Department of Chemical Engineering and Materials Science, University of California, Davis, CA956 16, USA ABSTRACT Carbon nanotubes have extraordinary characteristics not only in electrical and heat conductivities but also in mechanical properties. Therefore, they have recently emerged as potentially attractive materials for the reinforcement of ceramics, However, the potential application of carbon nanotubes in the reinforcement of ceramic composites has not been successfully demonstrated so far. In the present study, we have successfully realized this possibility in reinforcing nanocrystalline ceramic nanocomposites. Single-wall carbon nanotube/A1203 nanocomposites up to 10~01%with nanocrystalline alumina matrix (20Onm) have been fabricated at 1150°C in three minutes by spark-plasmasintering. A dramatic improvement in toughness of more than 200% as compared to pure nanocrystalline alumina has been achieved in the lOvol%SWCN/A1203 nanocomposite. INTRODUCTION The fabrication of nanocrystalline ceramics is an exciting theme in materials research because such bulk materials with nanocrystalline grain size less than 100 nm exhibit novel properties as compared with their microcrystalline However, the brittleness of nanocrystalline ceramics limits their potential and promise for use in structural application. The development of nanocomposites by addition of a second phase can overcome the inherent brittleness of nanoceramics. Many attempts have been made for improving the mechanical properties of nanocrystalline ceramics during our current investigation, such as by adding Sic whisker, dispersions of 2 1 0 2 particles and metallic particles that have been widely used for microcrystalline counterparts. Significant progress has been achieved in these nanocrystalline (-1 OOnm) alumina matrix nanocomp~sites.~
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
161
Recently, the unique electronic and mechanical properties of carbon nanotubes including single-wall carbon nanotubes (SWCN) and multiwalled carbon nanotubes (MWCN) have prompted intense research into a wide range of applications in materials, electronics, chemical processing and energy management: Single-wall carbon nanotubes possess extraordinary electrical conductivity with a resistivity of 104 ohm-cm at 300K,making them the most conductive fibers known. In addition, it has now been shown that they have a thermal conductivity at least twice that of diamond. The extraordinary properties of single-wall carbon nanotubes are not limited to electrical and thermal conductivities, but also include mechanical properties, such as stiffness, toughness, and strength. Single-wall carbon nanotubes are among the stiffest fiber known, with a measured Young’s modulus of 1400 GPa. They have an expected elongation to failure of 20-30%, which combined with the stiffness, projects to a tensile strength well above 100 GPa, by far the highest known. These properties point to a wealth of applications for exploiting them in advanced composites. For instance, their addition to a polymer matrix leads to a very low electrical percolation threshold and allows one to obtain, with only very small amounts of SWCN, and electrical conductivity sufficient to provide an electrostatic discharge. However, the utilization of the extraordinary mechanical properties of carbon nanotubes in ceramic composites has not been successfully realized so far. Flahaut et a15-6and Peigney et a17-*reported a novel catalytic route to develop insitu SWCN or MWCN-Fe-Al203 nanocomposites. It was found that the fracture strength of most carbon nanotubes-Fe-AlzO3 composites is only marginally higher than of alumina and markedly lower than those of the carbon-free Fe-Al203 composites. In addition, the fracture toughness is lower than or similar to that of alumina. Recently, Siege1 et a1 have reported that 24%improvement of fracture toughness over pure alumina could be obtained in lOvol%MWCN filled alumina nanocomposites. These results seem to indicate that it is difficult to realize the reinforcing effect by carbon nanotubes in the ceramic composites. Single-wall carbon nanotubes are polymers of pure carbon and can be reacted and manipulated using the tremendously rich chemistry of carbon. Structurally, a nanotubes is like a single graphitic sheet wrapped around a cylinder and capped at the ends. More significantly, single-wall carbon nanotubes are believed to be defect free, which means that they are free of propertydegrading flaws in the nanotube structure that is different from multiwalled carbon nanotubes. Therefore, we selected high-quality “ropes” of single-wall carbon nanotubes as a reinforcing phase for the present study. This paper will report the processing, microstructure, and mechanical properties of single-wall carbon nanotubes reinforced nanocrystalline alumina nanocomposites consolidated by spark-plasma-sintering(SPS).
162
Ceramic Nanomaterials and Nanotechnology
EXPERIMENTAL PROCEDUWS Purified single-wall carbon nanotubes (also called Buckytubes) produced by the continuous process was obtained from Carbon Nanotechnologies Incorporated (Houston, TX 77084). More than 90% of the catalyst particles have been removed in these purified continuous process Buckytubes. Buckytubes are single-wall carbon nanotubes, in which a single layer of graphite - grupherse - is rolled up into a seamless tube with either open or capped ends (Fig.la). The diameter of single-wall carbon nanotubes is 0.7 to 2 nrn (typically about 1.O nm) 100,000 times thinner than a human hair. Buckytube lengths are typically hundreds of times their diameters. Another structural aspect of tubes is their selforganization into "ropes," as shown in Fig. lb, which consist in many (typically, 10-100) tubes running together along their length in van der Waals contact with one another. Ropes are far longer than any individual tube in them: whereas tubes are typically about 100-1000 nm in length (and about 1 nm in diameter) ropes are essentially endless, branching off from one another, then joining others.
Fig. 1 (a) A schematic example of capped single-wall carbon nanotubes with 0.72 nm in diameter and 100-1000 nm in length (From Carbon Technologies Incorporated) and (b) High-resolution scanning electron micrograph of the asreceived single-wall carbon nanotubes dispersed by ethanol alcohol media. Note that each rope has -10 nm in diameter and endless in length. The a-AlzO3 nanopowder with an average particle size of 50 nm obtained from Nanophase Technologies Corporation (Darien, IL 60651) was used in the present study. In order to uniformly disperse SWCN in the alumina matrix, there is a special route to prepare the starting composite powders. Firstly, the asreceived single-wall carbon nanotubes in the "paper" form must be dispersed in to an ethanol alcohol using an ultrasonic probe. Secondly, alumina nanopowder was
Ceramic Nanomaterials and Nanotechnology
163
added into the dispersed SWCN alcohol media at a proper time and then some amounts of alcohol was removed. Finally, the composite powders were sieved using a 200-mesh and then mixed for 24 h in ethanol using zirconia ball media. In the present study, we have investigated two contents of SWCN at 5.7~01%and 1Ovol%. In order to obtain fully dense nanocomposites and retain nanocrystalline alumina, spark plasma sintering technique was employed in the present study. It is a pressure assisted fast sintering method based on high temperature plasma (spark plasma) momentarily generated in the gaps between powder materials by electrical discharge at the beginning of ON-OFF DC pulsing. It has been suggested that the ON-OFF DC pulse energizing method could generate: (3) spark plasma, (2) spark impact pressure, (3) Joule heating, and (4) an electrical field diffision effect. In this process, powders were loaded into a graphite die and were heated by passing an electric current through the assembly. The low heat capacity of the graphite die allows rapid heating and cooling thus promoting heat and mass transfer. Thus, SPS could rapidly consolidate powders to near theoretical density through the action of a rapid heating rate, pressure application, and proposed powder surface cleaning. In the present study, spark plasma sintering was carried out under vacuum in a Dr. Sinter 1050 SPS apparatus (Sumitorno Coal Mining Co., Japan). The powder mixtures were placed into a graphite die (20 mm in inner diameter) and cold-pressed at 200 MPa. The SPS processing parameters used in the present study were as follows: (1) an applied pressure of 63 MPa, (2) the pulse duration time of 12 ms and the interval between pulse of 2 ms, and (3) the pulse current of -5000 A and a voltage of 10 V. After applying the given pressure, samples were heated to 600 OC in 2 minutes and then ramped to 1150 O C for 3 minutes at a heating rate of 500°C/min. The temperature was monitored with an optical pyrometer that was focused on the “non-through” hole (0.5 mm in diameter and 2 mm in depth) of the graphite die. The final densities of the sintered compacts were determined by the Archimedes’ method with deionized water as the immersion medium. The theoretical densities of the specimens were calculated according to the rule of mixtures. Note that the density of graphite (2.25g/cm3) was usually used for SWCN5’*but the measured density for the present SWCN is 1.8 g/cm3 because the density of carbon nanotubes is a function both of their diameter and number of shells. The microstructurai observation was carried out using an FE1 XL30-SFEG high-resolution scanning electron microscopy with a resolution better than 2 nm and magnification over 600kX.Grain sizes were estimated from high-resolution SEM of fractured surfaces, Indentation tests were performed on a Wilson Tukon hardness tester with a diamond Vickers indenter. Bulk specimens were sectioned and mounted in epoxy, then polished though 0 . 2 5 ~diamond paste. The indentation parameters for fracture toughness (KIc) measurements were a 1.5 Kg
164
Ceramic Nanomaterials and Nanotechnology
load with a dwell of 15s. The fracture toughness was calculated by Antis equation." At least ten points were performed for each material.
RESULTS AND DISCUSSION Spark plasma sintering provides an attractive sintering process for nanocrystalline ceramic composites. The sintering temperature and dwell time were minimized to limit grain growth. The processing conditions and physical properties for pure alumina and nanocomposites are shown in Table I. It can be seen that the pure ahmina nanopowder could be consolidated by SPS at 1150 "C for 3 minutes to get full density. Table I Processing conditions and relative density of carbon nanotubes toughening alumina matrix nanocomposites Materials
Processing conditions
Relative density
Reference
(%)
SPSl 15OoC/3min Pure A1203 SPS 1 1 5O0C/3min 5.7~0l%sW CN/Al203 SPSl 150°C/3min 10~0l%SWCN/A1203 HP 1300°C/60min 1Ovol%MWCN/A1203 6.4vol%SWCN-2.5v01%Fe-/AlzO~ HP1475°C/15min HP1475°C/15min 1 1.6vol%SWCN-2.5v01%Fe-/Al~O/o~e-1~1203 4.7vo1%SWCN-5.3vo1%Fe-/Al2O3 HPl475"C/15min 1 7.2vol%SWCN-5.0vol%Fe-/A1203 HP 1475"C/15min 8.5vol%SWCN-4.3vol%Fe-/Al203 HP 15OO0C/15min 1OVol%SWCN-4.3vol%Fe-/A1z03 HP 1 5OO0C/15min Hp:hot-pressing.
100 100 100 100 91 97.5 97.8 99.2 88.7 87.5
Present Present Present 191 181
PI
181
PI
[51
-[51
The microstructure of pure A1203 consisted of equiaxed grains with an average value of 349 nm. Measured Vickers hardness and fi-acture toughness are 20.3 GPa and 3.3 M P a d 2 , respectively. It is very interesting to note that both 5.7vol%SWCN/A1203 and 1Ovol%SWCN/Al203 nanocomposites could also be successllly consolidated to their theoretical densities at the same sintering conditions as that for pure alunina, suggesting that addition of SWCN to the alumina matrix was not detrimental to the sintering. The mean grain sizes for both nanocomposites are around 200 nm. However, it was found that it was very dificult to obtain fully dense nanocomposites with truly nanocrystalline alumina matrix by conventional hot-pressing. For example, Laurent, Peigney, and Rousset et developed a novel catalytic route for the in-situ formation of a huge amount of single-wall and multi-wall carbon nanotubes in alumina matrix composite powder. Depending on the quality of carbon nanotubes, the frnest grain size of alumina matrix is 30Onm but the density of the composites is only 89%TD. Siege1 et a1 used the hot-pressing method to obtain a fully dense multiple-wall
Ceramic Nanomaterials and Nanotechnology
165
carbon nanotubes filled alumina nanocompositesbut the matrix grain size is more than 500 nm.
Fig.2 High-resolution scanning electron micrograph of (a) 5.7vof%SWCN/Al20~ nanocomposite and (b) 1Ovol%SWCN/A1203 nanocomposite consolidated by spark-plasma-sintering at 1 150°C for 3 minutes. Figure 2 shows the microstructures of high-resolution scanning electron micrograph of fractured surfaces for 5.7vol%SWCN/A1203 nanocomposite (Fig.2a) and for 1Ovol%SWCN/Al2O3 nanocomposite (Fig.2b), respectively. There are some very interesting features to be noted. First, it can be seen that carbon nanotubes are in general homogenously dispersed in the matrix for both 5.7vol%SWCNIA1203 and lOvol%SWCN/A1203nanocomposites although some agglomerations were observed when the content of carbon nanotubes was increased to 10% volume. Second, the fracture surfaces were quite textured and rough in both nanocomposites. Finally, no pullout of carbon nanotubes was observed but it was very interesting to note that these carbon nanotubes’ ropes were entangled with the alumina grains’ surfaces to form a network structure. These unique features for the present nanocomposites are quite different fiom InSifu carbon nanotubes reinforced alumina nanocomp~sites~-~ and MWCN filled alumina nanocomposites 9, therefore leading to different mechanical properties. In terms of toughening by carbon nanotubes in alumina nanocomposites, a dramatic improvement in fracture toughness has been achieved in the present nanocomposites (Table 11). Fig.3 shows the relationship between the toughness and carbon nanotubes content in the alumina nanocomposites. It can be seen that fracture toughness significantly increases with the content of single-wall carbon nanotubes. This is more than three times as tough as the pure nanocrystalline
166
Ceramic Nanomaterials and Nanotechnology
alumina in the lOvol%SWCN/Al2O3 nanocomposite. This is quite different from the other work. Laurent et a1 investigated the in-situ carbon nanotubes-Fe-A1203 nanocomposites and found that there was no toughening effect by adding carbon nanotubes into Fe-Al203 nanocomposites although the fracture toughness is similar to that of alumina. Moreover, there was a little increase in fracture toughness by improving the quality of carbon nanotubes in the latter work.' Siegle et a19 reported that 24% increase in toughness could be achieved in ~OVO~%MWCN/A~~O~ nanocomposites, value that is far lower than that seen in our present results. Table 11. Grain size and fi-acture toughness of carbon nanotubes reinforced alumina matrix nanocomposites Materials
Grain size
(nm) Pure Alz03 5.7voi%swcN1A1~o3 10v0l%SWcN/Al~O~ 1Ov01%MWCN/A1203
349 200 200 -500
6.4vol%SWcN-2.5~01%Fe-/A~~03 -500 1 1.6~ol%SWCN-2.5vol%Fe-/Al~O~ -500 -500 4.7vol%SWCN-5.3v01%Fe-/A1203 -500 17.2vol%SWCNJ .0v01%Fe-/A1203 300 8 .5vol%SWCN4.3~0l%Fe-/A1203 300 1 Ovol%SW CN-4.3vol%Fe-/Alz03
Fracture Reference toughness (MPNll'n> Present 3.3 Present 7.6 10.7 Present 4.2 191 4.8 PI 2.8 181 3.6 PI 2.7 PI 5.0 151 3.1
[51
The enhanced toughening effect observed in the present study is likely related to the following factors. First, it is due to extraordinary mechanical property and perfect structure of single-wall carbon nanotubes over multiwalled carbon nanotubes. Single-wall carbon nanotubes are predicted to have extremely high Young's modulus. This has been recently confirmed by experimental investigation. Yu et a14have studied the mechanical response of SWCN ropes and obtained the average mechanical properties of individual SWCN in these ropes. The mean value for the breaking strength and the mean value for the Young's modulus of ropes of single-wall carbon nanotubes are 30 GPa and 1002 GPa, respectively. Second, the toughening effect may be related to the unique network structure that is developed. As mentioned above, ropes of single-wall carbon nanotubes were entangled with alumina matrix and led to mainly intergranular fractured mode. Third, it may be related to the high quality ropes of single-wall carbon nanotubes used in the present study. It was reported that ropes containing about 10 SWCN or less will enable them to reach their optimum properties in terms of Young's modulus and strength due to the perimeter loading of ropes of
Ceramic Nanomaterials and Nanotechnology
167
SWCN! The ropes of SWCN used in the present study include about 10 SWCN and meet the above requirement, resulting in effective toughening effect. Thus, our present preliminary work gives a promising future for application of carbon nanotubes in reinforcing structural ceramic composites. Ongoing investigation to optimize the microstructure and to understand the toughening mechanisms is currently underway. For instance, it is necessary to optimize the mechanical properties by improving the preparation process for composite powder with more homogenously dispersion and with much higher volume fraction of SWCN. The interface between the ropes and the alumina matrix also needs to be investigated by high-resolution transmission electron microscopy.
-0- SWCN/A1203 [Prcsent study] MWCN fllled AD03 [SIegel et at] In-sltu SWCN-FeAu03 IWgney et all ’V-lnsftu SWCN-FeA1203 lFlPhnnt et PI1
2.
.
n
-0
I
I
I
I
I
I
I
I
I
2
4
6
8
10
12
14
16
18
20
Carbon nanotubes volume content (%)
Fig.3 Relationship between fracture toughness and carbon nanotubes volume content in carbon nanotubes toughening alumina matrix nanocomposites
Summary SWCN/A1203 nanocomposites up to 10~01%with nanocrystalline alumina matrix and superior mechanical properties have been fabricated at 1150°C in three minutes by spark-plasma-sintering. Fracture toughness of more than 10 MPam’” has been achieved in 1Ovol%SWCN/Al203 nanocomposite. This preliminary study demonstrated that ropes of single-wall carbon nanotubes are attractive materials for reinforcement of nanoceramics.
168
Ceramic Nanomaterials and Nanotechnology
Acknowledgements: This investigation was supported by a grant (#G-DAAD 19-00-1-0185) from US Army Research Office with Dr. William Mullins as the Program Manager.
References 1. M. J. Mayo, Superplasticity of Nanostructured Ceramics, in Mechanical Properties and Deformation Behavior of Materials Having Ultra-fine Microstructures, Eds.by M. Nastasi, D. M. Parkin, and H. Gleiter, Kluwer Academic Publishers, 1993, pp.361-380. 2. M. J. Mayo, “Nanocrystalline Ceramics for Structural Applications: Processing and Properties,” pp361-385 in Nanostructured materials, Ed., by G . M. Chow and N. I. Noskova, Kluwer Academic Publishers, The Netherlands, 1998. 3. G.-D. Zhan, J. Kuntz, and A IS. Mubegee, Interim report to US army research office, Durham, NC (April, 2002). 4. M.-F. Yu, B, S. Files, S. Arepalli, and R. S. Ruoff, “Tensile Loading of Ropes of Single Wall Carbon Nanotubes and Their Mechanical Properties,” Phys. Rev. Lett. 84,5552 (2000). 5. E. Flahaut, A. Peigney, C. Laurent, C. Matliere, E. Chastel, and A. Rousset, “Carbon Nanotubes-Metal-Oxide Nanocomposites: Microstructure, Electrical Conductivity, and Mechanical Properties,” Acta Mater., 48,3803-3812 (2000). 6. C. Laurent, A. Peigney, 0. Dumortier and A. Rousset, “Carbon NanotubesFe-Alumina Nanocomposites. Part 11: Microstructure and Mechanical Properties of the Hot-Pressed Composites,” J. Euro. Ceram. Soc., 18,2005-2013 (1998). 7. A. Peigney, C. Laurent, 0. Dumortier and A. Rousset, “Carbon NanotubesFe-Alumina Nanocomposites. Part I: Influence of the Fe Content on the Synthesis of Powders,” J. Euro. Ceram. Soc., 18, 1995-2004(1998). 8. A. Peigney, C. Laurent, E. Flahaut, and A. Rousset, “Carbon Nanotubes in Novel Ceramic Matrix Nanocomposites,” Ceram. Inter., 26,677-683 (2000). 9. R. W. Siegel, S. K. Chang, B.J. Ash, J. Stone, P. M. Ajayan, R W. Doremus, and L. S. Schadler, “Mechanical Behavior of Polymer and Ceramic Matrix Nanocomposites,” Scripta Mater., 44,2061-64 (2001). 10. G. R. Antis, P. Chantikul, B. R Lawn and D. B. Marshall, “A Critical Evaluation of Indentation Techniques for Measuring Fracture Toughness: I, Direct Crack Measurement,” J. Am. Ceram. Soc., 64 [9] 533-38(1981).
Ceramic Nanomaterials and Nanotechnology
169
This page intentionally left blank
PROCESSING AND MICROSTRUCTURE OF COMPOSITE FROM PLASMA-SPRAYED Al2Ti05
Al2O3/TiO2
NANO-
Julin Wan, Guo-Dong Zhan and Amiya K. Mukherjee Department of Chemical Engineering and Materials Science University of California Davis, CA 95616 Bernard H. Kear Department of Ceramics and Materials Engineering Rutgers-The state University of New Jersey Piscataway, NJ 08854 ABSTRACT Processing of A1203/TiO2 nanocomposite was investigated by high pressure sintering of an aluminum titanate powder prepared by plasma-spray Materials with varied density and grain-size/phase composition were produced with an applied pressure of lGPa, within 750-1000°C. . The development of the dual phase structure of Al203Ei02 in this approach depends on the decomposition of the metastable aluminum titanate, which is accelerated by the high sintering pressure, plays a central role in grain-size control. This study also highlights the necessity of a high-energy ball-milling of the powder, which serves to enhance sintering and promote decomposition reaction of the titanate. Nano-nano composite with grain size of around 70 nm can be fabricated by this technique. INTRODUCTION Nanocrystalline ceramics have been drawing attentions due to their novel and often favorable properties as compared with their microcrystalline brethrens. The conventional way of making nanocrystalline ceramics has been consolidating nano-scale powders. Due to the high driving force rooted in the extraordinarily large grain boundary area, grain-growth becomes explosively fast when the material approaches full density.' Therefore, although the technology of making nano-powders in some systems has been relatively ripe for many years, the effort in producing bulk nano-ceramics has not been very fruitful.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
171
Transformation assisted consolidation (TAC) is an emerging technology to make nanocrystalline materials. This approach includes two steps: (1) Forming a metastable micro-sized ceramic powder, by the means of plasma spraying; (2) High pressure sintering of the metastable powder to form nanocrystalline ceramics. This novel technology offers a route to overcome two vital barriers for conventional techniques: the grain size of the sintered materials is not limited by the crystallite size of the starting powder; the explosive grain growth during sintering with nano-sized powders can be avoided, due to the concurrent transformation. This technique does not require a nano-sized starting powder to make nanocrystalline bulk ceramics, therefore can be applied to material systems that nano-powders can not be processed with convenience, and also in systems that grain growth is difficult to control, like alumina based ceramics. TAC processing has been successfully applied to fabricate nano-A1203 and nano-Ti02. Single-phase materials with density >99%TD and grain size as small as 18 nrn have been produced. In this paper, the TAC technique is applied to the Al2O3/TiO2 system, in an effort to make alumindtitania nano-nano composites with enhanced superplasticity . EXPERIMENTAL
Fig. 1 Schematics of the furnace assembly in the high pressure sintering apparatus Commercially available mixtures of A1203 and Ti02 with 60/40 weight percentage (Metco 131) were used for the preparation of the metastable powder. &03/40Ti02 is a eutectic composition of alumina and titania, it can readily be melted and homogenized by fast melting processed like plasma spraying. The
172
Ceramic Nanomaterials and Nanotechnology
original powder was coarse-grained, with 45ym size for alumina and 5ym for titania. The powder mixture was fed into a plasma torch, melted and then rapidly solidified by quenching into water. The powder thus formed was subjected to high energy ball-milling (HEBM) with Spex 8000 mixerhill, using tungsten carbide balls and vial. The milling time is controlled at 24 hours. The ball-milled powder was then uniaxially pressed into green compacts with 6mm diameter and 6mm height. Sintering was conducted with a Boyd-England apparatus. A schematic description of the furnace set-up of the high pressure sintering is shown in Fig. 1. The furnace assembly is comprised of a 32mm long and 6.0mm inner diameter graphite furnace, surrounded by a sheath of CaF2 and Pb foil, and placed into the center of the WC-CO core of the pressure vessel. The CaF2 cell crumbles under pressure and acts as a pressure-transmitting medium. Once crushed it transforms uniaxial pressure into quasi-hydrostatic pressure. The Pb foil acts as a lubricant for the furnace assembly during the push-out after the test. The remaining volume of the furnace is filled by semi-sintered A203 filler rod. A D type thermocouple (W/Re) was placed at the top of the sample and temperature was regulated using a Eurotherm controller. Sintering was performed by applying a pressure of lGPa on the sample, which is provided by two separate hydraulic pumps fiom both ends of the assembly. AC electric current passes through the graphite furnace, providing a heating rate of 150°C/min to the setpoint. Once at temperature, samples were held for 1 hour before the pressure was reduced to -0.3GPa. The temperature and remaining pressure were then allowed to ramp down simultaneously. Samples were analyzed for density using the Archimedes method. Phase characterization was carried out using X-ray diffraction. Microstructural observation was conducted with a high-resolution field emission scanning electron microscope. RESULTS Phase-transformationin powders The as-sprayed and water quenched powder is composed of coarse particles (220 pm), mostly spherical in shape (Fig. 2(a)). High-energy ball-milling breaks the particles into finer irregularly shaped particles of about 200-600 nm in size, but tightly agglomerated to the size roughly identical to that of the starting powder (Fig. 2(b)). Heat-treatment was conducted to both the as-received powder and that milled for 36 hours, to study the effect of high-energy ball-milling on the phase transformation of the material. Each powder was hold at a temperature between 700-1000°C for one hour. X-ray diffraction analysis results are shown in Fig. 3 and Fig. 4.The as-received powder, as Fig. 3 shows, is a complex mixture of at least four phases. The major phase is orthorhombic p-Al2Ti05, which is the result
Ceramic Nanomaterials and Nanotechnology
173
Fig. 2 Morphology of the plasma-spayed aluminum titanate powder (a) As-received (b) HEBMed for 24 hours
of reaction between alumina and titania while being melted in the plasma flame. This phase is relatively coarse grained, as revealed by the sharp peaks in the Xray spectra. Small amount of a-Al2TiO5, which is a high temperature form of aluminum titanate (readily changes to p form on slowing cooling to 1820°C3), was found to exist as a fine-grained phase. This phase can exist in the plasmasprayed powder, owing to the rapid cooling of water-quenching. a-Al2O3 (corundum) and Ti02 (rutile) are the two other minor phases in this powder, they are either the residue of incomplete reaction during melting, or the decomposition product during cooling. Upon heat-treating, the decomposition reaction proceeds rather sluggishly in the as-sprayed powder. Virtually no change in phase proportions is observed up to 800°C. Apparent increase in amount of alumina and rutile starts at about 900°C. However, the majority of aluminum titanate is still retained in the lOOO"C/lhr treated sample. It is also noticed that the high temperature a-Al2TiO5 does not have an obvious conversion into P-Al2TiO5, the stability of this former phase might be related to its small grain size which adds an energetic barrier to the transformation. High energy ball-milling brings about several changes to the XRD pattern of powder, Fig 4. HEBM leads to significant broadening of diffraction peaks, especially that of P-Al2TiO5. This broadening is partly due to the smaller particle size as shown in Fig. 2(b). The high level of internal stress also surely has its contribution. The amount of a-Al2TiO5 seems to be increasing with milling time, as can be discerned from the emergence of the peak at 28=31.5". The peak at 28=35.8" is thought to be an overlapping peak of A1203 <104> and Ti02 < l o b , rise of this peak with HEBM indicates an enhanced decomposition of aluminum
174
Ceramic Nanomaterials and Nanotechnology
0
20
P-AI2TiO5
30
0
a-A12Ti05 A AI2O3
40
50
60
TiOp
70
W O )
Fig.3 XRD of the as-sprayed powder and its decomposition products with heat-treatment titanate at room temperature. These two new peaks induced by HEBM (20=3 1.5", 35.8"), however, disappear with a heat-treatment at 700°C. The most signifcant effect HEBM has on the decomposition of the aluminum titanate is dropping of the decomposition temperature. First noticeable decomposition happens at around 800°C. At 900°C the phase proportion of decomposition products (alumina and titania) is roughly similar to that of the un-milled powder heat-treated at 1000°C. Therefore the decomposition is accelerated by HEBM, resulting in a lowering of decomposition temperature for about 100°C. Another feature HEBM brings about is, Al2Ti7015 phase which does not exist in any of the un-HEBMed samples emerges upon heat-treating. This phase is probably an intermediate phase in the process of transformation from Al2TiO5 toward A1203 and Ti02. If this is true, the existence of Al2Ti7015 might indicate that, HEBM not only changes the kinetics of the aluminum titanate decomposition, but also changes the route by which the decomposition happens. High-pressure sintered microstructures It was found that HEBM greatly enhanced the sinterability of the plasmasprayed aluminum titanate. Under an applied pressure of lGPa, and with a fixed sintering time of lhr, the as-spayed powder can only be sintered to a density of
Ceramic Nanomaterials and Nanotechnology
175
o P-AI2TiO5 0 a-AIPTi05 A AI2O3
I
T
+
AI2Ti7Ol5 Ti02 1OOO°C/l hr
8o0°C/l hr
70O0C/l hr
HEBM 36hr
20
30
40
50
60
70
Fig. 4 XRD of HEBMed (36hr) powder and its decomposition product with heat-treatment 3.97g/cm3 (96.8% theoretical density) at 1000°C, while the HEBMed powder can be sintered to 4.1g/cm3 (-100%TD) at a temperature as low as 850°C. Representative XRD results of the high pressure sintered materials, for both the as-sprayed and HEBMed powders, are shown in Fig. 5. The as-sprayed powder, when subjected to high pressure sintering, were sintered to only about 80%TD at 800°C, and the phase-composition remained similar as the starting powder, no obvious decomposition occurs. When the sintering temperature is raised to 1000°C, the decomposition is complete, the material is composed of only A1203 and Ti02, no aluminum titanate residue was observed. Comparing with the 1000°C heat-treated sample in Fig. 3, it is quite obvious that the high appliedpressure promotes the decomposition reaction. For the HEBMed powder, the aluminum-titanate dominated structure of the powder can be stable up to 750°C. At 800°C, advanced decomposition has already taken place within one hour. The structure is dominated by A1203 and Ti02, but some of P-Al2TiO5 is still retained. The decomposition is complete at 85OoC, and from this temperature on the materials are composed only of A1203 and Ti02. It is also noted that the intermediate Al2Ti7015 phase which is present during the heat-treatment of the HEBMed powders, does not appear in the high pressure sintered samples.
