Availableonlineatwww.sciencedirect.com MATERALS DIRECT LETTERS ELSEVIER Materials Letters 58(2004)1288-1291 Preparation of mullite whiskers from topaz decomposition Ping Peng b School of Materials Science and Engineering. The University of New South Wales, Sydney NSw 2052, Australia bCFC Key Laboratory. National University of Defense Technology: Hunan 410073, China Received 10 September 2003: accepted 23 September 2003 An inexpensive, simple method to prepare the pure mullite whiskers from natural topaz has been introduced. XRD and sEM were used to alyse the structure and morphology. a possible growth mechanism for whiskers has been suggested. C 2004 Elsevier B. V. All rights reserved. Keywords: Whiskers; Mullite: Topaz:: Ceramics: SEM 1. Introduction whiskers. Moyer made a whisker with 100 um, but with aspect ratio of less than 15 [7]. More usually, they Whiskers are monocrystals in the form of filaments. possess an average length of less than 30 um and aspect Because of small diameters and crystalline perfection, they ratio of less than 35 with a rare exception to 100 um of possess extremely high strengths. In addition, their aspec length. Many methods are based on the mixture of ratios can be considerably high, which become an advan- Al2O3, SiO2 and AlF, with fluorotopaz being an inter- of the reinforcements. However their finne mediate product [7-14]. Meanwhile, natural topaz has creates the handling difficulty as well as potential hazard been found to be a source of mullite whisker after to health. The published data [1] points out that around 1 thermal decomposition since 1994 [19, 20]. The present m is just the size to lead to illness, while larger than 1 work is investigating a simpler and less expensive meth- um and much smaller sizes have less such problem. Thus, od toward producing more acceptable mullite whiskers making thick whiskers with the high aspect ratio is from natural topaz. expected to find wide application for fabricating composite materials as well as for thermal insulation and other high temperature environment. 2. Experiment Considering continued demands for high temperature eramic. composite, mullite whiskers, as a candidate for 2.1. Preparation reinforcement, entail great potential due to good refrac torinese, high creep resistance, low coefficient of thermal lydroxyl-bearing topaz concentrate sand from Torring expansion and chemical stability in its own right [2]. ton, Australia was used as the starting material, which Numerous methods have thus been developed to produce consists of 96-97 wt. topaz, with 2-3 wt. quartz and mullite whiskers [3-18 and most of them are expensive. about 1% of other minerals including micas and wolframite So far, whiskers with 50-200 um are reported growing [20]. The raw topaz was processed either by wet ball milling on alumina particle [5], which means they are not free several hou S WI ith 10-mm-high alumina balls, then dried and crushed into powders or just grinding by a mortar and pestle, followed by mixing with 0.5 wt. of AlF3 powders Corresponding author. lan Wark Research Institute, University of in ethanol and dried under about 100c. the resulted Adelaide sa 5095. Australia. Tel: +61-8 powder was pressed into ca. 15-mm-diameter pellet at a 3023692;fax:+61-8-83023683 pressure of 20 MPa, with the pressure being held for 15 s or E-mailaddress:ppeng21st(@yahoo.com.au(P.Peng just loosely packed in an alumina crucible. As shown in Fig 0167-577X/S- see front matter c 2004 Elsevier B V. All rights reserved. doi:10.1016 j. mallet2003.09
Preparation of mullite whiskers from topaz decomposition Ping Penga,b,*, Chris Sorrell a a School of Materials Science and Engineering, The University of New South Wales, Sydney NSW 2052, Australia bCFC Key Laboratory, National University of Defense Technology, Hunan 410073, China Received 10 September 2003; accepted 23 September 2003 Abstract An inexpensive, simple method to prepare the pure mullite whiskers from natural topaz has been introduced. XRD and SEM were used to analyse the structure and morphology. A possible growth mechanism for whiskers has been suggested. D 2004 Elsevier B.V. All rights reserved. Keywords: Whiskers; Mullite; Topaz; Ceramics; SEM 1. Introduction Whiskers are monocrystals in the form of filaments. Because