Ab initio Green-Kubo simulations of heat transport in solids: Method and implementation

TL;DR

This paper details practical strategies for applying ab initio Green-Kubo simulations to calculate lattice thermal conductivity in solids, emphasizing noise reduction, finite-size corrections, and automation for complex materials.

Contribution

It introduces methods for parameter selection based on dynamical properties to improve aiGK simulations, enabling semi-automated and high-throughput thermal conductivity calculations.

Findings

Successfully computed thermal conductivity of MgO at room temperature.

Accurately evaluated conductivity of strongly anharmonic CuI.

Demonstrated noise reduction and correction strategies in simulations.

Abstract

Ab initio Green-Kubo (aiGK) simulations of heat transport in solids allow for assessing lattice thermal conductivity in anharmonic or complex materials from first principles. In this work, we present a detailed account of their practical application and evaluation with an emphasis on noise reduction and finite-size corrections in semiconductors and insulators. To account for such corrections, we propose strategies in which all necessary numerical parameters are chosen based on the dynamical properties displayed during molecular dynamics simulations in order to minimize manual intervention. This paves the way for applying the aiGK method in semi-automated and high-throughput frameworks. The proposed strategies are presented and demonstrated for computing the lattice thermal conductivity at room temperature in the mildly anharmonic periclase MgO, and for the strongly anharmonic marshite CuI.