We analyze various aspects of the stress/strain present at the interface during the growth of the oxide which arise from the volume change associated with the transfer of Si atoms from the substrate to the oxide. This volume change is so high (126%) that it cannot be accommodated simply by elastic strain. Previously suggested viscoelastic stress relief models are incompatible with several experimental observations, require unrealistically low viscosity values, and have a number of conceptual difficulties. Analyzing the situation from a structural/chemical point of view, we suggest that structural effects are of primary importance during growth. According to our model the noncrystalline film grows on silicon with a high degree of short range order because of the influence of the crystalline substrate. The oxide near the interface resembles some high density crystalline polymorph of , e.g. , coesite; hence, this is a quasi‐epitaxial growth process. During subsequent growth, annealing takes place which results in a pseudo‐polymorphic transformation of the oxide to a lower density structure in a manner similar to the decompaction of pressure compacted silica glass. Structural rearrangement occurs preferentially normal to the interface, may involve variations in local conformations of the very flexible noncrystalline structure, and results in substantial accommodation of the oxidation strain. Stress relief can also occur on the silicon side of the interface by structural rearrangement in the upper layer of silicon atoms and/or by plastic flow. The latter could occur when the strain rate associated with oxide growth at a given temperature is so high that the yield point of silicon is exceeded.
No takes yet. Share an insight, caveat, or question.
Mrstik et al. (1987) studied this question.