Availableonlineatwww.sciencedirectcom ScienceDirect E噩≈RS ELSEVIER Journal of the European Ceramic Society 27(2007)679-682 www.elsevier.comlocate/jeurceramsoc Further tailoring of material properties in non-equimolar aluminium titanate ceramic materials LD. Alecu.r. stead Rojan Advanced Ceramics Pry Ltd, PO Box 7126, Spearwood, WA 6163, Australia Available online 26 May 2006 The authors started their r&d activities related to aluminium titanate(at) ceramics in the early 1990s by addressing the critical problems of aluminium titanate, such as thermodynamical instability and micro-cracking. After that, their focus has permanently been on improving, optimizing powders. This paper presents their latest investigations on AT ceramics produced by reaction-sinteringon quin and tailoring the properties of the aluminium titanate ceramic materials produced by reaction-sintering of ee n-equimolar mixtures of the two main ponents. The objective of these investigations has been to emphasize the effects of excesses of either alumina or titania on the internal stresses build up in these materials and the consequences these internal stresses could have on some material properties which are most important for the industrial applications of AT ceramics. The ways and the extent in which these internal stresses could be controlled/managed were investigated, with the ultimate goal of expanding the possibilities for further tailoring the thermo-mechanical properties of the aluminium titanate ceramics, so that they meet specific application requirements. o 2006 Elsevier Ltd. All rights reserved. Keywords: Non-equimolar; Al2 TiOs; Powders-solid state reaction; Strength; Structural applications 1. ntroduction As a consequence of a pronounced thermal expansion anisotropy, intergranular micro-cracking can occur in alu Rojan Advanced Ceramics, based in Perth, Western Aus- minium titanate. This can decrease the strength of the material tralia, has been involved in R&D activities related to aluminium on one side, and increase its toughness(and thermal shock resis- titanate for more than a decade and currently produces a wide tance)on the other range of AT materials and products which are exported through The authors found previously that, for average grain sizes out the world, mainly for applications in the non-ferrous metal- around, or slightly above the critical grain size for intergranular micro-cracking, both a reasonably good strength and a substan- Whole families of aluminium titanate ceramic materials were tial gain in toughness can be achieved veloped, based on compositions containing equimolar ratios According to these findings, a trade is possible between these of Al]O3 and TiO2. Small amounts of various additives were also antagonistic parameters, and specific AT materials with proper ed with the aim of improving the thermodynamic stability at ties tailored to suite particular applications can be developed intermediate temperatures within each of the families mentioned above Aluminium titanate has inherently a relatively low mechani- Further developments at Rojan were able to achieve increases cal strength. Although the AT strength is higher than that of some in the strength of our AT ceramics of up to 2.5 times, without competing materials, such as calcium silicate and fused silica, a penalty on the thermal shock resistance. This was attained by this strength is much lower than that of another main competitor, refining the powder processing and ceramic forming technolo- sialon gies, and, finally, by optimizing the firing programs. In addition, AThas still some advantages over sialon, as it is much cheaper a ceramic composite material based on aluminium titanate was and a much better thermal insulator. However, any increase in also developed with a mechanical strength five times higher than the strength of At is more than welcome the common at material. 2 This paper is presenting the results of a recent short-term Corresponding author. Tel: +61 1155: fax: +61 1156. R&d programme, which has investigated the effects of devia E-mailaddress:ioana@rojan.com.au(LD.Alecu). tions from the equimolar composition of aluminium titanate on 0955-2219/S-see front matter o 2006 Elsevier Ltd. All rights reserved. doi: 10. 1016/j-jeurceramsoc. 2006.04.082
