Weak coupling theory of topological Hall effect

TL;DR

This paper develops a weak-coupling theory for the topological Hall effect, revealing how the effective magnetic field and Hall conductivity depend on electron coupling strength, spin texture, and diffusion regimes, especially when the adiabatic approximation fails.

Contribution

It provides analytic expressions for the topological Hall conductivity across various weak-coupling regimes, including diffusive and ballistic, highlighting the role of spin diffusion and nonlocal effects.

Findings

In local, diffusive regime, THC is proportional to M.

In nonlocal regimes, THC depends on effective spins during diffusion.

THC decreases with increasing skyrmion density in nonlocal regime.

Abstract

Topological Hall effect (THE) caused by a noncoplanar spin texture characterized by a scalar spin chirality is often described by the Berry phase, or the associated effective magnetic field. This picture is appropriate when the coupling, $M$, of conduction electrons to the spin texture is strong (strong-coupling regime) and the adiabatic condition is satisfied. However, in the weak-coupling regime, where the coupling $M$ is smaller than the electrons' scattering rate, the adiabatic condition is not satisfied and the Berry phase picture does not hold. In such a regime, the relation of the effective magnetic field to the spin texture can be "nonlocal", in contrast to the "local" relation in the strong coupling case. Focusing on the case of continuous but general spin texture, we investigate the THE in various characteristic regions in the weak-coupling regime, namely, (1) diffusive and local, (2) diffusive and nonlocal, and (3) ballistic. In the presence of spin relaxation, there arise two more regions in the "weakest-coupling" regime: (1') diffusive and local, and (2') diffusive and nonlocal. We derived the analytic expression of the topological Hall conductivity (THC) for each region, and found that the condition for the locality of the effective field is governed by transverse spin diffusion of electrons. In region 1, where the spin relaxation is negligible and the effective field is local, the THC is found to be proportional to $M$, instead of $M^3$ of the weakest-coupling regime. In the diffusive, nonlocal regions (2 and 2'), the effective field is given by a spin chirality formed by "effective spins" that the electrons see during their diffusive motion. Applying the results to a skyrmion lattice, we found the THC decreses as the skyrmion density is increased in region 2', reflecting the nonlocality of the effective field, and shows a maximum at the boundary to the "local" region.