Embedding approach for dynamical mean field theory of strongly correlated heterostructures

TL;DR

This paper introduces an embedding method that efficiently applies dynamical mean field theory to inhomogeneous strongly correlated materials by reducing the explicit layers needed in calculations, demonstrated on Hubbard model heterostructures.

Contribution

The paper develops a localized basis embedding approach that simplifies DMFT calculations for inhomogeneous systems, including surfaces and heterostructures, by incorporating substrate effects via complex embedding potentials.

Findings

Efficient reduction in layers treated explicitly in DMFT calculations.

Successful numerical application to strongly correlated surfaces and heterostructures.

Demonstrated effectiveness on single-band Hubbard model systems.

Abstract

We present an embedding approach based on localized basis functions which permits an efficient application of the dynamical mean field theory (DMFT) to inhomogeneous correlated materials, such as semi-infinite surfaces and heterostructures. In this scheme, the semi-infinite substrate leads connected to both sides of the central region of interest are represented via complex, energy-dependent embedding potentials that incorporate one-electron as well as many-body effects within the substrates. As a result, the number of layers which must be treated explicitly in the layer-coupled DMFT equation is greatly reduced. To illustrate the usefulness of this approach, we present numerical results for strongly correlated surfaces, interfaces, and heterostructures of the single-band Hubbard model.