An international research consortium has identified two previously unknown superconductors with the help of machine learning, a breakthrough that scientists say could dramatically accelerate the global race to find materials that conduct electricity without any loss. The discovery, announced by Aalto University in Finland and partner institutions, marks one of the first times artificial intelligence has guided the confirmed discovery of new superconducting materials.
Superconductors carry electrical current with zero resistance, a property that underpins technologies such as MRI scanners, magnetically levitated trains and quantum computers. Until now, they have only worked at extremely low temperatures or under enormous pressure, and finding new candidates has traditionally relied on slow trial-and-error experimentation. The hunt for a material that superconducts at room temperature remains one of the most coveted goals in physics.
The team, working within the SuperC consortium led by Aalto University professor Päivi Törmä, combined machine learning methods with quantum physics calculations to screen candidate compounds. The approach flagged two materials, YRu3B2 and LuRu3B2, which the researchers then synthesized in the laboratory and confirmed as superconductors through experiments. Both compounds derive their properties from electrons forming so-called flat bands within a kagome lattice, a crystal structure that resembles a traditional Japanese basket-weaving pattern.
The results come from a collaboration spanning Aalto University in Finland, Rice University and Princeton University in the United States, Ruhr University Bochum in Germany and the Donostia International Physics Center in Spain. The findings appeared in the journal Physical Review Research, and the consortium said the same pipeline of prediction, synthesis and verification can now be applied at a far larger scale.
Researchers believe the machine learning approach could eventually push the number of candidate materials that can be screened into the billions, compared with the handful that laboratories can test by hand each year. The SuperC project, launched in 2023, has set itself the ambitious target of discovering a room-temperature superconductor by 2033, a material that would transform power grids, transportation and computing by eliminating energy losses.
The scientists cautioned that the two new compounds still operate at low temperatures and will not themselves revolutionize everyday technology. Their significance, the team noted, lies in proving that an AI-guided discovery pipeline works end to end, from computational prediction to experimental confirmation, providing a template that laboratories around the world can now adopt and scale up in the years ahead.
Comments