Machine-Learned Force Fields for Lattice Dynamics at Coupled-Cluster Level Accuracy

TL;DR

This paper develops machine-learned force fields trained on high-accuracy coupled-cluster data to improve predictions of lattice dynamics, phonon dispersions, and vibrational properties in solids.

Contribution

It introduces a delta-learning approach and charge-aware models to enhance MLFF accuracy at the coupled-cluster level for lattice dynamics.

Findings

MLFFs trained on CC data yield higher vibrational frequencies aligning better with experiments.

MLFFs improve the estimation of anharmonic effects in lithium hydride.

Compared to DFT, CC-based MLFFs provide more accurate phonon dispersions.

Abstract

We investigate Machine-Learned Force Fields (MLFFs) trained on approximate Density Functional Theory (DFT) and Coupled Cluster (CC) level potential energy surfaces for the carbon diamond and lithium hydride solids. We assess the accuracy and precision of the MLFFs by calculating phonon dispersions and vibrational densities of states (VDOS) that are compared to experiment and reference ab initio results. To overcome limitations from long-range effects and the lack of atomic forces in the CC training data, a delta-learning approach based on the difference between CC and DFT results, as well as a charge aware MLFF approach is explored. Compared to DFT, MLFFs trained on CC theory yield higher vibrational frequencies for optical modes, agreeing better with experiment. Furthermore, the MLFFs are used to estimate anharmonic effects on the VDOS of lithium hydride at the level of CC theory.