An international team of researchers has identified two new superconductors, YRu3B2 and LuRu3B2, both exhibiting superconductivity due to a unique atomic arrangement resembling a traditional Japanese kagome lattice. The discovery represents a substantial step toward realizing energy efficiency gains, achieved by utilizing machine learning to filter a large number of possible material combinations.
According to Aalto University Professor Päivi Törmä, who leads the SuperC consortium, “Our method uses machine-learning-based pre-screening followed by targeted calculations on the promising candidates.” This approach will greatly speed up superconductor discovery in the future. These materials, carrying electric current with zero resistance, could one day power technologies ranging from neuroimaging to fusion reactors.
Machine Learning Identifies YRu3B2 and LuRu3B2 Superconductors
YRu3B2 and LuRu3B2 exhibit superconductivity stemming from a unique atomic arrangement; both materials feature a kagome lattice, a hexagonal pattern inspired by traditional Japanese basket weaving. This structural characteristic is key to the flow of electrons within the materials, enabling the quantum phenomenon of superconductivity, where electrical resistance vanishes. The discovery, detailed recently in Physical Review Research, marks a change in how scientists approach the search for these elusive materials, moving beyond serendipitous finds toward a predictive, machine-learning driven methodology.
This computational approach addresses a longstanding bottleneck in superconductor discovery; traditionally, researchers have only been able to theoretically predict the viability of roughly 20 of the over 7,000 superconductors identified to date. The team’s algorithm pre-screens potential candidates, dramatically reducing the number of materials requiring detailed, computationally intensive calculations to confirm superconductivity.
These synthesized materials then underwent rigorous testing to confirm their superconducting properties, validating the predictions made by the machine-learning algorithm. The success of this workflow demonstrates the power of integrating computational prediction with experimental verification, a strategy the SuperC consortium intends to refine and expand. The consortium’s ultimate goal is ambitious; they aim to identify a room-temperature superconductor by 2033, a breakthrough that would revolutionize energy transmission and storage.
The arduous task of finding viable superconductors is compounded by the fact that many promising combinations prove unusable in practice, often due to difficulties in synthesis or scalability. Professor Törmä notes that even identifying the 7,000 known superconductors has largely been a matter of chance, a situation the SuperC consortium seeks to change.
Their machine-learning approach promises to move beyond this reliance on serendipity, enabling the processing of billions of materials and significantly increasing the odds of finding a scalable, room-temperature superconductor. “With machine learning, we may be able to process into the billions,” says Törmä, highlighting the potential for exponential growth in discovery rates. The implications of room-temperature superconductivity are far-reaching, extending beyond quantum computing to impact diverse fields like neuroimaging, fusion reactors, and magnetic levitation trains.
Currently, superconductors require extremely low temperatures to function, necessitating expensive and energy-intensive cooling equipment. Eliminating this requirement would unlock a vast range of applications and dramatically reduce energy consumption. Törmä emphasizes the potential for a significant reduction in the heat footprint of the information and communication technology sector if these materials could replace conventional conductors in computers and data centers.
The SuperC consortium’s work is underpinned by a recognition of the complex quantum mechanical theory governing superconductivity. This complexity has historically made the search for new superconductors a daunting challenge, requiring immense computational resources and often yielding limited results. The team’s combination of quantum geometry and machine learning provides a powerful new framework for navigating this complexity, focusing computational efforts on the most promising material candidates.
The discovery of YRu3B2 and LuRu3B2 is not merely an addition to the list of known superconductors; it represents a validation of a new methodology. The SuperC consortium’s success demonstrates the potential of machine learning to accelerate materials discovery, offering a pathway to overcome the limitations of traditional trial-and-error approaches.
The consortium’s research will be showcased at Aalto University’s Designs for a Cooler Planet exhibition, opening September 1st, 2026, in Greater Helsinki, Finland, providing a public platform to highlight the potential of these materials to address climate change. Funding for the SuperC consortium comes from a range of sources, including The Kavli Foundation, Klaus Tschira Stiftung, and Kevin Wells, as well as the Jane and Aatos Erkko Foundation, the Keele Foundation, and the Magnus Ehrnrooth Foundation and the Neste and Fortum Foundation.
This diverse funding base reflects the broad recognition of the importance of superconductivity research and the potential for transformative impact. The consortium’s coordinated global effort, combined with its innovative use of machine learning, positions it to continue the search for a room-temperature superconductor, a material that could reshape the future of energy and technology.
“Our method uses machine-learning-based pre-screening followed by targeted calculations on the promising candidates. This approach will greatly speed up superconductor discovery in the future.”
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
