Theoretical prediction of water’s infrared spectrum under external electric fields

Determine a complete and quantitatively accurate theoretical description of the infrared absorption spectrum of water, with particular emphasis on predictions in the presence of finite external electric fields, so that all major intramolecular and intermolecular spectral features are correctly captured across frequencies.

Background

Accurately predicting the infrared (IR) spectrum of water has been a longstanding challenge in computational chemistry and condensed-matter physics due to the complex interplay of hydrogen bonding, molecular dynamics, and electronic polarization. This difficulty is exacerbated when external electric fields are applied, which modify intermolecular interactions and vibrational band positions and intensities.

In this work, the authors introduce a framework that infers Born effective charges and polarization directly from machine learning interatomic potentials using the Latent Ewald Summation (LES) approach, trained only on energies and forces. They demonstrate strong agreement with density-functional theory for bulk water and show reasonable field-dependent trends in the IR spectrum. Nonetheless, they explicitly note that the theoretical prediction of water’s IR spectrum—particularly under external fields—remains not fully resolved, highlighting an outstanding problem for future theoretical and computational development.