It is well known that a solid solution forms in the a-SiC (2H)-AIN system over a wide composition range, because both of them have wurtzite (2H) structures [1]. The lattice parameters of the solid solution vary with the composition. It has been reported that their behaviour approximately follows Vegard's law [2]. Although SiC-A1N solid solutions are prepared by hot-pressing of both powders [3], they can also be obtained directly by the carbothermal reduction of oxides having the SiO2-A1203 compositions in nitrogen. In the carbothermal reduction process, the intimate mixing of SiO2, A1203, and carbon is desirable. So far, two attempts at the preparation of solid solutions by carbothermal reduction have been reported; SiC-A1N from the mixture of fumed silica, AI(OH)3, and starch [4, 5], and SiC-A1N-A12OC from acidtreated kaolinite and starch [5]. Aluminosilicate materials, for example kaolinite and zeolite, consist of silicon, aluminium and oxygen, which are mixed at the atomic level. Of these, montmorillonite shows a layered structure and takes up a number of organics in its interlayer space to form "intercalation compounds" [6]. In montmorilloniteorganic intercalation compounds, silicate and organic layers are interstratified alternately. Hence, the intimate mixing between SiO2, A1203, and carbon is expected after their thermal transformation into a nonoxidative atmosphere. We prepared n-alkylammoniummontmorillonite-polyacrylonitrile (PAN) intercalation compounds and applied these to the carbothermal reduction processes for fl-sialon [7-9] and SiC production [10]. In such processes, the intimate mixing and the two-dimensional ordering in the intercalation compounds affected the reactions to a great extent. In the present study, the preparation of a SiC-A1N solid solution from the montmorillonitePAN intercalation compound by carbothermal reduction was attempted. To investigate the effect of the mixing condition on solid solution formation, montmorillonite-carbon mixtures were also converted. The amount of carbon in the mixtures was varied to study its influence on solid solution formation. The method for the preparation of the intercalation compound was described elsewhere [7]. The compound after thermal pre-treatment at 220°C was characterized by X-ray powder diffraction (XRD) analysis (Rigaku, RAD II-A, Mn filtered FeK~ radiation) and thermogravimetry (TG) (Shimadzu, TGA-20); the basal spacing was 1.75nm and the amount of carbon from PAN approximately corresponded to that in the mixture containing 53 wt % carbon. The procedure for the preparation of the mixtures was the same as described elsewhere [10]. The carbon contents were 40, 50, and 60 wt %. These indicate that the calculated C/SiO2 molar ratios were 5.8, 8.2, and 12.0, respectively; the C/A1203 ratios were 21, 37, and 55, respectively. Both kinds of starting materials were placed in a graphite boat, and heated in a nitrogen flow (flow rate: 900 ml min~) for 0 to 10 h. Crystalline phases were identified by XRD analysis after decarbonization at 650 ° C for 6 h. The characterization of 2H compounds was based on the profile and the position of their (1 0 0) lines. The position was precisely measured by using NaC1 as an internal standard and compared with those calculated from the reported d values for a-SiC (2H) [11] and A1N [12] (a-SiC (2H), 42.56°; A1N, 42.13°). The composition was approximately estimated from the lattice parameter, a, which was calculated from the position of the (1 00) lines on the basis of the reported relationship [2]. The Si/A1 ratio in the products was obtained by inductively-coupled plasma emission spectroscopy (ICP) (Nippon Jarrell Ash, ICAP-575 II). The products were dissolved in water by alkali fusion. Details were described elsewhere [8]. The reaction temperature was fixed at 1670 ° C in the present study on the basis of our primary results. The presence of the boundary temperature above which SiC formed instead of Si3N4 was reported at ~ 1400 to 1600°C in the SiO2-C-N2 system [13, 14]. A similar one was observed in the carbothermal reduction process of kaolinite [15]. Hence, the reaction temperature was initially set at 1700°C in our primary runs. When the mixture containing 40% carbon was heated, a large amount of/~-SiC and smaller amounts of the 2H compounds were detected. Two (1 0 0) lines of the 2H compounds were observed and one was ascribed to almost pure A1N (detailed discussion for the assignment was similar to that shown below). As heating time increased, the peak intensity of the (1 0 0) line of A1N remarkably decreased; A1N seemed to decompose at 1700 ° C. Thus, the~eaction temperature was lowered to 1670°C. In the reactions at 1670°C, the peak intensity of the (1 00) line of A1N did not remarkably decrease. In the XRD patterns of the products obtained by heating the intercalation compound for 0 to 10h,