Unrestricted slave-boson mean-field approximation for the two-dimensional Hubbard model

TL;DR

This paper introduces an unrestricted slave-boson mean-field approach for the 2D Hubbard model, revealing reduced polarization and increased delocalization compared to Hartree-Fock, with insights into spin polarons and domain walls.

Contribution

The study develops an unrestricted slave-boson scheme that improves upon Hartree-Fock by reducing polarization and capturing more accurate inhomogeneous states in the Hubbard model.

Findings

Reduced polarization of inhomogeneous states compared to HF

Attractive interaction between spin polarons over a wide U range

Delayed crossover from vertical to diagonal domain walls at higher U

Abstract

The Kotliar-Ruckenstein slave-boson scheme is used to allow for an unrestricted variation of the bosonic and fermionic fields on the saddle-point level. Various inhomogeneous solutions, such as spin polarons and domain walls are discussed within the two-dimensional Hubbard model and compared with results of unrestricted Hartree-Fock (HF) calculations. We find that the present approach drastically reduces the polarization of these states and leads to increased delocalized wave functions as compared to the HF model. The interaction between two spin polarons turns out to be attractive over a wide range of the on-site repulsion U. In addition we obtain the crossover from vertical to diagonal domain walls at a higher value of U than predicted by HF.