Interaction effects on topological phase transitions via numerically exact quantum Monte Carlo calculations

TL;DR

This study uses numerically exact quantum Monte Carlo methods to investigate how interactions influence topological phase transitions in generalized Kane-Mele-Hubbard models, revealing non-perturbative shifts in phase boundaries and symmetry-dependent stabilization of topological phases.

Contribution

It provides the first numerically exact QMC analysis of topological phase transitions in these models, directly computing topological invariants from Green's functions without sign problems.

Findings

Correlation effects shift topological phase boundaries.

Quantum fluctuations stabilize or destabilize phases based on lattice symmetry.

Phase boundaries are nearly size-independent, but topological invariants show strong size dependence.

Abstract

We theoretically study topological phase transitions in four generalized versions of the Kane-Mele-Hubbard model with up to $2\times 18^2$ sites. All models are free of the fermion-sign problem allowing numerically exact quantum Monte Carlo (QMC) calculations to be performed to extremely low temperatures. We numerically compute the $\mathbb{Z}_2$ invariant and spin Chern number $C_\sigma$ directly from the zero-frequency single-particle Green's functions, and study the topological phase transitions driven by the tight-binding parameters at different on-site interaction strengths. The $\mathbb{Z}_2$ invariant and spin Chern number, which are complementary to each another, characterize the topological phases and identify the critical points of topological phase transitions. Although the numerically determined phase boundaries are nearly identical for different system sizes, we find strong system-size dependence of the spin Chern number, where quantized values are only expected upon approaching the thermodynamic limit. For the Hubbard models we considered, the QMC results show that correlation effects lead to shifts in the phase boundaries relative to those in the non-interacting limit, without any spontaneously symmetry breaking. The interaction-induced shift is non-perturbative in the interactions and cannot be captured within a "simple" self-consistent calculation either, such as Hartree-Fock. Furthermore, our QMC calculations suggest that quantum fluctuations from interactions stabilize topological phases in systems where the one-body terms preserve the $D_3$ symmetry of the lattice, and destabilize topological phases when the one-body terms break the $D_3$ symmetry.