Orbital-resolved DFT+U for molecules and solids

TL;DR

This paper introduces an orbital-resolved DFT+U method that significantly improves the accuracy of electronic, energetic, and structural predictions for molecules and solids with hybridized states, using linear-response calculations for Hubbard parameters.

Contribution

The paper develops an orbital-resolved extension of DFT+U that enhances predictive accuracy for complex materials with hybridized states, surpassing conventional shell-averaged approaches.

Findings

Improved predictions for bulk solids like FeS2 and MnO2.

Accurate modeling of Fe(II) molecular complexes.

Hubbard parameters obtained efficiently from linear-response calculations.

Abstract

We present an orbital-resolved extension of the Hubbard $U$ correction to density-functional theory (DFT). Compared to the conventional shell-averaged approach, the prediction of energetic, electronic and structural properties is strongly improved, particularly for compounds characterized by both localized and hybridized states in the Hubbard manifold. The numerical values of all Hubbard parameters are readily obtained from linear-response calculations. The relevance of this more refined approach is showcased by its application to bulk solids pyrite (FeS$_2$) and pyrolusite ($\beta$-MnO$_2$), as well as to six Fe(II) molecular complexes. Our findings indicate that a careful definition of Hubbard manifolds is indispensable for extending the applicability of DFT+$U$ beyond its current boundaries. The present orbital-resolved scheme aims to provide a computationally undemanding yet accurate tool for electronic structure calculations of charge-transfer insulators, transition-metal (TM) complexes and other compounds displaying significant orbital hybridization.