Statistical modelling of skyrmion magnetic materials with defects

Abstract

Skyrmions, a whirling magnetic texture, stabilised by the Dzyaloshinskii-Moriya interaction, emerge as promising candidates for next-generation data particles, owing to their compact size and topological stability. However, defects introduced during the manufacturing process can unpredictably affect the stability and dynamics of skyrmions, compromising data reliability. To address this, we conduct a comprehensive study using statistical physics-based modeling and machine learning data analysis to assess the impact of defects on the equilibrium properties of skyrmions. In our investigation, we uncover a novel disorder-driven continuous phase transition from a hexagonally-ordered arrangement of skyrmions (OSkL), in defect-free systems, to a disordered array of skyrmions (DSkL) at high defect levels. We pinpoint the OSkL-DSkL transition, representing the minimum defect level required to disrupt the OSkL, via the spin-spin correlation analysis. Due to the lack of appropriate order parameter formalism, we employ deep learning dimensionality reduction methods, which yield alternative transformations of the spin variables, to differentiate other phases present in such materials. In addition to phase classification, we use deep learning methods, based on the U-Net network, to estimate the Hamiltonian parameters and precisely identify defect locations. This more integrated approach, combining machine learning with numerical methods, provides insights into the complex interplay between defects and skyrmions, offering potential pathways for new experimentation and a wide range of technological applications.

More information

Identifiers

ORCID for Ondrej Hovorka: orcid.org/0000-0002-6707-4325

ORCID for Hans Fangohr: orcid.org/0000-0001-5494-7193

Catalogue record

Export record