Atomistic simulation of solid phase epitaxial regrowth in Si: A critical review

During ultra-shallow junction formation, pre-amorphization or dopant implantation may lead to the formation of an amorphous (a-)layer. In the first stage of annealing the solid phase epitaxial regrowth (SPER) of the a-layer takes place. During SPER, redistribution of the dopants occurs, and they are incorporated into crystalline Si (c-Si), either substitutionally or within clusters containing dopant atoms, self-interstitials (Is) and/or vacancies. The SPER process leaves beyond the original a/c interface the end-of-range (EOR) damage which contains an excess of Is. During further annealing, free Is are emitted from the EOR damage and diffuse towards the surface. In this period the diffusion of dopants is considerably enhanced. It is highly desirable to obtain the initial conditions for calculations of dopant diffusion and activation, i.e. the fractions of dopants on substitutional sites and in clusters as well as the cluster morphology, from atomistic simulations of SPER. This work gives a critical review of previous simulations that are based on classical molecular dynamics. It is shown that even for pure Si the experimental regrowth rates cannot be reproduced reasonably well. The main cause for the disagreement is the inaccuracy of the interatomic potentials used. Proposals for physically-based improvements are discussed. They are based on a better description of the amorphous phase using a modified potential, without changing the well-established potential for c-Si.