Journal of the European Ceramic Society 27 (2007) 679–682 Further tailoring of material properties in non-equimolar aluminium titanate ceramic materials I.D. Alecu ∗, R.J. Stead Rojan Advanced Ceramics Pty Ltd., P.O. Box 7126, Spearwood, WA 6163, Australia Available online 26 May 2006 Abstract The authors started their R&D activities related to aluminium titanate (AT) ceramics in the early 1990s by addressing the critical problems of aluminium titanate, such as thermodynamical instability and micro-cracking. After that, their focus has permanently been on improving, optimizing and tailoring the properties of the aluminium titanate ceramic materials produced by reaction-sintering of equimolar mixtures of alumina and titania powders. This paper presents their latest investigations on AT ceramics produced by reaction-sintering of non-equimolar mixtures of the two main components. The objective of these investigations has been to emphasize the effects of excesses of either alumina or titania on the internal stresses build up in these materials and the consequences these internal stresses could have on some material properties which are most important for the industrial applications of AT ceramics. The ways and the extent in which these internal stresses could be controlled/managed were investigated, with the ultimate goal of expanding the possibilities for further tailoring the thermo-mechanical properties of the aluminium titanate ceramics, so that they meet specific application requirements. © 2006 Elsevier Ltd. All rights reserved. Keywords: Non-equimolar; Al2TiO5; Powders-solid state reaction; Strength; Structural applications 1. Introduction Rojan Advanced Ceramics, based in Perth, Western Australia, has been involved in R&D activities related to aluminium titanate for more than a decade and currently produces a wide range of AT materials and products which are exported throughout the world, mainly for applications in the non-ferrous metallurgy. Whole families of aluminium titanate ceramic materials were developed, based on compositions containing equimolar ratios of Al2O3 and TiO2. Small amounts of various additives were also used with the aim of improving the thermodynamic stability at intermediate temperatures. Aluminium titanate has inherently a relatively low mechanical strength. Although the AT strength is higher than that of some competing materials, such as calcium silicate and fused silica, this strength is much lower than that of another main competitor, sialon. AT has still some advantages over sialon, as it is much cheaper and a much better thermal insulator. However, any increase in the strength of AT is more than welcome. ∗ Corresponding author. Tel.: +61 8 9437 1155; fax: +61 8 9437 1156. E-mail address: ioana@rojan.com.au (I.D. Alecu). As a consequence of a pronounced thermal expansion anisotropy, intergranular micro-cracking can occur in aluminium titanate. This can decrease the strength of the material on one side, and increase its toughness (and thermal shock resistance) on the other. The authors found previously that, for average grain sizes around, or slightly above the critical grain size for intergranular micro-cracking, both a reasonably good strength and a substantial gain in toughness can be achieved. According to these findings, a trade is possible between these antagonistic parameters, and specific AT materials with properties tailored to suite particular applications can be developed within each of the families mentioned above.1 Further developments at Rojan were able to achieve increases in the strength of our AT ceramics of up to 2.5 times, without a penalty on the thermal shock resistance. This was attained by refining the powder processing and ceramic forming technologies, and, finally, by optimizing the firing programs. In addition, a ceramic composite material based on aluminium titanate was also developed with a mechanical strength five times higher than the common AT material.2 This paper is presenting the results of a recent short-term R&D programme, which has investigated the effects of deviations from the equimolar composition of aluminium titanate on 0955-2219/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2006.04.082
680 L.D. Alecu, R.J. Stead Journal of the European Ceramic Sociery 27(2007)679-682 the material properties that are most relevant for the industrial 3.55 plications of AT ceramics. Some preliminary results were pre viously reported 34 2. Experimental mima o tania excesses ranging from 090 to 100% were investi-: E gated. Two series of water-based slurries were prepared by using T alumina and titania powder mixtures with 0%0, 5%0, 15%0, 30%0, 93.15 50%0, 75% and 100%o alumina, respectively titania excess Sta-3. izing additives were also used in the same amounts as for the Several prismatic blocks 40 mm x 40 mm x 100 mm were Percentage of alumina/titania excess pressure-cast from each slurry, and then let to dry at room tem- perature for about I week Fig. 1. Sintered bulk density vs alumina/titania excess. A first complete series of 14 blocks was then fired with a dwell of 4 h at 1400C. Because some blocks with higher compo- Eight to ten MoR bars were prepared from each nent excesses sintered under these conditions developed heavy equimolar AT material variety. Each MoR bar was then subjected cracks and, in the worst case, even disintegrated, repeat firir to two consecutive four-point bending tests. The average of the were performed on identical spare blocks with dwells of 6.8 16-20 measured values was used in the strength versus compo- and even 10h at 1400oC. until the sintered blocks remaine nent