176
Ceramic Nanomaterials and Nanotechnology
o P-AI2TiO5
I
P
0 a-A12Ti05
A A1203 U TiOp
HEMBed
As-spraved U
20
30
40
50
60
W O >
Fig. 5 XRD of the materials produced by high-pressure sintering at different temperatures, from the as-sprayed and HEBMed powders The microstructure of the sintered materials is shown in the examples of Fig. 6. In these micrographs, the lighter phase is A1203 and the darker phase Ti02. In the sample sintered from as-sprayed powder, although there are only two phases (alumina and titania), the grain morphology demonstrates a complex mixture of coarse and fine grains. A large quantity of the Ti02 phase resides in the microstructure as large grains with a size up to 5pm. Large grains of A1203 are less frequently seen, the size of these grains can range from 500 nm to lpm. The rest of the material is a mixture of fine-grained alumina and titania, the grain size in these regions ranges from 50-100 nm, with the phase proportion of alumina apparently higher than that of the overall proportion of alumina in the material. The HEBMed powder, however, shows a uniform dual-phase structure of nanocrystalline alumina and titania (Fig. 6(b)). The mean grain size in the 850°C sintered sample is about 72nm. Obviously, the high energy ball-milling not only enhanced the kinetics of the alumina-titanate decomposition, it also greatly promote the formation of nano-nano composite structure of A1203 and Ti02.
Ceramic Nanomaterials and Nanotechnology
177
Fig. 6 Microstructure of high-pressure sintered composites of Al203/Ti02 (a) From as-sprayed powder, 1000°C/1GPa/30min (b) From HEBMed powder, 850°C/lGPa/lhr DISCUSSION The major purpose of this research, as stated earlier, is to make Al203/Ti02 nano-nano composites with a prospect of enhanced superplasticity over the currently available alumina bases ceramics. Another as important purpose is to search out a way to make nanocomposites not from nano-scale powders, but from commercially available and inexpensive coarse powders. The results presented above indicate the feasibility of achieving these goals. The principle of this approach is to form a metastable state, in this particular case of Al203/40Ti02, this metastable state is a Al2TiO5 -dominated structure. Al2TiO5 can undergo a eutectoid-like decomposition:
which is an exothermic reaction with a negative entropy change.495Therefore alumina titanate is stable at higher temperatures while it tends to decompose at temperatures below about 1300°C. At below 7OO0C, although aluminum titanate is thermally unstable, the slow kinetics due to low temperature will usually prevent the decomposition from happening. As Al2TiO5 is a material known for its low thermal expansion coefficient and can be used as a thermal shock resistant material, the attention to this material has been on how to enhance its stability at lower temperatures.6However, our intention here is to utilize this decomposition, and to search for acceleration of this reaction to facilitate the condition that the decomposition being completed at as low a temperature as possible.
178
Ceramic Nanomaterials and Nanotechnology
The grain size of the material derived by the phase-transformation from a parent phase depends on the nucleation density as well as the rate of grain growth. Furthermore, a uniform dual phase nano-structure also required total consumption of the parent phase. Owing to a positive change in specific gravity upon decomposition (the density for P-Al2TiO5, a-Al2O3 and rutile TiO2 is 3.75, 3.98 and 4.3g/cm3, respectively), the nucleation density is increased by the high applied pressure, which provide a positive driving force for nucleation of the denser phases thus decreases critical size for nuclei f~rmation.~ Decomposition kinetics could be controlled by the diffusivityeither in the parent phase, or across the phase boundaries between the parent phase and product phases. Diffusivity can be greatly enhanced by high level of defect such as linear defects (dislocations) and point defects (vacancies). The plasma-spray is supposed to be able to provide “frozen-in” defects from the liquid state, in addition to serving the purpose of forming the metastable aluminum titanate. It is now clear that at least for the high pressure sintering condition used in this study (lGPa, 700-lOOO”C), plasma spraying alone is not enough to provide fast decomposition kinetics for the formation of nano-nano composite. In addition the residue alumina and titania phases in the as-sprayed material, which tend to be rather coarse, might serve as inhomogeneous nucleation sites for the similar phases, leading to “clustering” of the same phase. The accumulation of similar phases decrease the area of interphase boundaries, thus diminishes the mutual grain-growth restriction between phases and lead to large grain size. Therefore, the current results highlight the necessity of high-energy ballmilling. High energy ball milling is known to be able to bring about a number of effects on oxides which greatly enhances solid-state phase transf~rmation.~’~ The effect of HEBM on the microstructural formation in the Al2O3/TiO2 nanocomposite can be postulated as follows: (1) Breaking-up of the coarse Al2TiO5 particles, which leads to higher green density of the compacts before sintering; (2) Introducing of defects to the aluminurn titanate lattice, which leads to enhancement of decomposition by accelerated mass-transport; (3) Dispersing of the residual (or primary) alumina or titania, offering more dispersed nucleation site for inhomogeneous nucleation; (4) The high energetic level and highly localized internal energy distribution of the HEBMed powder might also provide a higher density of nuclei for homogenous nucleation. All these effects, and in combination with the effect fiom high applied pressure, leads to the optimized condition for forming of homogenous nano-nano composites of alumina and titania. It should be acknowledged that this is a preliminary effort to make kd203/40Ti02 nanocomposite by this technique. A number of vital questions, e.g., the detailed structure of the plasma-sprayed powder and its evolution during HEBM and sintering, remain to be answered. The understanding to the some of
Ceramic Nanomaterials and Nanotechnology
179
the observed phenomena, e.g., the role of individual effects brought about by HEBM, remains speculative and needs to be explored in further depth. Furthermore, the effort of characterizing the property (mainly superplasticity) of the nanocomposite make by this process, is still underway. SUMMARY Nano-nano composites of alumina and titania can be produced fi-om coarse starting powders, by a transformation assisted consolidation process that includes plasma spray, high energy ball-milling, followed by high pressure sintering. The present work on a eutectic composition revealed that the decomposition of metastable aluminum titanate, which is usually sluggish at lower temperatures, can be accelerated by introducing a high level of defects by high energy ballmilling. This effect, along with the enhanced consolidation and nucleation provided by high sintering pressure, are key factors in the formation of nanocrystalline microstructure in this material. ACKNOWLEDGMENT This work is sponsored by Office of Naval Research (Grant No.NO0014-01-C0370). REFERENCES H. Hahn, “Microstructure and properties of nanostructured oxides,” Nanostruct. Mater., 2[3] 251-65 (1993). B. H. Kear, J. Golaizzi, W. E. Mayo, S.-C. Liao, “On the Processing of Nanocrystalline and Nanocomposite Ceramics,” Scripta mater., 44 2065-68 (200 1). PDF card No. 18-0068, 1999JCPDS-InternationalCentre for Diffraction Data, PCPDJWIN v. 2.02. 4 E. Kato, K.Daimon and I. Takahashi, “Decomposition Kinetics of Al2TiO5 in Powdered State,” J. Am. Ceram. Soc., 63 355-6 (1980). W.R. Buessem, N. R. Thielke and R.V. Sarakauskas, “Thermal Expansion Hysteresis of Aluminum titanate,” Ceram. Age, 60[5] 38-40 (1952). I. J. Kim and H. S. Kwak, “Thermal Shock Resistance and Thermal Expansion Behavior with Composition and Microstructure of Al2TiO5 Ceramics,” CanadianMetall. Quart., 39[4] 387-396 (2000). D.R. Uhlman, J.F. Hays and D. Turnbull, Phys.Chem.Glasses, 7 159 (1966). D. Michel, L. Mazerolles, S. Begin-Colin, “Mechanical alloying of oxides,” Annales de Chimie, 22[6], 403-16 (1997) 9M. L. Panchula and J. Y. Ying, “Mechanical Synthesis of Nanocrystalline aA1203 Seeds for Enhanced Transformation Kinetics,” Nanostruct. Mater., 9 1614 (1997).
180
Ceramic Nanomaterials and Nanotechnology
MICROSTRUCTURAL CHARACTERIZATION OF SILICON NITRIDELBORON NITRIDE NANOCOMPOSITES
T. Kusunose, H. Kondo, Y. Yamamoto, M. Wada, T. Adachi, T. Sekino, T. Nakayama, and K. Niihara The Institute of Scientific and Industrial Research, Osaka University, 8-1 Mihogaoka, Ibaraki, Osaka 567-0047, Japan
ABSTRACT The Si3N4/13Nnanocomposites were fabricated by hot-pressing the a-Si3N4 powders covered partly with turbostratic-BN(t-BN), which were prepared by a unique chemical process. The microstructure and preferred orientation of Si3N4/BNnanocomposites and microcomposites were investigated. The effect of nano-sized h-BN particles on the microstructure evolution of Si3N4 ceramics was discussed.
INTRODUCTION Si3N4ceramics have excellent properties i. e., high strength, relatively high toughness and high wear resistance. The expansion of application of Si3N4 ceramics has been especially expected in the field of engineering ceramics. However, such properties as corrosion resistance to molten metal, thermal shock resistance and machinability are not sufficient for their practical uses to various machine parts. In order to improve these disadvantageous properties of monolithic Si3N4 ceramics, several attempts have been made to disperse hexagonal BN (h-BN) particles in the Si3N4matrix as a second phase[ 1-31. The h-BN possesses a
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication,reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
181
number of interesting properties such as high thermal conductivity, low coefficient of thermal expansion and chemical inertness. These properties have led to its use in a variety of specialized high-temperature areas including crucibles for molten metals, thermocouple protection tubes, and break rings for horizontal continuous casting of steels. Recently, many scientists and engineers have investigated nanocomposites in which the nano-sized particles are dispersed within the matrix grains andor at the grain boundaries [4-121. They revealed that the nanocomposites had finer and more homogeneous microstructure due to inhibition of abnormal gain growth by nano-sized dispersoids, compared to the monolith and the microcomposites. Also, in Si3N4-BN system, it is well known that both of BN and Si3N4grains easiIy orient during hot-pressing, because these materials have crystallographic anisotropy. Therefore, the effect of nano-sized BN on grain growth and anisotropy with respect to the microstructure is of great interest. The purposes of this study are to estimate the preferred orientation of Si3N4 grains in the Si3N4/BN nanocomposites and the microcomposites, and to investigate the influence of nano-sized BN on the microstructure.
EXPERIMENTAL 1.Sample Preparation To fabricate of Si3N.&N nanocomposites, the process presented in previous paper[l3] was adopted. Fig. 1 shows the conceptual schematics of a-Si3Nd/BN-precursor composite powders and Si3N4/BN nanocomposites. The starting powders for the nanocomposites were the commercially available a-Si,N4 (SNE-10 grade, Ube Industries, Japan) powder having an average particle size of 0.2 pm and boric acid and urea as the BN precursors. In this process, a-Si3N4powder, boric acid and urea were mixed by the conventional wet ball milling method. The dried mixtures were reduced at 300°C and 1100°C in H2, and then heated at 1500°C in N2 to produce the Si3N4-t-BN composite powder. The composite and 6 wt% Y203 powders were ball milled with the sintering aid of 2 wt% A1203 again. After second step of ball milling, the slurry was dried, and then the powders
182
Ceramic Nanomaterials and Nanotechnology
were dry ball milled. For comparison, the commercially available a-Si3N4 (SNE-10 grade, Ube Industries, Japan) and BN (GP grade, D E W KAGAKU KOGYO K.K.,Japan) powder with an average grain size of 9 pm was also used to fabricate Si3N4/BN microcomposites. The dried mixtures were hot pressed at 1750°C in nitrogen
Fig. 1. Fabrication image for BN dispersed nanocomposite.
atmosphere under 30 MPa of applied pressure. 2. Characterization The grain size of Si3N4was measured by the disintegration method. At least 2000 grains were measured for each sample. As p-Si3N4 grains have almost cylindrical shape, the diameter of each grain was directly determined from the shortest grain diagonal. In the present work, we proposed the new method to evaluate the degree of the preferred individual orientation of p-Si,N4 grains in the hot-pressed body. In this analysis, the three dimentional angle of long axis (c-axis) of elongated p-Si3N4 grain fkom perpendicular to the hot-pressing direction was Fig.2. Schematic diagram Of the intersected Si3N4 grain on the etched surfkce. The angle of the Si3N4 measured as a parameter Of the grain fiom perpendicular to the hot-pressing direction preferred orientation, as shown is Eq. 1 : e = sin-’ I- r,’/rm2sina .
U--
Ceramic Nanomaterials and Nanotechnology
1
183
in Fig. 2. The Si3N4grain cut on the planar section is seen as elliptical shape, but its shape is exactly hexagon. The size of major (r,) and short radius (rs)arid two-dimensional angle (a)of major radius fi-om perpendicular to the hot-pressing direction were deduced fiom SEM photographs of chemically etched surface parallel to the hot-pressing direction by using image analysis technique. The three-dimensional angle (9 of the Si3N4 grains from perpendicular to the hot-pressing direction was calculated from Eq. 1:0 = sin-'( d-sin
a) in
Fig. 2, assuming a,perfectly cyrindrical shape and high aspect ratio of Si3N4
grain[ 131. The averages of aspect ratio for all materials were more than 5.2, and these values were thought to be enough for this evaluation. At least 2500 grains were measured for each sample.
RESULTS AND DISCUSSION Fig. 3 shows the chemically etched surfaces perpendicular to the hot-pressing direction for the Si3Ndl5vol%BN microcomposite and the
Fig.3. SEM micrographs of polished and etched surfaces perpendicularto the hotpressing direction for SisNdlSvol%BNmicrocomposite (a) and nanocomposite @).
Fig. 4. SEM micrographs of polished and etched section parallel to the hobpressing direction for Si~NdlSvol%BNmicrocomposite (a) and the nanocomposite (b).
184
Ceramic Nanomaterials and Nanotechnology
nanocomposite. The microstructure of the microcomposite exhibited the confined large BN grains dispersed at grain boundaries of elongated p-Si3N4 grains. On the other hand, the nanocomposite showed the finer microstructure with small equiaxed BN particles and rod-like Si3N4 m a h p i n s . It is thought that the Fig. 5. Elongation of p-Si3N4 grains along h-BN plates in Si3Nd20vol%BNmicrocomposite. finer BN particle significantly inhibited grain growth of Si3N4matrix. Fig. 4 shows the SEM micrographs of polished and etched section parallel to the hot-pressing direction for the Si3N4/15vol%BN microcomposite and the nanocomposite. The larger elongated P-Si3N4grains particularly lay perpendicular to the hot-press direction in the section parallel to the hot-press direction. Furthermore, in the microcomposite large h-BN grains were also orientated preferentially. It is well-known that the c-axis of hexagonal p-Si3N4crystal with the rod-like shape preferentially oqented perpendicular to the hot-pressing direction[14-17]. And also, the basal plane of h-BN has a preferred orientation as the same direction because of the morphology of its crystallographic anisotropy[181. Since the grains of p-Si3N4and h-BN are rod-like and plate-like, respectively,both materials easily orient preferentially during hot-pressing. Goto et al. [19] used the ratio (I101A210)of X-ray intensity of (101) and (210) planes as a parameter of preferred orientation for p-Si3N4.This analysis can give the information on the entire bulk, but the information of each grain cannot be obtained. As seen from the two-dimensional micrographs in Fig. 5, the larger elongated P-Si3N4grains oriented preferentially, however, the relatively equiaxed and small grains are in random. To further understand the microstructure of Si3N4/BN composites, therefore, the information on correlation between individual grain shape and three-dimensional inclination is necessary. Fig. 6 shows the preferred orientation relating to the diameter of individual grain in the
Ceramic Nanomaterials and Nanotechnology
185
section parallel to the hot-pressing direction. The comparison of diameter is suitable for this evaluation, because this parameter can well express the grain growth and the development of microstructure by sintering. The average angle of the Si3Nd/20vol%BNnanocomposite and the microcomposite were 23.4"and 20", whereas that of largest 5 % of grains in diameter were 20.6' and 18.1', respectively. From this result, it was found that the grains with larger diameter in all materials oriented preferentially perpendicular to the hot-press direction. Especially in the microcomposite, not only large grains but also small grains oriented well, and this tendency would be more apparent at higher BN content. These differences of the preferred orientation between nanocomposites and microcomposites may be attributed to the size of h-BN as a second phase. In the hot-press sintering, h-BN plates firstly lie perpendicular to the hot-pressing direction, and fine equiaxed a-Si3N4grains grow into rod-like P-Si3N4 grains among h-BN plates. As h-BN content increased, the space between h-BN plates will be narrower. Consequently, the microcomposites containing a larger volume of h-BN indicated higher degree of the preferred orientation than the nanocomposites, since P-Si3N4 grains elongated along h-BN plates as seen in Fig. 5. In contrast, in the nanocomposites, the preferred orientation of the larger grains was suppressed because of fine and relatively equiaxed h-BN grain as a dispersoid.
80 n
&,
60
a3
3 Q)
40
P 4 20
0 0 0.4 0.8 1.2 1.6 2.0 2.4 Diameter (pm) Diameter (pm) Fig. 6. Three dimentional angle of elongated SisN4 grain fiom perpendicular to the hot-pressing direction as a function of grain diameter in the section parallel to the hot-pressing direction ((a) Si3Nd2Ovol%BN microcomposite, and (b) Si3N4/20vol%BNnanocomposite.). 0 0.4 0.8 1.2 1.6 2.0 2.4
186
Ceramic Nanomaterials and Nanotechnology
Thus, the observed preferred orientation will influence the anisotropy of mechanical properties.
CONCLUSION The Si3N4/BN nanocomposites, including the nano-sized BN up to 20
vol. %, were attained by novel chemical route with hot-pressing the a-Si3N4 powders covered partly with t-BN. As compared with the monolithic Si3N4 and the microcomposites, the nanocomposites showed homogeneous microstructure and low preferred orientation due to inhibition of grain growth by nano-sized BN particles. ACKNOWLEDGEMENT This study was supported by Industrial Technology Research Grant Program in '01 from New Energy and Industrial Technology Development Organization (NEDO) of Japan and Hosokawa Powder Technology Foundation.
REFERENCE 1. M. Iwasa S. Kakiuch, Yogyo-Kyokai-shi, 93,661-665 (1985). 2. K. S. Mazdiyansni and R. Ruh, J. Amer. Ceram Soc., 64,4 15-419( 1981). 3. K. Niihara, L. D. Bentsen and D. P. H. Hasselman, J. Am. Ceram. Soc., 64, c l 17-118(1981). 4. K. Niihara, J. Ceram. Soc. Jpn., 99 [ 101 974-82 (1991). 5. T. Ohji, T. Hirano, A. Nakahira, and K. Niihara, . IAm. Ceram. Soc., 79 [ 11 33-45 (1996). 6. A. Nakahira and K. Niihara, pp. 165-78 in Fracture Mechanics of Ceramics, Vol. 9. Edited by M.Sasaki, R. C. Bradt, D.P. H. Hasselman, and D. Mum. Plenum, New Tork, 1992. 7. T. Ohji, Y. -K.Jeong, Y. -H. Choa and K. Niihara, J. Am. Ceram. Soc., 77 [6] 1453-60 (1998) 8. J. Zhao, L. C.Steams, M. P. Harmer, H. M. Chan,G. A. Miller, and R. E. Cook, . IAm. Ceram. Soc., 76 [2 ] 503-10 (1993).
Ceramic Nanomaterials and Nanotechnology
187
9. I. A. Chou, H. M. Chan, and M. P. Harmer, 79 [9] 2403-9 (1996). 10. J.Dusm, P.Sajgalik, and M.Steen, J. Am. Chem. Soc.,82[12] 3613-20 (1999). 11.G. Pezzotti and M.Sakai, J. Am. Chem. Soc., 76 [5] 1313-20 (1993). 12. A.. M. Thompson, H. M. Chan, M. P.Harmer, and R. F. Cook, J. Am. Chem. SOC., 78 [3] 567-71 (1995). 13. T. Kusunose, T. Sekino, Y. -H. Choa, and K. Niihara, “Fabrication and Microstructure of SiIicon Nitridemron nitride Nanocomposites,” J. Ant. Chem. Soc., in print. 14.R.Kossowsky, J. Mater. Sci., 81,603-15(1973) 15. F. F. Lange, J. Am. Ceram. Soc., 56 [lO] 518-22 (1973). 16. K. Nuttall and D. P. Thompson, J. Mater. Sci., 9,850-53 (1974). 17. J. E. Weston, J. Mater. Sci., 15, 1568-76 (1980). 18. F.Lee and K.J. Bowman, J. Amer. Cerarn. Soc., 75 [7] 1669-72 (1992). 19. Y.Goto and A. Tsuge, J. Euro. Ceram. Soc., 6,269-272 (1990).
188
Ceramic Nanomaterials and Nanotechnology
CHARACTERIZATIONOF FIBRE COATINGS AND GLASS COMPOSITE INTERFACES BY ATOMIC FORCE MTCROSCOPY WM) P. Fehling, Th.Mache, and D.Hulsenberi Technische Universitiit Ilmenau Fachgebiet Glas-und Keramiktechnologie Gustav-Kirchhoff-Strasse 6 D-98693Ilmenau Germany
ABSTRACT Nanocrystalline oxidic fibres of Nextel 440m-type (3M Saint Paul, USA), meant for glass reinforcement, were coated with boron nitride or pyrolytic carbon by a thermal chemical vapour deposition (CVD) process. To study the thermal stability of the coating the surfaces of desized and boron nitride coated fibres were investigated by atomic force microscopy (AFM) before and after heat treatment over 5 hours. Thermal annealing of the fibres at 1000°C simulates the worst case of heat influence on the coatings and fibres which in practice is given by long time heat treatment at lower temperatures. For studying the interfacial properties in composites the uncoated, BN- or Ccoated fibres were put into a borosilicate glass matrix by a common glass powder slurry infiltration, followed by vacuum hotpressing. AFM investigations of the crosspolished samples were used to study the interface, the coating thickness, and the cross sectional area of the fibres. The results are useh1 for characterising the fibres and fibre coatings as well as the fibre matrix interface region as a function of the manufacturing and thermal annealing conditions to evaluate the efficiency for reinforcement. INTRODUCTION The aim of a current research project is the development of transparent glass matrix composites, that means the reinforcement of glass with fibres to prevent the brittle fiacture behavior, possibly without influencing the transparency of visible wavelength [l]. Therefore some stages have to be set relating to the properties of the components: The thermal expansion of the fibre should be in the range of the thermal expansion of the matrix, similarity should be given for the ratio of the Corresponding author: Prof. D.Hulsenberg, e-mail: [email protected] To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
189
refractive indices. The Young’s modulus of the fibre should be greater than that of the glass and, furthermore, the softening temperature of the fibre should be higher than that of the matrix [2,3]. An important role in realizing reinforcement mechanisms plays the interface between fibres and matrix. This interface should guarantee the effectivity of the reinforcement mechanisms as follows: A load on the composite leads primary to matrix cracking. The embedded fibres prevent the total failure of the material. If the load is increased the interface between fibre and matrix realizes debonding of the fibre fiom the matrix and the fibre is step-by-step pulled out the matrix. So a “pseudoductile’’ behavior is achieved. The interface design is done by coating the fibres with suitable materials. The structure of the materials is responsible for the sliding effect in the composite, e.g. a plain or amorpheous structure is required. Furthermore the interface may not interact with the visible wavelength. Such interfaces are realized by chemical vapour deposition on the fibre surface. In this case a boron nitride coating was choosen. The coated fibres were characterized by atomic force microscopy (AFM) before and after heat treatment by determine the surface roughness. In comparison with other high resolution methods for surface characterization AFM is simple to do, applicable on various substrates under similar conditions and there is barely any effort for sample preparation. Earlier investigations have shown the influence of the used analyzing method as well as the type of device on the determined roughness values [4,5 3. Comparable spatial resolutions of the method and devices are essential for comparable results. The characterized fibres were embedded in a glass matrix by a common slurry infiltration / hotpressing process. Crosspolished samples of the composites were investigated by AFM to analyze the fibre-matrix interface. The results will be given in the paper.
