Quantum criticality and optical conductivity in a two-valley system

TL;DR

This paper investigates how the topology of the Fermi surface in a two-valley system influences optical conductivity, revealing unique scaling behaviors and signatures of quantum criticality due to intervalley electron-electron interactions.

Contribution

It provides a microscopic derivation of optical conductivity in a two-valley Fermi liquid, highlighting the effects of intervalley drag and geometric constraints near a quantum critical point.

Findings

Intervalley drag restores Gurzhi-like conductivity scaling.

Distinct $ ext{Re}\sigma(\omega)$ scaling near quantum criticality.

Predicted large response at Lifshitz transition observable experimentally.

Abstract

We demonstrate that the optical conductivity of a Fermi liquid (FL) in the absence of umklapp scattering is dramatically affected by the topology of the Fermi surface (FS). Specifically, electron-electron (ee) scattering leads to rapid current relaxation in systems with multiple, or multiply connected, FSs, provided the valleys have different effective masses. This effect results from intervalley drag. We microscopically derive the optical conductivity of a two-valley system, both within the FL regime and near a quantum critical point (QCP) of the Ising-nematic type. In the FL regime, intervalley drag restores the Gurzhi-like scaling of the conductivity, $\mathrm{Re} \sigma(\omega) \sim \omega^0$. This dependence contrasts sharply with the previously identified sub-leading contribution to the conductivity of a two-dimensional FL with a single convex FS, where $\mathrm{Re} \sigma(\omega) \sim \omega^2 \ln |\omega|$. The vanishing of the leading term in the optical conductivity is a signature of geometric constraints on ee scattering channels, which are lifted for a multiply connected FS. A large differential response, $d \mathrm{Re} \sigma/d \mu$ with $\mu$ being the chemical potential, is predicted at the Lifshitz transition from a single-valley to a multi-valley FS, which should be observable within the experimentally accessible frequency range. Near a QCP, intervalley drag leads to a $|\omega|^{-2/3}$ scaling of $\mathrm{Re} \sigma(\omega)$ in 2D, thus providing a specific current-relaxing process for this long-standing conjecture.