excess and sintering temperature graphs Two other complete series of 14 prismatic blocks were subse- 3. Results and discussion quently fired to 1450 and 1500C, respectively, with dwell times of 4h. All sintered blocks, irrespective of the level of component The density and porosity values showed gradual variations imbalance remained intact this time with the alumina, respectively titania excess(see Figs. 1-3, The sintered prismatic blocks had first the dense and hard for samples fired at 1450 C/4 h; the theoretical density of the skin ground off, in order to expose the bulk of the material. equimolar aluminium titanate is Dr=3.68 kg/dm3) They were then cut into samples that were used for material Some preliminary XRD results were presented elsewhere. property investigations. About 10 MoR bars, as well as samples The results obtained so far show first some gradual change or XRD and SEM analyses were prepared from each of these in the lattice parameters with increasing component excesses blocks Then, for higher excesses, they indicate the presence of Density and porosity were measured by Archimedes method ondary phases formed by the component in excess. The complete according to the Australian Standard AS 1774.5-1989, by fol- results of the XRD investigations will make the object of afuture lowing the evacuation procedure he XRD investigations have been performed at Curtin SEM results showed that the 1400oC/4 h sintering resulted University of Technology in Perth, Western Australia, with a in at grain sizes between 1 and 4 um and no micro-cracks.The and they are still continuing only occasional intergranular micro-cracking. After sintering at The SEM samples were prepared at Rojan. They were pol- ished with diamond grits going down to 0. 25 um average particle size, by using a Kent 3 automatic lapping and polishing machine manufactured by Kemet International. The polishing process was checked by using a Kyowa(Tokyo, Japan)optical micro- scope. The SEM examinations, including EDAX analyses, have 2 3 been performed at the University of Western Australia in Perth, ith a Philips 505 scanning electron microscope, and they are still continuing The modulus of rupture(MoR)was measured in four-point bending tests at Rojan, by using an X-Tran machine and load cell, while the rupture load was registered by an Ozy-Dyn apparatus The dimensions of the mor bars were 3 mm x 4 mm x 50 mm The distance between the upper load points was 7 mm, while the lower load points were 20 mm apart. The width and thickness of Percentage of alumina /titania excess the bar at the rupture point were measured with digital callipers, -+Alumina excess.-Titania excess with an accuracy of 0.01 mm. Fig. 2. Apparent solid density vs alumina/titania excess
680 I.D. Alecu, R.J. Stead / Journal of the European Ceramic Society 27 (2007) 679–682 the material properties that are most relevant for the industrial applications of AT ceramics. Some preliminary results were previously reported.3 2. Experimental Samples of non-equimolar AT compositions with either alumina or titania excesses ranging from 0% to 100% were investigated. Two series of water-based slurries were prepared by using alumina and titania powder mixtures with 0%, 5%, 15%, 30%, 50%, 75% and 100% alumina, respectively titania excess. Stabilizing additives were also used in the same amounts as for the equimolar AT. Several prismatic blocks 40 mm × 40 mm × 100 mm were pressure-cast from each slurry, and then let to dry at room temperature for about 1 week. A first complete series of 14 blocks was then fired with a dwell of 4 h at 1400 ◦C. Because some blocks with higher component excesses sintered under these conditions developed heavy cracks and, in the worst case, even disintegrated, repeat firings were performed on identical spare blocks with dwells of 6, 8 and even 10 h at 1400 ◦C, until the sintered blocks remained intact. Two other complete series of 14 prismatic blocks were subsequently fired to 1450 and 1500 ◦C, respectively, with dwell times of 4 h. All sintered blocks, irrespective of the level of component imbalance, remained intact this time. The sintered prismatic blocks had first the dense and hard skin ground off, in order to expose the bulk of the material. They were then cut into samples that were used for material property investigations. About 10 MoR bars, as well as samples for XRD and SEM analyses were prepared from each of these blocks. Density and porosity were measured by Archimedes’ method according to the Australian Standard AS 1774.5-1989, by following the evacuation procedure. The XRD investigations have been performed at Curtin University of Technology in Perth, Western Australia, with a Siemens D500 diffractometer, by using the Cu K radiation, and they are still continuing. The SEM samples were prepared at Rojan. They were polished with diamond grits going down to 0.25 m average particle size, by using a Kent 3 automatic lapping and polishing machine