Capturing local atomic environment dependence of activation barriers in metals using cluster expansion models

Abstract

It is well known that surface diffusion in metals can proceed via multiple mechanisms, such as hop, exchange and other types of concerted moves. However, the manner in which kinetic rates associated with a mechanism can depend sensitively on local atomic environment is relatively less understood. We describe recent attempts in our research group to capture the atomic environment dependence using the cluster expansion model (CEM). In particular, we focus on hop and exchange moves at the (001) surface in homoepitaxy, and show that while CEM can work remarkably well in most cases, it can sometimes provide inaccurate predictions for concerted moves.