of small diameters and crystalline perfection, they possess extremely high strengths. In addition, their aspect ratios can be considerably high, which become an advantage of being the reinforcements. However, their fineness creates the handling difficulty as well as potential hazard to health. The published data [1] points out that around 1 Am is just the size to lead to illness, while larger than 1 Am and much smaller sizes have less such problem. Thus, making thick whiskers with the high aspect ratio is expected to find wide application for fabricating composite materials as well as for thermal insulation and other high temperature environment. Considering continued demands for high temperature ceramic composite, mullite whiskers, as a candidate for reinforcement, entail great potential due to good refractoriness, high creep resistance, low coefficient of thermal expansion and chemical stability in its own right [2]. Numerous methods have thus been developed to produce mullite whiskers [3 –18] and most of them are expensive. So far, whiskers with 50 –200 Am are reported growing on alumina particle [5], which means they are not free whiskers. Moyer made a whisker with 100 Am, but with the aspect ratio of less than 15 [7]. More usually, they possess an average length of less than 30 Am and aspect ratio of less than 35 with a rare exception to 100 Am of length. Many methods are based on the mixture of Al2O3, SiO2 and AlF3 with fluorotopaz being an intermediate product [7– 14]. Meanwhile, natural topaz has been found to be a source of mullite whisker after thermal decomposition since 1994 [19,20]. The present work is investigating a simpler and less expensive method toward producing more acceptable mullite whiskers from natural topaz. 2. Experiment 2.1. Preparation Hydroxyl-bearing topaz concentrate sand from Torrington, Australia was used as the starting material, which consists of 96– 97 wt.% topaz, with 2– 3 wt.% quartz and about 1% of other minerals including micas and wolframite [20]. The raw topaz was processed either by wet ball milling for several hours with 10-mm-high alumina balls, then dried and crushed into powders or just grinding by a mortar and pestle, followed by mixing with 0.5 wt.% of AlF3 powders in ethanol and dried under about 100 jC. The resulted powder was pressed into ca. 15-mm-diameter pellet at a pressure of 20 MPa, with the pressure being held for 15 s or just loosely packed in an alumina crucible. As shown in Fig. 0167-577X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.matlet.2003.09.046 * Corresponding author. Ian Wark Research Institute, University of South Australia, Mawson Lakes, Adelaide SA 5095, Australia. Tel.: +61-8- 83023692; fax: +61-8-83023683. E-mail address: ppeng21st@yahoo.com.au (P. Peng). www.elsevier.com/locate/matlet Materials Letters 58 (2004) 1288 – 1291
P. Peng, C. Sorrell/Materials Letters 58(2004)1288-1291 1289 Compressed Air the decomposition of topaz and the reactions are supposed 6AL2SO4(F075,OH025)2→2(3A2O3·2SiO2)↓ +2SiF4↑+HF↑+H2O (1) 5% NaOH 6AlSiO4( Fo7s, OH02s+ 54SiF4+ 27HF 27H.0 Dish ↑+42H2O个 i Mullite Tube Fig. 1. Schematic diagram of apparatus for producing mullite whisker. 2(3Al2O3·2SiO2)4+56siF41+28HF↑+28H2O个 and other possible reactions I, the decomposition experiments were carried out at a programmed heating profile with maximum temperature AlF,+H,0-AIOF +2HF T of air, which was supplied by compressed air pump and measured by counting the bubble in a paraffin oil, SiF4+2H2O←SO2+4HF↑ sometimes, water. NaoH water solution (5%) was used to absorb the off-gases from the system Chemical decomposition, phase mobility and kinetics of crystallisation controlled the formation mechanism of whis- 2.2. Characterisation ker. In topaz, the ratio of Al/Si is 2, and 3 for mullite. Thus the release of Si from topaz is necessary for mullite and size of the whiskers were observed by formation, which just happened as the above reaction, SEM ( JEOL JSM-840)and a chemical analysis was carried giving off SiF4. However, the proper retention of SiF4 can out by energy-dispersive X-ray analyser (link AN 10 000), stimulate the transformation of mullite from topaz according using sintered mullite(Commercial Minerals, Australia) as to Eqs. (2)and(3). So a slight flow of ambient