Rice Confirms 2 AI-Predicted Superconductors as SuperC Model Screens Untested Compounds
Updated
Updated · Futura · Sep 5
Rice Confirms 2 AI-Predicted Superconductors as SuperC Model Screens Untested Compounds
1 articles · Updated · Futura · Sep 5
Summary
Rice University physicists synthesized YRu₃B₂ and LuRu₃B₂ and verified that both behave as superconductors after an AI model first identified them as promising candidates.
Aalto University’s machine-learning system, built under the SuperC project launched in 2023, screened large sets of untested chemical structures and sent only top prospects for heavier quantum calculations and lab work.
Both compounds use a kagome lattice of ruthenium and boron that creates flat electron bands, strengthening the interactions needed for electron pairing and zero-resistance current flow.
Neither material is ready for grid use because both still need extreme cooling well below room temperature, but the result shows AI can sharply cut trial-and-error in the hunt for practical superconductors.
Researchers say the same predictive workflow could now be applied to wider materials searches, including room-temperature superconductors, denser batteries and industrial catalysts.
AI just found new superconductors, but could this exact algorithm finally unlock the holy grail of room-temperature zero-resistance energy?
If machine learning can accurately predict complex quantum materials, are traditional trial-and-error chemistry experiments becoming completely obsolete?