Oxidation of silicon (100): Experimental data versus a unified chemical model

It has been observed in many studies of oxidation kinetics that silicon dioxide growth in the thin regime (<30 nm) is faster than the predictions of the linear-parabolic Deal-Grove relationship. We have developed a conceptual and mathematical model for the thermal oxidation of silicon which is based on an initial dissociative chemisorption of molecular oxygen on the 2×1 silicon (100) surface followed by subsequent diffusion and dissociative reactions of molecular oxygen at the Si/SiO₂ interface. This model accounts for current experimental observations on the structural modification of the reconstructed surface to a 1×1 superlattice on limited exposure to molecular oxygen and provides a mechanistic rationale for the self-limiting mechanism of the oxide film at ∼0.6 nm during low-temperature oxidation processes. The rate-equation model, consistent with the proposed reactions at the Si/SiO₂ interface, has been refined to give an excellent fit to experimental data within all thickness regimes, but especially in the initial rapid growth regime where the growth rate is proportional to (P_o2)^1/2. The rate equation reduces to a linear dependence on oxygen partial pressure in the thicker regime, where it predicts classic Deal-Grove behavior. We present the development of the model for oxidations performed between 780-1100 °C consistent with observed oxide growth and film properties. The activation energies for the reaction-controlled regime and the diffusion-controlled regime are consistent with literature data. We observe a sharp transition in a characteristic length parameter at ∼950 °C which may be possibly due to the experimentally observed change in oxide film density. A detailed analysis and explanation is presented.