Mineral-Oxide-Doped Aluminum Titanate Ceramics with Improved Thermomechanical Properties

Investigations were carried out, on the effect of addition of kaolinite (2Al2O3⋅3SiO2⋅2H2O) and talc (Mg3Si4O10(OH)2) in terms of bulk density, XRD phases, microstructure, as well as thermal and mechanical properties of the aluminium titanate (AT) ceramics. AT ceramics with additives have shown enhanced sinterability at 1550C, achieving close to 99% of TD (theoretical density) in comparison to 87% TD, exhibited with pure AT samples sintered at 1600C, and found to be in agreement with the microstructural observations. XRD phase analysis of samples with maximum densities resulted in pure AT phase with a shi in unit cell parameters suggesting the formation of solid solutions. TG-DSC study indicated a clear shi in AT formation temperature with talc addition. Sintered specimens exhibited signi�cant reduction in linear thermal expansion values by 63% (0.42 × 10/C, (30–1000C)) with talc addition. ermal hysteresis of talc-doped AT specimens showed a substantial increase in hysteresis area corresponding to enhanced microcrack densities which in turn was responsible to maintain the low expansion values. Microstructural evaluation revealed a sizable decrease in crack lengths and 200% increase in �exural strength with talc addition. Results are encouraging providing a stable formulation with substantially enhanced thermomechanical properties.


Introduction
Aluminum Titanate (Al 2 TiO 5 , designated as AT) ceramics exhibit excellent thermal properties such as low thermal expansion coefficient in the range of 1.0-1.5 × 10 −6 /C (RT-1000 ∘ C) in combination with low thermal conductivity (∼1.5 Wm −1 K −1 ) and a high melting temperature of 1860 ∘ C [1][2][3][4]. is makes Al 2 TiO 5 , a material of choice for many refractory applications. Some of these applications include wall �ow �lters for diesel particulate emission control, exhaust port liners in automotive engines and thermal shock resistant refractory parts for nonferrous metallurgical industries [5][6][7][8]. One of the disadvantages of this ceramic material for practical applications is the low �exural strength due to the extensive microcracks generated whil processing [3,4,9]. Crystal structure of aluminum titanate ( -Al 2 TiO 5 ) is pseudo-brookite and is associated with the strong anisotropy in crystallographic axis while heating, which is responsible for microcracking on cooling [9,10]. e microcracking phenomenon is closely related to the material microstructure. Below a critical grain size, the elastic energy of the system is insufficient for microcrack formation during cooling and thus the mechanical properties are considerably enhanced. Additionally, eutectoid decomposition to its parent oxides such as -Al 2 O 3 and TiO 2 between 750 to 1280 ∘ C is also an issue leading to thermal instability of the Al 2 TiO 5 phase. is decomposition occurs when the adjacent Al 3+ (0.54 Å) and Ti 4+ (0.67 Å) octahedra collapse, because of the lattice site occupied by the Al 3+ ion is too large. e thermal energy available from this collapse allows Al 3+ to migrate from its position and causes structural dissolution to rutile (TiO 2 ) and corundum ( -Al 2 O 3 ) [9][10][11][12].
In view of the above, the objective of this paper is to elucidate systematically the effect of additives such as kaolinite and talc into Al 2 TiO 5 formulation with varying concentrations followed by the evaluation of thermomechanical properties. Accordingly, precursor oxide samples were compacted with and without additives and were subjected to sintering at varying temperatures to achieve close to theoretical density. Sintering behavior of AT was elucidated with dilatometric shrinkage curve and DSC studies. Sintered samples were characterized for their density, phase, microstructure, and CTE measurements. Further, the thermal hysteresis was recorded while heating and cooling using dilatometer. Flexural (3-point bend) strength was also determined and correlated with microcrack densities, observed with fractograph.

