Controlling Process of P Diffusion in Si Based on the Pair Diffusion Model

摘要

Pair diffusion models of vacancy and interstitial (Ⅴ and Ⅰ) mechanisms are applied to P diffusion in Si. Assuming that the Ⅰ mechanism is dominant, the I component of effective P diffusion coefficient, D_(p~+, Ⅰ)~(eff), and the controlling process of P diffusion are obtained. Not only the Ⅰ~- concentration, C_(Ⅰ~-), but also its gradient, d C_(Ⅰ~-) /d λ, affects D_(P~+,Ⅰ)~(eff) at high C_(P~+) . D_(P~+,Ⅰ)~(eff) is large at d C_(Ⅰ~-)/dλ<0 and small at dC_(Ⅰ~-) /dλ>0. By the dissociation of P-Ⅰ pair, P~+ and Ⅰ~- are generated. When the excess Ⅰ~- thus generated is removed, dC~(Ⅰ~-)/dλ<0 is obtained. dC_(Ⅰ~-)/dλ<0 is also obtained by the effect of the decrease in quasi self-interstitial formation energy. Because different diffusion models simulate the P diffusion profile well under an inert atmosphere, it is difficult to conclude which model is correct. By applying the controlling process to these models, the reason why they simulate the P diffusion profile well is investigated. By the simulation of the P diffusion profile under oxidation at _(P~+)~s > 1 X 10~(20) cm~(-3) (s: surface), it is possible to conclude whether the decrease in quasi self-interstitial formation energy is necessary or not.