Modeling electronic response properties with an explicit-electron machine learning potential

TL;DR

This paper introduces the eMLP, an explicit-electron machine learning potential that models short-range electronic interactions in molecules and crystals, enabling accurate prediction of electronic response properties with a semi-classical approach.

Contribution

The paper presents the eMLP, a novel explicit-electron force field utilizing machine learning for short-range interactions, improving the modeling of electronic response properties in molecules and solids.

Findings

eMLP accurately predicts dipole moments and polarizabilities.

eMLP reproduces IR spectra and response properties like stiffness.

Benchmarking shows high precision on new datasets.

Abstract

Explicit-electron force fields introduce electrons or electron pairs as semi-classical particles in force fields or empirical potentials, which are suitable for molecular dynamics simulations. Even though semi-classical electrons are a drastic simplification compared to a quantum-mechanical electronic wavefunction, they still retain a relatively detailed electronic model compared to conventional polarizable and reactive force fields. The ability of explicit-electron models to describe chemical reactions and electronic response properties has already been demonstrated, yet the description of short-range interactions for a broad range of chemical systems remains challenging. In this work, we present the electron machine learning potential (eMLP), a new explicit electron force field where the short-range interactions are modeled with machine learning. The electron pair particles will be located at well-defined positions, derived from localized molecular orbitals or Wannier centers, naturally imposing the correct dielectric and piezoelectric behavior of the system. The eMLP is benchmarked on two newly constructed datasets: eQM7, a extension of the QM7 dataset for small molecules, and a dataset for the crystalline $\beta$-glycine. It is shown that the eMLP can predict dipole moments, polarizabilities and IR-spectra of unseen molecules with high precision. Furthermore, a variety of response properties, e.g. stiffness or piezoelectric constants, can be accurately reproduced.