Electronic Spectra from TDDFT and Machine Learning in Chemical Space

TL;DR

This paper combines TDDFT and machine learning to accurately predict electronic spectra across chemical space, significantly improving the reliability of computational spectral predictions for small organic molecules.

Contribution

It introduces a machine learning correction approach trained on deviations from high-level CC2 spectra to enhance TDDFT predictions in chemical space.

Findings

ML models reduce prediction errors to within ±0.1 eV for 10,000 molecules

Prediction errors decrease with larger training sets

Spectral database analysis suggests potential for even higher accuracy

Abstract

Due to its favorable computational efficiency time-dependent (TD) density functional theory (DFT) enables the prediction of electronic spectra in a high-throughput manner across chemical space. Its predictions, however, can be quite inaccurate. We resolve this issue with machine learning models trained on deviations of reference second-order approximate coupled-cluster singles and doubles (CC2) spectra from TDDFT counterparts, or even from DFT gap. We applied this approach to low-lying singlet-singlet vertical electronic spectra of over 20 thousand synthetically feasible small organic molecules with up to eight CONF atoms. The prediction errors decay monotonously as a function of training set size. For a training set of 10 thousand molecules, CC2 excitation energies can be reproduced to within $\pm$0.1 eV for the remaining molecules. Analysis of our spectral database via chromophore counting suggests that even higher accuracies can be achieved. Based on the evidence collected, we discuss open challenges associated with data-driven modeling of high-lying spectra, and transition intensities.