Special CMSM Seminar No.314

Computational micromagnetics is widely used to design magnetic devices. It relies on the continuum theory of micromagnetism. This approach models magnetization on an intermediate length scale. It is small enough to resolve magnetic domain walls, but large enough to replace atomic spins with a continuous function.
This method works well for recording devices, sensors, and permanent magnets. However, classical micromagnetics reaches its limits for certain materials. It cannot simulate large-grain permanent magnets because of the vast difference between domain wall width and typical grain size. It is also restricted to temperatures below the Curie temperature and cannot model phase transitions.
In my talk, I will discuss how to overcome these two limits.

(1) I will introduce a coarse-graining approach to simulate large-grained permanent magnets. We treat the nucleation of reversed domains at interfaces and grain boundaries semi-analytically. We then compute the equilibrium magnetic states micromagnetically. For fast evaluation, we use graph neural networks trained on these simulations. They can rapidly predict how hysteresis properties vary with chemical composition, grain size, or hard-shell thickness in grain-boundary diffused magnets.
(2) I will present a computational framework for magnetic and structural phase transitions. We developed this by combining micromagnetism with Landau theory. Initial results show how thermal hysteresis changes with grain size and texture in Fe2P, Mn5SiB2, and Mn0.5Fe0.5(Ni,Cu)Si alloys. The model also calculates the local stress distribution and shows how the microstructure influences peak stress.

Hossein SEPEHRI-AMIN
Green Magnetic Materials Group
・E-mail