Approximating the impact of nuclear quantum effects on thermodynamic properties of crystalline solids by temperature remapping

Abstract

When computing finite-temperature properties of materials with atomistic simulations, nuclear quantum effects are often neglected or approximated at the quasiharmonic level. The inclusion of these effects beyond this level using approaches like the path integral method is often not feasible due to their large computational effort. We discuss and evaluate the performance of a temperature-remapping approach that links the finite-temperature quantum system to its best classical surrogate via a temperature map. This map, which is constructed using the internal energies of classical and quantum harmonic oscillators, is shown to accurately capture the impact of quantum effects on thermodynamic properties at an additional cost that is negligible compared to classical molecular dynamics simulations. Results from this approach show excellent agreement with previously reported path integral Monte Carlo simulation results for diamond cubic carbon and silicon. The approach is also shown to work well for obtaining thermodynamic properties of light metals and for the prediction of the fcc to bcc phase transition in calcium.