The Elliott-Yafet theory of spin relaxation generalized for large spin-orbit coupling

TL;DR

This paper extends the Elliott-Yafet theory to account for large spin-orbit coupling in heavy metals, providing analytic and numerical formulas that unify weak and strong SOC regimes, with implications for materials like gold.

Contribution

The authors develop a generalized Elliott-Yafet theory for large SOC, including analytic formulas and empirical pseudopotential methods, bridging the gap between weak and strong SOC regimes.

Findings

The generalized theory recovers the original EY theory for weak SOC.

In the strong SOC limit, spin-relaxation rate approaches momentum-relaxation rate.

Application to gold suggests the theory accurately describes real heavy-element metals.

Abstract

We generalize the Elliott-Yafet (EY) theory of spin relaxation in metals with inversion symmetry for the case of large spin-orbit coupling (SOC). The EY theory treats the SOC to the lowest order but this approach breaks down for metals of heavy elements (such as e.g. caesium or gold), where the SOC energy is comparable to the relevant band-band separation energies. The generalized theory is presented for a four-band model system without band dispersion, where analytic formulae are attainable for arbitrary SOC for the relation between the momentum- and spin-relaxation rates. As an extended description, we also consider an empirical pseudopotential approximation where SOC is deduced from the band potential (apart from an empirical scaling constant) and the spin-relaxation rate can be obtained numerically. Both approaches recover the usual EY theory for weak SOC and give that the spin-relaxation rate approaches the momentum-relaxation rate in the limit of strong SOC. We argue that this limit is realized in gold by analyzing spin relaxation data. A calculation of the $g$-factor shows that the empirical Elliott-relation, which links the $g$-factor and spin-relaxation rate, is retained even for strong SOC.