manufactured by Kemet International. The polishing process was checked by using a Kyowa (Tokyo, Japan) optical microscope. The SEM examinations, including EDAX analyses, have been performed at the University of Western Australia in Perth, with a Philips 505 scanning electron microscope, and they are still continuing. The modulus of rupture (MoR) was measured in four-point bending tests at Rojan, by using an X-Tran machine and load cell, while the rupture load was registered by an Ozy-Dyn apparatus. The dimensions of the MoR bars were 3 mm × 4 mm × 50 mm. The distance between the upper load points was 7 mm, while the lower load points were 20 mm apart. The width and thickness of the bar at the rupture point were measured with digital callipers, with an accuracy of 0.01 mm. Fig. 1. Sintered bulk density vs. alumina/titania excess. Eight to ten MoR bars were prepared from each nonequimolar AT material variety. Each MoR bar was then subjected to two consecutive four-point bending tests. The average of the 16–20 measured values was used in the strength versus component excess and sintering temperature graphs. 3. Results and discussion The density and porosity values showed gradual variations with the alumina, respectively titania excess (see Figs. 1–3, for samples fired at 1450 ◦C/4 h; the theoretical density of the equimolar aluminium titanate is DT = 3.68 kg/dm3). Some preliminary XRD results were presented elsewhere.4 The results obtained so far show first some gradual changes in the lattice parameters with increasing component excesses. Then, for higher excesses, they indicate the presence of secondary phases formed by the component in excess. The complete results of the XRD investigations will make the object of a future report. SEM results showed that the 1400 ◦C/4 h sintering resulted in AT grain sizes between 1 and 4 m and no micro-cracks. The 1450 ◦C/4 h sintering produced AT grains of 3–6 m size and only occasional intergranular micro-cracking. After sintering at Fig. 2. Apparent solid density vs. alumina/titania excess.
L.D. Alecl, R. Stead Joumal of the European Ceramic Society 27(2007)679-682 543 Eo 5050 Percentage of alumina /titania excess Percentage of titania excess -Alumina excess---Titania excess 1400c1450 Fig 3. Apparent(open) porosity vs alumina/titania excess. Fig. 5. Flexural strength vs titania exces 1500C for 4 h, the AT grain size was 5-10 um and micro-crack The corresponding samples with 75%o and 100%o alumina free domains of 40-50 um were surrounded by pronounced excess were heavily cracked after being sintered at 1400C for intergranular, and sometimes, trans-granular micro-cracks. Sec- 4 h only, so that it was not possible to measure their strength ondary phases were apparent in samples with high component The data represented on the same row of the graph in Fig 4 for excesses. The complete results of the SEM investigations will samples no 6 and no. 7 were obtained on repeat samples which make the object of a separate report were sintered at the same temperature, 1400C, but for 6 and The MoR results showed a general tendency of increase in 8h, respectively the four-point bending strength of the non-equimolar aluminium Fig 5 shows a chart representing the variation of the bending titanate materials with increasing alumina and titania excesses strength with the amount of titania excess. again, the row of darker columns represents the results on samples sintered at Fig 4 presents a chart showing the variation of the bending 1450C for 4 h, and shows a gradual increase in strength with strength with the amount of alumina excess. The row of darker increasing titania excesses for 4 h and shows a monotonous increase of the material strength e Once again, the other row of columns in the graph in Fig. 5 columns represents the results for samples sintered at 1450C with increasing alumina excess The data for the first three samples(with 0%o, 5%o and 15% titania The row of lighter columns in Fig. 4 represents the results excess) were obtained after a sintering at 1400C for 4 h and of the 1400C firings. The data for samples no. 1(zero alu- show a pronounced tendency towards significant increases in mina excess)to no 5(50%o alumina excess)represent the results strength with increasing titania excess obtained for the 1400 C/4 h sintering and show an increasingly The other samples with larger titania excesses did not allow pronounced tendency towards large increases in strength with any measurement of their strength, as they were heavily distorted increasing alumina excess. and cracked. After sintering at 1400C for I h only, a repeat sample with 100%o titania excess has kept exfoliating and hurling small chips meters away for several days, whilst sitting on a laboratory bench, until a small, heavily cracked, egg-shaped core The data represented in the row of lighter columns on the graph in Fig. 5 for samples no 4 to no.7 were obtained on repeat samples, which were sintered again at 1400C, but for 6h(sam- ples no 4 and no 5),&h(sample no 6) and even 10h(sample no. 7)in order to preserve their integrity It became clear that with increasing alumina/titania excess, larger and larger internal stresses can accumulate and remain preserved within these materials, leading in some cases to for- mation of large individual cracks and, in the worst cases to the complete destruction of the test pieces Percentage of alumina excess Aluminium titanate is an"artificial"material that is syn- 1400c1450c thetised at high temperature(1300C)by the direct reaction of alumina and titania. it is worth stressing here that during Fig. 4. Flexural strength vs alumina excess the AT forming reaction a volume expansion of about 10.7
