Routine calculation of ab initio melting curves: application to aluminum

TL;DR

This paper introduces a fast and reliable ab initio molecular dynamics method based on the two-phase thermodynamic model to compute melting curves, demonstrated on aluminum with high accuracy and lower computational cost.

Contribution

It presents a simplified, efficient approach to calculate melting curves using velocity autocorrelation functions within ab initio simulations, improving computational efficiency.

Findings

Achieves 5-10% accuracy in melting curve predictions

Demonstrates method on aluminum up to 300 GPa

Reduces computational cost compared to traditional methods

Abstract

We present a simple, fast, and reliable method to compute the melting curves of materials with ab initio molecular dynamics. It is based on the two-phase thermodynamic model of [Lin et al., J. Chem. Phys. 119, 11792 (2003)] and its improved version given by [Desjarlais, Phys. Rev. E, 88, 062145 (2013)]. In this model, the velocity autocorrelation function is utilized to calculate the contribution of the nuclei motion to the entropy of the solid and liquid phases. It is then possible to find the thermodynamic conditions of equal Gibbs free energy between these phases, defining the melting curve. The first benchmark on the face-centered cubic melting curve of aluminum from 0 to 300 GPa demonstrates how to obtain an accuracy of 5-10%, comparable to the most sophisticated methods, for a much lower computational cost.