Abstract
We describe a unified model for silicon oxidation in dry and wet conditions. It goes beyond the current models of kinetics and of ellipsometric and spectroscopic data by explicitly addressing the issues raised by isotope experiments. The model concentrates especially on the problems indicated in recent discussions, notably the very early stages of oxidation, the reactions at the Si/oxide interface, and the origin and nature of stress.
Three features emerge as central: first, the several roles of stress and stress relaxation; second, the specific properties of the oxide adjacent to the silicon and the concept of the ‘reactive’ layer different in structure and composition, and third, the processes which encourage the insertion of interstitial oxygen into the silicon/oxygen network of the oxide. This ‘reactive layer’ or ‘altered layer’ is central to our analysis of dry oxidation through its role in the conversion of interstitial oxygen molecules into network oxygens. This conversion can be stimulated both electronically and by the use of atomic oxygen. We believe the deviations from Deal-Grove predictions for thin oxides are mainly due to modification of oxygen diffusion, probably from structure changes rather than mere compression; the motion of network oxygen is certainly involved. In wet-grown oxide any region of altered oxide constitutes a much thinner modified layer adjacent to the silicon, with less effect on the thin-film behaviour than for dry-grown oxide, though many problems of interpretation remain.