air is standard. The crystalline phases in the fired samples were expected to function as a trap of off-gases from topaz and identified by XRD(Philip 1140/00) keep the essential moisture of the system for Eq(4).As a result, AlF3 is used to provide a supplement of Al in order to suppress polymerisation of the silica component(Eq (5)as 3. Results and discussion confirmed by the experiment. Moreover, Al2O3-rich mullite pellets was found to grow with whiskers in absence of Fig. 2a is the representative SEM image of mullite flowing air(Fig 2d) when most of SiF4 escaped the reaction whiskers, formed on the surface of fired pellet, with the zone as soon as they were formed On the other hand, when maximum length of 250 um or more. In addition, most of using water as a substitute of paraffin oil over moist air led them are thicker than 3 um, which correspond to the aspect to formation of a silica film via liquid phase on the surface ratio of ca. 30, while some whiskers were twinned crystals. of topaz without any whisker(Fig. 2e), where Eq(4)was The average composition ratio of Al/Si in whiskers was the domain reaction. the same as that of mullite standard. The XRD pattern Investigation was then made to the site of the growth of Fig. 3 indicates the characteristic orthorhombic mullite line whiskers. The results show that there were more AlO3 in (PDF#15-0776 these sites as shown in Fig. 2c and f. For each growing site. The high magnification SEM image(Fig. 2b and c) for delicate radiating growths, or starbursts, of mullite whiskers ig. 2a reveals that the shape of mullite is tetragonal with a with high aspect ratio, average ca. 100 um in length, are facet on the extremity, with two triangular faces connecting illustrated in Fig. 2f-h. It can be presumed that mullite together on sharing one edge nuclei may form randomly on the surface of topaz particles he crystal system of topaz is also orthorhombic [21], at the very beginning, but sample with free space will have which consists of chains parallel to the c-axis of AlO4 F2 high nucleus density. The as-resulting mullite crystal may octahedra which are cross-linked by independent SiO4 in turn protect the topaz beneath from further decomposi tetrahedra. Such an intimate mixture of Al, Si and F in tion. Thus, further decomposition was likely to happen or atomic scale provides a unique opportunity for pure mullite the opposite site of mullite crystal with high density on formation. Many research works have been devoted to study each considerable size particle of topaz, while small par-
1, the decomposition experiments were carried out at a programmed heating profile with maximum temperature 1300 jC for 4 h and some for 1 h under continuous flow of air, which was supplied by compressed air pump and measured by counting the bubble in a paraffin oil, or sometimes, water. NaOH water solution (5%) was used to absorb the off-gases from the system. 2.2. Characterisation The shape and size of the whiskers were observed by SEM (JEOL JSM-840) and a chemical analysis was carried out by energy-dispersive X-ray analyser (link AN 10 000), using sintered mullite (Commercial Minerals, Australia) as standard. The crystalline phases in the fired samples were identified by XRD (Philip 1140/00). 3. Results and discussion Fig. 2a is the representative SEM image of mullite whiskers, formed on the surface of fired pellet, with the maximum length of 250 Am or more. In addition, most of them are thicker than 3 Am, which correspond to the aspect ratio of ca. 30, while some whiskers were twinned crystals. The average composition ratio of Al/Si in whiskers was the same as that of mullite standard. The XRD pattern in Fig. 3 indicates the characteristic orthorhombic mullite lines (PDF #15-0776). The high magnification SEM image (Fig. 2b and c) for Fig. 2a reveals that the shape of mullite is tetragonal with a facet on the extremity, with two triangular faces connecting together on sharing one edge. The crystal system of topaz is also orthorhombic [21], which consists of chains parallel to the c-axis of AlO4F2 octahedra which are cross-linked by independent SiO4 tetrahedra. Such an intimate mixture of Al, Si and F in atomic scale provides a unique opportunity for pure mullite formation. Many research works have been devoted to study the decomposition of topaz and the reactions are supposed to be [20,22] 6Al2SiO4ðF0:75; OH0:25Þ2 ! 