Experimental Procedure
Basic raw materials such as alumina (Al 2 O 3 , Baikowski, France) and titania (TiO 2 , Qualigens) powders along with the additives kaolinite (2Al 2 O 3 ⋅3SiO 2 ⋅2H 2 O) and talc (Mg 3 Si 4 O 10 (OH) 2 ) were used for the investigations. XRD phase analysis of the raw materials was carried out by Bruker's D8 advanced system and morphology and particle size analysis by scanning electron microscope (S-4300SE/N, Hitachi, Tokyo, Japan). Physiochemical properties of the raw materials are depicted in Table 1. e concentration of silica (1.5, 3, and 4.5 wt%) in aluminum titanate and sample ID's (C 0 , C 1 , C 2 , C 3 , C 4 , C 5 , and C 6 ) were depicted in Table 2. e formulations were granulated with 2 wt% of poly vinyl alcohol as a binder and compacted into green compacts of 65 mm × 65 mm × 8 mm using a hydraulic press. e green density of the compacted samples was measured by the dimensional method and was found to be greater than 50% of the theoretical density (estimated by the rule of mixtures). In order to study, the extent of physiochemical changes occurring in green specimens over the processing temperature ranges, the specimens were subjected to the differential scanning calorimetric (DSC) analysis using TG-DSC analyzer (Netzsch, Germany) from room temperature to 1550 ∘ C. Basic formulation of Al 2 TiO 5 (C 0 ) has been subjected to dilatometry (Netzsch 402C, Germany) and the shrinkage pro�le was recorded with respect to the temperature. ree sintering temperatures of 1500 ∘ C, 1525 ∘ C, and 1550 ∘ C were selected for the sintering of all formulations and selection of optimum sintering temperature.
Densities of the samples sintered at different temperatures were evaluated by widely used Archimedes principle (ASTM 792) and phases were analysed by XRD (Bruker's D8 advanced XRD). e polished samples were thermally etched at a temperature, 50 ∘ C below the sintering temperature for microstructural observations using a scanning electron microscope (S-4300SE/N, Hitachi, Tokyo, Japan). e specimens were also subjected to dilatometric analysis using a push-rod-type dilatometer (Netzsch 402C, Germany) incorporating the sample holder correction to determine coefficient of linear expansion (CTE). ermal hysteresis was recorded for the samples of C 0 , C 3 , and C 5 formulations with additives which have exhibited the highest densities. Sintered samples with the highest densities were also machined to the rectangular specimens of 45 mm × 4 mm × 3 mm for the determination of the �exural strength using 3-point bend test following ASTM C-1161-02C (Instron). , and 1(d), respectively. XRD pattern of Al 2 O 3 indicated the coexistence of -phase (∼85%) as a major phase along with -phase (∼15%) and TiO 2 powder has shown single anatase phase. SEM micrographs of both the basic powders have shown an irregular morphology with an average particle size of 150 and 300 nm, respectively.

�ensi�cation of Al 2 TiO 5 with and without Additives.
Variation of bulk densities with sintering temperature for all the compositions is shown in Figures 2(a) and 2(b). It is evident that the bulk density of Al 2 TiO 5 formulation without additives has shown only a marginal increase from 84 to 85% of TD with increase of sintering temperature from 1500 to 1550 ∘ C. No signi�cant increase in density is observed even at a sintering temperature of 1600 ∘ C. Composition, concentration, and sintering temperature are to found have a signi�cant effect on the �nal density values. Concentrations of both the additives are �xed based on their silica content (Table 1) in such a way that the �nal formulation corresponds to a silica addition of 1.5, 3, and 4.5% (Table 2). Kaolinite (2Al 2 O 3 ⋅3SiO 2 ⋅2H 2 O) addition from 2.9 to 8.8% has resulted in the consistent increase in density from 90% to a maximum density of 99% of TD. Talc (Mg 3 Si 4 O 10 (OH) 2 ) addition from 2.5 to 5% increases the density from 92 to 98.63%; however, unlike kaolinite, a higher concentration of talc (Mg 3 Si 4 O 10 (OH) 2 ) has not shown any signi�cant increase in density beyond 92%.
