Dynamic magnetic susceptibility and electrical detection of ferromagnetic resonance

TL;DR

This paper develops a comprehensive theoretical framework for analyzing the dynamic magnetic susceptibility near ferromagnetic resonance, enabling more accurate interpretation of electrical detection signals and separation of different contributions.

Contribution

It introduces a general form of dynamic susceptibility for arbitrary samples, clarifies its tensor properties, and provides a method to determine key parameters from microwave absorption measurements.

Findings

Dynamic susceptibility is not a Polder tensor for arbitrary anisotropic materials.

Frequency dependence of susceptibility is characterized by six parameters; field dependence by seven.

A practical recipe for parameter determination from microwave absorption data.

Abstract

The dynamic magnetic susceptibility of magnetic materials near ferromagnetic resonance (FMR) is very important in interpreting dc-voltage in electrical detection of FMR. Based on the causality principle and the assumption that the usual microwave absorption lineshape around FMR is Lorentzian, general forms of dynamic susceptibility of an arbitrary sample and the corresponding dc-voltage lineshape are obtained. Our main findings are: 1) The dynamic susceptibility is not a Polder tensor for material with arbitrary anisotropy. Two off-diagonal elements are not in general opposite to each other. However, the linear response coefficient of magnetization to total rf field is a Polder tensor. This may explain why two off-diagonal elements are always assumed to be opposite to each other in analyses. 2) The frequency dependence of dynamic susceptibility near FMR is fully characterized by six numbers while its field dependence is fully characterized by seven numbers. 3) A recipe of how to determine these numbers by standard microwave absorption measurements for an arbitrary sample is proposed. Our results allow one to unambiguously separate the contribution of the anisotropic magnetoresistance to dc-voltage from that of the anomalous Hall effect. With these results, one can reliably extract the information of spin pumping and the inverse spin Hall effect, and determine the spin-Hall angle. 4) The field-dependence of susceptibility matrix at a fixed frequency may have several peaks when the effective field is not monotonic of the applied field. In contrast, the frequency-dependence of susceptibility matrix at a fixed field has only one peak. Furthermore, in the case that resonance frequency is not sensitive to the applied field, the field dependence of susceptibility matrix, as well as dc-voltage, may have another non-resonance broad peak. Thus, one should be careful in interpreting observed peaks.