EXPERIMENTAL: The matrix glass DURANm was obtained from SCHOTT (Germany), Nextel 440TM fibre from 3M (Germany). The composites were manufactured by uniaxial vacuum hotpressing at 850 “C / 30 min. and 5 MPa [6]. All composite samples were thinned to nearly 500 pm and polished with STRUERS diamond suspension (0,25pm). CVD coating of the fibres was performed by Chemnitz Technical University, Dept. of Physical Chemistry [7,8,9]. The heat treatment of the fibres took place in a muffel furnace under atmospheric conditions at various temperatures with regard to the hotpressing and employing conditions. For Atomic Force Microscopy (AFM)a TMX 2000 Explorer SPM from TopoMetrix was used. The measurement was carried out in the contact mode under atmospheric conditions. The determination of the surface roughness based on measurements of four representative areas for every temperature and thickness. The evaluation was performed by the softwaretool from TMX.
190
Ceramic Nanomaterials and Nanotechnology
RESULTS AND DISCUSSION Fibre studies The BN layers seemed to be stable in air until 750 “C. Generally a temperature above 500°C lead to a stepwise change of the surface roughness. The trend is shown on the scanned pictures (figure 1).
Figure 1: Surface of 300nm BN coated Nextel 440m-fibres before and after heat treatment, performed by AFM Thermal annealing of the “as received” fibres at 500°C leads to crystal growth and increasing surface roughness. While thermal annealing a gain growth of the fibre particles was aIso found for desized Nextel MOT -fibres [6]. A coating oxidation during heat treatment at 1000 “C is thought to be responsible for the fibre surface becoming relatively smooth and the disappearance of the nanocrystalline structure of the surface. Some single crystals with dimensions of 130-150 nm jut out fiom the smooth surface. It is supposed that the oxidation product of the layer (boron oxide) forms a glassy phase [ 10,111. These results were confirmed by SEM investigations (figure 2).
Figure 2: Surface of l o o m BN coated Nextel MOTMfibres before and after heat treatment, performed by SEM
Ceramic Nanomaterials and Nanotechnology
191
Quantitative investigations of determing the surface roughness by AFM have shown a trend but not signifkant changes (figure 3).
Figure 3: Surface roughness of desized and BN coated Nextel440TMfibres with different coating thickness before and after heat treatment The surface roughness Ra shown at the Y-axisis given by equation (1):
where x is the heighvdepth of peakshalleys and h is the number of peakshalleys per area (figure 4).
Figure 4: Surface structure (net shape) of Nextel 440m-fibre (30Onm BN, 5OO0C, 5h) after fibre curvature elimination 192
Ceramic Nanomaterials and Nanotechnology
The trend shows the increase of the surface roughness owing to fibre coating while thermal annealing at 1000°C decreases the roughness. It is established that crystal growth and coating oxidation at elevated temperatures change the surface roughness of the fibres in a contrary way. Composite studies Figure 5 shows the cross sectional area of a composite with a pyrolytic carbon interface. The fibre-coatin matrix transition and the elliptic cross sectional area of the Nextel 440’ fibres are clearly visible. According to the enlargement of the interface area in figure 5 coating thickness of 290 nm was determined. The analysis of the rotational distribution of the pyrolytic carbon coating has shown a good symmetry for the samples with a sharp and close interface.
Figure 5: Nextel 440m fibre/pyro-C/glass interface in DURANm A profile few in z-direction (figure 6) explains the results above. It is assumed that the mechanical properties of the pyrolytic carbon coating leads to a “polishing effect”. In consequence both the carbon coating and the fibre jut out of the matrix plane but the coating a little bit more than the fibre. Furthermore the mean values of the elliptical axis of the Nextel 44OW fibre were determined (figure 5) to be 2a=13,Opm and 2b=9,1pm. The diameter of the corresponding circle is 10,9pm.
Ceramic Nanomaterials and Nanotechnology
193
Figure 6: Nextel MOTMfibre/C-interface/glasstransition area in z-profile In contrast to this no interface has been detected in the composite with boron nitride coating on Nextel 44OW fibres in DURANm matrix (figure 7) using a Si3N4 AFM tip in contact mode. The fibre-matrix transition seems to be diffuse. Matrix inhornogenities were also established. These aspects are confirmed by the close-up of one fibre in figure 7 (right) and the contineous matrix-fibre-transitionin z-graph in figure 8.
Figure 7: Cross polished sample of a Nextel 44Om/3O0nm BNDURANTM composite
194
Ceramic Nanomaterials and Nanotechnology
Figure 8: Nextel 440m fibre/3OOnm BN interface/glass matrix transition area in z-profile
In spite of the continuous matrix-fibre-transition the composites have shown effective debonding and pull out. Only using a AFM supertip in lateral force contact mode the interfacial region is detectable (figure 9). In the special case of figure 9 the interface thickness is determined approximately with about 370nm.
Figure 9: 3 N interface characterization in a Nextel 440TM-DURANTM glass composite
CONCLUSIONS The AFM method is a useful too1 for characterizing both the changes of surface structure of fibres and the interfaces in glass matrix composites. The Ceramic Nanomaterials and Nanotechnology
195
determination of fibre dimensions with elliptical cross sectional area was realized. The interface region of crosspolished composite samples was detected and the coating thickness could be measured. The investigated boron nitride coatings seem to be stable under hot pressing conditions for composite manufacturing and the interface has proven itself in mechanical testing. Further investigations are necessary to confirm the observed facts.
ACKNOWLEDGEMENT The Authors are grateful to the Deutsche Forschungsgemeinschaft for supporting this study. REFERENCES 'D.Hulsenberg, P. Fehling (2001) Verbundwerkstoffeund Werkstoffierbunde, VCH-Wiley: 298-302 *P.Kangutkar, T. Chang, Y. Kagawa, M.J. Kocak (1993) Ceram. Eng. Sci. Proc. 14: 963 3K.K. Chawla (1998) Composite Materials, Springer: 227 4S.H. Cohen (Ed.): Atomic Force Microscopy/ScanningTunneling Microscopy, Plenum Press, New York, (1994), 28 1-299 %. Riidlein: Werkstofhdliche Beurteilung von Glasern und Schichten mittels Rastersondenmikroskopie, Habilitationsschrift TU Clausthal, 1999 6T.Leutbecher: Beitrag zur Entwicklung von oxidfaserverstiirkten Glasem, Shaker Verlag Aachen, 2002 7 G. Manc, P.W. Martin, N. Meyer, K. Nestler (1993) Fresenius J Anal Chem 346: 181 *D.Dietrich, S. Stockel, G. Man (1998) Fresenius J Anal Chem 361: 653-655 k.Weise, S. Stockel, M. Wirikler, K. Nestler, D. Dietrich, G. M m (2001) Verbundwerkstoffeund Werkstoffierbunde, VCH-Wiley: 359-364 '%. Jacobson, S. Farmer, A. Moore, H. Sayir: J.Am.Ceram.Soc. 82 [2] (1999) 393-98 "N.S. Jacobson, G.N. Morscher, D.R. Bryant, R.E. Tressler: J.Am.Ceram.Soc. 82 [6] (1999) 1473-82
196
Ceramic Nanomaterials and Nanotechnology
A STUDY ON THE PROCESSING OF OXIDE-BASED NANOCOMPOSITES Steven Mullens, Jos Cooymans, Carine Smolders and Jan Luyten Flemish Institute for Technological Research (VITO) Material Technology Boeretang 200 2400 Mol Belgium ABSTRACT This study deals with the design and processing of oxide-oxide ceramic nanocomposites. Two strategies to introduce ceramic nanoparticles in the ceramic matrix are applied, starting either from nanopowders or from colloidal precursor sols. For the system Zr02(nm0iA1203,a homogeneous distribution of nanoparticles in the matrix could be obtained. This resulted in a suppressed grain growth of the A1203 matrix. The microstructural analysis of Al2O3(,,ano)/zTo2 system revealed larger grain size for the incorporated particles, probably due to the strong agglomeration of the nanopowder. The addition of A1203 did not influence the average grain size of the yttrium stabilised 21-02. INTRODUCTION Recent advances in nanotechnology have opened a large number of promising potentials for ceramic materials. As particle size becomes smaller and the influence of grain boundaries increases, various material properties improve or alter drastically. With increasing capabilities to synthesize and process nanoparticles with precisely controlled size, composition and morphology, new applications in a wide variety of domains come into sight. One group of nanostructured materials are ceramic nanocomposites, in which a small fiaction of ceramic nanoparticles is homogeneously dispersed in a ceramic matrix. Since the report by Niihara et all, describing a considerable improvement of the mechanical properties for A1203 dispersed with small amounts of Sic nanoparticles, nanocomposites in general have been the subject of growing interest'". However, the underlying mechanism for the strengthening is not well understood to this day. A number of potential strengthening mechanisms in ceramic nanocomposites have been proposed? the incorporation of nanoparticles
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
197
will suppress the grain growth of the matrix particles, an elimination of processing flaws and grain pull out during machining, crack healing during annealing and residual stress originating from thermal expansion mismatch. This relatively new class of materials combines the benefits which result from the incorporation of nanoparticles with economical aspects as only a small percentage of (expensive) nanoparticles is needed. Various synthesis methods have been reported for nanocomposites: mechanical alloying and milling, in situ precipitation of nanoparticles by pyrolysis of precursors’, a nano article-coating processlo-ll, coprecipitation12 and several sol-gel technologies2B3y However, fwther progress regarding both synthesis, processing and shaping of nanocomposites is essential in order to fully exploit their superior behaviour and to enable industrial implementation. The present study examinines at the synthesis, processing and characterization of ceramic oxide-based nanocomposites. The most promising applications for these materials are in fields in which hardness, durability and wear resistance are essential properties. The mechanical and tribological characterization of nanocomposites prepared by these methods is planned in the near hture. A double strategy was applied for the synthesis of such materials. A first method consists in dispersing a commercial nanopowder in the submicron matrix by planetary ball milling. As one of the main issues during the synthesis of nanocomposites is a precise controllability in terms of homogeneity of the nanoparticle distribution and spacing, a strong agglomeration of the starting nanopowder poses serious limitations towards this route. A homogeneous dispersion of the nanopowders is of essential importance in view of revealing the high performance and new properties. Therefore, a second route will be considered. For this fabrication manner, sol-gel technology is used for the in situ generation of nanoparticles. A colloidal suspension of precursor particles is mixed with the matrix powder and will transform during the heat treatment to oxide nanoparticles. This way of manufacturing nanocomposites offers prospects for avoiding the obstacles posed by the agglomeration of the dry nanopowders. In this study, two oxide nanocomposite systems will be characterized: These materials could find applications as A1203(nmo)/Zr02and ZI-O~(,,~,,)/A~~O~. advanced ceramic component^'^, wear resistant parts and cutting materials12. The microstructure, the dispersion of the nanoparticles and the influence of several parameters in the synthesis procedure are investigated.
P’.
EXPEFUMENTAL PROCEDURE The experimental procedure which is used for manufacturing the different nanocomposite systems is outlined in figure 1. As a source for nanoparticles either commercially available nanopowders or colloidal precursors sols can be applied.
198
Ceramic Nanomaterials and Nanotechnology
In all cases, the concentration of nanoparticles is 5 vol%, unless otherwise specified.
: or : Precursor sol : - - - - - - - ' ---------'
/------..-9.
]-[
\. F] !
[--
I
Freezedrylng
v [
Thermaltreatments
/-.--a-----
Nanopowder
I
1
Figure 1. Flow chart for the fabrication of nanocomposites starting from dry nanopowder or colloidal precursor sol. As matrix powder 3 mol% yttrium stabilized zirconium oxide powder (TZ-3Y7 TOSOH Co., Japan) or aluminium oxide (AKP-30, Sumitorno Chemical Co., Japan) is used. The particle size, as listed by the supplier, and the specific surface, measured by BET analysis, are presented in table 1. Commercial A 1 2 0 3 and ZrO2 nanopowders were kindly supplied by respectively Baikowski Chimie (France) and Konig Technische Keramik (Germany). The particle size and specific surface area, according to the specifications of the manufacturer, are listed in table 1. A boehmite precursor sol was made starting from aluminium sec.-butoxide (Merck, Germany). Zirconium oxide and titanium oxide precursor nanoparticles were introduced by a colloidal solution starting fiom respectively zirconium oxychloride (Merck, Germany) and titanium isopropoxide (Acros Organics, Belgium). The particle size of the precursor sols are determined by laser diffraction (Coulter N4Plus) and are listed in table 1. After the introduction of nanoparticles, the mixture is wet milled for a total time of 1 h with the aid of a planetary ball mill using Zr02 balls. After freeze drymg, a dry powder is obtained containing the ceramic nanoparticles (in case of the dry nanopowders) or precursor particles (in case of the precursor sol) which are distributed in the submicron powder. This powder mixture can be cold pressed
Ceramic Nanomaterials and Nanotechnology
199
and subsequently sintered. The sample densities after sintering were measured using the Archimedes method. For the observation of the morphology of the grain structure, samples were polished using a suspension of diamond in ethanol to a 1 pm fmish. Subsequently, the surface of the samples was thermally etched. After depositing a Pd/Pt layer, the etched surface and the fracture section were examined using a field emission scanning electron microscope (FEGSEM, type JEOL JSM-6340F). Table 1. Specifications of starting materials. Matrix powder Nanopowder Precursor sol
A1203 zrO2
AKP-30 TZ-3Y
BET (m2/g) 8 16
A1203
Baikalox CR125 Grade B
99 35
zro2
A1203 ZrO2
Particle size (nm) -500 -300 20 30
85 45
RESULTS The system ZrO2(-,,)/AhO3. As already described, 21-02 nanoparticles can be introduced by means of a precursor sol or by commercially available dry nanopowder. When applying dry nanopowders as starting materials for the introduction of nanoparticles, agglomeration can hinder a homogeneous distribution of these particles in the matrix. Figure 2a presents a SEM image of the dry ZrO2 nanopowder. As clearly can be noticed, the powder is composed of spherical particles with an individual particle size between -20 and 70 nm. The agglomerates are structures of several hundred nanometer. This powder has been used for the fabrication of nanocomposites by the procedure as outlined in figure 1. The microstructural analysis of the surface reveals a homogeneous dispersion of the ZrO2 nanoparticles in the A1203 matrix (figure 2b). The average ZrO2 particle size is around 100 - 200 nm. Although this size is larger than the starting size of the individual particles, the milling procedure proved to be efficient for de-agglomerating the ZrO2 nanopowder to a large extent. Most of the nanoparticles are located at intergranular positions and may in this way influence the grain growth of the matrix particles. Some intragranular positions can be observed as well, probably due to a mechanism of entrapment by the A1203 grains during sintering. Dense components could be obtained by varying the press loading and the sinter procedure (table 2).
200
Ceramic Nanomaterials and Nanotechnology
Figure 2. SEM image of (a) agglomerated 21-02 nanopowder and (b) surface composite. microstructure of Zr02(nan0)lA1203 Table 2. Overview of nanoparticle content (in volume percentage), maximal sinter temperature and sintered density (as percentage of the theoretical density) for submicron powders and composites. Nanoparticle Sinter temperature Sintered density content (~01%) ("C) (%TD) Submicron powder A1203 1500 99 Zr02 1500 100 zr02(nano)/A1203
Nanopowder Precursor sol
5 2
5
1500 1500
97- 100 90-98 93-98
1500 1500 1500
96- 100 95-98 96-99
1500
M203(nano)/Zr02
Nanopowder Precursor sol
5 2 5
Figure 3a shows a SEM image of the surface microstructure after the manufacture of the ZrO2(,,,)/Al2O3 nanocomposite using the colloidal zirconium oxide precursor sol as a source for Zr02 nanoparticles (2 ~01%). The microstructure is very similar to the one obtained by the addition of the dry nanopowder. The nanoparticles are again homogeneously distributed throughout the Z r 0 2 matrix. The individual 2 1 - 0 2 particle size is between 100 and 200 nm.
Ceramic Nanomaterials and Nanotechnology
201
Figure 3. SEM image of surface microstructure of (a) Zr02(nm0)/A1203 (2 ~01%)and (b) pure AI203 sintered at the same conditions. When comparing this with the precursor particle size of 45 nm (table l), ZrO2 particles are larger then expected. Plausible mechanisms are either hydrolysis of the precursor sol particles during the processing procedure or by grain growth during sintering. Table 2 lists the sintered density of the components. The influence of the addition of a small concentration of nanoparticles in regard of the grain growth of the A1203 matrix is revealed by comparison of the microstructure of the pure A1203 component, using the same submicron powder and sintered at the same conditions as the nanocomposite (figure 3b). The microstructure of the pure A1203 is characterized by a non-uniform grain growth, resulting in a wide distribution of particle diameters. Some pores, mostly at intragranular positions, can be noticed. The presence of ZrO2 nanoparticles impedes grain growth to a large extent. This observation indicates and confirms the good prospects for these materials in view of their wear resistance. Apart from a more quantitative analysis of the reduction of the grain growth of A1203matrix using image analysis, mechanical and tribological characterization of this material is planned. The system Al203(,,)/ZrO2. Analogous to the previous systems, two strategies for the synthesis have been employed. The agglomeration of the dry A1203 nanopowder is shown in figure 4a. Although no individual particles can be observed by the SEM analysis, the agglomerates are several hundred nanometer and seem to be more compact compared to the ZrO2 nanopowder (figure 2a). The microstructure of the nanocomposite fiom this nanopowder is demonstrated in figure 4b. Although the homogeneity of the distribution is comparable to the previous system, not all the A1203 particles are in the nanometer range anymore.
202
Ceramic Nanomaterials and Nanotechnology
Figure 4. SEM image of (a) agglomerated A1203 nanopowder and (b) surface microstructure of A1203(,,)/ZrO2 composite. The average A1203 particle diameter is around 1 pm. At some places, two or more particles are bounded together and can still be distinguished. These observations point to a strong agglomeration of the individual A1203 particles in the powder and an insufficient milling procedure to break the strong interactions between the individual particles. As such, a submicron composite is formed. The density after sintering is listed in table 2. Starting fiom a colloidal boehmite sol, the problems regarding the agglomeration of powders might be circumvented. The average particle size in the precursor sol is 85 nm. During thermal treatment, the boehmite is in a first stage converted into ?-A1203 and finally at higher temperatures into cr-Al203. As the particles in the sol are not agglomerated, the milling procedure should be efficient for complete homogenization of the boehmite particles in the matrix. However, particle growth occurs during the processing procedure, resulting in a microstructure with A1203 particle sizes slightly smaller compared to figure 4b (figure 5a). A1203 particles have average diameters around 500 nm. Further analysis is needed to identify the mechanism behind the particle growth. Possibly, the sol particles are hydrolysed during milling or freeze drymg. Another mechanism could involve strong particle growth at some stage in the sinter process. The density after sinkring ranges fiom 95 to 98 TD%. The microstructure of the pure ZrO2 matrix at these sinter conditions shows grains sintered in an uniform manner with average particle diameter around 500 nm. The applied sinter conditions lead to full density (table 2). Compared to figure 5a, no substantial influence on the particle size of the 2 1 0 2 grains could be noticed by adding nanoparticles. As the grain growth is already limited for this ZrO2 power, the effect of the addition of nanoparticles may be restricted.
Ceramic Nanomaterials and Nanotechnology
203
Figure 5. SEM image of surface microstructure of (a) A1203(nmo)/ZT02 (2 ~01%)and (b) pure yttrium stabilised Zr02 sintered at the same conditions. The effect on the sinter process of coarser ZrO2 powder may be more substantial and will be the subject of further research. As a conclusion for this system, the apparent particle growth compared to the system Z1O2(~~t/Al203 leads to the formation of submicron composites instead of nanocomposites. Both starting from dry nanopowders or from boehmite precursor sol, particle growth andor insufficient de-agglomeration in case of the dry nanopowder account for this behaviour. CONCLUSIONS This preliminary study demonstrates the feasibility of manufacturing oxide-based ceramic nanocomposites starting either from dry nanopowders or fiom colloidal precursor sols. As a strong agglomeration of the dry nanopowders hinders a homogeneous dispersion of nanopowders in the matrix, the latter strategy could offer some benefits concerning this matter. However, the hydrolysis reaction in the precursor sol must be carefully controlled. For the system Zr02(nmo)/A1203,both routes are capable of fabricating nanocomposites with average ZrOz particle size around 100 to 200 nm. The nanoparticles are homogeneously dispersed Within the matrix. Moreover, grain growth of A1203 matrix is suppressed to a considerable extent. This observation arouses expectations for their use as wear resistant materials. On the other hand, in the Al2O3~,,,,)/ZrO2 system, larger grains of incorporated particles can be observed. This can be explained by a strong agglomeration of the dry nanopowder and the insufficient milling procedure. For the synthesis route based on the precursor sol, hydrolysis reactions of precursor particles during the
204
Ceramic Nanomaterials and Nanotechnology
further processing procedure of the nanocomposite or grain growth during the sinter process are responsible. In the near future, research will focus on the reactions that are occurring in the precursor sol during the processing steps with the aim of impeding the hydrolysis reactions. Also, several nanopowders will be investigated with regard to their agglomeration. A more quantitative analysis of grain size distribution will enable a full description of the limitation of grain growth by incorporating nanopowders. Mechanical and tribological analysis will give a view on possible applications for these materials. ACKNOWLEDGMENTS The authors gratefully acknowledge M. Schoeters for carefully polishing the samples and R. Kemps for the extensive SEM analysis. This work is supported by the Flemish Government through GBOU project no 010059, granted by IWT. REFERENCES ‘K. Niihara, ‘Wew design concept of structural ceramics: ceramic nanocomposites”, Journal of the Ceramic Society Japan, 99 974-982 (199 1). 2M. Balasubramaniam, S . Malhotra and C. Gokularathnam, “Sintering and mechanical properties of sol-gel derived alumina-zirconia composites”, Journal Materials Processing Technology, 67 67-70 (1997). 3S. Bruni, F. Cariati, M. Casu, A. Lai, A. Misinu, G. Piccaluga and S. Solinas, “IR and NMR study of nanoparticle-support interactions in Fe203-Si02 nanocomposite prepared by a sol-gel method”, Nunostructured Materials, 11 [5] 573-586 (1999). 4H. Wang, L. Gao and J. Kuo, “The effect of nanoscale Sic nanoparticles on the microstructure of A1203 ceramics”, Ceramics International, 26 39 1-396 (2000). 5T.Ohji, Y. Jeong, Y . Choa and K. Niihara, “Strengthening and thoughening mechanisms of ceramic nanocomposites”, Journal of the American Ceramic Society, 81 [6] 1453-1460 (1998). 6 J. Pkrez-Rigueiro, J. Pastor, J. Llorca, M. Elices, P. Miranzo and J. Moya, “Revisiting the mechanical behavior of alumina/silicon carbide nanocomposites”, Acta Materialia, 46 [15] 5399-5411 (1998). 7 H. Wan and W. Yang, “Thoughening mechanisms of nanocomposite ceramics”, Mechanics of Materials, 30 111- 123 (1998). 8 H. Wu, S . Roberts and B. Derby, “Residual stress and subsurface damage in machined alumina and alumidsilicon carbide nanocomposite ceramics”, Acta Materialia, 49 507-5 17 (2001). ’D. Li, D. Wu, X. Wang, L. Lu and X. Yang, “Rapid preparation of porous FezO3/SiOz nanocomposites via an organic precursor”, Matevials Research Bulletin, 36 2437-2442 (2001).
Ceramic Nanomaterials and Nanotechnology
205
‘‘A. Boulle, 2. Oudjedi, R. Guinebretikre, B. Soulestin and A. Dauger, “Ceramic nanocomposites obtained by sol-gel coating of submicron powders”, Acta Materialia, 49 8 1 1-8 16 (200 1). “A.RuYs and Y. Mai, “The nanoparticle coating process: a potential sol-gel route to homogeneous nanocomposites”, Materials Science and Engineering, A265 202-207 (1 999). ”J. Hong, S. DelaTorre, K. Miyamoto and L. Gao, “Crystallization of A1203/ZrO2 solid solution powders prepared by coprecipitation”, Materials Letters, 37 6-9 (1998). I3M. Yoshimura, S. Oh, M. Sand0 and K. Niihara, “Crystallization and microstructural characterization of 21-02nanosized powders with various A1203 contents”, Journal of AlZoys and Compounds, 290 284-289 (1999).
206
Ceramic Nanomaterials and Nanotechnology
Nanoscale Phenomena in Glasses, Glass=Ceramics, and Glass=Containing Composites
This page intentionally left blank
INTERSTITIAL NANOSTRUCTURES IN ENGINEERED SILICATES Lilian P. Davila, Subhash H. Risbud, aqd James F. ShackeZford Department of Chemical Engineering and Materials Science University of California, Davis, California 95616
ABSTRACT Interstices in silicates play an important role in various technologies such as gas transport in glasses and catalysis in zeolites. The interstices are generally in the nanometer-range and, in non-crystalline silicates, are important components of “medium-range” structure (larger than the tetrahedral building blocks). The atomic structures of crystalline silicas such as quartz and cristobalite are analogs of vitreous silica and can be described as a packing of filled oxygen polyhedra (Si04 tetrahedra) and empty ones (interstices of various shapes). Computer simulation software has been used to describe the interstices quantitatively. In addition, the energetics of high-silica zeolites is directly related to the internal surface area of these interstices. Computer simulation also allows the interstitial surface area in vitreous silica to be compared to that for the common silica polymorphs (quartz and cristobalite) as well as a wide-range of the high-silica zeolites. INTRODUCTION Silicate glasses are known for their permeability to various gases [l], and zeolites are widely used as catalysts and molecular sieves [2]. These material applications depend on nano-scale interstitial structure, which is in turn a useful structural descriptor. As in Bernal’s canonical hole model of liquids [3], interstitial voids are defined by connecting the centers of adjacent atoms. Amorphous metals [4] and metallic grain boundaries [S, 61 have been rnodeled in this way. An advantage of the stacking of polyhedra as a structural descriptor is that the technique does not “break down” as one goes from crystalline to defective to completely non-crystalline structures. The interstitial structure of cristobalite helped to illustrate the nature of gas transport in vitreous silica [7]. Oxygen polyhedra (edge-shared CaO6 octahedra) defined medium-range ordering in calcium silicate glass, for which wollastonite is a crystalline analog [S]. A wide range of silicates has s i k a tetrahedra intricately linked to non-silicon metal-oxide polyhedra, often assembled as planar sheets of edge-shared octahedra (e.g., wollastonite) [9]. In identiqing crystalline analogs To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
209
of silicate glasses, one can identify the shapes of the unfilled polyhedra (interstices) as well as the filled ones (e.g., SiO4 and CaO6) 1101. The interstitial structure of wollastonite (distorted tetrahedra, square pyramids, and triangular prisms) has been identified as part of a general description of the canonical hole set for non-metallic solids, such as silicates [l 11. Based on the constructable, convex polyhedra identified by Zalgaller [12], the canonical hole set for silicates [l 11 consists of 44 simple polyhedra. This number is larger than the eight simple, Bernal holes for metals [ 5 ] due to the different bonding. (Some covalent, directional bonding in ceramics allows more open, cage-like structures in contrast to the densely packed nature of metallic, non-directional bonding).