2ð3 Al2O32SiO2Þ # þ 2SiF4z þ HFz þ H2Oz; ð1Þ and other possible reactions: AlF3 þ H2O ! AlOF þ 2HFz ð4Þ SiF4 þ 2H2O X SiO2 þ 4HFz ð5Þ Chemical decomposition, phase mobility and kinetics of crystallisation controlled the formation mechanism of whisker. In topaz, the ratio of Al/Si is 2, and 3 for mullite. Thus the release of Si from topaz is necessary for mullite formation, which just happened as the above reaction, giving off SiF4. However, the proper retention of SiF4 can stimulate the transformation of mullite from topaz according to Eqs. (2) and (3). So a slight flow of ambient air is expected to function as a trap of off-gases from topaz and keep the essential moisture of the system for Eq. (4). As a result, AlF3 is used to provide a supplement of Al in order to suppress polymerisation of the silica component (Eq. (5)) as confirmed by the experiment. Moreover, Al2O3-rich mullite pellets was found to grow with whiskers in absence of flowing air (Fig. 2d) when most of SiF4 escaped the reaction zone as soon as they were formed. On the other hand, when using water as a substitute of paraffin oil over moist air led to formation of a silica film via liquid phase on the surface of topaz without any whisker (Fig. 2e), where Eq. (4) was the domain reaction. Investigation was then made to the site of the growth of whiskers. The results show that there were more Al2O3 in these sites as shown in Fig. 2c and f. For each growing site, delicate radiating growths, or starbursts, of mullite whiskers with high aspect ratio, average ca. 100 Am in length, are illustrated in Fig. 2f– h. It can be presumed that mullite nuclei may form randomly on the surface of topaz particles at the very beginning, but sample with free space will have high nucleus density. The as-resulting mullite crystal may in turn protect the topaz beneath from further decomposition. Thus, further decomposition was likely to happen on the opposite site of mullite crystal with high density on each considerable size particle of topaz, while small parFig. 1. Schematic diagram of apparatus for producing mullite whisker. ð2Þ ð3Þ P. Peng, C. Sorrell / Materials Letters 58 (2004) 1288–1291 1289
P. Peng, C. Sorrell Materials Letters 58(2004)1288-1291 828318kvX1:8918H,M028 927418KV3,3981ysWD28 845428KV83,50018W029 e1122eKvX2,28818ymM028 e17720KX901evW028 (g2010301090(0606 012820Kvx130010yaWD20 835420Kx8581FnW026 082728KvX1,68818DMD2e 013528KvX1,3816M028 (1) Fig. 2. SEM image of as-fired samples (a-c, f-h) Pellet surface, ball milling topaz, 1300C for 4 h under flowing air. (d) Pellet surface, ball milling topaz, 1300C for 4 h without flowing air.(e)Loose pack, ground in mortar, 1300C for 4 h under flowing air passing through water. (i)Pellet cross section, ball milling topaz, 1300C for I h under flowing air. O) Loose pack, ball milling topaz, 1300C for 4 h under flowing air. (k)Vertical cross section, ball milling topaz, 1300C for 4 h under flowing air. (I) Horizontal cross section, ball milling topaz, 1300C for 4 h under flowing ai ticles would decompose thoroughly with mullite joining nomenon. At the later stage, due to the less SiF4 releasing that on the large particles at earlier stage. On the other from few topaz residue and the loss of SiF4, a phase with hand, once the decomposition built up, the reaction was high alumina formed as appeared in the experiments. The self-catalysed by SiF4, which can move up and down fracture section(Fig. 21) of the incomplete decomposition depending on the flowing air and the concentration gradient pellet, fired at 1300 for only 1 h, can be evidence of the around the sample. At this stage, the reduction of surface above assumption. The increase of the free spac energy became a domain effect on mullite deposition create more crystal seed and result in forming within gas zone, thus newly produced small mullite crystal whisker as shown in Fig. 2j, where the sample is is likely to bond to the formed larger crystal rather than packed for firing form new crystal. Consequently, the as-formed whiskers The fracture sections are found to be highly porous and developed in both length and thickness. Besides, tempera- often have difficulty in taking a clear photo due to their ture in already formed crystal should be lower than the site poor conductivity. Fig. 2k is the SEM image of vertical undertaking decomposition, which also create an additional cross section, and Fig. 21 for horizontal cross section. It is ondition for crystal formation due to the saturation phe- obvious that whiskers also grow inside the pellet and it