Dilatometric curve of Al 2 TiO 5 depicted in Figure 3(a) clearly shows an initial shrinkage corresponding to the in-situ reaction sintering of the starting mixture of Al 2 O 3 and TiO 2 followed by an expansion regime corresponding to formation of Al 2 TiO 5 and is in good agreement with endothermic peak at 1375 ∘ C (Figure 3(b)). is expansion regime in  Clay addition Talc addition F 2: (a) Density variation with temperature of C 0 , C 1 , C 2 , C 3 , and (b) Density variation with temperature of C 0 , C 4 , C 5 , and C 6 . fact o�sets the densi�cation of AT matrix by around 20%, resulting in poor sintered density of 87% even at a sintering temperature of 1600 ∘ C. Endothermic peak at 1386 ∘ C for C 3 , indicate that there is no signi�cant shi in Al 2 TiO 5 phase formation compared to undoped (C 0 ) formulation ( Figure  3(b)). Al 2 TiO 5 phase formation temperature is shied to a lower temperature range of 1274 ∘ C for C 5 compared to 1375 ∘ C observed with (C 0 ) un-doped formulation as is evident from the DSC peak (Figure 3(b)). XRD pattern of the maximum dense sample (C 3 ) depicted in Figure 4 has not shown any additional peaks and all the peaks could be indexed with standard Al 2 TiO 5 phase. However, a shi in unit cell constants is observed in comparison to the pure Al 2 TiO 5 samples, especially in the case of " " parameter [10,17]. e cell parameters and volume of C 0 sample are 4315 Å, 63 5 Å, 3 5 0 Å, and 326.35 (Å) 3 , where as cell parameters and volume of sample C 3 are 4342 Å, 6536 Å, 3 5 40 Å, and 327.32 (Å) 3 . e change in unit cell parameter is probably due to silicon (Si 4+ , ionic radius 0 41 Å) replacing Al 3+ in the lattice of Al 2 TiO 5 matrix. is may result in the multivalent state for titanium that is, Ti 3+ /Ti 4+ , corresponding to a stoichiometry of ((Al 3+ ,Ti 3+ ) 2 (Ti 4+ ) 1 (O 2− ) 5 ) which in turn enhances the sintering process. A closer look at the micrographs also reveals a change in the porous lamellar type of structure in to relatively dense faceted grains. Further, the microcracks are found visible and crack lengths are reduced to a greater extent in comparison to C 0 with no additives.
XRD pattern of C 5 sample with maximum density depicted in Figure 4 has shown ∼96% of Al 2 TiO 5 phase with minor quantities of Al 2 O 3 aer sintering at 1550 ∘ C. Further, the TiO 2 peaks are completely absent in the pattern. It is well known that talc (2MgO⋅2Al 2 O 3 ⋅5SiO 2 ) transform to clinoenstatite (MgSiO 3 ) close to 1100 ∘ C and at high temperatures, it decomposes to MgO and SiO 2 . In the case of C 5 formulation, Mg 2+ (MgO: 1.5%) and Si 4+ (SiO 2 : 3%)   ions undergo simultaneous lattice substitutions for Al 3+ to stabilize Al 2 TiO 5 stoichiometry. is leads to change in unit cell parameters due to addition of talc. e unit cell parameters and volume of sample C 0 are 9.4315 Å, 9.6385 Å, 3.590 Å, and 326.35 (Å) 3 . Whereas in sample C 5 unit cell parameters and volume are 9.4651 Å, 9.6715 Å, 3.5981 Å and 329.37 (Å) 3 , which displaces the interplanar spaces. e dilation along crystallographic c-axis in talc added samples, further expected to improve phase stability of AT [17]. A larger distortion from the C 0 composition can be attributed to the simultaneous substitution of Si 4+ (ionic radius 0.41 Å) and Mg 2+ (ionic radius 0.65 Å) replacing Al 3+ in the lattice of Al 2 TiO 5 matrix. Al 2 TiO 5 formation process was led by nucleation and growth of grains and �nally di�usion of reactant through the matrix and is controlled by the very slow reacting species. In addition to the multivalent state for titanium, that is, Ti 3+ /Ti 4+ as a result of Si 4+ substitution, the presence of Mg 2+ ions  variations in the grain size with both the additives (C 3 , C 5 ). However, considerable reduction in intergranular pores is observed with the addition of talc (C 5 ). e presence of additives enhanced the grain growth in AT ceramics.

