Unified Simulation of Diffusion in Silicon and Silicon Dioxide

Abstract

We present a unified simulation of diffusion in silicon (Si) and silicon dioxide (SiO2) that is based on the diffusing dopant species and point defects that primarily contribute to the diffusion. We first present the simulation of phosphorus (P) diffusion in Si based on the integrated diffusion model that we have developed and elucidate the mechanism of the appearance of the anomalous P in-diffusion profile. The vacancy mechanism governs P diffusion in the plateau region, while the kick-out mechanism governs it in the deeper region, where Si self-interstitials dominate in the kink region and P interstitials dominate in the tail region. Next, we present the simulation of boron (B) diffusion and Si self-diffusion in SiO2. We examined the co-diffusion of implanted B and 30Si in thermally grown 28SiO2, which shows increasing diffusivities with decreasing distance between the diffusers and Si/SiO2 interface and with higher B concentration in SiO2. We propose a model in which SiO molecules generated at the interface and diffusing into SiO2 enhance both B diffusion and Si self-diffusion. The simulation showed that the SiO diffusion is so slow that the SiO concentration at the B and 30Si region critically depends on the distance from the interface. In addition, the simulation predicts the possibility of time-dependent diffusivities for B and Si because more SiO molecules should be arriving from the interface with time, and this time dependence was experimentally observed. Moreover, based on the B concentration dependence, the simulation result indicates that B and Si atoms in SiO2 diffuse correlatively via SiO; namely, the enhanced SiO diffusion by the existence of B enhances B diffusion and Si self-diffusion.