The formation of thin silicon carbide films for various uses is of technological intercst but still presents some problems. In many cases. the film produced has an cxccss or deficiency of carbon or silicon.[lJ One reason for this is that the precursor gas generally used is a mixture of several components, for example monosilane and methane or similar compounds.[2, 31 Different boiling points, vapor pressures, and diffusion coefficients are reasons for the problems e n ~o u n t e r e d , ~~] which can be avoided by the use of a single compound already having the correct silicon-to-carbon ratio.['* 61 Ideally, this compound should be a liquid, easy to evaporate, not self-igniting and stable with respect to air and moisture. The decomposition temperatwe leading to crystalline S i c should be as low as possible.
Most of these conditions are fulfilled by disilane derivatives. Because ethyl groups are groups that leave easily during a decomposition, an ethyldisilane with a silicon-to-methyl ratio of 1 : 1 would seem to be a suitable precursor for pyrolytic decomposition. First we tried diethyldimethyldisilane. Its boiling point (78 "C) is in a very convenient region, and the compound is stable in air and easy to handle. However, it was previously unknown and we had to synthesize it for the first time (see the Experimental section).
We carried out chemical vapor deposition (CVD) investigations on diethyldimethyldisilane. With a very simple apparatus consisting of a flask containing the precursor substance, a graphite tube oven and a small cooling trap, we started by decomposing 1,2-diethyl-l,2-dimethyldisilane at 1250 K. To avoid oxygen entering the system the whole apparatus was placed under a helium atmosphere before being heated. We had to use a graphite tube reactor because previous experiments carried out in a quartz tube showed that the films formed contained a large amount of Si-0, which pos- sibly originated from the quartz glass. The low boiling liquid (b.p. 78 'C) was evaporated in a 50 "C water bath and carried into the preheated graphite tube oven, which contained some MgO single-crystal wafers of well-known crystallographic [