Artificial intelligence is giving scientists a faster and more focused way to search for new superconductorsโmaterials that can carry electricity without resistance. A new study shows that machine learning can help researchers quickly sort through huge numbers of possible material combinations and identify the most promising candidates for discovery.
The breakthrough was led by an international team connected to the SuperC consortium, headed by Professor Pรคivi Tรถrmรค of Aalto University. Their approach combines machine learning, quantum physics, theoretical calculations, and laboratory experiments to speed up the search for superconducting materials.
Superconductors are remarkable because they allow electrical current to flow without energy loss. They are already important in advanced technologies such as quantum computers, medical imaging systems, fusion research, maglev trains, and sensitive scientific instruments. However, most known superconductors only work at extremely low temperatures, often close to absolute zero, which makes them expensive and difficult to use widely.
A practical superconductor that works at room temperature would be revolutionary. It could transform power transmission, reduce energy waste, improve computing systems, and cut the heat produced by data centers and other high-energy technologies.
According to Tรถrmรค, room-temperature superconductors could permanently change how the world uses energy. If ordinary electrical conductors could be replaced with superconducting materials in computers, communication systems, and data centers, global energy consumption and heat production could be greatly reduced.
AI and Quantum Physics Join the Search
Finding new superconductors is extremely difficult because there are almost limitless combinations of chemical elements that could form new materials. Only a very small number of these combinations become superconducting, and discovering them has often depended on slow trial-and-error methods.
To overcome this challenge, the research team used machine learning to scan large numbers of possible elemental combinations. The AI system helped narrow the search to materials with the strongest potential. These selected candidates were then examined using detailed quantum calculations to determine whether they were likely to become superconductors.
This method led the researchers to two new superconducting compounds: YRuโBโ and LuRuโBโ. Both materials gain their superconducting behavior from electrons arranged in flat bands within a kagome lattice, a geometric pattern named after a traditional Japanese basket-weaving design.
After the theoretical predictions were made, collaborators at Rice University, led by Professor Emilia Morosan, synthesized the compounds in the laboratory. Experimental tests then confirmed that both materials are indeed superconductors.
The study demonstrated superconducting critical temperatures of about 0.81 K for YRuโBโ and 0.95 K for LuRuโBโ, confirmed through magnetization, specific heat, and electrical transport measurements. Both compounds also showed nearly complete superconducting volume fractions, confirming bulk superconductivity.
A Faster Route to Future Discoveries
Over the years, scientists have identified more than 7,000 superconductors, but many were found by chance rather than by targeted prediction. The theoretical search for new superconductors has remained slow because it requires complex quantum mechanical calculations that demand major computing power.
The new AI-guided method changes this process. Instead of performing expensive calculations on every possible material, machine learning first filters the list and highlights only the most promising candidates. Researchers can then focus detailed calculations and experiments on those selected materials.
This could make future searches dramatically faster. According to the research team, the method may eventually allow scientists to screen billions of material combinations, bringing them closer to the long-standing goal of discovering a room-temperature superconductor.
Although YRuโBโ and LuRuโBโ do not operate at room temperature, their discovery proves that the AI-guided strategy works. It also shows how combining artificial intelligence with quantum theory and experimental chemistry can accelerate progress in one of the most important areas of materials science.
The research marks an important step toward a future where superconductors could help reshape energy systems, computing, transportation, and advanced technologies.
Journal Reference:
Mustaf, R. A., Sajilesh, K. P., Mishra, S., Deng, J., Jiang, Y., Hiorth, K. H., Lamponen, E. O., Gutierrez-Amigo, M., Tรถrmรค, P., Marques, M. A. L., Bernevig, B. A., & Morosan, E. (2026). Machine-learning-guided discovery of kagome superconductors YRuโBโ and LuRuโBโ. Physical Review Research, 8, 023308. https://doi.org/10.1103/lpqj-7hyg