Revisiting single-crystal silicon oxidation: A comprehensive analysis of crystal structural transformation to SiO2

Abstract

In this study, we revisit the oxidation reactions of a single-crystal silicon wafer and compare the reported crystal structures of the formed oxides with the original diamond structure of single-crystal silicon. It is commonly assumed that interstitial silicon atoms are always emitted during oxidation at the reactive interface between silicon and the formed oxide due to volume differences. However, it is important to acknowledge that this phenomenon may not always be observed. Thermal equilibrium studies have revealed that a certain percentage of silicon atoms in the diamond structure remains even after surface oxidation [Kamiyama and Sueoka, J. Appl. Phys. 134, 115301 (2023)]. These retained silicon atoms undergo a transformation into a different crystal structure, presenting as β-cristobalite (space group: P41212) rather than the anticipated non-ideal cristobalite (space group: Fd-3m). Our ab initio calculations indicated that the latter remains stable next to a quartz-based structure, exhibiting optimal compatibility with the Si (001) surface. This quartz-based structure is formed through the emission of a Si atom during the oxidation of single-crystal silicon, finally forming a quartz/Si (001) interface. Therefore, we propose a coexisting model involving an alternative β-cristobalite and a quartz crystal structure originating from the surface oxidation of single-crystal silicon. This model offers an explanation for why thermally oxidized films derived from single-crystal silicon exhibit an amorphous nature. In addition, studies have revealed that the oxide precipitates observed in Si crystals are cristobalite and coesite. Hence, the qualitative differences in SiO2/Si interfaces between surface and internal oxidations in metal-gettering effectivity shown in experimental literature require clarification. We also discuss the conditions that prevent the emission of Si atoms.