Core Excitations with Excited State Mean Field and Perturbation Theory

TL;DR

This paper evaluates excited state mean field and perturbation theories for predicting K-edge positions and X-ray peak separations, finding that mean field is surprisingly accurate and perturbation corrections can be optimized for better results.

Contribution

It introduces and tests a new perturbation theory approach for core excitation energies that is computationally efficient and competitive with high-accuracy methods.

Findings

Mean field theory is highly accurate for K-edge predictions.

Electron-counting biases affect some core-valence separation schemes.

Perturbation theory improves accuracy and scales better than coupled cluster methods.

Abstract

We test the efficacy of excited state mean field theory and its excited-state-specific perturbation theory on the prediction of K-edge positions and X-ray peak separations. We find that the mean field theory is surprisingly accurate, even though it contains no accounting of differential electron correlation effects. In the perturbation theory, we test multiple core-valence separation schemes and find that, with the mean field theory already so accurate, electron-counting biases in one popular separation scheme become a dominant error when predicting K-edges. Happily, these appear to be relatively easy to correct for, leading to a perturbation theory for K-edge positions that is lower scaling and more accurate than coupled cluster theory and competitive in accuracy with recent high-accuracy results from restricted open-shell Kohn Sham theory. For peak separations, our preliminary data show excited state mean field theory to be exceptionally accurate, but more extensive testing will be needed to see how it and its perturbation theory compare to coupled cluster peak separations more broadly.