Multi-million atom Monte Carlo simulation of oxide materials and solid solutions

JA Purton and NL Allan, COMPUTATIONAL MATERIALS SCIENCE, 103, 244-249 (2015).

DOI: 10.1016/j.commatsci.2015.03.016

We propose a method for the simulation of large ionic systems with long- range forces using the Monte Carlo method. This method employs a domain decomposition strategy for subdividing the simulation cell and parallelisation of these subdomains using a thread based strategy. This is thus ideally suited to modern day multi-core architectures. Evaluation of the long range interactions that is incompatible with a domain decomposition strategy has been replaced by the direct calculation of the Coulomb sum (Fennell and Gezelter, 2006). We compare this approach with that of "standard" Monte Carlo simulations that employ the Ewald technique. A relatively large two-body cutoff is required to reproduce the Ewald results accurately. Finally, as a pilot application, we demonstrate that our novel approach can be applied to very large simulation cells (>1 million atoms); results for enthalpies, are presented for a typical non-ideal oxide solid solution (MnO-MgO) as a function of composition and highlight the formation of nano-sized domains in the very large simulation cells. Well defined structures such as exsolution lamellae are not observed. (C) 2015 Elsevier B.V. All rights reserved.

Return to Publications page