ticles would decompose thoroughly with mullite joining that on the large particles at earlier stage. On the other hand, once the decomposition built up, the reaction was self-catalysed by SiF4, which can move up and down depending on the flowing air and the concentration gradient around the sample. At this stage, the reduction of surface energy became a domain effect on mullite deposition within gas zone, thus newly produced small mullite crystal is likely to bond to the formed larger crystal rather than form new crystal. Consequently, the as-formed whiskers developed in both length and thickness. Besides, temperature in already formed crystal should be lower than the site undertaking decomposition, which also create an additional condition for crystal formation due to the saturation phenomenon. At the later stage, due to the less SiF4 releasing from few topaz residue and the loss of SiF4, a phase with high alumina formed as appeared in the experiments. The fracture section (Fig. 2i) of the incomplete decomposition pellet, fired at 1300 for only 1 h, can be evidence of the above assumption. The increase of the free space may create more crystal seed and result in forming thinner whisker as shown in Fig. 2j, where the sample is loosely packed for firing. The fracture sections are found to be highly porous and often have difficulty in taking a clear photo due to their poor conductivity. Fig. 2k is the SEM image of vertical cross section, and Fig. 2l for horizontal cross section. It is obvious that whiskers also grow inside the pellet and it Fig. 2. SEM image of as-fired samples. (a – c, f – h) Pellet surface, ball milling topaz, 1300 jC for 4 h under flowing air. (d) Pellet surface, ball milling topaz, 1300 jC for 4 h without flowing air. (e) Loose pack, ground in mortar, 1300 jC for 4 h under flowing air passing through water. (i) Pellet cross section, ball milling topaz, 1300 jC for 1 h under flowing air. (j) Loose pack, ball milling topaz, 1300 jC for 4 h under flowing air. (k) Vertical cross section, ball milling topaz, 1300 jC for 4 h under flowing air. (l) Horizontal cross section, ball milling topaz, 1300 jC for 4 h under flowing air. 1290 P. Peng, C. Sorrell / Materials Letters 58 (2004) 1288–1291
P. Peng, C. Sorrell/Materials Letters 58(2004)1288-1291 5000 Acknowledgements One of the authors(Ping Peng) received a fellowship from China Scholarship Council(CSC). This is gratefully acknowledged. The authors would like to thank Mr John 1000 Sharp, Mr. Peter Legge, Ms. Viera Piegerova and Dr Charlie Kong for their help in SEM analysis. Ms. Inna Bolkovsky, Ms. Cathy Lau and Ms. Jane Gao are also Degree 2-Theta thanked for their assistance in several aspects XRD pattern of mullite 4000 References [1] A.R. Bunsell, M H. Berger, Fine Ceramic Fibers, Marcel Dekker, [2] H. Schneider, K. Okada, J. Pask, Mullite and Mullite Ceramics, Wi- 3] M.F. Souza, J. Yamamoto, I. Regiani, C.O. Paiva-Santos, D P.F. Sou za, P. Sao, J. Am. Ceram. Soc. 83(1)(2000)60. Degree 2-Theta 44 W.M. Kriven, M.H. Jilavi, D. Zhu, J.K. R. Weber, B Cho, J. Felten, XRD pattern of whisker P.C. Nordine, Ceram Microstruct. Control At Level, Proc. Int Ma- ter.symp,(1996)169 5 H Katsuki, H Ichinose, S Furuta, J. Ceram. Soc. Jpn. 104(8)(1996) 6A. Sayir, S.C. Farmer, Mater Res. Soc. Symp Proc. 365(1995)11 [7 J.R. Moyer, NN. Hughes, J Am Ceram Soc. 77(4)(1994)1083 18JR. Moyer, P.R. Rudolf, J. Am. Ceram Soc. 77(4)(1994)1087 9]K Okada, N.T. Otsuka, J. Am. Ceram Soc. 74(10)(1991)2414 seems that the fracture is prone to happen on the whiskers- (10 G. Talmy, D.A. Haught, EP 0428565,(1991). [11] D.A. Haught, I.G. Talmy, D Divecha, S. Karmarkar, Mater. Sci. Eng implies the high fracture strength for whiskers from the A144(1991)207. single crystal structure. [12]G. Talmy, D.A. Haught, US 4911902, (1990) [13] J.R. Moyer, M.S. Labarge, B.D. Brubaker, NN. Hughes, Proc. Sat ellite Symp. 2, Advanced Structural Inorganic Composites, 7th Int Meeting on Modern Ceramics Technologies, Mater. Sci. Monogr. 