Universality of the Mott-Ioffe-Regel limit in metals

TL;DR

This paper demonstrates that the Mott-Ioffe-Regel limit remains a universal boundary for metallic conductivity, even in strongly correlated metals, by analyzing optical conductivity and scattering rates.

Contribution

It provides evidence that the MIR limit is universal in strongly correlated metals and links spectral weight loss to the extended resistivity behavior in bad metals.

Findings

Coherent quasiparticle conductivity disappears at the MIR limit.

Scattering rate saturates at the MIR limit in bad metals.

Spectral weight shedding explains high-temperature resistivity behavior.

Abstract

The absence of resistivity saturation in many strongly correlated metals, including the high-temperature superconductors, is critically examined from the viewpoint of optical conductivity measurements. Coherent quasiparticle conductivity, in the form of a Drude peak centred at zero frequency, is found to disappear as the mean free path (at $\omega$ = 0) becomes comparable to the interatomic spacing. This basic loss of coherence at the so-called Mott-Ioffe-Regel (MIR) limit suggests that the universality of the MIR criterion is preserved even in the presence of strong electron correlations. We argue that the shedding of spectral weight at low frequencies, induced by strong correlation effects, is the primary origin of the extended positive slope of the resistivity to high temperatures observed in all so-called "bad metals". Moreover, in common with those metals which exhibit resistivity saturation at high temperatures, the scattering rate itself, as extracted from optical spectra, saturates at a value consistent with the MIR limit. We consider possible implications that this ceiling in the scattering rate may have for our understanding of transport within a wide variety of bad metals and suggest a better method for analysing their optical response.