INTERIOR NANOSTRUCTURES IN CRYSTALLINE SILICAS - ANALOGS FOR VITREOUS SILICA Images of silica polymorphs and associated interstitial structure were created using software from Accelrys (formerly known as Molecular Simulations Inc. [MSI] and BIOSYM Inc.) on Silicon Graphics workstations [lO]. Interstitial polyhedral volume calculations were performed using simple vector analysis. The process involved calculating an interstitial polyhedron volume as the s u m of tetrahedral volume elements. To confirm the accuracy of the method, the sum of volmes of Si04 tetrahedra and the interstitial polyhedra were shown to be equivalent to the total volume of the corresponding unit cells. Figure 1 shows three adjacent interstitial polyhedra (truncated tetrahedra) in high cristobalite, the simplest of the silica polymorphs. In this crystalline analog for vitreous silica, there are eight interstitial polyhedra along with eight SiO4 tetrahedra in the cristobalite unit cell [7]. Figure 1 illustrates the utility of this approach for monitoring diffusional paths in relatively open network silicates. Six-membered rings (hexagons) serve as doorways between adjacent truncated tetrahedra. Although cristobalite is an appropriate crystallhe analog for vitreous silica [7], the range of ring sizes in the non-crystalline material requires that some smaller interstices also exist. The interstices in the common, higher-density polymorph low quartz provide some indication of the nature of the smaller polyhedra expected in silica glass [lO]. The linkage of SiO4 tetrahedra in low quartz follows a double helix when viewed along the c-axis [131. It is interesting to note that the interstices found in low quartz are of the same type found in wollastonite [111, viz. distorted tetrahedra, square pyramids, and triangular prisms. Of course, the exact nature of the distortion is somewhat different in the two cases. Quartz, like wollastonite, is a relatively tight structure with interstitial space represented by a limited number of relatively small oxygen polyhedra. These relatively s d l interstices shown in Figure 2 are part of the complete set of polyhedra for non-metallic solids given in Table I.
210
Ceramic Nanomaterials and Nanotechnology
Figure 1 A difisional path in high cristobalite illustrated by three, adjacent interstices (each being a truncated tetrahedron). The doorway between adjacent polyhedra is a six-membered, hexagonal ring.
Ceramic Nanomaterials and Nanotechnology
211
Figure 2 With the c-axis in the plane of the page, one can see how a set of interstices (two distorted tetrahedra and two distorted square pyramids) packs into the space between two adjacent silica tetrahedra along the quartz double helix. Also, the interstitial space dong the c-axis is filled with distorted triangular prisms. Table I. Polyhedra sets for interstices in metals and nonmetals [11 Metalsa Nonmetalsb
8 polyhedra (with up to 20 triangular faces)
44 polyhedra 28 simple, convex regular polyhedrac 8 prisms 8 antiprisms
"Ref. [5], k e f . [l11. 'Ref. [12] In Table 11, the sizes of the interstices found in the crystalline analogs of vitreous silica are compared with the distribution of interstitial solubility site sizes determined by the analysis of gas transport in vitreous silica [141. Assuming an oxygen radius corresponding to a 50 - 75% covalent nature of the Si-0 bond [151, the inscribed sphere diameters for regular polyhedra and the ''doorways'' into those polyhedra are the appropriate comparison for the "sizes" of interstices determined by gas probe atoms. The values of the doorway sizes in Table I1 are in good agreement with the range of interstitial sizes determined by gas transport (0.1 to 0.4 nm) [14].
212
Ceramic Nanomaterials and Nanotechnology
Table LI. Size of interstices and their doorways (as inscribed spheresa)for quartz and cristobalite [101 Interstice
Interstice dia. [nm] Doorway Doorway dia. [nm]
Tetrahedron Square pyramid Triangular prism Truncated tetrahedron
0.147 - 0.190 0.197 - 0.241 0.227 - 0.271 0.442 - 0.486
Triangle Square Square Hexagon
0.128 - 0.172 0.197 - 0.241 0.197 - 0.241 0.275 - 0.319
"Using an oxygen radius of 0.967 - 0.089 nm corresponding to 50 - 75% covalent nature of the Si-0 bond [151. The next phase of the current research will focus on systematically cataloging all interstices in a vitreous silica model equivalent to that of Fueston and Garofalini [161. The linkage of tetrahedra for the Fueston-Garofdini model is illustrated in Figure 3, for which a wide variety of interstitial geometries are evident. The Fueston and Garofalini model is especially interesting as it is the basis of an independent study by Chan and Elliott [17] who showed that the distribution of interstitial site sizes closely follows a log-normal distribution consistent with that based on gas solubility data [7].
INTERIOR NANOSTRUCTURES IN HIGH-SILICA ZEOLITES - A COMPARISON WITH VITREOUS SILICA The Cerius2 modeling environment is an Accelrys simulation product that allows the complex interior surfkce area of interstitial regions to be mapped, as illustrated for a zeolite structure in Figure 4. Moloy, et al. have shown that the formation enthalpy of a wide range of high-silica zeolites displays a linear relationship with this internal surface area [Z]. The slope of the regression line through the data corresponds to an internal surface enthalpy with a value of 0.093 + 0.009 Urn2, in good agreement with the experimental value for external silica surfaces. It is interesting to apply the technique used by Moloy, et al. [2] to vitreous silica. Figure 5 shows the interior surface area of the structure in Figure 3 mapped out by a 0.096 nm diameter probe atom, the same size as used in the study by Moloy, et al. In Figure 6, one can see that the interior surface area of vitreous silica follows the Same general trend shown by the relatively open, cage structures of the zeolites, as well as the common (and more dense) silica polymorphs quartz and cristobalite. The horizontal axis in Figure 6 is the framework density, the number of tetrahedra per unit volume.
Ceramic Nanomaterials and Nanotechnology
213
Figure 3 The linkage of tetrahedra in a vitreous silica model equivalent to that of Fueston and Garofalini [161.
214
Ceramic Nanomaterials and Nanotechnology
Figure 4 Example of (a) a zeolite structure for which (b) an interior surface area is defmed by tracing out accessible space by a probe sphere of radius = 0.05 nm. Figure 7 shows that the enthalpy of transition (AHm) versus internal surface area for a wide range of silica structures follows a linear trend, as noted by Moloy, et al. [2]. The enthalpy of transition is defined as the enthalpy difference between the given silica and that of a-quartz. One can note that the value for vitreous silica is consistent with this linear trend exhibited by the wide range of crystalline silicas reported by Moloy, et al. ACKNOWLEDGMENTS We thank T.E. Allis of the University of California, Davis and J.M. Newsam, N. Khosrovani, and A. Amiti of Molecular Sirnulations, Inc. and Accelrys for experimental help. Professor Stephen Garofalini of Rutgers University has provided numerous useful discussions, as well as the atomic coordinates corresponding to Figure 3. Professor A. Navrotsky and E.C. Moloy of the Thermochemistry Facility at the University of California, Davis provided especially helpful discussions regarding the energetics of various silicas. One of us (LPD) was supported by the UCLLNL Student-Employee Graduate Research Fellowship (SEGRF) Program. REFERENCES 1. J.F. Shackelford, “Gas Solubility in Glasses - Principles and Structural Implications,” J. Non-Crystalline Solids 253 23 1-24 1 (1999).
Ceramic Nanomaterials and Nanotechnology
215
Figure 5 The interior surface area of the structure in Figure 3 mapped out by a 0.096 nm diameter probe atom.
216
Ceramic Nanomaterials and Nanotechnology
160 140
i
Y
60
20
0 0
5
10
15
20
25
30
F.D. (#/nm*3)
Figure 6 Interior surface area in a wide variety of silica structures (mostly highsilica zeolites, along with quartz, cristobalite, and vitreous silica) versus framework density (the number of tetrahedra per unit volume). Note that the interior surface area of vitreous silica follows the overall trend for the wide range of crystalline silica.
2. E.C. Moloy, L.P. Davila, J.F. Shackelford, and A. Navrotsky, “High-Silica Zeolites: A Relationship between Energetics and Internal Surface Areas,” Microporous and Mesoporous Materials 54 { 1-21 1 - 13 (2002). 3. J.D. Bernal, “The Structure of Liquids,” Proc. R. Soc. (London) 280 299322 (1964). 4. J.L. Finney, and J. Wallace, “Interstice Correlation Functions: A New Sensitive Characterization of Non-Crystalline Packed Structures,” J. NonCrystalline Solids 43 165-1 87 (1 98 1). 5. M.F.Ashby, F. Spaepen, and D. Williams, “The Structure of Grain Boundaries Described as a Packing of Polyhedra,” Acta Metallurgica 26 16471663 (1 978). 6. M.R. Fitzsimmons, and S.L. Sass, “The Atomic Structure of the I:= 13 (0 = 22.6”) [OOlJ Twist Boundary in Gold Determined Using Quantitative X-ray Difiaction Techniques,” Acta MetalZurgica 37 1009-1022 (1989).
Ceramic Nanomaterials and Nanotechnology
217
100
50
150
SA (mA2 x E3/mol)
Figure 7 Enthalpy of transition in a wide variety of silica structures (mostly highsilica zeolites, along with quartz, cristobalite, and vitreous silica) versus internal surface area. Note that the datum for vitreous silica follows the overall trend for the wide range of crystalline silicas reported by Moloy, et al. [2).
7. J.F. Shackelford and J.S. Masaryk, “The Interstitial Structure of Vitreous Silica,“J. Non-Crystalhe Solids 30 127-139 (1978). 8. P.H. Gaskell, M.C. Eckersley, A.C. Barnes, and P. Chieux, “MediumRange Order in the Cation Distribution of a Calcium Silicate Glass,” Nature 350 675-677(1991). 9. F. Liebau, Structural Chemistry of Silicates, Springer, Berlin, 1985. 10. L.P. Davila, S.H. Risbud, and J.F. Shackelford, “QuantiQing the Interstitial Structure of Non-Crystalline Solids,” Recent Res. Devel. NonCrystalline Solids,1 73-84 (2001).
218
Ceramic Nanomaterials and Nanotechnology
1 1. J.F. Shackelford, “The Interstitial Structure of Non-Crystalline Solids,” J. Non-Crystalline Soli&-,204 205-21 6 (1 996). 12. V .A. Zalgaller, Convex Polyhedra with Regular Faces, Consultants Bureau,New York, 1969. 13. D.C. Palmer, Silica - Physical. Behavior, Geochemistry and Materials Applications, P.J. Heaney, C.T. Prewitt, and G.V. Gibbs, Eds., Mineralogical Society of America, Washington, D.C., 1994, p. 85. 14. G.S. Nakayama and J. F. Shackelford, “Solubility and Diffusivity of Argon in Vitreous Silica,” J Noncrystalline Solids 126 49-54 (1990). 15. J.F. Shackelford, A.G. Revesz, and H.L. Hughes, “The Effect of Covalency on Interstitial Structure in Vitreous Silica Reaction Layers,” in: Reactivity of Solids, P . Barret and L.-C. Dufour, Eds., Elsevier Science Publishers, Amsterdam, 1985, pp. 279-283. 16. B.P. Fueston and S.H. Garofalini, “Empirical Three-Body Potential for Vitreous Silica,” J: Chem. Phys. 89 5818-5824 (1988). 17. S.L. Chan and S.RElliott, “Theoretical Study of the Interstice Statistics of the Oxygen Sublattice in Vitreous SiOa,” Phys. Rev. B 43 4423-4432 (1991).
Ceramic Nanomaterials and Nanotechnology
219
This page intentionally left blank
RECENT ADVANCES IN LAS-GLASS CERAMICS Wolfgang Pannhorst Schott Glas Hattenbergstr. 10 55 122 Mainz Germany
ABSTRACT Examples of glass ceramic developments in recent years are presented which address either existing commercial products or product developments which look favourable to establish themselves in the market. Raman spectroscopy has been proven to be an effective tool to describe the nucleation of model compositions of lithium aluminosilicate low expansion glass ceramics quantitatively. These glass ceramics still form the dominant composition family for glass ceramics products. Product developments concerning cooktop panels, telescope mirror substrates, and more recent applications in microlithography and in beamers are discussed. INTRODUCTION Since Coming's public announcement of the invention of glass ceramics in 1957 more than 40 years have passed. The scientific community has investigated many glass systems for their potential to provide new glass ceramic materials with a combination of properties which otherwise cannot be achieved. While the researchers were very successful in developing many new glass ceramics the commercial success of these materials was so far rather limited. The low expansion glass ceramics still stand as the most often cited reference for the commercial success of the material class of glass ceramics. In [11 I have looked at the development of tough glass ceramics and on sintered ones and at the applications in which these glass ceramics are more or less used. I have also pointed out that glass ceramics have a cost disadvantage compared with other materials because of the additional ceramisation process; as a result glass ceramics have to have a unique feature to be competitive. The almost zero coefficient of thermal expansion of the low expansion glass ceramics is such a feature and explains why they still form the most important class of glass ceramics commercially; new glass ceramics in the areas of information technology and telecommunication may become as important in the near future. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
221
LOW EXPANSION GLASS CERAMICS BASED ON THE LASCOMPOSITION FIELD Glass ceramics are glasses which can be crystallised in a very well controlled manner; this control is attained when the two processes of nucleation and crystal growth can be sufficiently se arated and when the nucleation rates are sufficiently high, i.e. in the range of 10 nuclei/(mm3s). These conditions are fulfilled in the Li20-A1203-Si02-(LAS)-systemwhen Ti02 andor ZrO2 are added as nucleating agents. The LAS-system itself is of particular interest since the major crystalline phase that forms during crystallisation in a certain field of the phase diagram displays an unusual physical behaviour; the high-quartz solid solutions (SA) crystals that form during devitrification display strong negative volume expansion with increasing temperature (121, [3]). By adjusting the amounts of the crystalline phase with negative volume expansion and of the residual glass phase with positive volume expansion it is possible to design materials that display low expansion behaviour, at least in certain temperature intervals [4]. The development of the microstructure in the LAS glass ceramics during ceramisation, i.e. during the heat treatment for nucleation and crystal growth, has been investigated many times but the intriguing question which up to today has not been answered in a convincing manner is a quantitative description of the phase formation during the ceramisation process of glass ceramics, especially for the low expansion glass ceramics of the LAS-system. The lack of this description is related to the difficulty to find characterisation methods which span nuclei andor crystal densities fiom 1 to 10"/cm3. Investigations by the two stage development method have revealed the initial stages of nuclei formation and their kinetics [53;but, as this method makes use of the detection of crystals by optical microscopy only densities up to about 106/cm3 can be investigated. On the other hand X-ray analysis is not sensitive enough to describe quantitatively the development of phases with an accuracy in the range of 0.1 mol%, especially at concentrations below Imol%. Furthermore, the aforementioned methods are not suited to detect glass in glass phase separation processes which are suggested from TEM-investigations [6] for the commercial glass ceramic Zerodur for optical precision applications. These TEMinvestigations also reveal a relatively good separation between nucleation and crystal growth. Fig. 1 shows a TEM micrograph of a Zerodur base glass sample heated with constant rate of 1.67Wh fiom 620 to 700°C and then quenched in air.
P
222
Ceramic Nanomaterials and Nanotechnology
Fig. 1 :TEM micrograph of heat-treated Zerodur base glass showing (Zr,Ti)02-nuclei formation. Left-hand side: bright field image; right-hand side: dark field image (after ref. 6)
The bright field image on the left-hand side displays small nuclei which are more clearly visible in the dark field image on the right-hand side. The size of the nuclei is approximately 5nm; no larger crystals are detected. A fblly ceramised sample
Fig. 2: TEM micrograph of the typical microstructure of Zerodur glass ceramic (after ref. 6)
Ceramic Nanomaterials and Nanotechnology
223
shows a remarkably different microstructure, Fig. 2. In addition to the small nuclei crystals larger crystals with a narrow size distribution are visible. These crystals are high quartz S.S. crystals and fonn the major crystalline phase of the glass ceramic with it volume fraction between 60 to 70%. QUANTITATIVE DESCRIPTION OF THE CERAMISATIONPROCESS Raman scattering is a method which can be applied to crystals as well as to glasses and, consequently, there are a few investigations of the ceramisation process of LAS-glass ceramics [7,8]; but until recently no quantitative description of the phase formation during the ceramisation process was attempted in these investigations. In [9,10] it was shown that such a quantitative description is possible for the model system LiAlSi308 + 4.0 mol% TiO;!. The measured rnultiphase Raman spectra were interpreted to be a linear superposition of the spectra of up to 8 single phases, the spectra of which were measured separately. For the description of the phase formation and phase decomposition the Johnson, Mehl, Avrami [11,121 expression was used. Figure 3 gives an example of the kind of results that can be achieved.
I
..., I
.
.
,
....., . . .
. . . . . , I
I0
I00
,
. . .....I IOW
1 ... 1 )
time [min]
Figure 3: Kinetics of the formation of pseudobrookite , Al, Ti treatment at 740 OC;after [9]
.
..._....,
....., 10
loo
time [min]
'...__.I 1wO
02,and anatase during nucleation heat
This quantitative description of the phase formation in the model system LiAlSi30~+ 4.0 mol% Ti02 is a great improvement in our understanding of the ceramisation process of low expansion glass ceramics. Recently [13], the investigations have been extended to a more complex system with two important changes: a) the molar ratio between Li20 and A1203has been changed fiom 1:1 to 0.8:1.2 so that there is a surplus of A13+ions; this ratio is closer to that used in commercial LAS glass ceramics; it is expected that these additional A13+ions will not be incorporated into the crystalline high quartz S.S. phase during the
224
Ceramic Nanomaterials and Nanotechnology
ceramisation process thus modifying the composition of the residual glass phase; b) the nucleating agent of 4.0 mol% TiO2 has been replaced by two nucleating agents, 2mol% Ti02 and lmol% ZrOz; although there are commercial glass ceramics which contain only TiO2 as nucleating agent the combined use of both nucleating agents is preferred. Also in this case it is possible to describe the segregation of single oxides from the glass and the parallel formation of crystalline phases quantitatively. Fig. 4 shows as an example the segregation of
1
4
' " I
I
i
I
.., 10
I
I
100
1000
c
nucleation time [min]
Figure 4: Kinetics of the segregation of 4fold co-ordinated TiN02 and Zrl"02 and parallel formation of ZrTi04; after [13J
4fold co-ordinated TiTvOzas well as ZrwOz fiom the glass and the parallel formation of ZrTiO4 nuclei. Even though these new investigations indicate that the phase changes induced during ceramisation in these more complex glasses can be described surprisingly well, still quite some work is necessary to prove that the same approach can be applied successfully to the investigation of a commercial composition which consists of about 10 oxides instead of up to 5 for the model systems investigated so far.
Ceramic Nanomaterials and Nanotechnology
225
LARGE TELESCOPES In precision optics applications one area of ongoing challenges is the production of large telescope mirrors. For telescope mirrors in the range of 4- I Om in diameter different approaches have been used [ 141; spin casting has been used for monolithic menisci up to 8.2m [15]; other large telescopes like the two 1Om Keck telescopes [ 161 have segmented mirrors. This approach is also used in the new 1lm GTC (Gran Telescopio Canarias) telescope 1171. A particularly challenging request is to build a testing lens which will be used to verify the specifications of the secondary mirror. The secondary mirror is a light weighted convex Be minor with 1.17 m diameter. To test the figure of this mirror a high precision lens with concave/convex shape and with 1.33 m in diameter is required where the convex contour is aspherical. As the middle of the lens is 189 mm thick good transmission of the glass ceramic material is required. The specification was set to 15%; choosing a particularly homogeneous and clear piece of Zerodur a value of 29% was achieved. The machined lens has been produced within 4 weeks (Fig. 5).
Figure 5: Machined lens for testing the secondary mirror of the GTC telescope.
226
Ceramic Nanomaterials and Nanotechnology
MICROLITHOGRAPHY Low expansion glass ceramics for precision optics have played an important role in the form of wafer stages and reticle stages in microlithography steppers for many years. While the sizes of structures in memory chips as well as in processors continue to decrease from one generation to the next, not only new, more precise optics are required, but also the positioning of the masks and the wafers requires better precision. This has led to the use of low expansion glass ceramics for these components. The use of light weighted mask reticles and wafer stages has become a stable business. The exact design of these light weighted structures are considered to be proprietary knowledge. Fig. 6 therefore shows an older version of a wafer stage.
Figure 6: Wafer stage made of optical low expansion glass ceramic Zerodur
Optical transmission lithography has been the dominant process for several decades in the production of chips. Currently, lithography steppers work at the wavelength of 193nm and are able to produce line width of 13Onm and, probably, will serve the lOOnm generations. The industry further hopes that it will be able to provide steppers operating at 157nm, thereby shifiing the principle of optical
Ceramic Nanomaterials and Nanotechnology
227
transmission lithography even to the 7Onm resolution. Even if 157nm steppers will be able to provide 7Onm resolution it is generally agreed that the industry will need a next generation lithography (NGL) tool which is based on reflective rather than refractive optics. Although 5 different technical approaches are currently under development, the extreme-ultraviolet lithography (EWL) looks very promising to go into production in about 2006, at which time 7Onm resolution is required [ 181. Fig. 7 shows schematically the test tool which is currently under development at EUV LLC, a company founded by several chipmakers in 1997 to promote the development of EUVL steppers. The operating wavelength is 13.4nm.
Figure 7: EUVL test tool as developed by EUV LLC (taken from [191)
While the substrate material currently used for the mask is silicon, low expansion materials will be used in production tools because even with the best multilayer mirrors more than 30% of the EUV radiation is absorbed and causes heating of the mask which is difficult to cool in vacuum. There are two candidate materials, the optical low expansion glass ceramic Zerodur produced by Schott
228
Ceramic Nanomaterials and Nanotechnology
and ultra low expansion glass (ULE)produced by Coming. Both materials and their processing will have to be improved to meet the requirements for E W L steppers [203. Very demanding requirements include surface roughness with 0.15nm rms over various spatial wavelengths, a coefficient of thermal expansion (CTE) of OkSppbK within the narrow temperature interval of 19 to 25"C, and a CTE homogeneity of I 6ppbK over the whole substrate, which will be a 6" square with a thickness of 0.25". While it is believed that these requirements can be met by the glass ceramic Zerodur, developmental efforts are needed to demonstrate that they can be met in stable production processes. COOKTOP PANELS The dominant product application for low expansion glass ceramics are still cooktop panels for cooktops [21]. Their high attraction is their flatness without any unattractive expansion joints (cooktop panels with solid disks) or holes (gas stoves) so that they can be cleaned easily; easy cleanability is highly rated by the customer. In recent years a new trend has developed; designers would like to extend the area of the glass ceramic beyond that of the heating zones to create larger, more integrated components with functions like raised areas for control elements or ridges surrounding bore holes for specifically adapted gas burners. Many other features are currently evaluated. All these components require that the planar glass ceramic plate is reshaped in an additional production step. This is a very demanding requirement as the glass ceramic base glass is designed to crystallise in a controlled manner rather rapidly when heated from room temperature to about 150°C above T,. A process has been developed which is very well suited to fabricate three-dimensionally shaped cooktops. Fig. 8 shows an example in which a decorated area has been reshaped.
Ceramic Nanomaterials and Nanotechnology
229
Figure 8: Cooktop panel in which a decorated area has been reshaped
REFLECTORS FOR PROJECTION TECHNIQUES New portable projection techniques like beamers, DVD- or video-projection require that the power of the light sources continuously increases. This means that more and more heat generated in the discharge lamps is in part stored and transported through the reflector. Especially during switching operations steep gradients form within the reflectors. These are especially large in mid-sized or large reflectors. Typical temperature gradients which form in a 80mm reflector after lhour operation of a 120W lamp over the whole reflector are larger than 350OC;the steepest gradient forms close to the inner hole of the reflector with a gradient close to 200°C over a distance of 5-10 mm. As a result low expansion glass ceramic reflectors have been developed. They now start to fulfil1 the market demands in addition to the borosilicate glass reflectors in these applications. Their advanta e is their lower coefficient of thermal expansion of at least a factor of 3 (3.3 10-f/K -> 1 10-6/K) yielding smaller stresses at the critical points. The main challenges for the development of the glass ceramic are a) a composition and processes which ensure high quality inner surfaces and stable parabolic inner contour to secure high light output for the projection and b) a composition with high IR transmission so that as much IR radiation is transmitted through and not
230
Ceramic Nanomaterials and Nanotechnology
absorbed by the reflector to reduce the heating of the reflector. Especially, the first challenge requires attention at several process steps.
Figure 9: Low expansion glass ceramic reflectors for digital projection
A very precise contour has to be pressed and maintained through cooling of the glass body and during ceramisation; during ceramisation a compromise has to be found between a fast, economical process and too rapid heating which will result in too low viscosities and, thus, in deformations of the reflectors; in addition the reflectors shrink by about 3 vol.% during ceramisation; it has to be secured that this process occurs evenly over the whole volume. Fig. 9 shows a pair of the new glass ceramic reflectors.
SUMMARY Using Raman spectroscopy the understanding of the nucleation process in LAS glass ceramics has been improved by describing quantitativelythe desolution and formation of phases during heat treatment. This description so far has been developed for model compositions only but there are good chances that the methodical approach can be extended to commercial compositions. Low expansion glass ceramics are viable products which will maintain their position also in the foreseeable future. Due to the general trend of miniaturisation new precision applications will appear to make use of the stability low expansion
Ceramic Nanomaterials and Nanotechnology
231
glass ceramics can provide; there will be continuous product development. Current developments which are evidence for this trend are glass ceramic reflectors for beamers and mirror substrates for the Extreme-UV-Lithography
.
