An international team has used artificial intelligence to identify two new superconducting materials, YRu3B2 and LuRu3B2. The approach combines machine learning with quantum calculations to accelerate the search for materials that conduct electricity without resistance.
Researchers first applied a machine learning algorithm to screen vast numbers of possible chemical combinations. They then performed targeted quantum calculations on the most promising candidates before collaborators at Rice University synthesized and experimentally confirmed the materials.
The work was led by Professor Päivi Törmä of Aalto University through the SuperC consortium, established in 2023. Both compounds exhibit superconductivity due to electrons forming flat bands in a kagome lattice structure.
"Our method uses machine-learning-based pre-screening followed by targeted calculations on the promising candidates," Törmä said. "This approach will greatly speed up superconductor discovery in the future."
The proof-of-concept study was published in Physical Review Research. The findings are scheduled to appear in an Aalto University exhibition in Greater Helsinki from September 1 to October 30, 2026.
