Disorder-driven magnon anomalies: from bandstructure blur to spin-stiffness attenuation

We leverage quantum-accurate, spin-informed machine-learning models to investigate how atomic defects and disorder affect the properties of magnetic materials, from the microscopic magnon dispersion to macroscopic observables such as spin stiffness.

We elucidate how non-uniformities in the spatial distribution of vacancy defects control blurring of effective magnon bandstructures and related density-of-state anomalies.

We overcome the challenge of accurately describing atomic-scale disorder by developing a highly-efficient “Magnon“ software package, which enables us to quantitatively predict exchange interactions and linear-spin-wave-theory magnon dispersions in solids with arbitrary composition, ranging from perfect crystals to defective solids — simulated explicitly with tens of thousands of atoms. We show applications in simulating the well-defined magnon bandstructure of ordered crystals with wavevector-commensurate collinear or non-collinear spin ordering, as well as how it shifts, broadens, and blurs upon increasing defect concentrations.

Finally, we show that our methodology interfaces with state-of-the-art foundational spin-informed models, and is thus a step towards automating magnon simulations in materials with arbitrary composition and disorder.

References: https://magnon.readthedocs.io