REFERENCES W. Pannhorst, “Glass ceramics: State-of-the-art” J. Non-Cryst. Solids 219 198-204 (1 997) G. Miiller: “Volumen und thermische Ausdehnung von AluminosilikatMischkristallen mit H-Quarz-Struktur;“ Furtschr. Mineral. 63,7-20 (1 985) A. I. Lichtenstein, R. 0. Jones, S. de Gironcoli and S. Baroni: ,,Amsotropic thermal expansion in silicates: A density functional study of B-eucryptite and related materials” Phys Rev BA62(1 7) 1 1487-11493 J. Petzold and W. Pannhorst, “Chemistry and structure of glass ceramic materials for high precision optical applications” J. Non-Cryst. Solids 129 191-198 (1991) U. Schifker and W.Pannhorst, “Nucleation in a precursor glass for Li20AlzO3-Si02-glass ceramic, part 1. Nuleation kinetics” Glastech. Ber. 60 21 1-221 (1987) V. Maier and G. Muller,: “Mechanism of oxide nucleation in lithium Aluminosilicate glass Ceramics” Commun. Amer. Ceram. Soc. 70 176-178 (1 987) R. Feile and E. Rodek, “ Phase transformation in a glass-ceramic observed by laser spectroscopy”Appl. Phys. A 45 185-187 (1 988) 0. S. Dymshits, A. A. Zhilin, V. 1. Petrov, M. Ya. Tsenter, T. I. Chuvaeva and V. V. Golubkov “A Raman spectroscopic study of phase transformations in titanium-containing lithium aluminosilicate glasses” J. Phys. Chem. 100 79 (1998) R. Sprengard, K. Binder, U. Fotheringham and W. Pannhorst “Characterisation of nuclei formation in TiOz nucleated LAS-gIass ceramics by R a m spectroscopy” Proc. XVIIZth Int. Cong. Glass; Amer. Ceram. SOC., CD-ROM E06 52-57 (1 998) R. Sprengard, K. Binder, U. Fotheringham and W. Pannhorst “Titaniaactivated nucleation in lithium-aluminosilicate glass ceramics investigated by Raman spectroscopy” Proc. 6‘h Int. Otto Schott Colloquium, Glastech. Ber. Glass Sci. Technol. 71C 1 17-124 (1 998) W. A. Johnson and R. F. Mehl “ Trans. Amer. Inst. Min.(Metall.) Engrs.135 416 (1939) A. Avrami “Kinetics of phase changes. I General theory” J. Chem. Phys. 7 1103-1 112 (1939)
232
Ceramic Nanomaterials and Nanotechnology
F. Gabel, G. Miiller, F. Raether; W. Kiefer, W. Pannhorst, 0. Soh and R. Sprengard, “Quantitative charaterisation of nuclei formation by Raman spectroscopy of lithium-aluminosilicate glass ceramics doped with titania and zirconia”, submittedfor publication A. J. Marker 111, H. Fuhrmann, H. Tietze, W. Froehlich, “Lightweight large mirror blanks of Zerodur“ SHE 571 5 1-59 (1985) W. Pannhorst, R. Muller, H. Hones, H. Morian, H. Tietze and V. Wittmer “Status of the production of 8m Zerodur mirror blanks“ SPIE 2018 226-236 (1 993) T. S. Mast and J. Nelson “The Status of the W. M. Keck observatory and ten meter telescope” S H E 571 226-232 (1985) L. Jochum, Y. Castro and N. Devaney, “Gran Telescopio Canarias: Current status of its optical design and optornechanical support system” SPIE 3352 621-631 (1998) K. Diefendofl, “Extreme Lithography - Intel Backs E W for Next Generation Lithography” Cahners Microprocessor Report, December 20, 2000LNTEL-Homepage Ch. W. Gwyn, “Extreme Ultraviolet Lithography for Next Generation IC’s“ Cahners MDR MicroprocessorForum 2000 (October 1 1,2000) SEMI, “Specifications for extreme ultraviolet lithography mask substrates” SEMI dra$ document 3148 (October 16,2000) H. Scheidler and E. Rodek “Li2O-Al203-Si02glass ceramics” Arner. Ceram. SOC. 68 1926-1930 (1989)
FIGURES CAPTIONS TEM micrograph of heat-treated Zerodur base glass showing (Zr,Ti)Oz1 nuclei formation. Left-hand side: bright field image; right-hand side: dark field image (after ref. 6) TEM micrograph of the typical microstructure of Zerodur glass ceramic 2 (after ref. 6) Kinetics of the formation of pseudobrookite, AlxTi(l-x)02,and anatase during 3 nucleation heat treatment at 740 “C;after [9] Kinetics of the segregation of 4fold co-ordinated Tiw02 and Zrrv02 and 4 parallel formation of ZrTiO4; after [ 131 Machined lens for testing the secondary mirror of the GTC telescope 5 Wafer stage made of optical low expansion glass ceramic Zerodur 6
Ceramic Nanomaterials and Nanotechnology
233
7
8 9
234
EUVL test tool as developed by EUV LLC (taken from [ 191) Cooktop panel in which a decorated area has been reshaped Low expansion glass ceramic reflectors for digital projection
Ceramic Nanomaterials and Nanotechnology
NANOPHASE FORMATION IN DIFFERENT GLASS-CERAMIC SYSTEMS
Marml Schweiger Ivoclar Vivadent AC3 Bendererstrasse 2 FL-9494 Schm Principality of Liechtenstein
Wolfrcun HOland Ivoclar Vivadent AG Bendererstmsse 2 FL-9494 Schaan Principality of Liechtenstein
Volker Rheinberger Ivoclar Vivadent AG Bendererstrasse 2 FL-9494 Schaan Principatity o'f Liechtenstein
ABSTRACT The formation of nanophases was observed in the SiO~-Li~O-K2O-~O-CaOP205-F system prior to the crystallization of needle-like fluoroapatite. The size of the nanophase was ir2 the mge of 30 to 60 nm.The nanophase was detected as fluoroapatite. Subsequent heat treatment lead to the fornation of needle-like crystals 0.1 to 0.4 p in length and with an aspect ratio of up to 6, The glassceramic showed an opalescent effect. The base glass in the Si02-Al203-IC20N~~O-C~O-P~OS-F system demonstrated phase separation. The phase separation led to the f o d o n of a primary crystal phase of NaCaP04 with diameters smaller than 200 m.Fluoroapatite was precipitated at 700°C as spherulitic crystallites measuring approximately 100 nm in diameter. Prior to the apatite formation an unknown crystalline was precipitated. The microstructure of the fluoroapatite glass-ceramic resulted in highly btanslucefit glass-ceramics with an opalescent effect. The opalescence was of interest in the field of esthetic dentistry. The material was used in a sinteres form to build up the incisal edge of dental restorafions such as mwns or bridges. Spherical nanophases with a diameter of 50 to 100 nm were derived by heat treatment at 585 to 630°C in the SiO2-MezOCaO-F system.The nanocrystalline phase was detected as CaF2. The crystallites were uniformly distributed in the glassy matrix. The glass-ceramic was transparent and showed an opalescent effect. The nanophases in these different glass-ceramic systems were characterized in SEM, XRD and SEM investigations. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
235
INTRODUCTION Nanostructured glass-ceramics offer new possibilities for combining material properties such as strength and transparency. Microstnrctures consisting of nanophases also have the potential to improve the wear behavior and the esthetics of dental restorations. The demands on glass-ceramics for dental applications are manifold. The esthetic properties are essential as well as the mechanical strength and biocompatibility, Dental restorations are highly functionalized. Therefore, Werent glass-ceramic systems are used for dental applications. Dental restorations consist of different layers and components with specifically tailored properties. The substructure has to achieve the mechanical strength and the layering materials have to match the optical properties, such as translucency and opalescence, of natural teeth. Natural enamel of hwnan teeth consists of needlelike apatite crystals measuring approx. 200 to 500 nm in length [I]. This crystalline structure serves as a model for the development of new biomaterials for dental restorations. In order to achieve the desired microstructure, it is essential to have full control of the nucleation and crystallization processes. Solid state reactions lead to the final microstructure which can be controlled by the subsequent heat treatment. The mechanism for the nucleation and growth of oxyfluoroapatite in a wollastonite glass-ceramic was described by Kokubo et al. [2]. HCTland et al. [3,4] investigated the apatite krmation in mica glass-ceratnics and apatite-leucite glass-ceramics. In this publication, the formation of the nanocryds in different glass-ceramicwill be d y z e d . It is the goal of this paper to compare the microstructure and the crystallization mechanisms of nanosized and needle-like Buoroapatite in two different glass systems SiO2-Li2O-K2O-ZnO&O-P205-F and Si02-Al203-KzO-Na20-CaO-P205-F. An important regulating variable in the crystallization of needle-like apatite in these glass-systems is nucleation. In addition, the formation of nanosized CaF2 crystallites in a third glass system will be discussed. The CaF2 nanocrystals were the precursor phase for the futber crystallizationprocesses of Ca2SioZF2 and crisbbalite. EXPERIMENTAL METHODS Glass in the Sio2-Li2o-K2U-Zno-C~O-P205-Fsystem A glass (AO) with the composition of 59.6 wto/o Si02,13.1 wf?! K20,4.2 wt?? Li20, 10.2 wt?! ZnO, 6.1 wt???CaO, 2.8 wt% P205, 1.6 wto/o AlzQ3, 2.0 wt'?! 0.4 wt??F was melted at M O O T for 2 h in a platinum crucible. The glass melt was poured directly into water in order to prevent any phase separation or nucleation processes. The quenching rate was approximately 500 Ws. Bulk samples of the glass fiit were used for the further thermal studies on the nucleation and crystallizationprocesses in the temperature range of 520 to 700T. Furthermore, samples were prepared using glass A0 in a sintering process. The glass fiit A0 was isothermally heat treated at either 650 or 700OC for 1,2 or 4 h. 2102,
236
Ceramic Nanomaterials and Nanotechnology
Tbe glass-ceramic was subsequently milled to a fine powder in an agar mill. In addition, the powder was sieved < 9Op. The final bulk q l e was achieved by subsequent sintening at 820°C for 1 minute with a heating rate of 60 Wmin. The sinkring process corresponded to the processing of a dental product in a dental laboratory. A fhrther series in this glass-system was investigated by varying the CaO and P2O5 contents. The compsition of 61.4 wt5h Si02, 13.6 wt!? K20,4.3 Wtb! Li20,10.6 wt% ZnO, 6.1 W? CaO, 3.5 wt??P2O5,O.S wt?! F was chosen as the base glass (Al). Two other glass compositions A2 and A3 with an increased content of CaO and P205 were produced. The concentration ratio of the other components was kept constant. Glass A2 consisted of 6.7 wt% CaO and 3.9 wt% P205 and glass A3 had concentrations of 8.7 wt?? CaO aad 5.1 wt% P2O5. The molar ratio of Ca/P was kept constant at 2.18. The glasses were heat treated in a two-stage procedure. The glass fiit was first nucleated at 520°C for 4 hours and subsequently heat treated at 7OOOC for lh. The resulting microstructure was examined with a scanning electron microscope (LEO DSM962, Germany). The calorimetric measurements were conducted using differential scanning calorimetsy @SC) (Netzsch Type 404, Germany) with a heating rate of 10 Wmin in a Pt-crucible and a sample weighing approx. 50 mg. The measurement was done under N2 atmosphere. The crystal phase was identified using X-ray difEaction analysis (XRD) with CuKa radiation (Bruker, Type D5005,Germany). The XRD investigations were carried out at room temperature on heat treated samples. Glass in the Si02-A1~03-K20-Na20-CaO-P2O~-F system A glass with the composition of 54.6 Wto? SiO2, 10.7 wto/o K20, 8.4 wt% NazO, 0.2 wt% Li20, 5.0 wt?! CaO, 4.0 wt'?? P2O5, 14.2 wt% 4 2 0 3 , 0.3 wt?! B2O3, 0.9 wt% ZrO2, 0.2 wto/o Ti02, 0.8 wt!!! CeO2, 0.7 wt?! F was melted at 1480°C for 100 minutes and cast into monolithic blocks. The monolithic glass samples were heat treated at 580°C for 15 minutes and at 700°C for 8 h, The microstructure was examined using a scanning electron microscope (SEM) (LEO, DSM 962, Germany). The qualitative determination of the crystal phase was conducted on samples by X-ray difbction analysis (XRD) using Cu €&radiation (Bruker D5005, Germany). The XRD investigations were carried out at room temperature. The crystalline surfice layer of the heat treated samples consisted of leucite. It was removed by grinding. The crystalline phases formed by the mechanism of volume crystallizationwere therefore investigated.
Glass in the SiO2-Me20-CaU-F system A glass with the composition of 53.0 wt% Si@, 42.0 wt% CaO, 5.0 wt?? Me0 and 8.0 wt?? F (-8% fluor equivalent oxygen) was melted at 1250°C for 1 hour and poured into preheated steel moulds. The bulk samples were slowly cooled
Ceramic Nanomaterials and Nanotechnology
237
down from 500°C to room temperature. The heat treatment was conducted at 585, 600,615 and 630°C for 1 h. Additional heat treatment was carried out at 630°C €or 2, 4 and 8 h. The microsbnrcture wiits examined using a scanning; electron microscope (SEM) (LEO, DSM 962, Germany). The qualitative determination of the crystal phase was conducted on samples by X-raydifhction analysis p) using Cu & radiation @&er D5005, Germany). The calorimetric measurement was carried out using differential Scanning cdorimetry @SC) (Netzsch Type 404, Germany) with a heating rate of 1 Wrnin in a Pt-crucible and samples weighing approx. 50 mg. The measurement was conducted under N2 atmosphere, RESULTS & DIScuSSION Glass in the Sio2-Li2O-K20-zno-CaO-P2~~-F system Varied content of CaO and P205: The differential scanning calorimetry measurements of the glasses Al, A2 and A3 (Figure 1) revealed that the exothemic peak shifted towards lower temperatures with increasing combined content of CaO and P205 (Table I).
Table I. Exothennic peak of DSC measurement in dependence on the (CaO+P205) content Glass composition (CaWP20s) (wt??) A1 9.6
A2 A3
10.6 13.8
Molar ratio Ca/P 2.18 2.18 2.18
Exothermic peak ("C) 659 625 575
Figure 1: DSC measurements of the glasses Al, A2 and A3; heating rate 1OWmin; curves are deliberately displaced along the vertical axis for better clarity.
An explanation for the temperature shift could be the role of P205 as the nucleating agent for heterogeneous nucleation. As described by Cramer von Clausbruch et al. [S] for lithium &silicate glass-ceramics, an increasing amount of P2O5 causes a shift of the first crystallization peak towards lower temperatures. The microstructure of the three different glasses Al, A2 and A3 were quite similar
238
Ceramic Nanomaterials and Nanotechnology
after a two-step heat treatment at 52OOC for 4 h and 700°C for 1 h. The nanosized crystals were homogeneously and uniformly precipitated in the volume of the sample. The crysta€line phase was detected by XRD as fluoroapatite caio(po4)&. Glasseramic A1 showed crystal sizes fiorn 30 to 60 run (Figure 2a). Glassceramics A2 and A3 exhibited nearly the same microstructure (Figure 2b and c). The crystal size was in the range of 30 to 60 run. The fluoroapatite crystals were densely precipitated and uniformly distributed. It was interesting to nofe that the microstructure formed after the heat treatment at 520°C/4 h + 700°C/1 h was not influenced by the variation of the content of CaO and P2O5.Higher concentrations of CaO and P205 neither increased the size nor the volume percentage of the fluoroapatite phase.
Figure 2a-c:Microstructure of glass Al,
A2 and A3 aRer heat treatment at 520°C/4 h +700°C/1 h. SEM (10 s, 3% €E+)
Varied isothermal heat treatment of glass AO: The glass f i t A0 was heat treated at 650°C and 7OOOC for 1 to 4 h. The glassceramic was then ground and subsequently sintered at 820°C for 1 minute to obtain dense samples for filrther characterization. DSC investigations on the base glass A0 revealed an exothermic peak at 708OC. Tg was measured at 516°C. The heat treatment at 650°C for different periods revealed that there was no crystal growth with increasing time. The diameter of the spherical crystals remained constant in a range of 30 to 60 nm. Figure 3 shows the typical microstructure for the entire
Ceramic Nanomaterials and Nanotechnology
239
series. The XRD investigation revealed that the crystalline phase was fluorapatite (Figure 4). The crystalline phase grew according to the mechanism of volume crystaltization.
Figure 3: Microstructure of glass A0 (650°C/lh + 820°/lmin). SEM (I0 s, 3% HF)
-
?-Theta scale
Figure 4: XRD pattern of glass-ceramics A0 to A3, reference pattern ICDD 150876 (fluorapatite) The crystal growth of needle-like fluoroapatik was obsented at 700°C by SEM. These results on monolithic glass-ceramics demonstrate that after short-tern thermal treatment, the fluoroapatite crystals were smaller and the number of crystals was higher than after long-term heat treatments at 700°C.One demand for the crystallization process of Ostwald ripening is fUilled with this result. The fluoroapatite crystals grew anisotropicdly in one direption to form elongated crystals. The diameter of the elongated crystals remained constant at approximately 60 nm.The length hcreased from 120 to 400 nm. The aspect ratio
240
Ceramic Nanomaterials and Nanotechnology
o f the fluomapatite crystals hcreased steadily fiom 2 to 6. Table I1 and Figure 5
give an overview of the results fiom the heat treatment at 700'C. Table II. Morphology of crystals of sample A0 isothemaly heat treated at 700°C lh
2b
Min. and max. crystal length 120-180
180-240
Min. a d m a ~ aspect . ratio
3-4
2-3
4h 240400 .
3-6
Figure 5 a-b. Microstsucture of shtered samples of glass A0 after heat treatment at a: 700°C/lh, b:70O0C/4h. SEM (10 sec, 3% HF) Glass in the Si02-~~03-K20-Na2eCaO-P205-F system: The nucleation and crystallizationbehavior was studied Using monolithic glass blocks. In this system two crystalline phases were derived. Leucite crystals grew according to the mechanism of surface crystallization. The formation of apatite, however, proceeded according to the mechanism of volume crystallization. The following results are focused on the formation of apatite crystals and were described by Haland et al. [l, 41. Liquid-liquid phase separation of the base glass constituted the nucleation process. The phase separated glass consisted of a glassy matrix and a droplet shaped gfass phase measuring 150 to 400 nm in diameter. Subsequent heat treatment of this base glass led to the primary crystal formation of NaCaPO4. The crystals were stable in the temperature range of 550 to 610°C. After heat treatment at 580°C for 25 minutes, the NaCaP04 crystals measured approx. 100 nm in diameter (Figure 6). The crystallization of nanosized apatite was observed after heat treatment at 700°C for 8 h. Figure 7 shows the very high precipitation density of fluoroapatite crystals and a yet unidentified X-phase. The maximum diameter of the spherical crystallites was measured as 100 nm.Recent
Ceramic Nanomaterials and Nanotechnology
241
double resomce NMR investigations by C h a et al. [6] cont'lrmed the formation of fluorinefke precursor phases such as NacaPo4 and a sodium calcium phosphate silicate phase. Subsequent heat treatment at l0SO"C led to the formation of needle-like fluoroapatite. The formation of needle-like fluoroapatite is the result of a series of solid state reactions. The formation of the nanosized crystals at 700°C seems to play a key role in the desired formation of needlelike fluorapatite. The mechanism of crystal growth of needle-like apatite is diffision controlled. This growth is characterized by OstwaId ripening [7]. The needle-like fluorapatite is particularly suitable for dental applications. Apatite crystals in naturalteeth have a similar morphology.
Figure 6: Microstmctweconsisting of NaCaP04 and liquid-liquid phase separation at 58OoC/15min. SEM (10 s, 3% HF)
Figure 7: Microstructure consisting of fluorapatite and an unidentified X-phm at 7OO0C/8h. SEM (10 s, 3% HF)
Glass in the Si02-Ca0-Me20-Fsystem The base glass has been used as a reactive filler for dental composites. The leaching characteristics for calcium and fluorine ions were described by Schweiger et al. [S]. The DSC exadnations revealed three exothem and one endotherm. The exothedc peaks were located at 583OC, 6 4 4 O C and 895°C. An endothermic peak occurred at 994°C (Figure 8). The main interest of this paper is the formation of the f W crystallinephase formed at 583OC. The crystalline phase was identified as CaF2 (ICDD:35-0816)by XRD. The further crystalline phases were calcium fluoride silicate 'CazSi02Fz) (ICDD 35-0002;exo peak at 664°C) and cristobalite(ICDD39-1425; ex0 peak at 895OC).
242
Ceramic Nanomaterials and Nanotechnology
Figure 8: DSC measurement of the base glass (heating rate The microstcucture of the glassceramic &er beiug heat treated at 585OC and 630°C for 1 h is j shown in Figure 9a and b. The spherical C a F 2 crystals measwed 60-b 90 nm in diameter m e a s d after the heat treatment at 585OC. The crystalline phase was derived by the mechanism of volume crystallization.At this stage, the spherical crystals showed a tendency towards agglomeration. The agglomaates measured 0.3 to 0.5 pn (Figure 9a). Increasing the temperature to 60OoC led to a more homogeneous distribution of the z?anocrystds. The CaF2 crystals were 30 to 90 nm in diameter (Figure 9b). Despite the very small size of the crystals, C a 2 was clearly detected by XRD investigations. huther increases of the tempratme to 615°C and 630°C showed no influence on the microstmctwes. CaF2 constituted the main crystalline phase. The spherical crystals measured 60 to 90 nm in diameter. J
~l*~1.c-
I
~-
F i g m 9 a-b: Micromcm of CaF2 g l a s s - c e d c after heat treatment at 600°C, and 630°C for 1 h. SEM (10 s 3% HF)
CONCLUSIONS The formation of nanophases plays a key role in the formation of the microstructure of difTerent glass-ceramics. Nanosized fluoroapatite with a spherical morphology was derived in the Si0~-Li~0-K~0-Zn0-CaO-P~05-F system. The crystatlites measured 30 to 60 nm in diameter. Subsequent heat treatment at higher temperatures lead to the formation of nde-like fluoroapatite with a length of 0.4 prn and an aspect ratio of up to 6. The initial spherical
Ceramic Nanomaterials and Nanotechnology
243
naflocrystalsdid not grow in diameter with increasing time of the heat treatment at 630°C. The precipitation of needle-like apatite in the Si02-A&-K2O-Na20C~O-P~OS-F system was nucleated by a heterogeneous reaction of primary crystals of NaCaPO4 m&or an intetface reaction of a glass droplet-glass matrix. The diffusion-controlled growth of needle-like apatite was characterized by Ostwald ripening. In the SiO2-CaO-MQO-F system, the precipitation of spherulitic CaF2 was achieved after heat treatment in the range of 585 to 630°C. The crystallites measured 30 to 90 nm in diameter.
REFERENCES 'W. Hliland, V. Rheinberger, S. Wegner, M. Frank, "Needle-likeapatite-leucite glass-ceramics as a base material for the veneering of metal restorations in dentistry", J. Mat. Sci.: Mat. Med., 11 11-17 (2000). 2T. Kohbo, "AW GIass-Ceramic: Processing and Properties"; pp. 75-88 in An Introduction to Bioceramics, Edited by L.L. Hench and 3. Wilson. World Scientific, Singapore, 1993. 3W. Heland, W.Gistz, G. Cad, W. Vogel, l'Micros~ctureof Mica GlassCeramics and Interface Reactions between Mica Glass-Ceramicsand Bone': Cells Mater., 2 105-112 (1992). 4W.Hziiand, V. Rheinbrger, M. Frank, "Mechanism of Nucleation and Controlled Crystallization of Needle-like Apatite in Glass-Ceramics of the SiO2Al20&20-CaO-P20~ System",J Non-Cys. Sol.,253 170-177 (1999). 'S. Cramer von Clawbruch, M. Schweigeq W.HGland, V. Rhehherger, "The Effect of P2O5 on the Crystallization and Microstructure of Glass-Ceramicsin the SiO2-Li20-K2O-ZnO-P205system", J. NOB-Cryst.Sol., 263&264 388-394 (2000). 6J.C,C Chan, R Ohnwrge, K. Meise-Gresch, H. Eckert, W. H6land, V. Rheinberger, "Apatite Crystallization in an Aluminosilicate Glass Matrix: Mechanistic Studies by X-ray Powder DifEaction, Thermal Analysis, and Multinucleat. Solid-state NhrlR Spectrmcopyl', Chem, Mater., 13 4 198-4206 (2001). 'R. MWer, L.A. Abu-Hilal, S. Reinsch, W. Hcilmd, "Coarsening of Needleshaped Apatite Crystals in Si02-A1203-Na20-K20-CaO-P205-F Glass",J. Mater. Sci., 34 65-69 (1999) 8M.Schweiger, P. &thing, L. Schlapbach, W. Holand, V. Rheinberger, "Thermal and Chemical Properties of a Glass in the SiOz-CaO-F System for Dental Applications", . I Therm. Anai'y. CaZory., 60 1009-1018 (2000).
244
Ceramic Nanomaterials and Nanotechnology
UNIAXIAL PLASTIC DEFORMATION IN THEZTRCONIA-BASED NANOCRYSTALLN CERAMICS CONTAINING A SILICATE GLASS
R Chaim and R Rammoorthy Department of Materials Engineering Technion-Tsrael Institute of Technology Haifa 32000, Israel
A. Goldstein Israel Ceramic and Silicate Institute Technion City Haifa 32000, Israel
I. Eldror and A. Gurman Metallurgy Group, Engineering Center Israel Aircraft Industries Ben-Gurion International Airport 70100, Israel ABSTRACT Nanocrystalline yttria-stabilized tetragonal zirconia polycrystal (nc-Y-TZP) powders coated with silica-based glasses were cold isostatically pressed and sintered near to the full density (98 to 99%). Two glasses with different compositions were used: 93 Si02 - 1 Na20 6 SrO (mole%) (designated as SNS glass) and 58 SiO;! 29 A1203 - 13 SrO (designated as SAS glass). Uniaxial compression tests of the pure (glass-free) nc-Y-TZP samples yielded strain rates as high as 2 ~ 1 O s-' - ~ under 60 MPa at 1300°C. Comparable strain rates were measured in the SNS glass-containing samples, with the maximum of 3-104 s-l at 1300°C under the stress of 80 MPa (5~01%SNS glass content). Compression tests under 100 MPa exhibited relatively high strain rates of 5 . 10-4and lW4at 1300°C and 1200"C, respectively, in the 15~01%SAS glass samples. The strain rates measured in the SAS glass-containing samples were achieved at temperatures lower by 100°C compared to the similar strain rates in the glass-free and SNS glass-containing samples. Microstructure of the deformed samples was similar to those before deformation, within which the ultrafine and equiaxed character of the grains were preserved. Clear evidence for cooperative grain boundary sliding was observed in the SAS glass-containing samples.
-
-
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
245
INTRODUCTION Superplastic deformation of ceramics is a well known phenomena and mainly has been related to their ultra-fine grain size.'-3 However, as was discussed by Dominguez et al., 475 the expected increase by several orders of magnitude in the strain rate due to the grain size refinement was not observed in the ultra-fine glass-free ceramics. On the other hand, based on the microstructural observations, there is a general agreement that the grain boundary sliding acts as the main mechanism for the superplastic deformation in these Therefore, engineering of the grain boundary properties such as incorporation of the grain boundary glassy phase with controlled properties should be appropriate for promoting the superplastic deformation characteristics. The present work describes the superplastic deformation behavior of nanocrystalline Y-TZP ceramics that contain varying amounts of two different glass compositions as grain boundary glassy phases.