ermal Expansion and ermal Hysteresis.
Dilatometric expansion curve and thermal hysteresis were recorded while heating and cooling of the C 0 , C 3 , and C 5 samples were shown in Figures 6 and 7, respectively. Table 3 lists the thermal expansion values of C 0 , C 3 , and C 5 formulations from 30-1000 ∘ C. It is evident that, the formulation C 0 has a CTE value of 1.09 × 10 −6 / ∘ C followed by C 3 with a marginally lower CTE value of 0.94 × 10 −6 / ∘ C(13% less than C 0 ) and the lowest value of 0.42 × 10 −6 / ∘ C (63% less than C 0 ) for C 5 formulation. ermal expansion curves of the entire samples initially contract till 450 ∘ C. ermal expansion curves of C 0 exhibit a dissimilar behavior in comparison with C 3 and C 5 formulation. C 0 sample exhibited a low expansion of −1.97 × 10 −6 / ∘ C in comparison to C 3 and C 5 for which CTE values are −1.71 × 10 −6 / ∘ C and −1.74 × 10 −6 / ∘ C, respectively. All samples have shown an expansion behavior beyond 450 ∘ C and C 0 exhibited a signi�cant slope change leading to the maximum expansion value. e temperature, beyond which the expansion becomes prominent, was regarded as the temperature at which healing of microcracks occurs that compensates the expansion effectively. Further, the curves corresponding to C 0 are tapered into a plateau beyond 900 ∘ C unlike other samples with a positive slope.
Cooling curves also behaved differently with all the three samples. C 0 formulation has shown a negative slope tapering into a plateau followed by the expansion. Initial contraction had values of 6.08, 7.75, and 3.04 × 10 −6 / ∘ C (temperature regime, 1000-450 ∘ C) and �nal expansion region had values of −4.37, −7.75, and −2.3 × 10 −6 / ∘ C (temperature regime, 450-100 ∘ C) for C 5 , C 3 , and C 0 respectively. Expansion below 450 ∘ C can be attributed to the reintroduction of micro-cracks healed while heating.
It is obvious that thermal properties of doped and undoped Al 2 TiO 5 ceramics are mainly governed by the presence of microcracks and are affected by the crack closure or healing. Areas of the thermal hysteresis were recorded during heating up to 1000 ∘ C and cooling up to 100 ∘ C for formulations of C 5 , C 3 , and C 0 were 95, 68 and 91 cm 2 , respectively. Unlike other samples, C 3 showed two closed loops where �rst loop is in the temperature regime of 1000 to 300 ∘ C, which had 0 value less than 0 value of sample while heating, as expected. However, the second loop formed at 300 ∘ C, exhibited a higher 0 value which is unusual. Probably this may be due to the transformation of microcracks into macro cracks due to high silica content (4.5%) leading to a relatively low hysteresis area of 68 cm 2 observed with C 3 sample in comparison to 91 cm 2 observed with C 0 sample having close CTE values [17]. C 5 sample exhibited hysteresis area of 95 cm 2 , which correlates well with the low CTE value (0.42 × 10 −6 / ∘ C) as a result of compensation of expansion values. Microcrack density (number of microcracks/unit area estimated using several SEM images) of C 5 sample was almost 3 times of that of the C 0 sample con�rming the above observations.

Flexural Strength and Hardness Measurements.