I.D. Alecu, R.J. Stead / Journal of the European Ceramic Society 27 (2007) 679–682 681 Fig. 3. Apparent (open) porosity vs. alumina/titania excess. 1500 ◦C for 4 h, the AT grain size was 5–10 m and micro-crack free domains of 40–50 m were surrounded by pronounced intergranular, and sometimes, trans-granular micro-cracks. Secondary phases were apparent in samples with high component excesses. The complete results of the SEM investigations will make the object of a separate report. The MoR results showed a general tendency of increase in the four-point bending strength of the non-equimolar aluminium titanate materials with increasing alumina and titania excesses (see Figs. 4 and 5). Fig. 4 presents a chart showing the variation of the bending strength with the amount of alumina excess. The row of darker columns represents the results for samples sintered at 1450 ◦C for 4 h and shows a monotonous increase of the material strength with increasing alumina excess. The row of lighter columns in Fig. 4 represents the results of the 1400 ◦C firings. The data for samples no. 1 (zero alumina excess) to no. 5 (50% alumina excess) represent the results obtained for the 1400 ◦C/4 h sintering and show an increasingly pronounced tendency towards large increases in strength with increasing alumina excess. Fig. 4. Flexural strength vs. alumina excess. Fig. 5. Flexural strength vs. titania excess. The corresponding samples with 75% and 100% alumina excess were heavily cracked after being sintered at 1400 ◦C for 4 h only, so that it was not possible to measure their strength. The data represented on the same row of the graph in Fig. 4 for samples no. 6 and no. 7 were obtained on repeat samples which were sintered at the same temperature, 1400 ◦C, but for 6 and 8 h, respectively. Fig. 5 shows a chart representing the variation of the bending strength with the amount of titania excess. Again, the row of darker columns represents the results on samples sintered at 1450 ◦C for 4 h, and shows a gradual increase in strength with increasing titania excesses. Once again, the other row of columns in the graph in Fig. 5 represents the results obtained on samples sintered at 1400 ◦C. The data for the first three samples (with 0%, 5% and 15% titania excess) were obtained after a sintering at 1400 ◦C for 4 h and show a pronounced tendency towards significant increases in strength with increasing titania excess. The other samples with larger titania excesses did not allow any measurement of their strength, as they were heavily distorted and cracked. After sintering at 1400 ◦C for 1 h only, a repeat sample with 100% titania excess has kept exfoliating and hurling small chips meters away for several days, whilst sitting on a laboratory bench, until a small, heavily cracked, egg-shaped core was left. The data represented in the row of lighter columns on the graph in Fig. 5 for samples no. 4 to no.7 were obtained on repeat samples, which were sintered again at 1400 ◦C, but for 6 h (samples no. 4 and no. 5), 8 h (sample no. 6) and even 10 h (sample no. 7) in order to preserve their integrity. It became clear that with increasing alumina/titania excess, larger and larger internal stresses can accumulate and remain preserved within these materials, leading in some cases to formation of large individual cracks and, in the worst cases, to the complete destruction of the test pieces. Aluminium titanate is an “artificial” material that is synthetised at high temperature (≥1300 ◦C) by the direct reaction of alumina and titania. It is worth stressing here that during the AT forming reaction a volume expansion of about 10.7%
