Orbital and spin contributions to the $g$-tensors in metal nanoparticles

TL;DR

This study investigates the mesoscopic fluctuations of $g$-tensors in metal nanoparticles, revealing how spin and orbital contributions vary with spin-orbit coupling and wavefunction character, and highlighting discrepancies with Random Matrix Theory predictions.

Contribution

It provides a semi-realistic tight-binding analysis of $g$-tensor fluctuations, emphasizing the role of orbital character and challenging existing RMT-based expectations.

Findings

Spin contribution aligns with RMT predictions.

Orbital contribution depends on wavefunction character, leveling off for $d$ states.

Including orbital effects reduces the predicted $g$-factor for $d$-dominated orbitals.

Abstract

We present a theoretical study of the mesoscopic fluctuations of $g$-tensors in a metal nanoparticle. The calculations were performed using a semi-realistic tight-binding model, which contains both spin and orbital contributions to the $g$-tensors. The results depend on the product of the spin-orbit scattering time $\tau_{\textrm{\small so}}$ and the mean-level spacing $\delta$, but are otherwise weakly affected by the specific shape of a {\it generic} nanoparticle. We find that the spin contribution to the $g$-tensors agrees with Random Matrix Theory (RMT) predictions. On the other hand, in the strong spin-orbit coupling limit $\delta \tau_{\textrm{\small so}}/\hbar \to 0$, the orbital contribution depends crucially on the space character of the quasi-particle wavefunctions: it levels off at a small value for states of $d$ character but is strongly enhanced for states of $sp$ character. Our numerical results demonstrate that when orbital coupling to the field is included, RMT predictions overestimate the typical $g$-factor of orbitals that have dominant $d$-character. This finding points to a possible source of the puzzling discrepancy between theory and experiment.