Universal Theory for Thermal Oxidation of Silicon

Hiroyuki Kageshima, Kenji Shiraishi, and Masashi Uematsu
Device Physics Laboratory

When crystal silicon is exposed to oxygen gas at temperatures around 1000℃, the surface silicon reacts with the oxygen to become silicon oxide. Since this phenomenon (silicon thermal oxidation) has various useful properties, it plays quite an important role in the fabrication of semiconductor devices as ULSI chips which are the heart of today's electronic information and communication hardware such as cellular phones, fax machines, electronic communication switching systems, and personal computers. In similar reasons, the silicon thermal oxidation will no doubt play essential roles in the fabrication of new-generation semiconductor devices (e. g. silicon single-electron-transistors), which bear the future highly advanced information and communication society.
The fabrication of these new-generation devices inevitably requires atomic control. This control is also necessary in thermal oxidation, but is quite difficult to achieve. Observations on the atomic scale reveal mysterious phenomena in thermal oxidation, such as "initial enhanced oxidation" and "pattern dependent oxidation", which cannot be explained by the conventional thermal oxidation theory. We have therefore been studying silicon thermal oxidation in the atomic world by solving quantum dynamics (motion law in the atomic world) with a supercomputer (much faster than a personal computer).
We found that in silicon thermal oxidation, in addition to oxygen sinking into the silicon to form surface oxide, a lot of silicon also sinks into the surface oxide [1]. This silicon penetration into the oxide enables us to explain the "initial enhanced oxidation" naturally and consistently [2]. Furthermore, the silicon penetration can be expected to enable us to predict and control the "pattern dependent oxidation" with atomical precision, as well as predict and control the physical and electrical properties of the formed oxide. These possibilities are now under assiduous investigation.
[1] H. Kageshima and K. Shiraishi, Phys. Rev. Lett. 81 (1998) 5936.
[2] H. Kageshima, K. Shiraishi, and M. Uematsu, Jpn. J. Appl. Phys. 38 (1999) L971.


Fig. 1. Schematic view of silicon thermal oxidation and efficiency of our theory.