EXPERIMENTAL PROCEDURE Fabrication of the Samples Commercial nanocrystalline Y-TZPpowders (Tioxide) were coated with two different silica-based glasses through the sol-gel technique. Detailed description of the coating process followed by sintering and the resultant microstructural evolution were described elswhere.I2.l 3 The glass compositions were determined by x-ray fluorescence spectroscopy to be 93 Si02 - 1 Na2O - 6 SrO (mol %) (designated as SNS glass) and 58 Si02 - 29 A1203 - 13 SrO (designated as SAS glass). Appropriate amounts of the coating solution were applied in order to reach the glass contents of 5,lO and 15~01%in the final compacts. The nc-Y-TZP powders with and without the glass were compacted into cylindrical pellets of 15 m m xl5 mm, followed by cold isostatic pressing at 250 MPa. The 55% dense green compacts were sintered at 1400°C for l h to near full density as summarized in Table I. Table I: Characteristicsof the nc-Y-TZP specimens sintered at 14OOOC for 1 h Glass Content* Relative Phase Content [%I** Mean Grain [VOlYO] Density [Yo] m Size [m] ---0 98 100 --196 83 17 140 5 99 69 31 110 10 98 15 99 74 26 100 *SNS glass;
246
** 1and
refer to tetragonal and monoclinic phases, respectively.
Ceramic Nanomaterials and Nanotechnology
Rectangular specimens of 3 x 3 x 4 mm3 were cut and used for uniaxial compression tests in the constant load regime. The compression test set-up was heated during 3 h to the deformation temperature. Then, the load was increased in steps at the constant temperature. The specimen contraction was recorded within the accuracy of k 2pm. Compression tests were performed using the homemade machine and within the temperature range 115OOC to 13OOOC. The average test duration was about 2.6 h. According to the high temperature deformation tests (Table 11) selected specimens were chosen for microstructural characterization prior to and after the deformation, using scanning electron microscopy (SEM-XL30) and transmission electron microscopy (TEM-2000FX); the specimens were prepared by the conventional techniques. Mechanical Tests Application of constant load in uniaxial compression results in a decrease of stress during the plastic deformation. Thus, the superplastic forming (SPF) strain, stress and strain rate are given by:
;=($ where: E and CT
-1 (3)
- the true strain and stress, respectively.
L. ,and h - the time dependent strain rate and height, respectively.
go, ho and 00 - the initial strain rate, height and stress, respectively.
m - the SPF parameter equals to the reciprocal of the stress exponent - n. The solution of this set of equations gives: h = (I + where: t - time.
ho
2)
Ceramic Nanomaterials and Nanotechnology
-m
(4)
247
At the begining of each load increase step, the stress is calculated by dividing the load to the actual area, i.e.:
-
-
where: P load, and SO the initial area under the stress. Simultaneously, the strain rate is calculated from the slop of the plot: = -$In(
t)]
RESULTS Deformation Tests The strain rate versus stress at different temperatures in pure (glass-free) ncY-TZP samples is shown in Fig.la. The strain rate increases with temperature and ranges fiom 510a to above 104 s-'. The data points were fitted in the logarithmic scale to the linear lines, the slops of which (n values, where n=l/m) were determined to be about 1.7. However, at 1300°C and stresses below 50 MPa the slope was about 4.0. The significance of these slops is for determination of the deformation mechanisms. The strain rate-stress diagram for the 15~01%SNS glass-containing samples (Fig. 1b) revealed almost similar strain rate values (although slightly higher) with respect to temperature and stress, compared to the pure nc-Y-TZP (Fig.la). The slop values were 1.6 and 4.0 at high and low stress levels, respectively. On the other hand, higher strain rates were recorded at the respective temperatures and stress levels for the 15~01%SAS glass-containing samples, as shown in Fig.lc. Comparison of the last two diagrams (Fig.lb and Fig.lc) demonstrates the effect of the glass composition on the strain rate. In the last system, high strain rates (-10-4 s-') were measured at relatively lower temperatures (I 200°C). In addition, the slops of the linear curves were 0.6 and 1.7 at the lower and higher temperatures, respectively. The effect of the glass content (~01%)on the strain rate at 1300°C for the SNS glass-containing samples are shown in F i g 2 Samples with higher glass contents (10 and 15~01%)revealed similar behavior while their n values changed fiom 1.6 at higher stresses to 4 at the stresses below 50 MPa. Nevertheless, the n value of the sample with 5~01%SNS glass was constant (n = 2.0) over the applied stress range.
-
248
Ceramic Nanomaterials and Nanotechnology
Table 11: Uniaxial Deformation test conditions and results #
Composition Deformation Deformation TotalStrain MeanGrain Remarks pal%] Temp.[oC] Duration[h] [%] Size# [nm]
1
nc-Y-TZP
2 3
-'-
4
5 6
-'-
-.-.-
1150 1200 1250 1250 1250 1300
3.7 3.2 1.3 1.5
2.0 2.8
84 75 71 55 67 45
196 196 196 196 196 196
**
*
* TEM observation; ** SEM observation. ## before deformation. The effect of temperature on the strain rate in the 10~01%SAS glasscontaining samples are shown in Fig.3. Compared to the 15~01%SAS glass samples (Fig.lc), decrease in the glass content was found to lower the corresponding strain rate values, but at the same time caused also to the change in the deformation mechanism. In this respect, the n values at 1300°C and 1200°C increased to 3.5 and 2.0, respectively (compared to 1.7 and 0.6 in 15~01%glass sample).
Ceramic Nanomaterials and Nanotechnology
249
c
1 0')
I
I
n4.6
n=4
3i 10-=
.
(a) Pun nc-Y-TZP
*
100
1 0
@) nc-Y-TZP+lS% SNS Qhss
10''
I
I
10
100
Stress [MPa] 1 0-
Stress [MPal
0 ,
i
n
C l 0' b0.6
f
0
(c) nc-Y-TZP+lB% SAS glass 1 010
100
Stress [MPa]
Fig.1: Strain rate versus stress and temperature dependence of the (a) glass-fiee nc-Y-TZP and alloys containing (b) 15~01% SNS glass and (c) 15~01% SAS glass. t
11.11 +1200 +1250
11.12.0
1 0'':
n51.6
1 0'
1300°C
: nc-YTZP+lO vol%SAS QkSS
10-6L-
10
I
100
Stress [MPa]
Fig. 2: Effect of the SNS glass content (~01%)on the strain rate versus stress at 130OOC.
1
10-4
I
100
I 0
Stress
I
[MPa]
Fig. 3: Effect of the deformation temperature on the strain rate versus stress in the nc-YTZP alloys with 10~01% SAS glass.
Arrhenius-type plots of the corresponding strain rates (at constant stress) versus reciprocal temperature exhibited linear line in the regime where n 1.6 to 2.0 as shown in Fig.4. Activation energyies of 297k15 FJ/mol] and 275f15 [kJ/mol]
-
250
Ceramic Nanomaterials and Nanotechnology
were determined for the glass-free nc-Y-TZP and the glass-containing samples, respectively. U
'O"
4
I
0.76
.
0.78
.
.
(
.
.
0.8
lfr
.
,
.
0.02
..
a . . . , . . .
0.04
0.66
0.06
oc-lxlo - s J Fig. 4: Activation energies for the plastic deformation at the regime where n 1.6 to 1.7 were determined as 297k15 p/mol] and 275+15 w/mol] in the pure and the glass-containingsamples, respectively. [
-
THE MCROSTRUCURE SEM observations The microstructure of selected samples prior to and after the deformation were characterized, as marked in Table 11. SEM observations were performed on two type of surfaces: a) the surfaces subjected to the compressive stress that were in contact with the compressing rod. b) free surfaces that were perpendicular to the swrfaces in (a) and subjected to the tensile and shear stresses. The following series of the SEM images represent the microstructure at these two type of surfaces. Low magnification image of the 15~01%SAS glass-containing sample surface in compression deformed at 1300°C is shown in Fig.5a and consists of homogeneous microcracks. The microcracks also originated from existing porosities at the surface. SEM observation at higher magnifications (Fig.5b) revealed that these microcracks were formed along the grain boundaries of the grain clusters. Similar microstructure, albeit with far less crack density, was observed in samples deformed at 1200°C (Fig.6a). However, the surfaces subjected to tensilehhear stresses revealed different microstructure. These surfaces at 1300°C appeared to be homogeneously corrugated, i.e. the surface roughness was due to many single grains as well as grain clusters (Fig.5~and 5d). No significant cavitation, crack growth or tearing effects were observed. The only few cavities were present and comparable in their size to the grain diameter (Fig.5d). In contrast, the similar surfaces at 1200°C exhibited nonhomogeneous microstructure with significant tearing (Fig.6b). At higher magnifications the absence of homogeneous deformation as well as locallized crack growth were clearly visible (Fig.6~).
Ceramic Nanomaterials and Nanotechnology
25 1
Fig. 5 : SEM images from the surfaces (a) and (b) under compression, and (c) and (d) under tensiIe/shear stresses. 15% SAS hss-cnntnininp nc-Y-TZP defnrmed nt 13nl)OCI. The rnmnressinn directions were shnwn hv arrows.
Fig. 6: SEM images fiom the s u r f “ (a) under compression, and (b) and (c) under tensiledshearstresses. 15% SAS glass-containing nc-Y-TZP deformed at 1200°C.The compression axis in (a) is perpendicular to the imagewhile in (b) 1 and ( c ) is alom the vertical direction.
SEM images from the compressive and tensile surfaces of the 10~01%SAS glasscontaining sample deformed at 1250°C are shown in Fig.7a and 7b, respectively. The microstructure at the as-compressed surfaces was fairly flat and contained traces of the glassy phase that apparently detached fiom the contact surface with the compressing rod (Fig.7a). No pore growth or continuous micro-cracks were observed. On the other hand, the tensilehhear surfaces (Fig.7b) exhibited two types of regions with different microstructures. First, pore growth and crack opening were observed at the regions buckled upwards from the surface (marked “up” in Fig.7b). These regions (Fig.7~)most possibly subjected to high tensile component, perpendicular to the compression axis. The neighbor regions were buckled inwards fiom the surface (marked “in” in Fig.irb), and most possibly were subjected to high shear component, in angle to the compression axis. The last regions exhibited a unique microstructure of grain clusters that slipped over each other (Fig.7d). To our best knowledge, this microstructural observation is the first evidence in the published literature that confirms the validity of the cooperative in ceramics. High magnification image of grain boundary sliding mechani~m’~ such region (Fig.7e) clearly revealed that these micrometer size clusters contain nanometer-size grains. This finding is in agreement with previous microstructural
Ceramic Nanomaterials and Nanotechnology
253
254
Ceramic Nanomaterials and Nanotechnology
Fig.7e: Cooperative sliding of micrometer-size polyhedral-shape clusters, each composed of hundreds of ultrafine grains. The slip at the cluster boundaries are accomodated by slip of single grains.
observation of the sintered where the glass did not promote grain growth. Comparison of the microstructures developed in the last three samples indicates that the excess glass may be detrimental in providing ‘weak’ sites for cavity nucleation under the tensile stresses. Consequently, the volume fraction of the glass needs to be optimized. SEM images from the surfaces under compression in the 15101% SNS glasscontaining sample deformed at 1250OC are shown in Fig.8. Low magnification image (Fig.8a) revealed microcrack pattern similar to that found in the 15~01% SAS counterpart sample (FigSa). Images at higher magnifications showed that these microcracks were developed and propagated along the grain boundaries (Fig.8b). Observation at the tensilehhear surfaces (Fig.&) revealed a complex microstructure that was composed of homogeneous microcracks (perpendicular to the tensile stress direction) and shear bands (parallel to the tensile stress and perpendicular to the surface). At higher magnifications (Fig.8d) the shear bands were consisted of the nanocrystalline grains. Nevertheless, this microstructure was not typical for the cooperative grain boundary sliding. Appropriate SEM images from the pure (glass-free) nc-Y-TZP samples deformed at 125OOC (as a reference sample) are shown in Fig.9. Homogeneous
Ceramic Nanomaterials and Nanotechnology
255
Fig. 8: SEM images fiom the sudaces (a) and (b) under compression, and (c) and (d) under tensile/ shear stresses. 15% SNS glass-containing nc-Y-TZP deformed at 125OOC. The compression axis in (a) and (b) is perpendicular to the image while in (c) and (d) is along the vertical direction.
microcracking was also observed at the cornpressive surfaces of this sample (Fig.9a). However, the tensilekhear surfaces contained traces of parallel straight lines, typical for the slip bands (Fig.9b). At higher magnifications, these bands were composed of nanocrystalline grains (Fig.9c) apparently slipped over each other (no cooperative grain boundary sliding was observed). This microstructure represents plastic deformation by grain boundary sliding mechanism.
TEM observations Typical TEM images from the as-sintered samples prior to the deformation showed equiaxed grains, either with faceted grain boundaries (Fig.lOa) (in glassfiee nc-Y-TZP) or spherical grains sorrounded by the glassy layer (Fig.lOb) (in glass-containing Y-TZP's). Increase in the glass content caused to increase in the roundness of the zirconia grains as was reported previously.'2 The microstructure of the pure, glass-free nc-Y-TZP as well as that with 15~01%SNS glass after more than 40% plastic deformation are shown in Fig. 11a and 11b, respectively. Generally, the microstructure of the deformed specimens
256
Ceramic Nanomaterials and Nanotechnology
Fig. 9: SEM images from the surfaces (a) under compression, and (b) and (c) under tensile/shear stresses. Glass-free nc-Y-TZP deformed at 1250°C. The compression axis in (a) is perpendicularto the image while in (b) apd (c) is along the vertical direction.
Ceramic Nanomaterials and Nanotechnology
257
Fig. 10: TEM images from the as-sintered samples prior to the high temperature plastic deformation. (a) Faceted equiaxed grains in the glass-free nc-Y-TZP. (b) Rounded grains in the 15~01%SNS glass-containing sample.
Fig. 11: TEM images fiom the samples after more than 40%plastic deformation. (a) glass-freencY-TZP. (b) 15% SNS glass-containing sample. The microstructures are similar to those prior to the deformationas in Fig.10.
were identical to those prior to the deformation with respect to the grain size and morphology. The main difference was formation of twins which may occur for stress relaxation or due to the martensitic phase transformation of the tetragonal grains to the monoclinic symmetry (during the cooling). No dislocations were resolvable due to the heavily twinned nature of the monoclinic grains. On the other hand, in the twin-free tetragonal grains, also no dislocation activity was found, irrespective of the sample composition. In addition, the glass-containing samples exhibited pore growth and the grains were rounded (Fig.1 lb). These
258
Ceramic Nanomaterials and Nanotechnology
findings do not add more information to the understanding of the deformation mechanism but confirm the grain boundary sliding to be the dominant deformation mechanism in these alloys.
DISCUSSION Superplastic deformation was reported by many investigators for submicron grain size Y-TZP’s having different impurity Comparison of the present data to the literature shows that the n values found here are comparable to the range of the n values (1.3 to 3.8) reported for the submicron grain size Y-TZP systems! A variety of strain-rate controlling mechanisms such as solutionreprecipitation, interface reaction controlled creep, grain boundary sliding accomodated by the intragranular slip and others were postulated to be operative. Nevertheless, it was shown by Dominguez et al? that single deformation mechanism could be used to explain the plastic deformation in submicron Y-TZP ceramics, assuming a threshold stress. In this respect, low purity samples exhibit n = 2 whereas higher purity samples exhibit transition from n E 2 to values higher than 3 while decreasing the stress. It should be noted that the exponent n = 2 is often reported for superplastic deformation in metals. Therefore, the n values above 1.5 in the present research could be related to the diffusive gain boundary sliding mechanism26. In this respect, the Gifkins’ mantle model is the most appropriate for description of the deformation mechanism in the present samples, as originally was suggested by Okamoto et a1?8 This model is in agreement with the morphological changes that were observed at the tensilehhear surfaces of the deformed samples. The value of the exponent n was found to increase to high values (between 3 to 4) at the transition stress with lowering the stress. Nevertheless, this transition occured in the pure as well as the SNS glass-containingnc-Y-TZP samples only at 1300°C at stresses of about 50 MPa. In the SAS glass-containing samples, this transition occured already at 1200OC. On the other hand, as the n values tend to decrease towards 1.0, one can expect for a linear dependence of the strain rate with stress. This type of behavior is consistent with the grain boundary sliding aided by either Newtonian viscous flow in the glass-containing samples or by diffusional creep2’ in the pure submicron size Y-TZP. The lowest n value of 0.6 which was observed in the 15 vol% SAS glasscontaining samples deformed at lower temperatures may be explained by the nonNewtonian viscous flow of the sample, by which the viscosity dependence on the shear stress is parabolic. However, verification of this postulation neccessitates characterizationof the viscosity of the grain boundary glassy phase. The published activation energies for superplastic deformation in the Y-TZP ceramics span over the wide range of values, i.e. from 360 to 720 [kJ/m01]~’~~.
Ceramic Nanomaterials and Nanotechnology
259
-
Activation energy of 500 [W/mol] is often reported for creep experiments of coarse-grained Y-TZP". The activation energy in the glass-free and the SNS glass-containing samples is very close to that for the grain boundary diffision of cations (290 @J/mol] for Y 3+ and 3 10 [kJ/mol] for Zr 4+)30,3'. The corresponding activation energies for the bulk diffusion of the same cations were measured, through the diffusion ex eriments as 460 [kJ/mol] and 5 10 [kJ/mol] for the Y3' and Zr", respectively 32y3. In addition, the activation energy for viscous flow in the silica-based glasses (i.e. diffision in viscous silica) is about 190 [kJ/mol]. Comparison of the measured activation energy (297 k 15 and 275 f 15 [kJ/mol]) to the data fiom the literature indicates that the plastic deformation mechanisms are controlled primarily by the grain boundary diffision processes. Nevertheless, in pure glass-free nc-Y-TZP sampfes the deformation rate is controlled by diffbsion of either or Zr'4 cations along the grain boundaries. On the other hand, appropriate calculations for the impure nc-Y-TZP ceramics have shown the interface reaction controlled processes (at the glass/crystal interfaces) to be the rate controlling rather than the diffusion processes within the glass." Addition of appropriate glass, i.e. SAS glass in the present investigation has led to the change in the deformation mechanism as was evidenced at lower deformation temperatures. In this respect, the mutual solubility between the lassy phase and the Y-TZP grains may change significantly at high temperatures! Apparently, the SAS glass has lower solubility (due to the A1203 content) than the SNS glass in the zirconia grains. Therefore, one may expect for the superplastic strain rate to depend on the solubility of the glass at the deformation temperature. This effect may change, in turn,the superplastic deformation mechanism as was observed here.
-
v3
SUMMARY AND CONCLUSIONS Nanocrystalline Y-TZP ceramics containing 5 to 151101% of two different glass compositions were uniaxially deformed at different temperature-stress conditions. The dependence of the strain rate on temperature, stress, glass content and its composition were determined. The highest strain rates were observed for the SAS glass-containing samples. The deformed microstructures clearly confirmed the cooperative grain boundary sliding and the grain boundary sliding mechanisms to be operative in the glass-containing and the glass-free nc-Y-TZP samples, respectively. ACKNOWLEDGMENTS The authors thank the Israel Ministry of Science, Culture and Sports for supporting this research through the infrastructure grant #1090-1-98.
260
Ceramic Nanomaterials and Nanotechnology
REFERENCES F. Wakai, S. Skaguchi, and Y.Matsuno, “Superplasticity of yttria-stabilized tetra onal ZrO2 polycrystals,” Ahanced Ceramic Materials, 1,259-263 (1986). ‘T. G. Nieh, C. M. McNally, and J. Wadsworth, “Superplastic behavior of a yttria-stabilized tetragonal zirconia polycrystal, Scripta metallurgica,22, 1 2971300 (1988). T. G. Nieh, J. Wadsworth, and 0. D. Sherby, pp. 91-99 in “Superplasticity in metals and ceramics”,Cambridge Solid State Science Series, Cambridge, 1997. F. Gutierrez-Mora, A. Dominguez-Rodriguez, M. Jimenez-Melendo, R. Chaim, B. G. Ravi, and J. L. Routbort, “Current understanding of superplastic deformation of Y-TZPand its application to joining,” Materials Science and EngineeringA, 302 154-161 (2001). F. Gutierrez-Mora, A. Dominguez-Rodriguez, M. Jimenez-Melendo, R. Chaim, and M. Hefetz, “Creep of Nanocrystalline Y-SZP Ceramics,” NanoStructured Materials, 11,53 1-537 (1999). 6 M. Jimenez-Melendo, A. Dominguez-Rodriguez, and A. Bravo-Leon, “Superplastic flow of fine-grained yttria-stabilized zirconia polycrystals: constitutive equation and deformation mechanisms,” Journal of the American CeramicSociety, 81,2761-2776 (1998). 7 Chokshi A. H., “Superplasticity in fine-grained ceramic and ceramic composites: current understanding and future prospects,” Materials Science and Engineering A, 166,119- 133 (1993). 8 T. G.Langdon, “The characteristics of superplastic-like flow in ceramics,’’ pp. 251-268 in Plastic Deformation of Ceramics, Edited by R. C. Bradt, C. A. Brooks, and J. L. Routbort, Plenum Press, New York, 1995. 9 G. M. Pharr, and M. F. Ashby, “On creep enhanced by a liquid phase,” Acta metallurgica,31, 129-138 (1 983). l0 J. R. Dryden, D.Kucerovsky, D. S. Wilkinson, and D. F. Watt, “Creep deformation due to a viscous grain boundary phase,” Acta rnetaZlurgica,37,20072015 (1989). R. Chaim, “Plastic deformation in impure nanocrystalline ceramics,” Journal of Materials Research, 14,2508-25 17 (1999). l2 R. Ramamoorthy, and R. Chaim, “Sintering and microstructure of glasscoated nanocrystalline Y-TZPpowder: Effect of the glass,” Journal of Materials Research, 16 (1) 296-302 (2001). 13 R. Ramamoorthy, and R. Chaim, “Microstructural evolution in the nanocrystalline Y-TZPcontaining a silicate glass,” Journal of the European CeramicSociety, 21,2895-2902 (2001). l4 M. G. Zelin, “Cooperative grain boundary sliding in materials with nonuniform microstructure,” Journal of Materials Science Letters, 15, 2068-2070 (1 996).
’
’’
Ceramic Nanomaterials and Nanotechnology
261
C-M. J. Hwang, and I-W. Chen, “Effect of a liquid phase on superplasticity of 2-mol%-Y~03-stabilized tetragonal zirconia polycrystals,” Journal of the American Ceramic Society. 73,1626-1 632 (1990). l6 Y-I. Yoshizawa, and T. Sakuma, “Role of grain-boundary glass phase on the superplastic deformation of tetragonal zirconia polycrystal, Journal of the American CeramicSociety, 73,3069-3073 (1990). 17 M. Gust, G. Goo, J. Wolfenstine, and M. L. Mecartney, “Influence of amorphous grain boundary phases on the superplastic behavior of 3-mol%-yttriastabilized tetragonal zirconia polycrystals (3Y-TZP),” Journal of the American CeramicSociety, 76,168 1-1690 (1 993). l 8 T. Rouxel, and F. Wakai, “The brittle to ductile transition in a Si3NdSiC composite with a glassy grain boundary phase,” Acta metdlurgica and rnaterialia, 41,3203-3213 (1993). J. R. Seidensticker, and M. J. Mayo, “Superplasticity in 3Y-TZP Doped with Small Amounts of Copper Oxide”, Scripta metallurgica and materialia, 31, 1749-1754 (1994). 2o K. Kajihara, Y. Yoshizawa, and T. Sakuma, “The enhancement of superplastic flow in tetragonal zirconia polycrystals with SiO2-doping,” Acta metallurgicaand materialia, 43, 1235-1242 (1995). 21 J. L. Shi, G. Q. Zhu, and T. R. Lai, “Compressive deformation behaviour of superplastic Y-TZP based ceramics: Role of grain boundary phases,” Journal of the European Ceramic Society, 17,851-858 (1997). 22 S. Tekeii, and T. 3. Davies, “Influence of a transition metal oxide (CuO)on the superplastic behaviour of 8 mo1% yttria-stabilised cubic zirconia polycrystal (SV-CSZ),” Journal of Materials Science, 33,3267-3273 (1 998). 23 P. Thavorniti, Y. Lkuhara, and T. Sakuma, “Microstructural characterization of superplastic SiOz-doped TZP with a small amount of oxide addition,” Journal of the American Ceramic Society, 81,2927-2932 (1998). 24 L. Chen, T. Rouxel, R. Chaim, H. Vesteghem, and D. Sherman, “Superplasticity and creep in monociinic and Y-PSZnano-grained zirconia,” Materials Science Forum,243-245,245-250 (1997). 25 A. Bravo-Leon, M. Jimenez-Melendo, A. Dominguez-Rodriguez A., and A. H. Chokshi, “The Role of a Threshold Stress in the Superplastic Deformation of Fine-Grained Yttria-Stabilized Zirconia Polycrystals,~’Scripta materialia, 34 (7) 1155-60 (1996). 26 S. P. S. Badwall, F. T. Ciacchi, M. V. Swain, and V. Zelizko, “Creep Deformation and the Grain-Boundary Resistivity of Tetragonal Zirconia Polycrystalline Materials,” Journal of the American Ceramic Society, 73 (8) 2505507 (1 990). 27 R. C. Gifkins, “Grain-Boundary Sliding and its Accomodation During Creep and Superplasticity,”Metal/urgicalTransactions 7A (8) 1225-32 (1 976).
262
Ceramic Nanomaterials and Nanotechnology
2* Y. Okamoto, J. Ieuji, Y. Yamada, H. Hayashi, and T. Nishikawa, “Creep Deformation of Yttria-Stabilized tetragonal Zirconia (Y-TZP),” pp.565-7 1 in Advances in Ceramics, Vol. 24, Edited by S. Somiya, N. Yamamoto, and H. Yan ida, The ACerS, Westerville, Ohio, 1988. ‘P. E. Evans, “Creep in Yttria- and Scandia-Stabilized Zirconia,” Journal of the American Ceramic Society, 53 (7) 365-69 (1970). 30 Y. Sakka, Y. Oishi, and K. Ando, “Zr-Hf Interdiffusion in Polycrystalline Y203- (Zr+Hf) 02,” Journal of Materials Science, 17,3 101-105 (1 982). 31 Y. Oishi, K. Ando, and Y. Sakka, “Lattice and Grain-Boundary Diffision Coefficients of Cations in Stabilized Zirconias,” pp. 208-219 in Advances in Ceramic, vol. 7, Edited by M. F. Yang and A. H. Heuer, The ACerS, Columbus, Ohio, 1983. 32 H. Solmon, J. Chamat, C. Dotin, and C. Monty, ccZr,Y and 0 SelfDiffision in Yx02-m(x=O.17),”pp. 175- 184 in Ceramic Transactions, Vol. 24, Point Defects and Related Properties of Ceramics,Edited by T. 0. Mason and J. L. Routbort, The ACerS, Westerville, Ohio, 1991. 33 F.R. Chien, and A. H. Heuer, “Lattice Diffusion Kinetics in Y203Stabilized Cubic-ZrO2 Single Crystals: A dislocation Loop Annealing Study,” Philosophical Magazine A 73 (3) 681-699 (1996). 34 Y. Ikuhara, Y. Nagai, T. Yamamoto, and T. Sakuma, “High-temperature behavior of SiOz at grain boundaries in TZP,”Inter$ace Science, 7,77-84 (1999).
