Nonlinear silicon oxide growth patterns in a gold-silicon system
- PDF / 1,549,766 Bytes
- 7 Pages / 576 x 792 pts Page_size
- 51 Downloads / 176 Views
Louis Paquin and Fabrice Baratay Centre de recherche en microelectronique, Universite de Sherbrooke, 2500, boul. Universite, Sherbrooke, Quebec, Canada]IK2R1 (Received 3 September 1991; accepted 19 May 1992)
Oxide films are grown on a silicon wafer at low temperatures through the catalytic action of a thin gold film. Our results indicate that the oxide film thickness and morphology vary with the initial gold film thickness but do not significantly depend on the temperature. A Fourier analysis of the film structure suggests that the growth mechanism may include a spinodal decomposition where a binary alloy undergoes a phase separation. It is argued that gold silicide is the most likely candidate for spinodal decomposition. I. INTRODUCTION
II. EXPERIMENTAL 12
It has been known for quite some time ' that a thin film of gold can catalyze the oxidation of silicon at low temperature (125-300 °C). Hiraki3'4 has presented a model that accounts for most of the phenomenology connected with this oxidation, including the influence of factors such as crystal orientation, temperature, ambients, and gold film thickness on the rate of growth and the final thickness of the oxide film. The main phenomenological aspect of this model is that when a thin film of Au is evaporated on c-Si, Si atoms diffuse uniformly through the Au film (unless the wafer is initially covered by a nonuniform oxide layer) to reach its surface where they are oxidized. This process goes on as long as the Au concentration at the Au-oxide interface is high enough to catalyze the oxidation reaction. Hewett and Lau5 have shown that the Si atoms in co-deposited films of Au-Si alloy will be oxidized at 200 °C as long as the Au concentration in the alloy exceeds 40%. Chang6 has systematically studied the effect of ambients on the kinetics of diffusion in thin metallic films and found that diffusion is the rate-determining step in this oxidation process. Extensive surface studies have been performed on Au-Si systems and are compatible with the Hiraki model.7"12 Similar oxidation has also been reported for hydrogenated amorphous silicon substrate. 1314 Although it is not well established whether the oxide is SiO, SiO 2 , or nonstoichiometric SiO*, the evidence seems to favor SJO21"5'14 and we will subsequently refer to the oxide as SiO 2 . The reader should thus keep in mind that SiO2 stands for any silicon oxide. We present here some new results about the influence of temperature and gold film initial thickness on the final thickness and morphology of the oxide film. Some of these results strongly suggest that nonlinear phenomena are taking place, apparently contradicting the Hiraki model. The possible nature of these phenomena is then discussed. 2458
http://journals.cambridge.org
J. Mater. Res., Vol. 7, No. 9, Sep 1992
Downloaded: 07 Apr 2015
Gold films with respective thicknesses of 2.5, 5.0, 7.5, 10, 15, 20, 30, 40, and 50 nm were evaporated at a rate of 0.02 nm/s on monocrystalline, {100} orientation, n-type, phosphor-doped silicon wafers. The wafers were
Data Loading...