Typical load-displacement curves obtained from the specimens C 0 , C 3 and C 5 which were subjected to 3-point bend loading and is shown in Figure 8. ough a minimum of 4-5 specimens were tested in each case, for the sake of clarity, only one loaddisplacement curve for each condition are shown. Flexural strength ( ) is calculated from the load displacement data as = 3 2PL/bd 2 , where max is the maximum load, the span length, the width, and the thickness of the specimens. e �exural strength values are given in Table  4. C 5 with talc doping exhibited the peak �exural strength before fracture (average, 28 ± 3 MPa) which is ∼200% higher than the corresponding �exural strength values (average 8 ± 4 MPa) of the C 3 and C 0 samples. It is evident from the data in Figure 8 that C 0 and C 3 materials show a similar unstable crack extension with load drop aer attaining peak �exural stress. A close look at the load-displacement curves obtained from the �exural loading of the C 3 and C 0 samples (shown in Figure 8) has shown several load excursions while on attaining peak load depicting fracture over a wide range of strain followed by a gradual load drop. Poor strength in fracture behavior even with a high density (99% of TD) can be attributed to increase in crack length and more number of macrocracks (crack length up to 30 m, Figure 5(b)). It is obvious that poor density (85%) and the high porosity are the factors that contribute in addition to the microcracks to the identical fracture behavior observed with C 0 compositions.
Unlike the other samples, C 5 composition exhibited initial rapid increase in the stress, reaching a peak value followed by the fracture. It is clear that, prior to the attainment of the peak �exural stresses, the material showed almost a linear increase in �exural stress with strain, indicating elastic deformation, before �nal failure. is can be attributed to the association of the existing large number of micro cracks ( Figure 5(c), microcracks are restricted within the grain with maximum crack length <10 m) leading to the stable crack extension. Finally, the material fails with sudden load drop, which is an indication of rapid propagation of the macrocracks that resulted from coalescence of large number of micro cracks under local tensile loading. Hardness of the specimens C 3 and C 5 has shown an enhancement 20% and 33.3% with respect to C 0 . A relatively high hardness under identical densities could attribute to the marginal decrease in grain size observed with C 5 samples.
SEM fractographs obtained from the specimens C 0 , C 3 , and C 5 tested till the failure under �exural (3-point bend loading) testing are shown in Figures 9(a)-9(f). ough large number of fractographs at different magni�cations is obtained in each of the specimen, for the sake of clarity, only one representative fractograph is depicted. ese fractographs show an identical transgranular fracture. As discussed above, C 5 composition is associated with more microcracks and the crack path while propagation needs to relocate for further crack extension as it encounters more and more microcracks. Such process needs higher fracture energy as compared to the C 0 and C 3 where microcrack densities were low.

Conclusions
Addition of kaolinite (2Al 2 O 3 .3SiO 2 .2H 2 O) and talc (Mg 3 Si 4 O 10 (OH) 2 ) in relatively low concentration of 8.8 wt% and 5 wt%, respectively, resulted in enhanced densi�cation leading to a substantial increase in percentage theoretical density to 99% in comparison with pure Al 2 TiO 5 with 87% processed under identical conditions. Kaolinite and/or talc substitution results in multivalent titanium (Ti 3+ /Ti 4+ ) and oxygen vacancies in Al 2 TiO 5 formulations promoting the enhanced diffusion and densi�cation. Microstructural evaluations revealed transformation of porous lamellar type of structure of pure Al 2 TiO 5 into relatively dense faceted grains with kaolinite and talc con�rming higher density values. XRD studies have shown an improvement in Al 2 TiO 5 phase content from 92% to a maximum of 98% with the addition of kaolinite. e enhancement in phase content was moderate in case of talc with a maximum of 95.5%; however, DSC studies indicated a drop of ∼85 ∘ C in phase formation temperature. Substantial improvement in thermomechanical properties was observed with talc addition in comparison to kaolinite. A decrease in thermal expansion value of Talc-doped AT by 63% and an enhancement of �exural strength value by 200% is demonstrated in the present study. ese properties found to have a good correlation with the presence and the mode of microcracks as revealed by microstructure and thermal hysteresis. Talc-doped Al 2 TiO 5 formulation presently developed with superior thermomechanical properties can be explored for various potential applications.