Efficient Calculation of the Lattice Thermal Conductivity by Atomistic Simulations with Ab Initio Accuracy

Abstract

High‐order force constant expansions can provide accurate representations of the potential energy surface relevant to vibrational motion. They can be efficiently parametrized using quantum mechanical calculations and subsequently sampled at a fraction of the cost of the underlying reference calculations. Here, force constant expansions are combined via the hiphive package with GPU‐accelerated molecular dynamics simulations via the GPUMD package to obtain an accurate, transferable, and efficient approach for sampling the dynamical properties of materials. The performance of this methodology is demonstrated by applying it both to materials with very low thermal conductivity (Ba8Ga16Ge30, SnSe) and a material with a relatively high lattice thermal conductivity (monolayer‐MoS2). These cases cover both situations with weak (monolayer‐MoS2, SnSe) and strong (Ba8Ga16Ge30) pho renormalization. The simulations also enable to access complementary information such as the spectral thermal conductivity, which allows to discriminate the contribution by different phonon modes while accounting for scattering to all orders. The software packages described here are made available to the scientific community as free and open‐source software in order to encourage the more widespread use of these techniques as well as their evolution through continuous and collaborative development.

GPUMD: A package for constructing accurate machine-learned potentials and performing highly efficient atomistic simulations

“…In all the cases, the validity of the heat current expressions has been numerically confirmed in terms of energy conservation. [62][63][64][65] Equation ( 27) has been recently used in an on-the-fly MLP. 66 Since the only assumptions for the derivations 60 of these expressions are the locality properties of the interatomic potentials, our heat current expressions generally apply to the multi-body cluster potentials as considered in Ref.…”

“…40 Recently, support was also added for machine-learned force constant potentials constructed using the hiphive package 74 . 65 The most recent addition are the various versions of NEP as developed in the previous papers 41,42 as well as the current one.…”

The amplitudons of dynamical stable binary cubic crystals for hot-electron cooling in BaTe, CdTe, SrTe and SnTe for enhanced photo voltaic effect

“…Gibbs free energy G(T, P) at given temperature, T and pressure, P is obtained from Helmholtz free energy F(T; V) through the transformation. Phonon lifetime is crucial in search of thermoelectric materials, transistors, and heat management systems since it could explain how long it takes phonon mode vibrations to dissipate 19 . Peaks in Fig.…”

Thermal transport of glasses via machine learning driven simulations

“…One route is to leverage datasets of mechanical and functional glass properties to discover new materials (Mauro et al, 2016;Onbaşlı et al, 2020), or to predict end-properties such as solubilities (Brauer et al, 2007), dissolution rates (Anoop Krishnan et al, 2018), or transition temperatures (Cassar et al, 2018), to name a few (Liu et al, 2021). By taking a microscopic approach, one can build surrogate models of ab initio interatomic potentials able to drive simulations to sample all kinds of properties of amorphous materials that can be accessed by atomistic simulations (Allen and Feldman, 1989;Sosso et al, 2012;Deringer and Csányi, 2017;Paruzzo et al, 2018;Sosso et al, 2018;Sivaraman et al, 2020;Deringer et al, 2021;Islam et al, 2021;Brorsson et al, 2022;Guo et al, 2022;Langer et al, 2023;Liu et al, 2023;Xie et al, 2023).…”