Physics-enhanced AI framework for discovering new classes of unconventional superconductors
The purpose of this project is to develop a physics-enhanced AI framework to discover unconventional superconductors and quantum materials. The strategy integrates high-accuracy simulations, machine learning, automated workflows, materials synthesis, and neutron scattering into a closed-loop discovery cycle.
Neutron-scattering experiments at ORNL’s High Flux Isotope Reactor (HFIR) and Spallation Neutron Source (SNS) will determine crystal and magnetic structures to validate predictions. These experimental results will feed directly back into the AI workflow, completing the closed loop of theory, computation, artificial intelligence, and physical experiments.
The discovery of new superconductors with higher operating temperatures remains a major challenge in materials science and one of DOE’s highest priorities. Such materials could ultimately enable more efficient power transmission, advanced energy technologies, high-field magnets, quantum devices, and new scientific instruments. However, unconventional superconductors are difficult to predict because their properties emerge from a complex interplay among magnetism, electronic structure, lattice vibrations, and chemical composition. Existing materials databases and artificial intelligence models generally do not capture these coupled effects with sufficient fidelity, limiting their ability to identify materials beyond already known families. This project will develop a physics-enhanced AI framework to discover unconventional superconductors and quantum materials.
