An empirical potential was developed which represented a considerable improvement over existing models with regard to the description of local bonding for bulk defects and disordered phases. The model involved 2- and 3-body interactions with theoretically based functional forms that reflected chemical and physical trends. The numerical parameters in the functional forms were obtained by fitting them to a set of ab initio results of quantum-mechanical calculations that were based upon density-functional theory in the local-density approximation. The potential was tested by applying it to the relaxation of point defects, to the core properties of partial dislocations and to the structure of disordered phases; none of which had been used during the fitting procedure. In the case of dislocations, the model gave predictions that were in excellent agreement with ab initio and tight-binding calculations. It was the only potential that was known to describe both the 30º and 90º partial dislocations in the {111} glide set. The structural and thermodynamic properties of the liquid and amorphous phases were also in good agreement with experimental and ab initio results. The potential was able to simulate directly quenching from the liquid to the amorphous phase, and the resultant amorphous structure was more realistic than that obtained when using existing empirical methods. Force evaluation in this model was also faster than in the case of the Stillinger-Weber potential, thus permitting reliable simulations of very large atomic systems to be performed.

Interatomic potential for silicon defects and disordered phases J.F.Justo, M.Z.Bazant, E.Kaxiras, V.V.Bulatov, S.Yip: Physical Review B, 1998, 58[5], 2539-50