The periodic Anderson model from the atomic limit and FeSi

TL;DR

This paper develops an approximate method to analyze the periodic Anderson model using the atomic limit, successfully explaining experimental properties of FeSi as a Kondo insulator with good agreement.

Contribution

It introduces an approximation of the effective cumulant based on the atomic limit to study the periodic Anderson model, linking theory with experimental data for FeSi.

Findings

Spectral density shows Kondo peak near the Fermi surface.

The model reproduces temperature-dependent conductivity and susceptibility of FeSi.

Hybridization causes conduction electron relaxation in the model.

Abstract

The exact Green's functions of the periodic Anderson model for $U\to \infty $ are formally expressed within the cumulant expansion in terms of an effective cumulant. Here we resort to a calculation in which this quantity is approximated by the value it takes for the exactly soluble atomic limit of the same model. In the Kondo region a spectral density is obtained that shows near the Fermi surface a structure with the properties of the Kondo peak. Approximate expressions are obtained for the static conductivity $% \sigma (T)$ and magnetic susceptibility $\chi (T)$ of the PAM, and they are employed to fit the experimental values of FeSi, a compound that behaves like a Kondo insulator with both quantities vanishing rapidly for $T\to 0$. Assuming that the system is in the intermediate valence region, it was possible to find good agreement between theory and experiment for these two properties by employing the same set of parameters. It is shown that in the present model the hybridization is responsible for the relaxation mechanism of the conduction electrons.