Designing Fe‐Based Amorphous Alloys With both Ultra‐High Magnetization and Ultra‐Low Coercivity Through Artificial Intelligence

Abstract Designing soft magnetic alloys with high magnetization and low coercivity is of special interest for application in high‐frequency and high‐power electric and electronic components. In this work, high‐precision machine‐learning models based on 536 different Fe‐based amorphous alloys are developed. It reveals that the electronegativity difference (δ_χ) and mixing enthalpy (ΔH_mix) of the alloying elements play critical roles in determining the saturated magnetization (B_s) of amorphous alloys. Specifically, smaller δ_χ can strengthen the biased distribution of spin‐up and spin‐down electrons as is revealed by ab initio simulations. Based on these findings, a series of advanced amorphous/nanocrystalline alloys with B_s higher than 1.90 T and coercivity (H_c) as low as 1.2 A m⁻¹ are designed, which also have good amorphous forming ability owing to the suitable mixing enthalpy. The designed alloys with high B_s and low H_c hold promising application potentials in electronic components of high power density and low energy loss.