Theory of oxide defects near the Si-SiO2interface

We have analyzed the (100) Si-${\mathrm{SiO}}_2$ interface based on the crystalline structural model proposed by Ourmazd et al. The quantum-mechanical modified intermediate neglect of differential overlap (MINDO/3) technique was employed to investigate the electronic properties and atomic configurations of the interface region with various oxide defects. We find that oxygen vacancies in the near-interface region may explain the existence of several silicon oxidation states which have been observed in core-level photoemission experiments. Our calculations indicate that the properties of the oxygen vacancies are a strong function of their locations. A positively charged vacancy in the first oxygen monolayer is predicted to be unstable against the formation of a neutral Si–Si bond at the vacancy along with the formation of a positively charged ${\mathrm{P}}_{\mathrm{b}}$-like defect in the silicon substrate. However, a positively charged vacancy in the second oxygen monolayer is predicted to behave very much like an ${\mathrm{E}}_1^{\prime }$ center in α-quartz. The energy levels associated with these vacancies are predicted to lie close to the Si valence-band edge, probably within the valence band. The positive charge state might therefore be neutralized by tunneling from the Si valence band.