To address the challenge of modelling magnetic effects at small grain sizes, Dr Skelland and Prof Hrkac produced a new methodology for modelling the preferential shape of grains.

The method starts by finding the most preferential grain surfaces (either in literature, or through calculation) and using them to build approximate grain shapes. The surface energies are used to decide what faces make up these grain shapes. For example, if the (111) faces have the lowest energy these will be used to cut the grain into shape, generally resulting in them comprising a majority of the surface.

Each grain shape, once defined, can be built across a range of sizes. At each size the grains can be "compositionally matched", giving them the same number of total atoms, and the same number of each element. Being compositionally identical means they are rearrangements of the same group of atoms and are therefore comparable by Boltzmann Factors. By selecting several possible shapes and building them across a size range it is possible to compare their preferability.

To explain their preferability their surface area is broken down into faces with the same normal. Grouping them in this manner allows the surface energy of each face to be calculated. Surface energy can be used to explain higher or lower preferability for the different grains. Measurement of the role the surface plays can offer insight into preferential grain shapes beyond those investigated in the model, giving a general idea of likely grain shapes.

This method has been presented in an abstract by Dr Skelland at the MMM/Intermag 2022 joint conference, using FePt L10 as a model structure. FePt L10 is the prime candidate for future high density recording media owing to its stability at small grain sizes even at high temperatures. It is expected that grains will have to be on the order <10nm in order to keep increasing the density of permanent magnetic recording media. Investigation of this structure has shown that at such small sizes its shape is dominated by surface effects, and the expected grain shapes will be those with a significant amount of (111) surfaces. This produces octahedrons and truncated octahedrons, and it is these shapes that should be used as input to micromagnetics or atomistic spin modelling when attempting to produce accurate simulations.

This case study is the subject of a future paper, which will be added here once published. In the future the model will be extended to build realistic polycrystalline structures, which will be used as input to atomic spin magnetic modelling.