Most of ceramics for high temperature applications such as refractory, ceramics filters and so on are damaged by a rapid temperature change, which largely reduces the usability and productivity because the heating and cooling rate should be less than the critical heating/cooling rates of the thermal shock [1]. In general, it takes several to tens hours just for setting the furnaces/device to required temperatures. Different degree of thermal expansion along with the thermal gradient in the ceramics causes such thermal shocks, therefore low-expansion ceramics are widely studied to reduce the energy usages and to improve the productivity.

Aluminum titanate (AT) ceramic is known to be an excellent thermal shock resistant ceramics [2]. Therefore, many applications as refractory materials have been expected for the AT ceramics. The use of AT ceramics as refractory is, however, restricted because of the low fracture strength and thermal decomposition in the temperature range from 800˚C to 1280˚C. Many attempts to improve the thermal stability have been made so far [3-13]. The present authors have reported that the AT ceramics doped with alkali feldspar ((Na_y,K_1−y)AlSi₃O₈, y = 0.5 - 0.8) exhibited not only low thermal expansion coefficient comparable to non-doped AT ceramics but also high thermal stability, high refractoriness, and relatively large fracture strength [14]. However, even with the feldspar-doped AT ceramics, prolonged thermal treatment at decomposition temperature range over several hundred hours results in the complete degradation into alumina and titania.

It is reported that MgTi₂O₅ (MT) shows excellent resistivity to the thermal decompositions [15]. However, its thermal expansion coefficient is reported as ~5 × 10⁻⁶ K⁻¹, which is much larger than AT ceramics. Attempt was made to reduce the thermal expansion coefficient without a lack of its thermal stability by a formation of AT-MT solid solutions [16-20]. Hereafter, AT-MT ceramic is abbreviated as MAT in this article. Prolonged heat treatment for over several hundred hours at 1200˚C - 1400˚C eventually induces complete decomposition of MAT ceramics. Accordingly, decomposition free MAT ceramics with low thermal expansion coefficient and high mechanical strength have been strongly desired. In the present study, the structure and properties of the feldspar-doped MAT ceramics were investigated. The effects of the feldspar addition on the improvement of mechanical strength and thermal stability are discussed.

Starting oxide powders of MgCO₃ (Kamishima, Japan), TiO₂ (rutile, Sakai Kagaku, Japan) and Al₂O₃ (corundum, Sumitomo Chemical, Japan) were weighed in a molar ratio of (1 − x)Al₂TiO₅-xMgTi₂O₅ (x = 0 - 1) and 4 wt% of alkali feldspar (Fukushima Choseki, Japan) was added to the mixture. The chemical composition of alkali feldspar used in the present study was analyzed as (Na_0.6,K_0.4) AlSi₃O₈ by fluorescence X-ray analysis (Rigaku ZSX- 100e, Rh source). Hereafter, (1 − x)Al₂TiO₅-xMgTi₂O₅ is abbreviated to MAT100x (x value in %, indicating a fraction of MT component) and feldspar-doped MAT ceramics is to f-MAT100x. For example, f-MAT25 indicates the ceramics of feldspar-doped 0.75Al₂TiO₅- 0.25MgTi₂O₅ composition. Alkali feldspar is simply called feldspar for convenience. The mixture of the starting reagents with 0.5 wt% peptizer Aron A-6114 (Toa Gosei, Japan) and 30 wt% water was put into the alumina pot with alumina balls, and mixed for 5 hours by using a planetary mill (Fritsch Pulversette 5, Germany). After being dried, the mixture with 10 wt% of a binder M30 (Kyoeisha Kagaku, Japan) was molded into 10 ´ 10 ´ 40 mm shape under a pressure of 60 MPa and subjected to drying at 180˚C for 2 hours. The molded samples were calcined at 340˚C for 4 hours and successively at 700˚C for 2 hours to remove organic matters completely. Then, the samples were sintered at 1500˚C for 2 hours to form MAT ceramics.

A degree of decomposition was estimated by the method reported in Ref. [14]. SEM-EDX analysis of f-MAT75 was carried out with JEOL 6500F with 15 kV acceleration voltage. Three point flexure strengths of the MAT and f-MAT ceramics were evaluated with Minerva TG-10kN testing machine according to JIS-1601. The values are estimated by averaging five measurements. Experiment error was estimated below 10%. Porosities were estimated by a conventional Archimedian method according to JIS R2205-74. Kerosene was used as a liquid media. The values were estimated by averaging three measurements and error was estimated below 5%. Thermal expansion coefficients (CTE) of the ceramics were measured by a thermal mechanical analysis equipment (Rigaku TMA 8227, Japan). CTE value was estimated from a difference of sample length at 20˚C and 800˚C. Formation temperatures of MAT0, MAT50, f-MAT0 and f-MAT50 were measured with differential scanning calorimeter (Rigaku DSC 8270, Japan).