Accelerating grain boundary modelling and simulation by automated pre-evaluation of the irreducible macroscopic representation

Abstract

A major challenge in the modelling and simulation of grain boundaries (GBs) is the conflict between the system size and computation capability, which inherently restricts the computationally accessible boundary characters. We present an algorithm to find the irreducible macroscopic representation of any given coincident-site-lattice GB character in the cubic lattice, and thus determine the irreducible size of its supercell. The algorithm is compared with the conventional orthogonal supercell and a published calculation method to assess its merits in saving computational resources. This supercell size parameter can be used to predict which GB character is relatively special across the vast 5D space, as those GBs possess small structural units are likely to exhibit particular structure-property relationships that are worthy of attention. The prediction is confirmed by examining the energy and mobility trends of aluminum mixed GBs obtained from the atomistic simulations.