Local Self-Energies for V and Pd Emergent from a Non-Local LDA+FLEX Implementation

TL;DR

This paper introduces a combined LDA+FLEX approach to calculate k- and ω-dependent self-energies, providing insights into correlation effects in V and Pd, and showing improved agreement with experiments over LDA+DMFT for Pd.

Contribution

The paper develops a perturbative LDA+FLEX method incorporating momentum-resolved spin fluctuations, reducing computational complexity and improving correlation effect predictions for transition metals.

Findings

Self-energies in V and Pd are k-independent.

LDA+FLEX yields better mass enhancement agreement for Pd than LDA+DMFT.

Method captures physics of spin fluctuations missing in GW.

Abstract

In the spirit of recently developed LDA+U and LDA+DMFT methods we implement a combination of density functional theory in its local density approximation (LDA) with a $k$- and $\omega -$dependent self-energy found from diagrammatic fluctuational exchange (FLEX) approximation. The active Hilbert space here is described by the correlated subset of electrons which allows to tremendously reduce the sizes of matrices needed to represent charge and spin susceptibilities. The method is perturbative in nature but accounts for both bubble and ladder diagrams and accumulates the physics of momentum resolved spin fluctuations missing in such popular approach as GW. As an application, we study correlation effects on band structures in V and Pd. The d-electron self-energies emergent from this calculation are found to be remarkably k-independent. However, when we compare our calculated electronic mass enhancements against LDA+DMFT, we find that for a long standing problem of spin fluctuations in Pd, LDA+FLEX delivers a better agreement with experiment, although this conclusion depends on a particular value of Hubbard $U$ used in the simulation. We also discuss outcomes of recently proposed combinations of k-dependent FLEX with DMFT.