When can localized spins interacting with conduction electrons in ferro- or antiferromagnets be described classically via the Landau-Lifshitz equation: Transition from quantum many-body entangled to quantum-classical nonequilibrium states

Abstract

Experiments in spintronics and magnonics operate with macroscopically large number of localized spins within ferromagnetic (F) or antiferromagnetic (AF) materials, so that their nonequilibrium dynamics is standardly described by the Landau-Lifshitz (LL) equation treating localized spins as classical vectors of fixed length. However, spin is a genuine quantum degree of freedom, and even though quantum effects become progressively less important for spin value $S>1$, they exist for all $S < \infty$. While this has motivated exploration of limitations/breakdown of the LL equation, by using examples of F insulators, analogous comparison of fully quantum many-body vs. quantum (for electrons)-classical (for localized spins) dynamics in systems where nonequilibrium conduction electrons are present is lacking. Here we employ quantum Heisenberg F or AF chains of $N=4$ sites, whose localized spins interact with conduction electrons via $sd$ exchange interaction, to perform such comparison by starting from unentangled pure (at zero temperature) or mixed (at finite temperature) quantum state of localized spins as the initial condition. This reveals that quantum-classical dynamics can faithfully reproduce fully quantum dynamics in the F metallic case, but only when spin $S$, Heisenberg exchange between localized spins and $sd$ exchange are sufficiently small. Increasing any of these three parameters can lead to substantial deviations, which are explained by the dynamical buildup of entanglement between localized spins and/or between them and electrons. In the AF metallic case, substantial deviations appear even at early times, despite starting from unentangled N\'{e}el state, which therefore poses a challenge on how to rigorously justify wide usage of the LL equation in phenomenological modeling of antiferromagnetic spintronics experiments.