Simulating trapping sites with accelerated random diffusion methods

Abstract

Many structural evolution is governed by diffusion of atoms. If the diffusion is random, accelerated kinetic Monte Carlo methods based on random-walk statistics can be used to model the structural evolution on 10 + year / μm scales. However, diffusion in practical materials is usually not random due to the presence of various trapping defects such as vacancies, impurity / alloy solutes, dislocations, and grain boundaries. If these defects are modeled with the conventional kinetic Monte Carlo methods, the computation efficiency can easily drop by more than 10 orders of magnitude. In this work, we show that the trapping energy of any trapping site can be arbitrarily modified without changing the trapping thermodynamics provided that we can modify the entropy of the trapping site to recover its trapping Gibbs free energy. Since we can set the trapping energy of trapping sites to zero, random-walk statistics can still be applied to incorporate trapping defects. Our new method will enable future accelerated kinetic Monte Carlo methods to be developed to simulate the evolution of realistic microstructures on 10 + year / μm scales.