Magnets and Magnetic Materials

Electric-vehicle motors, wind-turbine generators, and industrial drives all depend on high-performance permanent magnets. NdFeB and SmCo deliver exceptional energy products but rely on rare-earth elements exposed to supply-chain concentration, price volatility, and geopolitical risk. Developing rare-earth-free or rare-earth-reduced magnets with competitive performance is one of the decade's most pressing materials challenges. The design space for new magnetic phases is enormous: combinations of transition metals, metalloids, and light elements across various crystal structures create billions of potential candidates, and the physics linking composition and structure to magnetic performance involves complex exchange interactions, magnetocrystalline anisotropy, and domain structure that defy simple screening rules.

Current magnet discovery relies on DFT calculations of magnetic moments and anisotropy energies for proposed crystal structures, combinatorial thin-film synthesis of narrow composition ranges, or ML models trained on magnetic-property databases. DFT is accurate but expensive and evaluates only the structures someone has already proposed. Combinatorial synthesis produces thin-film samples whose magnetic properties may not transfer to bulk magnets. ML models interpolate within known families but lack the structural constraints to generate valid candidates in novel chemical spaces; they may predict high magnetic moments for compositions that cannot form stable crystals.