An international collaboration between scientists in the United Kingdom and Japan has engineered the strongest iron-based superconducting magnet to date, powered by artificial intelligence. The findings appear in the scientific journal magazine NPG Asian Materials (NPG AM).
Superconducting magnets generate exceptionally strong, stable magnetic fields with minimal energy input. They play crucial roles in magnetic resonance imaging (MRI) and magnetic levitation technologies. Today, the field relies on large magnets made from niobium-ttin alloys, which are powerful but costly and bulky, limiting practical deployment.
The research team employed a machine learning system named BOXVIA to design a magnet that is both more economical and efficient. BOXVIA recognized patterns that lead to improved performance and guided the material choices and structural configuration that underpin the magnet’s strength.
Without this AI-driven approach, the researchers estimate that months would be spent designing and testing each prototype to reach comparable performance levels.
Further analysis revealed that the BOXVIA-designed superconductors exhibit a distinctive microscopic structure, featuring larger iron crystals. This structural arrangement contributed to the creation of the strongest iron-based superconducting magnet achieved so far, pushing the boundaries of what this material class can deliver.
Earlier work in related fields has explored how artificial intelligence can assist in astronomical discoveries, including the search for Earth-like planets in deep space, highlighting AI’s broader potential across science.