Ceramic Nanomaterials and Nanotechnology
263
This page intentionally left blank
TRANSPARENT GALLATE SPINEL GLASS-CERAMICS L.R. Pinckney, B.N. Samson, G.H. Beall, J. Wang, and N.F. Borrelli Coming Incorporated Science and Technology Division SP-FR-5 Coming, NY 14831 ABSTRACT Transparent, nanocrystalline glass-ceramics based on the aluminogallate spinel crystals Li(Ga,Al)sOg and “.y-(Ga,Al)203”can be obtained in the Si02Ga203-A1203-K20-Na20-Li20 system. The glass-ceramics are self-nucleating via amorphous phase separation. Their microstructuresconsist of 10-20 nrn spinel crystals dispersed throughout a stable aluminosilicateglass, with total crystallinity ranging fiom 5-25%. Because gallium-rich spinel crystals contain large Ga3’ ions in both their octahedral and tetrahedral sites, these sites provide a lower crystal field strength environment than that obtained with conventional aluminate spinels. Thus, when doped with transition metal ions such as Ni2+,Co2+,and Cr3+,gallate spinels can yield fluorescence spectra that are significantly shifted toward the infrared compared with those of conventional spinels. Glass-ceramic fibers based on Ni2+-dopedaluminogallate spinel have demonstrated strong and broad fluorescence with peak wavelengths of 1200-1250 nm and emission lifetimes of >300 ps. INTRODUCTION Spinels have cubic crystal structures based on approximately close-packed oxygen atoms. They have the general chemical formula AB204, where A is a tetrahedrally-coordinated,typically divalent metal and B is an octahedrallycoordinated, usually trivalent metal. Spinels with divalent ions in the tetrahedral sites and trivalent ions in the octahedral sites, such as MgGa204and ZnGa204, are known as “normal”spinels. When trivalent ions are split between the octahedral and tetrahedral sites and the divalent ions are in the octahedra, they are known as “inverse” spinels. Most natural spinels are intermediate between the two extremes.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
265
Spinels high in gallium include the cubic gallate end member known as yGa203, the lithium gallate spinel LiGas08, and high-gallium solid solutions of these phases, particularly toward aluminate spinels. y-Ga2O3 has been shown to be isostructural with the spinel forms of aluminum and iron oxides: y-Ala03and y-Fe203.I The y-Ga2O3 structure can be described as a spinel with some vacant cation sites and is sometimes called a "defect1'spinel? The unit cell contains 32 close-packed oxygen ions arranged exactly as in spinel, with 2 1 1/3 gallium cations distributed randomly over the 24 normally-occupied sites. Charge balance is maintained by these vacancies. Its lattice constant is 0.822 nm. LiGa508 has an inverse spinel-like structure and can be considered a derivative of "Mg2Ga40i' (MgGa204), where Li+and Ga3' replace two Mg2+ ions.3 One half of the Ga3' ions occupy tetrahedral sites, while the Li+ ions, together with the other half of the Ga3' ions, occupy the octahedral sites. The LiGasOs phase is isostructural with LiAlsOs and LiFesOs. Its lattice constant is 0.833 nm. The X-ray diffraction powder patterns for y-Ga2O3 and LiGa508 are virtually identical. The optical properties of both aluminate and gallate spinel crystals, when doped with various transition metal ions, have been described in the Various potential applications have been suggested, including photoluminescent phosphors as well as tunable solid state lasers and saturable absorbers for visible and near-infrared wavelength^.^-'^ These studies have been carried out on single crystals or polycrystalline powders. A number of workers have described transparent, transition metal-doped spinel glass-ceramics, including C$+-doped ZnAl2O4 and LiGa508 glass-ceramics 13-15 and a Co2+-dopedZnA1204spinel glass-ceramic.16Tanaka et al. compared the optical properties of Co2+-dopedaluminate (ZnAl204) and gallate (LiGasOg) spinel glass-ceramics.I' The latter material comprised a bulk composition with molar ratio of 10 LizO * 20 Ga203-70 Si02 * 0.1 COO and a microstructure of LiGa5Og crystals dispersed in a silicate glass. EXPERIMENTAL PROCEDURES Glass-Ceramic Preparation Glass batches were prepared by mixing appropriate amounts of low-iron sand, alumina, reagent grade oxides of gallium, magnesium, lanthanum, nickel, and cobalt, and carbonates of potassium, lithium, and sodium. The glasses were melted in platinum crucibles for 16 hours at 1600°C and then cast as free patties or pressed with a steel plate to a thickness of -3 mm. The glasses were annealed at 550"-650"C for l h and subsequently heat treated at temperatures of 700-900°C for 1-2h to promote crystallization of the spinel phase. Glasses with more than 45 wt% Ga203 typically crystallize to transparent glass-ceramics during the
266
Ceramic Nanomaterials and Nanotechnology
annealing step, making further heat treatment undesirable. After visual examination of the sample, the crystalline phases were identified using X-ray powder diffraction (Phillips XRG 3 100, Cu Ka radiation). Images of the crystalline microstructure were obtained with a JEOL 2000FX scanning transmission electron microscope on ion-milled thin sections. Several glass-ceramic compositions doped with Ni, COand Cr were fabricated into clad fibers using a 'hard liquid (or rod)-in-tube' fiberization approach. The fibers were subsequently heat-treated to promote crystallization within the core. The clad tube had a boro-germano-silicate composition and was fabricated inhouse using an outside vapor deposition process. The fiber draw was generally performed under the condition of high-tensiodhigh-speed for minimal coreklad interaction, at temperatures just above the liquidus of the core glass. Fibers (125 pm diameter) were made with core diameters of 4 to 12 pm. Spectroscopy Optical properties of transition metal-doped glass-ceramics were determined from absorption and emission spectroscopy. Absorption measurements of Ni2+and Co2+-dopedglass-ceramics were performed on 1-2 mm thick polished samples on a Cary 3E UV-Vis spectrophotometer. Ni2+fluorescence spectra were measured with a SPEX Fluorolog-2 using 980 nm excitation. Fluorescence lifetimes were acquired by curve fitting an exponential decay fwnction to the experimental data. Measurements were obtained by pulsing a 980nm laser diode and measuring the resulting fluorescence decay curves with a fast (51s system response) photodetector and oscilloscope. Table I. Composition and heat treatment of gallate spine1 glass-ceramics Oxide (wt%) (A) 38.7 42.3 7.7 1.3 10.0 -
H.T. "C-h
850"-2
(B)
(C)
(D)
(E)
39.6 31.0 15.9 2.0
41.3 22.8 21.8 2.1
38.6 25.1 16.2
37.3 30.6 13.3
11.5
12.0
15.1 1.o 4.0
12.3 1.1 5.3
750"-8 900"-2
750"-8 900"-2
850"-2
850"-2
-
Ceramic Nanomaterials and Nanotechnology
267
RESULTS AND DISCUSSION Glass-Ceramic Compositions Examples of representative compositions and heat treatments that yield transparent aluminogallate spinel glass-ceramics are listed in Table I. The glassceramic microstructures consist of spinel nanocrystals, 10-20 nm in size, dispersed throughout a stable, continuous aluminosilicate glass, with total crystallinity ranging from about 5%-25%. The glass stability of these multicomponent compositions is superior to those of simple ternary glasses, providing an advantage for melting, forming, and fiberization of these glasses. Rapid quenching of these glasses is not required. The glass-ceramics rely on nucleation promoted by amorphous phase separation, presumably into alkali aluminosilicate-rich and higher gallidaluminarich regions. X-ray diffiaction analyses of glasses after annealing generally show no evidence of crystallinity. Figure 1 shows the microstructure of a spinel glassceramic (composition D in Table I) after heat treatment at 750°C for lh. The Xray diffiaction pattern for this material shows broad gallate spinel peaks. Many of these transparent glass-ceramicsdisplay remarkably little haze, even to strong collimated white light. The nickel-doped materials undergo a dramatic change in color upon ceramming, from brown to blue-green, as the Ni2+ions partition into the higher field strength octahedral sites of the spinel crystals.
Figure 1. Microstructure of transparent y-(Ga,A1)203 glass-ceramic. Bar = 0.1 pm.
268
Ceramic Nanomaterials and Nanotechnology
Because gallate-rich spinel crystals contain large Ga3' ions in both their octahedral and tetrahedral sites, these sites provide a weaker crystal field environment than that obtained with conventional aluminate spinels. Thus, when doped with transition metal ions such as octahedral Ni2+and tetrahedral Co2+, gallate spinels can yield fluorescence spectra that are significantly shifted toward the infrared compared with those of conventional spinels. (VI)Ni2' Spectroscopy Ni2+can reside in both the octahedral and tetrahedral sites of spinel, although octahedral coordination is preferred.' Moncorge showed that when gallium substitutes for aluminum in Ni2+-dopedlanthanum hexaluminates the reflectance spectra show an increased amount of Ni2+in the octahedral sites.'' Moreover, the Ni2+-dopedhexaluminogallates exhibit broadband fluorescence centered at 1200 nm, with fluorescence lifetimes one order of magnitude longer than in pure Ni:hexaluminate. In the competition for octahedral sites Ni2+shows a stronger preference for the octahedral environment than does Ga3! Moncorge also demonstrated a shift in peak emission wavelength toward longer wavelength (lower energy) when Ga3' replaces A13+. It is reasonable to expect that gallate or aluminogallate spinels might also have higher Ni2+in octahedral coordination than do aluminate spinels. Figure 2 shows the absorption and emission spectra of two nickel-doped (0.5 wt% NiO) transparent aluminogallate spinel glass-ceramics (compositions C and D) compared with that of a comparably doped aluminate spinel glass-ceramic.* The latter has higher crystallinity (closer to 30%) so peak heights of the gallate and aluminate spinel absorbance curves can not be compared directly. It is clear, however, that the absorbance peak positions of the aluminogallate spinels are shifted significantly toward shorter wavelength (higher energy) compared with those of the aluminate spinel. The aluminogallate absorbance peaks are also narrower (FWHM) than those of the aluminate spinel glass-ceramic. The aluminogallate spinel glass-ceramicsalso demonstrate strong, broadband fluorescence centered at 1250 nm, while the aluminate glass-ceramic shows weaker fluorescence centered at slightly shorter wavelength. These results are not readily understood. A detailed study of the spectroscopy of Ni-doped gallate, aluminate, and aluminogallate spinel glass-ceramics is undenvay and will be described separately.
* Gahnite glass-ceramic composition (wt%): SiOz: 55.0, A1203:20.2, ZnO: 16.0, KzO: 2.3, ZrOz: 6.5, NiO: 0.5.
Ceramic Nanomaterials and Nanotechnology
269
0.20
fluorescence from
I i
0.15 Q)
0 C
a
e 2
0.10
2 0.05
- .-
Ni:ZnAl,O, spinel
0.00 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 7
Wavelength (nm)
Figure 2. Absorbance and fluorescence spectra of Ni'+-doped aluminogallate and aluminate spinel glass-ceramics. A number of Ni-doped glass-ceramics were fabricated into clad fiber using the rod-in-tube technique. Figure 3 compares the absorbance of a Ni2+-dopedglassceramic fiber with that of the bulk glass-ceramic and its precursor glass (all doped with 0.5 wt% NiO). The energy level assignment for the glass-ceramic was obtained by comparison with published data for single crystal MgA120+20Ni2' ions in the precursor glass (lower field energy) yield absorption peaks at longer wavelengths. The spectroscopic features observed in the doped bulk sample are clearly reproduced in the fiber form, with the main absorption band at 1OOOnm (3A2-3T2)corresponding to around 3.5 dB/cm. The peak position of the l O O O n m absorption band is identical for the cerammed fiber and the bulk glass-ceramic. Although this concentration of Ni ions is not high when compared with that used in bulk single crystals:" the relatively low crystallinity (-1 5%) in the glassceramic, coupled with good partitioning of the Ni2+ions into the crystals, means that the ion concentration in these nanocrystals can be quite high. For this reason, the fluorescence properties of fibers containing only O.O5wt% NiO were found to be greatly superior to those of higher-Ni compositions. Figure 4 compares the fluorescence spectra of two glass-ceramic fibers doped with O.O5wt% NiO. One
270
Ceramic Nanomaterials and Nanotechnology
fiber was undercerammed (resulting in a poorly crystalline, glassy core) and the other was given the proper heat treatment for optimal crystallinity. The optimized spectrum has -250nm FWHM peaking at wavelength of 1200 nm. By comparison, the as-drawn glass fiber containing Ni ions in an amorphous environment would show no measurable fluorescence and a significantly different absorption spectrum (similar to that shown in Figure 3.)
-
glass-ceramic fiber
bulk glass bulk glass-ceramic
E7 a 6
s
/\
600
800
1000
1200 1400 1600 1800 2000 2200
1100
1200
1300
1400
1500
1600
Wavelength (nm)
Wavelength (nm)
Figure 3. Absorbance spectra of Ni2+doped glass-ceramic fiber, bulk glassceramic, and precursor glass.
Figure 4. Fluorescence spectra of under-cerammed and fully-cerammed Ni2+-dopedspinel glass-ceramic fiber.
As the optimum heat treatment for the fiber is approached, the fluorescence efficiency and room temperature fluorescence lifetime for the active Ni ions dramatically increases, as seen in Table 11. The emission shifts toward higher energy as the optimal ceramming occurs, from underceramming (Tl) to higher crystallization temperatures (T2-T3), where Ni2+ions most efficiently partition into the spinel crystals. These systematic changes in the Ni2+ion spectroscopy are associated with the change in the electron-phonon coupling of the ions as they become incorporated within the spinel crystals compared with the amorphous environment. The Nizf-doped fibers demonstrate strong and broad fluorescence with peak wavelength centered at 1200-1250 nm and emission lifetimes of >300 P.
Ceramic Nanomaterials and Nanotechnology
271
Table 11. Dependence of Ni ion lifetime in glass-ceramic fibers as a function of crystallizationtemperature. Peak T. T1 T2 T3
1‘‘ “e” folding time’ 2nd“e” folding time Peak Wavelength 210 ps 400 ps 1350 nm 780 ps 1.2 ms 1250 nm 1.1 ms 1.3 ms 1200 nm
(IV)CO~+ Spectroscopy Co2+ions can reside in both tetrahedral and octahedral coordination. In addition to strong absorption bands in the visible portion of the spectrum (500700 nm), crystals that contain Co2+in tetrahedral coordination typically possess very broad absorption features in the near infrared 1200-1700 nm. This is assigned to the 4A2-4T2transition. It has been shown that the Co2+ions substitute for the tetrahedrally coordinated Ga3’ ions in the LiGa508 lattice and occupy sites of ~3 point group symmetry.I2 Figure 5 shows the absorbance spectra of a Co2+:aluminogallateglass-ceramic batched with increasing levels of Co304from 0.005% to 0.20%, as well as the corresponding spectrum for a doped, uncerammed glass. The spectrum of the precursor glass has a shape quite distinct from those of the glass-ceramics in both the visible and the infiared. The doped glass-ceramics provide strong absorption across the entire telecommunicationsbandwidth; their absorbance curves are particularly flat between 1400 and 1600 nm. These results are identical to those described for a Co2+-dopedsingle crystal of LiGa508,11 and evidence good partitioning of the Co2+ions into the tetrahedral sites of the aluminogallate spine1 crystals. Strong absorption through these wavelengths is also observed in Co2+doped aluminate spinel16,but in the aluminates the absorbance shifts to shorter wavelengths.
The ‘e’ folding times are obtained by fitting an exponential decay curve to the experimental data. The 1’‘ ‘e’ folding time refers to the decay constant for the first part of the curve. A singleexponential decay would indicate that all Ni2+ions were fluorescing with the same lifetime.
272
Ceramic Nanomaterials and Nanotechnology
1.o
0.8
0.6
0.2%C0304
/'
0.4
8
c
CO
e
5:
0.2
2
I
0.0 400
600
800
loo0
1200
1400
1600
Wavelength (nm)
Figure 5. Absorbance spectra of Co*+-dopedy-(Ga,A1)203 glass-ceramic batched with C03O4 levels of 0.005% to 0.20% by weight. Dark line shows precursor glass batched with 0.05% Co304.
These materials could be useful as saturable absorber media, particularly for passive Q-switching of erbium-doped optical amplifier glasses at 1540 nm. While a number of such potential materials have been described, these glass-ceramics are readily fiberized and so might offer an advantage for such applications. CONCLUSIONS Transparent, nanocrystalline glass-ceramics based on the aluminogallate spinel crystals Li(Ga,Al)508 and y-(Ga,A1)203 can be obtained in the Si02-Ga203A1203-&0-Na20-Li,O system. The glasses are stable and fiberizable and are self-nucleating via amorphous phase separation. Their glass-ceramic microstructures consist of 10-20 nm spinel crystals dispersed throughout a stable aluminosilicate glass. Total crystallinity ranges from 525%. Because galliumrich spinel crystals contain large Ga3' ions in both their octahedral and tetrahedral sites, these sites provide a lower crystal field strength environment than is
Ceramic Nanomaterials and Nanotechnology
273
obtained with conventional aluminate spinels. Thus, when doped with transition metal ions such as Ni2+and Co2+,gallate-rich spinels can yield absorption and fluorescence spectra that are significantly shifted with respect to those of conventional spinels. Glass-ceramic fibers based on Ni2+-dopedaluminogallate spinel demonstrate strong and broad fluorescence with peak wavelengths of 12001250 nm and emission lifetimes of >3OO ps. Compared with aluminate spinels, Ni2+-dopedaluminogallate glass-ceramics also provide a larger AA between their peak absorption and emission wavelengths, possibly allowing for the avoidance of the excited state absorption that prevents lasing in Ni:MgAl204 spinel. Co2+:aluminogallateglass-ceramics provide broad absorption from 1300-1650 nm. Potential applications for transparent, transition metal-doped aluminogallate spinel glass-ceramics include broadband light sources2’as well as saturable absorber and laser media for the telecommunications wavelength band. REFERENCES K. Pohl, “Hydrothermale Bildung von y-Ga203,” Naturwissenschaften 55 82 (1968). R.C. Evans, An Introduction to Crystal Chemistry,2nd edition, Cambridge University Press, Cambridge, 1964. R.K. Datta, “Polymorphism of LiGa508 and of LiGa508-MgGa204solid solutions.” J. Am. Ceram. Soc. 54 262-265 (1971). J. Ferguson, D.L. Wood, and L.G. Van Uitert, “Crystal-field spectra of d3?7 ions. V. Tetrahedral Co2+in ZnA1204spinel.” J. Chem. Phys. 51 2904-2910 (1969). J.F. Donegan, F.J. Bergin, G.F. Imbusch, and J.P. Remeika, “Luminescence from LiGa5Og:Co.” J. Lumin. 31/32,278-280 (1984). J.F. Donegan, F.J. Bergin, T.J. Glynn, G.F. Imbusch, and J.P. Remeika, “The optical spectroscopy of LiGa50g:Ni2+.”J. Lumin. 35,57-63 (1986). T. Abritta and F.H. Blak, “Luminescence study of ZnGa204:Co2+.”J. Lumin. 48/49 558-560 (1991). ‘N.V. Kuleshov, V.P. Mikhailov, V.G. Scherbitsky, P.V. Prokoshin, and K.V. Yumashev, “Absorption and luminescence of tetrahedral Co2+ion in MgAl204.” J. Lumin. 55 265-269 (1993). J.F. Donegan, G.P. Morgan, T.J. Glynn, and G. Walker, “New materials for tunable lasers in the near infrared.” J. Modern Optics 37 769-777 (1990). K.V. Yumashev, N.N. Posnov, and V.P. Mikhailov, “Excited-state absorption and stimulated emission cross-section of tetrahedral Co2+ion in LiCasO8.” Appl. Phys. B 69 41-44 (1 999).
’
’
274
Ceramic Nanomaterials and Nanotechnology
11
K.V. Yumashev, “Saturable absorber Co2+:MgA1204crystal for Q-switching of 1.34 pm Nd3+:YA103and 1.54-pm Er3+:glasslasers.” Appl. Optics 38 63436346 (1999). 121.A. Denisov, M.I. Demchuk, N.V. Kuleshov, and K.V. Yumashev, “Co2+:LiGaj08saturable absorber passive Q switch for 1.34 pm Nd3+:YA103and 1.54 pm Er3+:glasslasers.” Appl. Phys. Lett. 77 2455-2457 (2000). 13R.Reisfeld, A. Kisilev, A. Buch, and M. Ish-Shalom, “Transparent glassceramics doped by chromium (111): Spectroscopic properties and characterization of crystalline phases,” J Non-Cryst. Sol. 91 333 (1987). 14Cz.Koepke, K. Wihiewski, M. Grinberg, and G.H. Beall, “Excited state absorption in the gahnite glass-ceramics and its parent glass doped with chromium,”Spectrochim. Acta A 54 1725-1734 (1 998). 141. Yamaguchi, K. Tanaka, K. Hirao, and N. Soga, “Preparationand optical properties of transparent glass-ceramics containing LiGa50&?+,“ J. Mater. Sci. 31 3541-3547 (1996). 16R.M.Boiko, A.G. Okhrimchuk, and A.V. Shestakov, “Glass ceramics Co2+ saturable absorber Q-switch for 1.3-1.6 pm spectral region.” OSA TOPS 19, Advanced Solid State Lasers 185-188 (1998). 17 K. Tanaka, T. Mukai, T. Ishihara, K. Hirao, N. Soga, S. Sogo, M. Ashida, and R. Kato, “Preparation and optical properties of transparent glass-ceramics containing cobalt (11) ions.” J. Am. Ceram. Soc. 76 2839-2845 (1993). 18 R. Pappalardo, D.L. Wood, and R.C. Linares, Jr. “Opticalabsorption spectra ofNi-doped oxide systems. I.” J. Chem Phys. 35 1460-1478 (1961). 19 R. Moncorgd, J. Thkry, and D. Vivien, “Enhancement of fluorescence from octahedrally coordinated Ni2’ in LaMgAl,, 0 1 9 materials by A13+/Ga3+ion substitution,” J. Lumin. 43 167-172 (1989). 2%.V. Kulashev, V.G. Shcherbitsky,V.P. Mikhailov, S. Kiick, J. Koetke, K. Petermann, and G. Huber, “Spectroscopy and excited-state absorption of Ni2+doped MgA1204,” J. Lumin. 71 265-268 (1997). 21B.N.Samson, L.R. Pinckney, G.H. Beall, J. Wang and N.F. Borrelli, “Nickel-doped nanocrystalline glass-ceramic fiber”, CLEO 2002, CMD 1 (2002).
Ceramic Nanomaterials and Nanotechnology
275
This page intentionally left blank
ELECTRONIC STRUCTURE OF INTERFACES IN CU-GLASS NANOPHASE COMPOSITES Monika Backhaus-Ricoult Dept. of Materials Science and Eng. Cornell University Ithaca NY 14853 USA
Marie-France Trichet, Fabien Maurel, Alain Dezellus Centre de Chimie Metallurgique CNRS 94 407 Vitry, France
Lolwa Samet Facult6 des Sciences Universite6 de Tunis 34 512 Tunis Tunesia
Dominique Imhoff Lab. de Physique des Solides, CNRS Universit6 d ’Orsay 91 607 Orsa France
ABSTRACT For a wide range of temperature and oxygen activity conditions, internal oxidation of (Cu,Si) and (Cu,Si,Al,Ca) alloys produces nanosize spherical precipitates of silica and, for (Cu,Si,Ca,Al) alloys, in addition, precipitates of ternary glass and crystalline phases. F or various oxidation conditions, the atomic and electronic structure of the glass precipitate interfaces has been studied by high -resolution electron microscopy and by electron energy loss spectroscopy with high spatia resolution. The interfaces of the glass precipitates are smoothly curved and show no faceting at an atomic level. Interfacial bonding changes with the reaction conditions. At high oxygen activity, EELS analysis reveals an important hybridization of Cu 3d and 0 2p states as typically found or oxide bonding. At lower oxygen activity, neither this hybridization, nor any other major modifications in the interfacial electronic structure compared to the bulk phases are found. In case of mixed Si, Ca, Al-oxide glass precipitates, no considerable segregation of the different cations to the glass -Cu interface is detected and silicon, calcium and aluminum do not show any modifications in their electronic structure at the interface.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Ceramic Nanomaterials and Nanotechnology
277
INTRODUCTION W e it is known that, within the thermodynamic stability range, the loca equilibrium chemistry at interfaces between crystalline ternary phases undergoes important changes, such knowledge is not present for glass/metal phases. For a crystalline oxide in contact with a metal phase, large modifications in t e interfacial composition occur by Gibbs’ adsorption or by segregation from the bulk to the interface. These compositional changes respect the constraints imposed by the crystalline structure of the oxide and the global electroneutrality i the system. Therefore, any Gibbs’ adsorption to the interface is associated with a local redox-reaction at the interface: local metal oxidation in case of oxygen excess or local reduction of the oxide for oxygen deficiency. A phenomenological model was developed by one o f the authors to describe the evolution of the interfacial chemistry with oxygen chemical potentia The model was applied to MgO-Cu and Al203-C~interfaces and compared to experimental results, which allowed todirectly determine the interfacial chemis ry (atomic and electronic interfacial structure determined by high resolution electron microscopy and electron energy loss spectroscop 27374) or gave indirect indications on modifications of the interfaces (evolution of precipitate Wulff shape576). For gl ss-metal interfaces, there is no long-range order in theglass, and it is unclear if the interfacial plane is a clearly defined atomic plane as encountered in the case of planar interfaces between metals and crystalline oxides. In contrast to dense-packed crystalline oxides, such as magnesia or alumina, the open glass network structure of a silicate glass allows to easily accommodate excess atoms a the interface. However, since electroneutrality must be respected at the glassmetal interfaces, any segrega ion or Gibbs’ adsorption of species to the interface is accompanied by some stabilizing interaction with the metal or other glass components and then follows again the same scheme as observed atinterfaces with crystalline oxides. For a crystalline oxide, the site density of the terminating plane defines the maximum oxygen excess. The open glass network structure may allow farther-reaching interactions across the interface and promote higher excess concentrations compared to a crystalline oxide. As a conseqence, larger decreases in interfacial energy could be achieved, which could be technically explored to fabricate metal coatings onglass with very strong adhesion or to optimize spreading and infiltration behavior in glass -metal composites. In the present work, we have studied the atomic and electronic structure o interfaces between glass and metal, in the goal to determine the interfacia bonding as function of oxygen chemical potential and glass composition and compare the interface behavior to that o crystalline oxide-metal interfaces. We have chosen copper as metal, because it is a quite noble metal, and any changes in its electronic structure are very clearly reflected in its absorption edge fine structure. Pure silica, alumino-calcio-silicate and te rnary glasses close to the
’.