L.D. Alecu, R.J. Stead Journal of the European Ceramic Sociery 27(2007)679-682 takes place. This goes against the volume shrinkage(35%) Some studies(see, e.g. Segadaes et al. )showed that a com- associated with sintering. The competition of these two antago- positional imbalance, specifically an alumina excess, can accel nistic processes leads to development of internal stresses in the erate the decomposition of AT. The explanation offered by some reaction-sintered aluminium titanate authors was that the excess alumina acts as nucleation centres There is yet another prominent cause for internal stresses for the reaction of decomp ing built up in the reaction-sintered aluminium titanate. Both This is, most probably, true. But the driving force for decom- alumina and titania start sintering at temperatures lower tha lan position in such cases is related, in our opinion, to the internal the 1300C threshold for aluminium titanate reaction of for- stresses caused by the compositional imbalance, as explained mation. That means that some degrees of sintering take place here above. When no or not enough stress relaxation is provided within the two main components, alumina and titania, before these internal stresses remain preserved within the reaction- the reaction of AT formation takes off. The volume increase due sintered AT material, and there will be a natural tendency for to this reaction is thus constrained by the rigid structures of the it to decompose and thus eliminate the stresses and minimize partly sintered components the intemal energy of the system. Alternatively, when an effec- Another series of sintering experiments was performed on tive relaxation of the internal stresses takes place during the equimolar alumina and titania mixtures, in which two different reaction-sintering cycle, a more thermodynamically stable alu ramp rates were used between 1000 and 1300C minium titanate should result Our interpretation is also consistent with early studies, which low rate: 150 C/h, which would allow some sintering of the showed that higher sintering temperatures and longer dwell times yielded thermodynamically more stable AT materials.6 high rate: 300C/h, aimed at less sintering of the components prior to the onset of the AT forming reaction. 4. Conclusion The low rate sintering experiments systematically delivered These studies open the way for further possibilities of AT materials with higher strength values, in agreement with our loring the aluminium titanate ceramic materials according to expectations requirements of specific applications. Again there is a trade pos This effect will be amplified if there is a significant composi- sible between higher strength on one side and higher toughness tional imbalance of the two main components, i.e. when there is (and thermal shock resistance)on the other, as well as between a large excess of either alumina or titania. Titania starts sintering higher strength on one side and higher thermodynamic stability at lower temperatures than alumina and that is why this effect is on the other more pronounced in high titania excess, reaction-sintered AT. Such internal stresses, if preserved within the sintered mate- Referenc rial, will have on one hand a hardening effect, with a possible increase in mechanical strength. On the other hand. there will 1. Alecu. L.D. and Stead. R. J. Aluminium titanate ceramics--an attractive certainly be an increased brittleness and a higher probability of solution. In Proceedings of the 30th Internat. Symp. on Automotive Technology cracks and failure and Auto-motion, ISATA 97, Paper 97NM061, June 1997 n order to avoid the preservation of high internal stresses 2. Alecu. I. D. Cilia. R.A., Dean, G. A. Reuben, R. Stead, R. J. and win R F, New developments in aluminium titanate ceramics and refractories In some measures had to be taken that would promote stress relax- Proceedings of the Seventh ECerS Conference. Key Engineering Materials ation. This was attempted in repeat sintering experiments for vols.206-23.2002,pp.1705-1710 samples with large alumina/titania excesses, by either provid- 3. Alecu, I D and Stead,R J. Non-equimolar aluminium titanate In Proceed- ing for longer dwell times for sintering at lower temperatures ings of the Australasian Ceram. Soc. Cony. AUSTCERAM 2004, December (1400C), or, alternatively, by using a higher sintering temper ature(1450or1500°C) 4. Alecu, L.D., Stead, R.J. and Hart, R D, XRD investigations on non-equimolar uminium titanate. In Proceedings of the Australasian Ceram. Soc. Co Increased sintering times seemed to allow a significant stress AUSTCERAM 2004, December 2004 relaxation, so that samples with high alumina/titania excesses 5. Segadaes, A. M, Morelli, M. R. and Kiminami, R. G. A, Combus. were no longer subject to cracking. However, as expected, the tion synthesis of aluminium titanate. J. Eur Ceram Soc., 1998. 18. 771 strength of these samples was somewhat diminished. Similar 6. Thomas. H.A. and Stevens. R. Aluminium titanate-a literature review. changes were obtained when sintering was performed at higher Part 2. Engineering properties and thermal stability. Br Ceram. Trans. J temperatures(see Figs 4 and 5) 1989,88(4),184-190
