Thermodynamic calculations using reverse Monte Carlo: A computational workflow for accelerated construction of phase diagrams for metal hydrides

TL;DR

This paper extends a reverse Monte Carlo-based thermodynamic calculation framework to metal hydrides, enabling rapid and accurate phase diagram construction with significantly fewer configurations than traditional methods, thus accelerating materials discovery.

Contribution

The authors develop an efficient computational workflow for thermodynamic analysis of metal hydrides using RMC, significantly reducing computational effort for phase diagram construction.

Findings

Phase diagrams can be constructed in minutes using <10 configurations.

The approach accurately captures complex behaviors like volume expansion and phase transitions.

Compared to grand canonical Monte Carlo, the method requires orders of magnitude fewer configurations.

Abstract

Metal hydrides are promising candidates for hydrogen storage applications. From a materials discovery perspective, an accurate, efficient computational workflow is urgently required that can rapidly analyze/predict thermodynamic properties of these materials. The authors have recently introduced a thermodynamic property calculation framework based on the lattice reverse Monte Carlo (RMC) method. Here the approach is extended to metal hydrides, which exhibit significant volume expansion, strong interaction between hydrogen and the host atoms, lattice strain, and a phase transition. We apply the technique to the nickel hydride (NiH_x) system by calculating the pressure-composition-temperature (PCT) isotherm and constructing its phase diagram. An attractive feature of our approach is that the entire phase diagram can be accurately constructed in few minutes by considering <10 configurations. In contrast, a popular technique based on grand canonical Monte Carlo would require sampling of several million configurations. The computational workflow presented paves the way for the approach to be used in future for wider materials search and discovery.