Working on cleaning the data we have available and seeing what we've got for a MAE prediction model. This resource was nice and had all the raw files uploaded so that you can process them yourself and not need to scrape for it.
This paper describes the open Novamag database that has been developed for the design of novel Rare-Earth free/lean permanent magnets. The database software technologies, its friendly graphical user interface, advanced search tools and available data are explained in detail. Following the philosophy and standards of Materials Genome Initiative, it contains significant results of novel magnetic phases with high magnetocrystalline anisotropy obtained by three computational high-throughput screening approaches based on a crystal structure prediction method using an Adaptive Genetic Algorithm, tetragonally distortion of cubic phases and tuning known phases by doping.
As with NovoMag, this is primarily a DFT generated database of theoretical materials. Additionally, structured were obtained by Adaptive Genetic Algorithm Structure Prediction Methods (USPEX+VASP). I'm not sure of the exact software Novomag used but it was AGA too.
Check out my deep dive into NovoMag:
Also known as the Magnetic Materials Database. I came to this database looking for magnetocrystalline anisotropy energy data for permanent magnet design. After scraping the data from the app, which is
There are about 1600 entries total and around 600 of them have MAE data! There is a small section (20 materials) of micromagnetic simulations with coercivity and Curie temperature data. We can come back to that some other time.
For now, the 600 we have is pretty good sized! These all should come with CIF data so we can really work them in computational environments.
Let's take a look at some example data:
There's a lot of good detail, but we'll be focusing on the magnetic section. The data reports the magnetic data of interest like so:
For uniaxial systems in this dataset:
MAE = K₁ (in MJ/m³)
Given directly as first element of magnetocrystalline_anisotropy_constants
Represents energy difference between hard and easy axes
Calculate using:
κ = √(K₁/μ₀Ms²)
Where:
K₁ = first anisotropy constant in J/m³ (multiply MJ/m³ by 10⁶)
μ₀ = 4π × 10⁻⁷ T·m/A
Ms = Js/(4π × 10⁻⁷) in A/m, where Js is the given value in Tesla
This was really surprising at first, but it makes a lot of sense now that I've learned more about it. In the raw data, we see this comment made for all the data with MAE calculated:
A tetragonal distortion was imposed on cubic material by changing the lattice constant in a brute force way, just to find out how the material would change under strain
At first, I thought this would make all of this data useless as it wasn't true to the material. I'll still need to see how compatible it is with our Novomag data and if we can train a model that merges both datasets.
Applied 1% compressive strain to all materials
Maintained constant volume (material expands perpendicular to compression)
Created tetragonal distortion (breaks cubic/hexagonal symmetry)
Kept structure unrelaxed (atoms can't move to minimize energy)
Natural cubic materials (Fe, Co, Ni) have zero MAE
1% strain creates artificial but measurable anisotropy
Shows which materials are most "magnetically responsive" to strain
The paper mentions how these materials could be good candidates for doping interstitial atoms such as H, B, C, and N and how that can lead to substantial tetragonal distortions -> higher MAE.
See Section 4.2.2, Tetragonally distortion of cubic phases.
Once again we're at a stopping point because of our inability to effectively predict MAE. Our AI discovery agents have discovered materials that have all the properties we can currently predict. This
Automated recap of the latest activity in #permanent-magnets, created by @hermes.