Fermionic and bosonic ac conductivities at strong disorder

TL;DR

This paper investigates the ac conductivity of strongly localized fermions and bosons in disordered systems, revealing how interference effects depend on particle statistics and magnetic fields, with explicit calculations assuming constant density of states.

Contribution

It provides a detailed analysis of interference corrections to ac conductivity in disordered localized systems, highlighting differences between fermions and bosons and including explicit formulas under specific assumptions.

Findings

Local dipole excitations dominate main conductivity contribution.

Interference correction depends on particle statistics and magnetic field.

Explicit logarithmic formulas for corrections assuming constant density of states.

Abstract

We study the ac conduction in a system of fermions or bosons strongly localised in a disordered array of sites with short-range interactions at frequencies larger than the intersite tunnelling but smaller than the characteristic fluctuation of the on-site energy. While the main contribution $\sigma_0(\omega)$ to the conductivity comes from local dipole-type excitations on close pairs of sites, coherent processes on three or more sites lead to an interference correction $\sigma_1(\omega)$, which depends on the statistics of the charge carriers and can be suppressed by magnetic field. For bosons the correction is always positive, while for fermions it can be positive or negative depending on whether the conduction is dominated by effective single-particle or single-hole processes. We calculate the conductivity explicitly assuming a constant density of states of single-site excitations. Independently of the statistics, $\sigma_0(\omega)=const$. For bosons $\sigma_1(\omega)\propto \log(C/\omega)$. For fermions $\sigma_1(\omega)\propto\log[\max(A,\omega)/\omega]-\log[\max(B,\omega)/\omega]$, where the first and the second term are respectively the particle and hole contributions, $A$ and $B$ being the particle and hole energy cutoffs. The ac magnetoresistance has the same sign as $\sigma_1(\omega)$.