278
Ceramic Nanomaterials and Nanotechnology
anorthite composition are the glasses we selected. Since our study requires chemically clean interfaces, we chose in -situ fabrication of the interfaces b precipitation. A very adapted approach is the internal reduction of glasses containing transition metals. We actually reduced a number of different glasses, obtained very small copper inclusions in the glass, but could not study the interfacial electronic structure, because of severe charging problems and consequent loss of the atomic spatial resolution in the microscope. Therefore, we decided to use the inverse composite with glass precipitates in a copper matrix as model system. Thermodynamicdly we -defined glass-metal interfaces were produced by internal oxidation of copper alloys at defined oxygen chemical potential. In this composite, the problem of charging under the electron beam was overcome.
EXPERIMENTAL PART Four Cu alloys were prepared by melting 99.999% copper, >99.999% grade silicon, 99.9 grade calcium and 99.99 grade aluminum under purified argon in a copper crucible. Alloy compositions as determined by flame induced absorption spectroscopy analysis are indicated in Table 1
Alloy 1 Alloy 2 Alloy 3
Si content [wt%] Al content [wt%] Ca content [wt%] 1.2 0.22 0.11 0.16 0.30 0.06 0.10
Polished alloy slices were oxidized at 900°C for 50 hours at different oxygen activities within the coexistence range of Cu and SiO 2. The highest possible oxygen activity, approximately 10-*at 9OO0C, was set by the solid-state buffer of a mixture of silica, copper and cuprite powder inside an evacuated quartz ampoule. Intermediate oxygen activities were established by buffering C O K Q gas mixtures with a continuous flow rate of 2 Ik through a gas tight furnace. After oxidation, sample surfaces and polished cross-sections were investigated by optical microscopy , analytical scanning electron microscopy (SEM/EDX) and X-ray diffi-action. For conventional, analytical, and high resolution transmission electron microscopy (TEM) investigations, 3mm diameter disks were mechanically polished to a thickness of 30pn and then ionmilled in a cold stage at 5kV, O S m A , at an angle of 12". Thin foils were investigated in a conven tional transmission electron microscope (CTEM JEOL 2000FX) for general microstructural characterization. The atomic structure of selected interfaces was
Ceramic Nanomaterials and Nanotechnology
279
investigated in a JEOL 4000EX high -resolution microscope operated at 300kV. Lattice images were recorded close to the Scherzer focus. The electronic structure of the interfaces was studied in a VG HB501 scanning transmission electron microscope (STEM) with cold field emission gun operated at 100kV. Electron energy loss spectroscopy (EELS) data were acquired by a CCD camera from a Parallel Electron Energy Loss Spectrometer (PEELS) for the different characteristic absorption edges. The collection se -angle of the spectrometer was 15mrad. A probe size of 0.22nm and an energy dispersion of about 0.4eVkh were used. Series of spectra were acquired while the probe was stepped in the line scan mode across the interface with an inter -step distance o 0.25nm. Spectra were explored by manual extraction of the characteristic edges. Background under the characteristic edges was subtracted by using a smooth power law fit.
EXPERIMENTALRESULTS Generals on precipitates While for higher silicon contents in the alloy interconnected fractal branches o silica glass and a minor fiaction of crystalline Si02 were found 7, in the present binary alloy with low silicon content, only isolated silica glass precipitates formed. Their size was in the nanometer range, typically varying between 201-1 and 500nm. Spherical precipitate morphology was observed, as shown in the high resolution SEM view of an electrolytically polished section of an alloy oxidized a 900°C, a02=10-~,Figure la. TEM observations showed, that even at the atomic level the precipitates did not show any faceting. Figures lb, c, d present typical TEM micrographs of such spherical precipitates with details of the interface and an electron diffraction pattern of the precipitate and the oriented copper matrix, which demonstrates the glass character of the precipitates. In the quaternary alloys, the same type of spherical g lass precipitates was encountered, but two very distinct compositions were found. Precipitates with very homogeneous contrast were composed of a silica -rich glass with less than 2% of aluminum and calcium, see Figure 2a. The other type of glass precipitates showed very inhomogeneous image contrast revealing thereby their decomposition in chemically different glasses, Figure 2b. The composition o such glass precipitates ranged from anorthite CaA 2Si2O8 to almost pure quartz, but was often close to a calciu -poor glass with equivalent portions of silicon and aluminum. Due to the small size of the homogeneous regions in such precipitates and their overlapping within the sample thickness, local compositions could not be precisely determined by TEM/EDX analysis. In the quaternary alloys, i addition, a considerable fraction of small crystalline precipitates was found. They were identified by X-ray diffi-action, electron diffraction and TEMEDX analysis as quartz, anorthit , gehlenite, lime and a small amount of y-alumina. Upon alloy
280
Ceramic Nanomaterials and Nanotechnology
processing, calcium was found to strongly segregate to the outer surface. Upon oxidation, it segregated to grain boundaries and surfaces and formed large platelets of different phases, which, besides CaO, included all minor impurities present in the alloy, such as phosphorus, sulfur.. .
Figure 1.Microstructure of a Cu-silica glass composite obtained by oxidation of Alloy I at 900°C. a ) SEM high resolution of an electropolished section showing the distribution of spherical glass precipitates in copper, b) brightfield image of a precipitate, c) selected area difraction pattern of precipitate and copper matrix in [OlI ] projection
Si
Si
A1
Ca Figure 2. TEM brightfield images of spherical glass precipitates and corresponding EDX analysis of a) precipitates with homogeneous image contrast, b) precipitates with inhomogeneous image contrast in a Cu-glass composite obtained by oxidation of Alloy 2 at 900°C
Ceramic Nanomaterials and Nanotechnology
281
Atomic structure of the interfaces HREM analysis of the interfaces showed that the interfaces were rough at an atomic level. Cu was always indirect contact with theglass, no intermediat crystalline silicate phases were formed. TEMEDX analysisdid not reveal an formation of extended solid solutions on both sides of the interfa ce; within the EDX detection limit (l%),both phases were “pure”. The copper lattice parameter was the one ofbulkfcc copper; however, close to the interface, an importan structural relaxation was observed in the metal. This finding was independent of theglass composition and the oxidation conditions. While liquids and even gases show ordering in the first few atomic layers on top of a crystalline solid, no such ordering was observed in the present glass. A possible reason might be found in the fact tha the materials were very dissimilar. For oxide glass grain boundary films in silicon nitride, it was found that theglass organized on top of the nitride and, over a few atomic layers, adopted the structure of crystalline silicates, which structurally well-matched the silicon nitride substrate 8. For inert gas inclusions in metals, similar findings were reported9. From those observations, it is expected, that glass also orders on a copper surface. A structure close to that of idealized hexagonal cristobali te could form, which according to ab initio calculations o copper-silica interfaces 10 shows good structural accommodation with metallic copper. However, a very important structural relaxations is expected to occur in such an organized layer, because mos cristobalite planes are not atomically flat, but wavy. Consequent structural relaxation may cause sufticient disorder in the terminating glass planes, that high-resolution imaging cannot reveal any ordering. Detailed image analysis of the glass struc ure may provide further insight and
Figure 3. a ) High angular bright and b) corresponding dark field image and c) high resolution image of a copper -glass inteface obtained by oxidation of Alloy 1 at a 02=10-~, [Oll]C, projection
282
Ceramic Nanomaterials and Nanotechnology
reveal a structural anisotropy close to the interface, which is related to structura ordering in the glass. With the current image analysis techniques (phase imaging and Bragg filtering), however, we did not reveal ordering in the glass
Electronic Structure of the interfaces A common feature was observedfor allglass/Cu interfaces. It consists of a decrease in intensity or a total loss of the three characteristic oscillations in the fine structure of metallic copper in the proximityof the interface. This modification in the copper spectra appeared strongest at the interface, and decreased slowly in importance over several atomic layers. A typical modified copper spectrum is presented with spectrum (12) in Figure 4 and can be compared to the regular spectru of metallic copper, spectrum (A). Since no shift in energ or increase in L2,3 intensity was noticed, the loss of the oscillations was interpreted as a perturbation in the local structure of copper. This fits the HREM observations of the atomic structure of the interfaces, which also revealed a certain structura disorder in copper in proximity of the interface. The ELNES observations confirm this statement. The interfacialelectronic structure was found to depend on the equilibrium oxygen chemical potential, therefore, we will separately present the results for the different glasses at two representative oxygen activities, a 02=1 0-8(representing the upper part of the stability range of the interface) and a 02=lO-l4 (representing the central part of the coaisting range).
Bulk phases; Absorption edges of the bulk phases of precipitates and matrix are presented i the figures 4-7 together with interfacial spectra. The shape of the C u - b edge ~ is very sensitive to the oxidation state of copper and to the configuration of the surrounding atoms. Figure 4 shows spectra of copper metal (A) and bulk cuprite (B). Metallic copper Cu(0) (dlOsl) presents a typical step-like spectrum with three oscillations following the first step threshold at 932.5,936.3 and 94 OeV. Cu20 shows no chemical shift compared to the metal and an edge shape comparable to that of metallic copper, which in addition shows the C U - L ~white ,~ lines at 932.5eV and a weakCu -L1 line at 953eV. The appearance of any white line intensity at the CO pper-glass interfaces can be used as indication for a local oxidation of copper. Oxygen 1s absorption edges of different bulk oxides and glasses are presented in Figure 5. The absorption edge of all glasses exhibited very simple shape with a fxst intense peak at 538.5eV (corresponding to the 1s-2p transition), followed by
Ceramic Nanomaterials and Nanotechnology
283
a much wider second peak at 562eV. Spectra (C) and (T) of a pure quartz glass precipitate and a ternary glass precipitate illustrate the typical spectral features. In cuprite Cu20, the first maximum in the 0 - K edge is shifted to 532.5eV, see spectrum (A). Therefore, a low energy shoulder on the silica 0-K edge will indicate any local oxidation of copper at the interface. Independently on the exact glass composition, glass precipitates always showed a Si-2p edge with peaks at 106 eV, 108 eV and 115 eV , typical for silicon in its SiO4 tetrahedral coordination with oxygen. The spectrum of a quartz glass precipitate is presented in Figure 6 (C). The ALL edge of the ternary glasses and crystalline precipitates was usuall composed of two characteristic sets of peaks located at around 80eV and 98eV. According to 11, the exact shape of these peaks, their energy location, width and intensity depend on the A -coordination: Aluminum in octahedral coordination produces a very sharp intense peak at 80eV with a broad high energy shoulder (saddle) of lower intensity, while aluminum in tetrahedral coordination shows a much wider peak at 80eV with an intense low energy shoulder, which shifts the peak threshold to slightly lower energy. The second spectral feature is composed of a broad peak being located between 96 and 98eV depending on the environment". Figure 7 illustrates these differences with the spectra of gehlenite (G) and a ternary glass (T). For the mixed silicates, the tail of the AIL edge is superposed by the Si-L edge. Inglasses with aluminum in network forming positions, aluminum is mainly found in tetrahedral coordination, showing then the two broad peaks typical for tetrahedral coordination. Spectra of our ternary glass precipitates (T) show higher tetrahedral coordination than crystalline mullite (M), but also reveal a large portion ofaluminum in higher coordination. This is consistent with Raman result 12, which showed variations i n the aluminu coordination with composition in such ternary glasses, ranging from mainly tetrahedral coordination for glasses with CdAl = 1 to important contributions of higher coordinated aluminum in glasses with a CdAl ratio c 1. The Ca-L edge in the ternary glasses was found to be very similar to that o pure CaO, spectra (A) and (T) in Figure 8. The spectrum is generally dominated by a set of white lines located at 349eV and 352eV. For different glass compositions, only the intensity of the L23 edge changed, but not its characteristic features.
Silica ghss/Cu interfaces at ao2=1f18; A typical interfacial spectrum of a silica glasskopper interface obtained a 9OO0C, ~ 2 = 1 0 -is* shown in Figure 4 (11). Compared to metallic copper, Figure 4
284
Ceramic Nanomaterials and Nanotechnology
(A), a significant increase of the first maximum of the Cu -2p edge at 932.5eV is noticed, without any detectable shift in energycompared to metallic copper. When discussing above the spectra of the different bulk phases, it was underlined that this ELNES feature is characteristic of copper in a Cu state and similar to that observed for bulk cuprite, see Figure 4 (B).
'+
The 0-1s absorption edge was also modified at the interfaces. Figure 5 (11) presents a typical interfacial spectrum. The main peak showed system tically a shoulder of varying intensity on its low energy loss side. Deconvolution of the signal revealed that this shoulder indicated the presence of a second peak with maximum at approximately 532eV, the characteristic peak energy of the corresponding cu prite peak. This shoulder has then to be interpreted in terms of the interaction of copper with oxygen, which was already noticed in the Cu -2p edge. Unfilled d-states in partially oxidized copper allowed hybridization of -2p and Cu-3d states and produced transitions to such hybridized states, which had lower energy than the silica states. The Si-2p edge at the interface was not modified compared to bulk silica. typical interfacial spectrum is shown in Figure 6 (11). Since all features of the silica S -2p edge were conserved, it was assumed that the tetrahedra configuration, responsible for this E N S feature, was preserved up to the interface.
Pure silica ghss/Cu interfaces at a 02=1014 At this oxygen activity condition in the central part of the CO pper-silica coexistence range, no modification in the edges of 0-K, Si-L and Cu-L compared to the bulk phases were observed. This is illustrated by the typical spectra (12) in Figures 4-6. Ternary ghss with low calcium and aluminum content/Cu interfaces; The alumino-calcio-silicate glass followed the pattern of pure silica glass. The concentrations of aluminum and calcium were too small to allow a precise comparison ofedge shapes at the interface with those in the bulk glass. It was only stated that no m ajor segregation of these minor elements to the interface occurred. Al-rich ternary ghss/Cu interfaces at a 02=1 Om8 Modifications in the Cu -L and 0 - K edges similar to those already described for the Cu-quartzglass interfaces at a 02=10-* were also noticed for the ternary glass at high oxygen activity. Representative spectra for the interface, which are chosen from the stepped scans across the interfaces, are shown in Figures 4 -8 as
Ceramic Nanomaterials and Nanotechnology
285
spectra (13). The Cu -L edge in Figure 4 (13) shows again an increased L 23 intensity at the interface and the interfacial O K edge in Figure 5 (13) shows an intense low energy shoulder, together with a shift of the second peak to lower energy. For better appreciation of the differences in the interfacial spectra compared to the bulk spectra, the spectra have been superposed by a spectrum in dotted lines of the matrix, which was normalized in intensity. The S -L interfacia absorption edge is shown in Figure 7 (13). It was superposed by the Cu -M edge, shown in the bottom part of Figure 7 for metallic Cu, since copper contributes to the signal at the interface. It was difficult to extract any modification in the edge features compared to the bulk glass. The interfacial Ca -b3edge is presented i Figure 8 (13). It showed no distinguishable features compared to the bulk glass. The series of spectra acquired while stepping with a small size probe across the interfaces did not reveal any segregation of calcium, aluminum or silicon to the interface.
Figure 4: Cu-L edge for the bulk copper matrix (A), bulk cuprite (B), at a Cu Si02 interface obtained at a 0 2 = ~ ~(II), - 7 at a Cu-SiO, interface obtained at a02=1014 (12), at an interface between Cu and the ternary glass obtained at a02=la8 (13). For better comparison, a spectrum oj metallic copper is superposed in dotted lines to the interfacial spectra.
920
930
940
950
960
970
980
energy loss in e V
286
Ceramic Nanomaterials and Nanotechnology
Al-rich ternary glass/Cu interfaces at ao2=I @14 In an intermediate oxygen activity range, interfacial spectra did not show an modification compared to the bulk phases. Since the spectra do not provide further insight, they are not depicted. As for higher ox ygen activity, no segregation of calcium, aluminum or silicon to the interface was detected.
L
*.
: 530
540
550
560
570
580
energy loss in eV Figure 5: 0-Kedge for bulk cuprite (B), the quartz precipitate center (C), the center of a ternary glass precipitate (T), at a Cu -SO2 intelface obtained at ao2=f@ (II), at a Cu -SO2 inte$ace obtained at a02=10-14(I2), at an inte$ace between Cu and the ternary glass obtained at a 02=10-8 (13). For better comparison, a spectrum in dotted lines of quartz glass is superposed to th e intelfacial spectra.
Ceramic Nanomaterials and Nanotechnology
287
Figure 6: Si -L edge for bulk of a quartz precipitate (C), at a Cu -SO2 interface obtained at a02=lO-~ (Il), at a Cu-SiO2 interface obtained at a02=10-14(12).
100 105 110 115 120 125 130 135 140 energy loss in eV I " " l " " l " " l " " l " " l " " 1 " "
AI-L
Si-L
U)
c C
a
0
Figure 7: Si -L edge for a bulk ternary glass precipitate (T),for a crystalline gehl enite precipitate (G) and at an interface between Cu and the ternary glass obtained at a 0 ~ = 1 0 (13). - ~ For the interface, in addition, first and second derivatives are presented. The Cu-M edge of bulk copper is also depicted..
Q)
.-m c
2
energy loss in eV
288
Ceramic Nanomaterials and Nanotechnology
Figure 8: Ca-L edge for bulk CaO (A), bulk ternary glass precipitate (0) and for an inte~acebetween Cu and the ternary glass obtained at aO2=IO-' (13).
340
350
360
370
380
390
400
e n e r g y l o s s in eV
DISCUSSION A number of results on the atomic and electronic structure of glasscopper interfaces were obtained on oxidized alloys. They indicate that the local interfacia chemistry strongly depends on the oxygen chemical potential. For all investigated glasses, pure silica glass, glass with low calcium and aluminum content and ternary glasses, special interfacial electronic states were observed in the high oxygen activity range. The corresponding features in theCu -L and O-K absorption edges were very distinct from those of the bulk glasses and bul copper. Increased L23 white line intensity in the Cu-L ELNES and appearance of a low energy shoulder in the O-K edge indicated a hybridization of Cu 3d and 0 2p states. According to the ELNES intensities, the density of hybridized states was important. The characteristic interfacial features in the Cu -L edge were very similar in shape and intensity distribution to those of bulk cuprite. In an intermediate oxygen activity range, no changes in interfacial ELNES compared to ELNES of the bulk phases were observed. From this we conclude that no considerable hybridization of Cu 3d and 0 2p states or any other major interaction across the interface took place under these conditions. It is interesting to notice that no segregation of the different glass cations to the interface occurred even though their oxide formation enthalpies differ (CaO being the less stable oxide). Relative oxygen excess at high oxygen chemical potential is achieved either by adsorption of excess oxygen to the interface or by desorption
Ceramic Nanomaterials and Nanotechnology
289
of the oxide metals fiom the interface. If excess oxygen is adsorbed, new copper bonds form at the interface and copper U ndergoes charge transfer, while a cations preserve the same oxide bonding in the glass and at the interface. In thi case, there is no driving force for cation segregation. If relative oxygen excess is achieved by desorption of the oxide metals, then desorption of the most noble metal (calcium) would take preferentially place, producing then a calciu depletion at the interface. The latter case is expected to occur for dense packed oxide lattices and interfaces, which do not provide enough space for accommodating excess oxygen. A typical example are (001) fcc oxide//meta interfaces. In the present case, however, the open glass network easily accommodates excess oxygen. Therefore, no segregation of cations was observed at the interfaces. In the low oxygen activity range, cation segregation is expected to occur, because then metallic bonding establishes across the interface and new -formed bonds between copper and aluminum, silicon or calcium contribute with different energies. Thisdifference in bonding ene rgy provides then a driving force for cation segregation. Unfortunately, we did not study interfaces at such extreme1 low oxygen activities. Both, the intensity of the Cu b 3 white line and the intensity of the low energ shoulder of the D K edge indica e hybridization of Cu 3d and 0 2p states in the band gap. A comparison of interfacial spectra for Cu -MgO, Cu-gamma alumina, Cu-quartz glass and Cu -ternary glass at the same temperature and oxygen activity revealed pronounced differences in the intensity o the white line and the oxygen low energy shoulder at the interface. The largest effect was observed for interfaces between copper and gamma alumina and interfaces between copper and pure silica glass. For the ternary glass, it was noticed that the spreadin energy of the interfacial states seemed to be larger than for pure quartz glass. By applying the phenomenological model of Gibbs’ adsorpti to glass-Cu interfaces, predictions on the interfacial chemistry are obtained7. Those predictions and the expe rimental results on copper matri -glass precipitate interfaces can be directly transposed to nanocomposites made of a glass matrix and small copper precipitates. Electrical and optical properties of such glass composites have been studied, but most studies have been realized for glasses with silver particles, because metallic silver is stable in air. The present results give indications on the space charge layer in such composites, their character, charge and thickness and its dependency on the oxygen chemical potential.
290
Ceramic Nanomaterials and Nanotechnology
ACKNOWLEDGEMENTS The authors are grateful for financial support provided by Corning Inc. for this research. REFERENCES Backhaus-Ricoult, M., Phil. Mag. 7, 1759 (2001) Imhoff, D., S.Laurent, C.Colliex and M.Backhaus-Ricoult, Eur. Phys. Journal AP 5 , 9 (1999) Backhaus-Ricoult, M., S. Laurent, J. Devaud, M. Hytch, D. Imhoff, S. Hagkge, Journal de Physique IV 9(P4), 13 ( 1999) Backhaus-Ricoult, M., Samet, L., Trichet, M -F., Hytch, M., Imhoff, D, submitted in March 2002 Backhaus-Ricoult, M. and S.Laurent, Proc. iib 98 : Intergranular and interphase boundaries in materials, ed. P.Lejcek, V.Paidar, Trans tech publ. 294-296, 173 (1998) Backhaus-Ricoult, M.. Acta Materialia, 49( 10). 1747 (2001) Backhaus-Ricoult, M, Samet, L., Trichet, M-F., Imhoff, D, submitted to Acta Mater. in Dec 2001 * Klebe, H.J., Cinibulk, M.K., Tanaka, I., Bruley, J., Vetrano, J.S., Ruehle, M., Properties of silicon nitride ceramics , p 259-274, ed. Hoffmann and Petzow, Kluwer Acadm> Publisher, Netherland, (1994) Donnelly, S.E., Birtcher, R.C., Allan, C.W., Morisson, I., Furuya, K., Song, M., Mitsuishi, K., Dahmen, U., Nature, March (2002) l0 Ashcroft et al, private communication l 1 R. Brydson et al., Ultramicroscopy 59, 81-92 (1995) l2 Sato, R.K., McMillan, P.F., Dennison, P, Dupree, R., Physics and Chemistry o Glasses, 32(4), 149 (1991)
Ceramic Nanomaterials and Nanotechnology
291
This page intentionally left blank
KEYWORD AND AUTHOR INDEX Adachi, T., 181 Agglomeration, 197 Allard Jr., L.F., 3 Alumina, 83, 161, 171, 197 Aluminum nitride, 143 Aluminum titanate, 171 Apatite, 235 Atomic force microscopy ( A M ) , 189 Backhaus-Ricoult, M., 277 Ball mill, high-energy, 129, 171 Ban, Z., 129 Barium titanate, 59 Beall, G.H., 265 Boron nitride, 93, 181 Borrelli, N.F., 265 Carbon, 161 Chaim, R., 245 Chassagneux, F., 93 Chen, T.-W., 23 Chip miniaturization, 59 Chu, M.S.H., 59 Coating, 143, 189 Cobalt, 129 Collins, R.T., 83 Composite, 189 Cooymans, J., 197 Copper, 277 Cornu, D., 93 Cristobalite, 209
ELNES, 277 Epicier, T., 93 EUVL mirror, 221 Fehling, P., 189 Feng, X., 59 Ferro, G., 93 Fiber, 189 Fluoroapatite, 235 Foschini, C.R., 75 Foster, B.C., 59 Gallate, 265 Garay, J., 161 Glass, 189,245 Glass ceramic, 221, 235, 265 Glass composite, 277 Goldstein, A., 245 Green, D.L., 33 Gurman, A., 245 Hafnia, 83 Harris, M.T., 3, 33, 83 He, J., 143 Holand, W., 235 Hot pressing, 181 Hu, M.Z., 3, 33, 101 Hulsenberg, D., 189
Imhoff, D., 277 Interface, 277 Interstices, 209 Jayasundara, S., 33
Davila, L.P., 209 de Almeida, V.F., 101 Dezellus, A., 277 Dielectric, 59 Diffusion, 101
Kear, B.H., 171 JShatri, L., 3 Kondo, H., 181 Kuntz, J., 161 Kusunose, T., 181
Eldror, I., 245
Ceramic Nanomaterials and Nanotechnology
293
Lam,Y-F, 33 Lin, J.S., 33 Lithium aluminosilicate,221 Luyten, J., 197 Mache, Th., 189 Maurel, E, 277 Mechanical activation, 129 Microlithography, 22 1 Microstructure, 181, 197, 235 Miele, P., 93 Monosize, 23 Mukherjee, A.K., 161, 171 Mullens, S., 197 Multilayer ceramic capacitor (MLCC), 59 Multiphase electrodispersion, 83 Nakayama, T., 181 Nanocable, 93 Nanocomposite, 197 Nanotube, 161 Nanowire, 93 Nickel oxide, 83 Nickel, 143 Niihara, K., 181 Nuclear Magnetic Resonance (NMR), 33 Nucleation, 22 1
Reflectors, 22 1 Ren, R., 129 Rheinberger,V., 235 Rietveld method, 75 Risbud, S.H., 209 Samet, L., 277 Samson, B.N., 265 Saulig-Wenger, K., 93 Schoenung, J.M., 143 Schweiger, M., 235 Sekino, T., 181 Shackelford, J.F., 209 Shaw, L.L., 129 Silica, 23, 33, 209 Silicate, 245 Silicalite-1, 3 Silicon carbide, 93 Silicon nitride, 181 Smolders, C., 197 Sol-gel, 23 Spark plasma sintering, 161 Spinel, 265 Stabilized zirconia, 75 Statistical mechanics, 101 Stojanovic, B.D., 75
Quartz, 209
Templated synthesis, 3 Terry, T.L., 83 Thermal spray coating, 143 Titania, 171 Titania, 83 Transmission Electron Microscope (TEM), 3923 Transparent, 265 Transport properties, 101 Trichet, M.-E, 277 Tungsten oxide, 129
Ramamoorthy, R., 245 Rast, H.E. 59
Varela, J.A., 75 Virtual processing, 101
Paiva-Santos, C.O., 75 Pannhorst, W., 221 Payzant, E.A., 3 Perazolli, L., 75 Pinckney, L.R., 265 Plasma spraying, 171 Plastic deformation, 245 Pozhar, L.A., 101
294
Ceramic Nanomaterials and Nanotechnology
Vitreous, 209 Wada, M., 181 Wan, J., 161,171 Wang, J., 265 Wei, W.-C. J., 23
Yamamoto, Y., 181 Yang, Z., 129 Zeolite, 3, 209 Zhan, G.-D., 161, 171 Zinc oxide, 83 Zirconia, 75, 197, 245
Ceramic Nanomaterials and Nanotechnology
295