4. Conclusion vol 68, Elsevier, Amsterdam, 1990, p. 57 [14 G. Talmy, D.A. Haught, US 4948766, (1990) Mullite whiskers with an average length of 100 um [5] M.G. M.U. Ismail, H Arai, Z Nakai, T. Akiba, J Am Ceram Soc. 73 and maximum length of 250 um can be prepared from decomposition of the natural topaz. The suggested method [16 K Okada, N.T. Otsuka, J Mater. Sci. Lett. 8(9)(1989)1052. [7]H. Katsuki, S. Furuta, H. Ichinose, H. Nakao, J. Ceram. Soc. Jpn. 96 is competitive compared with many other methods due to (11)(1988)108 its low cost and simplicity. The whiskers are proved to be [18] D.S. Perera, G. Allott, J Mater. Sci. Lett. 4(10)(1985)1270 orthorhombic mullite by XRD, while the morphology of [19] D.M. Buck, Int Ceram Monogr. I(1994)612 the whiskers in different conditions provided some clues [20] R.A. Day, E.R. Vance, D.J. Cassidy, JS.Hartman, JMater Res 10 for its growth mechanism. A further study is needed to (11)(1995)2963 explore the detailed mechanism of whisker formation 221] C. Klein, C.S. Hurlbut, Manual of Mineralogy, Wiley, New York, []X. Miao, C.C. Sorrell, J. Mater. Sci. Lett. 17(1998)2087
seems that the fracture is prone to happen on the whiskersgrowing site, rather than the whiskers themselves, which implies the high fracture strength for whiskers from the single crystal structure. 4. Conclusion Mullite whiskers with an average length of 100 Am and maximum length of 250 Am can be prepared from decomposition of the natural topaz. The suggested method is competitive compared with many other methods due to its low cost and simplicity. The whiskers are proved to be orthorhombic mullite by XRD, while the morphology of the whiskers in different conditions provided some clues for its growth mechanism. A further study is needed to explore the detailed mechanism of whisker formation later. Acknowledgements One of the authors (Ping Peng) received a fellowship from China Scholarship Council (CSC). This is gratefully acknowledged. The authors would like to thank Mr. John Sharp, Mr. Peter Legge, Ms. Viera Piegerova and Dr. Charlie Kong for their help in SEM analysis. Ms. Inna Bolkovsky, Ms. Cathy Lau and Ms. Jane Gao are also thanked for their assistance in several aspects. References [1] A.R. Bunsell, M.H. Berger, Fine Ceramic Fibers, Marcel Dekker, New York, 1999. [2] H. Schneider, K. Okada, J. Pask, Mullite and Mullite Ceramics, Wiley, Chichester, 1994. [3] M.F. Souza, J. Yamamoto, I. Regiani, C.O. Paiva-Santos, D.P.F. Souza, P. Sao, J. Am. Ceram. Soc. 83 (1) (2000) 60. [4] W.M. Kriven, M.H. Jilavi, D. Zhu, J.K.R. Weber, B. Cho, J. Felten, P.C. Nordine, Ceram. Microstruct.: Control At. Level, Proc. Int. Mater. Symp., (1996) 169. [5] H. Katsuki, H. Ichinose, S. Furuta, J. Ceram. Soc. Jpn. 104 (8) (1996) 788. [6] A. Sayir, S.C. Farmer, Mater. Res. Soc. Symp. Proc. 365 (1995) 11. [7] J.R. Moyer, N.N. Hughes, J. Am. Ceram. Soc. 77 (4) (1994) 1083. [8] J.R. Moyer, P.R. Rudolf, J. Am. Ceram. Soc. 77 (4) (1994) 1087. [9] K. Okada, N.T. Otsuka, J. Am. Ceram. Soc. 74 (10) (1991) 2414. [10] G. Talmy, D.A. Haught, EP 0428565, (1991). [11] D.A. Haught, I.G. Talmy, D. Divecha, S. Karmarkar, Mater. Sci. Eng., A 144 (1991) 207. [12] G. Talmy, D.A. Haught, US 4911902, (1990). [13] J.R. Moyer, M.S. Labarge, B.D. Brubaker, N.N. Hughes, Proc. Satellite Symp. 2, Advanced Structural Inorganic Composites, 7th Int. Meeting on Modern Ceramics Technologies, Mater. Sci. Monogr., vol. 68, Elsevier, Amsterdam, 1990, p. 57. [14] G. Talmy, D.A. Haught, US 4948766, (1990). [15] M.G.M.U. Ismail, H. Arai, Z. Nakai, T. Akiba, J. Am. Ceram. Soc. 73 (9) (1990) 2736. [16] K. Okada, N.T. Otsuka, J. Mater. Sci. Lett. 8 (9) (1989) 1052. [17] H. Katsuki, S. Furuta, H. Ichinose, H. Nakao, J. Ceram. Soc. Jpn. 96 (11) (1988) 1081. [18] D.S. Perera, G. Allott, J. Mater. Sci. Lett. 4 (10) (1985) 1270. [19] D.M. Buck, Int. Ceram. Monogr. 1 (1994) 612. [20] R.A. Day, E.R. Vance, D.J. Cassidy, J.S. Hartman, J. Mater. Res. 10 (11) (1995) 2963. [21] C. Klein, C.S. Hurlbut, Manual of Mineralogy, Wiley, New York, 1977. [22] X. Miao, C.C. Sorrell, J. Mater. Sci. Lett. 17 (1998) 2087. Fig. 3. P. Peng, C. Sorrell / Materials Letters 58 (2004) 1288–1291 1291