Micromagnetic simulations of thermally activated magnetization reversal of nanoscale magnets

TL;DR

This paper uses micromagnetic simulations incorporating thermal fluctuations to study magnetization reversal in nanoscale magnets, revealing the roles of thermal activation, end cap dynamics, and defect formation during reversal processes.

Contribution

It introduces a detailed simulation approach combining stochastic Landau-Lifshitz-Gilbert equations with the Fast Multipole Method for dipole interactions, enabling nanoscale magnetization dynamics analysis at finite temperatures.

Findings

Magnetization reversal occurs via end cap nucleation and propagation.

Thermal fluctuations significantly influence reversal at fields below coercivity.

Reversal times follow a statistical distribution, highlighting thermal activation effects.

Abstract

Numerical integration of a stochastic Landau-Lifshitz-Gilbert equation is used to study dynamic processes in single-domain nanoscale magnets at nonzero temperatures. Special attention is given to including thermal fluctuations as a Langevin term, and the Fast Multipole Method is used to calculate dipole-dipole interactions. It is feasible to simulate these dynamics on the nanosecond time scale for spatial discretizations that involve on the order of 10000 nodes using a desktop workstation. The nanoscale magnets considered here are single pillars with large aspect ratio. Hysteresis-loop simulations are employed to study the stable and metastable configurations of the magnetization. Each pillar has magnetic end caps. In a time-dependent field the magnetization of the pillars is observed to reverse via nucleation, propagation, and coalescence of the end caps. In particular, the end caps propagate into the magnet and meet near the middle. A relatively long-lived defect is formed when end caps with opposite vorticity meet. Fluctuations are more important in the reversal of the magnetization for fields weaker than the zero-temperature coercive field, where the reversal is thermally activated. In this case, the process must be described by its statistical properties, such as the distribution of switching times, averaged over a large number of independent thermal histories.