Evolutionary kinetic Monte Carlo: Atomistic rates of surface-mediated processes from surface morphologies

Abstract

Surface-mediated processes, such as epitaxial growth, heterogeneous catalysis, and etching, are typically modeled by Kinetic Monte Carlo (KMC) methods. Traditionally, the KMC simulations are based on a top-down approach, where the simulation parameters-the rates for the corresponding atomistic processes-are obtained by manually fitting the simulation output to the experiment. More recently, following the development of Density Functional Theory (DFT), an alternative bottom-up approach has been developed, obtaining the atomistic rates from activation energies and attempt frequencies procured by DFT. Nevertheless, the procedure still requires a labor-intensive fine-tuning of the rates to improve the match between simulation and experiment. Accordingly, we propose to modify the traditional top-down and bottom-up approaches by automating the search of the atomistic rates with the help of an evolutionary algorithm. On the basis of a power spectral density analysis of both the experimental and simulated images, the procedure is applied to characterize wet etching of silicon and epitaxial growth of silver as examples of typical surface-mediated processes. © 2014 American Chemical Society.