Covalency and the metal-insulator transition in titanate and vanadate perovskites

TL;DR

This study uses advanced computational methods to analyze the electronic structure of titanate and vanadate perovskites, revealing limitations of standard approaches and proposing adjustments to better match experimental observations.

Contribution

It demonstrates that standard DFT+DMFT with FLL double counting fails to predict insulating behavior and p-d splitting accurately, suggesting the need for an ad hoc correction and specific U values.

Findings

Standard DFT+DMFT predicts metallic states for LaTiO₃ and LaVO₃.

FLL double counting causes discrepancies in p-d level splitting.

Adjusting double counting and U yields results consistent with experiments.

Abstract

A combination of density functional and dynamical mean-field theory is applied to the perovskites SrVO$_3$, LaTiO$_3$ and LaVO$_3$. We show that DFT+DMFT in conjunction with the standard fully localized-limit (FLL) double-counting predicts that LaTiO$_3$ and LaVO$_3$ are metals even though experimentally they are correlation-driven ("Mott") insulators. In addition, the FLL double counting implies a splitting between oxygen $p$ and transition metal $d$ levels which differs from experiment. Introducing into the theory an \textit{ad hoc} double counting correction which reproduces the experimentally measured insulating gap leads also to a $p$-$d$ splitting consistent with experiment if the on-site interaction $U$ is chosen in a relatively narrow range ($\sim 6\pm 1$ eV). The results indicate that these early transition metal oxides will serve as critical test for the formulation of a general \textit{ab initio} theory of correlated electron metals.