Spin-memory loss due to spin-orbit coupling at ferromagnet/heavy-metal interfaces: Ab initio spin-density matrix approach

TL;DR

This paper introduces a quantum-mechanical approach using spin-density matrices combined with density functional theory to analyze spin-memory loss at various metal interfaces, revealing microscopic origins of SML phenomena.

Contribution

It develops a microscopic framework to quantify spin-memory loss at interfaces, linking spin-orbit coupling effects to experimental observations through ab initio calculations.

Findings

SML is nonzero at Co/Cu interface due to interfacial spin-orbit coupling.

Large SML observed at Pt/Cu interface explained by spin textures.

SML occurs even without disorder, intermixing, or magnons.

Abstract

Spin-memory loss (SML) of electrons traversing ferromagnetic-metal/heavy-metal (FM/HM), FM/normal-metal (FM/NM) and HM/NM interfaces is a fundamental phenomenon that must be invoked to explain consistently large number of spintronic experiments. However, its strength extracted by fitting experimental data to phenomenological semiclassical theory, which replaces each interface by a fictitious bulk diffusive layer, is poorly understood from a microscopic quantum framework and/or materials properties. Here we describe an ensemble of flowing spin quantum states using spin-density matrix, so that SML is measured like any decoherence process by the decay of its off-diagonal elements or, equivalently, by the reduction of the magnitude of polarization vector. By combining this framework with density functional theory (DFT), we examine how all three components of the polarization vector change at Co/Ta, Co/Pt, Co/Cu, Pt/Cu and Pt/Au interfaces embedded within Cu/FM/HM/Cu vertical heterostructures. In addition, we use ab initio Green's functions to compute spectral functions and spin textures over FM, HM and NM monolayers around these interfaces which quantify interfacial spin-orbit coupling and explain the microscopic origin of SML in long-standing puzzles, such as why it is nonzero at Co/Cu interface; why it is very large at Pt/Cu interface; and why it occurs even in the absence of disorder, intermixing and magnons at the interface.