AI Uses Crystal Symmetry to Find New Antiferromagnets for Spintronics (opens in new tab)

Researchers developed SG-CDVAE, a symmetry-aware generative AI framework that embeds space-group information into the generation of crystal structures to improve the search for high-symmetry antiferromagnets. From 110,000 generated structures, the workflow identified 80 high-symmetry AFM candidates, of which 4 were computationally validated as thermodynamically and dynamically stable.