In the present study, we investigate the performance of efficient pair potentials in comparison to accurate ab initio potentials as energy descriptions for Monte Carlo simulations of solid-state precipitation. As test scenario, we take the phase decomposition kinetics in binary Fe1-xCux. In a first effort, we predict thermodynamic equilibrium properties of bcc-rich Cu precipitates in an Fe-rich solution with a temperature and composition dependent Cluster Expansion. For this Cluster Expansion, combined ab inito and phonon calculations for various configurations serve as input. Alternatively, we apply the Local Chemical Environment approach, where the energy is described by computationally efficient pair potentials, which are calibrated on the first principles cluster expansion results. We observe that these fundamentally different approaches provide similar information in terms of the precipitate radius, chemical composition and interface constitution, however, the computational effort for the Local Chemical environment approach is significantly lower.
Large-scale ab initio simulations of MAS DNP enhancements using a Monte Carlo optimization strategy