682 I.D. Alecu, R.J. Stead / Journal of the European Ceramic Society 27 (2007) 679–682 takes place. This goes against the volume shrinkage (∼35%) associated with sintering. The competition of these two antagonistic processes leads to development of internal stresses in the reaction-sintered aluminium titanate. There is yet another prominent cause for internal stresses being built up in the reaction-sintered aluminium titanate. Both alumina and titania start sintering at temperatures lower than the ∼1300 ◦C threshold for aluminium titanate reaction of formation. That means that some degrees of sintering take place within the two main components, alumina and titania, before the reaction of AT formation takes off. The volume increase due to this reaction is thus constrained by the rigid structures of the partly sintered components. Another series of sintering experiments was performed on equimolar alumina and titania mixtures, in which two different ramp rates were used between 1000 and 1300 ◦C: • low rate: 150 ◦C/h, which would allow some sintering of the components; • high rate: 300 ◦C/h, aimed at less sintering of the components prior to the onset of the AT forming reaction. The low rate sintering experiments systematically delivered AT materials with higher strength values, in agreement with our expectations. This effect will be amplified if there is a significant compositional imbalance of the two main components, i.e. when there is a large excess of either alumina or titania. Titania starts sintering at lower temperatures than alumina and that is why this effect is more pronounced in high titania excess, reaction-sintered AT. Such internal stresses, if preserved within the sintered material, will have on one hand a hardening effect, with a possible increase in mechanical strength. On the other hand, there will certainly be an increased brittleness and a higher probability of cracks and failure. In order to avoid the preservation of high internal stresses some measures had to be taken that would promote stress relaxation. This was attempted in repeat sintering experiments for samples with large alumina/titania excesses, by either providing for longer dwell times for sintering at lower temperatures (1400 ◦C), or, alternatively, by using a higher sintering temperature (1450 or 1500 ◦C). Increased sintering times seemed to allow a significant stress relaxation, so that samples with high alumina/titania excesses were no longer subject to cracking. However, as expected, the strength of these samples was somewhat diminished. Similar changes were obtained when sintering was performed at higher temperatures (see Figs. 4 and 5). Some studies (see, e.g. Segadaes et al.5) showed that a compositional imbalance, specifically an alumina excess, can accelerate the decomposition of AT. The explanation offered by some authors was that the excess alumina acts as nucleation centres for the reaction of decomposition. This is, most probably, true. But the driving force for decomposition in such cases is related, in our opinion, to the internal stresses caused by the compositional imbalance, as explained here above. When no or not enough stress relaxation is provided, these internal stresses remain preserved within the reactionsintered AT material, and there will be a natural tendency for it to decompose and thus eliminate the stresses and minimize the internal energy of the system. Alternatively, when an effective relaxation of the internal stresses takes place during the reaction-sintering cycle, a more thermodynamically stable aluminium titanate should result. Our interpretation is also consistent with early studies, which showed that higher sintering temperatures and longer dwell times yielded thermodynamically more stable AT materials.6 4. Conclusion These studies open the way for further possibilities of tailoring the aluminium titanate ceramic materials according to requirements of specific applications. Again there is a trade possible between higher strength on one side and higher toughness (and thermal shock resistance) on the other, as well as between higher strength on one side and higher thermodynamic stability on the other. References 1. Alecu, I. D. and Stead, R. J., Aluminium titanate ceramics—an attractive solution. InProceedings of the 30th Internat. Symp. on Automotive Technology and Auto-motion, ISATA’97, Paper 97NM061, June 1997. 2. Alecu, I. D., Cilia, R. A., Dean, G. A., Reuben, R., Stead, R. J. and Wing, R. F., New developments in aluminium titanate ceramics and refractories. In Proceedings of the Seventh ECerS Conference, Key Engineering Materials, vols. 206–213, 2002, pp. 1705–1710. 3. Alecu, I. D. and Stead, R. J., Non-equimolar aluminium titanate. In Proceedings of the Australasian Ceram. Soc. Conf., AUSTCERAM 2004, December 2004. 4. Alecu, I. D., Stead, R. J. and Hart, R. D., XRD investigations on non-equimolar aluminium titanate. In Proceedings of the Australasian Ceram. Soc. Conf., AUSTCERAM 2004, December 2004. 5. Segadaes, A. M., Morelli, M. R. and Kiminami, R. G. A., Combustion synthesis of aluminium titanate. J. Eur. Ceram. Soc., 1998, 18, 771– 781. 6. Thomas, H. A. J. and Stevens, R., Aluminium titanate—a literature review. Part 2. Engineering properties and thermal stability. Br. Ceram. Trans. J., 1989, 88(4), 184–190