An international group of quantum researchers demonstrated how machine learning may be used to filter a nearly endless number of possible material combinations to uncover candidates for superconductivity. The study was published in Physical Review Research.
YRu3B2 and LuRu3B2 gain their superconductivity from electrons forming flat bands in a kagome lattice, named after a hexagonal Japanese basket-weaving pattern. Image Credit: Esa Kapila
According to Aalto University Professor Päivi Törmä, who heads the SuperC consortium behind the research, the innovation will allow for the considerably faster discovery of novel superconductors.
Superconductors transport electric current with no resistance due to a quantum phenomenon that occurs only at very low temperatures. They power not only quantum computers, but also neuroimaging, fusion reactors, and maglev trains.
Nevertheless, it is quite difficult to detect these unicorn materials. Any infinitely variable combination of elements might be a superconductor, but few actually are. What's more, the ones that have already been found need costly cooling equipment to reach the temperatures close to absolute zero that give them their quantum characteristics.
Finding a scalable superconductor that operates at ambient temperature is a race on for scientists worldwide.
Superconductive materials that can operate at room temperature would forever change the way we consume energy. If such a material could replace regular conductors in applications like computers and data centers, global energy consumption could be slashed and the heat footprint of the ICT sector vastly reduced.
Päivi Törmä, Professor, Aalto University
Arriving at Proof of Concept
Professor Törmä and a group of eminent physicists founded the SuperC consortium in 2023, united by a shared goal of using quantum physics to help combat climate change. Their mission is to discover a room-temperature superconductor by 2033. SuperC represents the first coordinated global effort dedicated to discovering new superconducting materials.
Törmä says SuperC's integration of machine learning and quantum geometry provides a strong foundation for its research. Both of the recently discovered materials, YRu3B2 and LuRu3B2, exhibit superconductivity through electrons forming flat bands within a traditional arrangement known as a kagome lattice, which serves as the foundation for this latest discovery.
The scientists narrowed down possible elemental combinations using machine learning to find the two novel superconductors, and performed extensive calculations to identify which materials would be superconductive after pre-screening these using a special method.
SuperC colleagues at Rice University began synthesizing the samples after theoretical validation. Professor Emilia Morosan oversaw this intricate procedure, which entails chemically mixing raw materials to create new molecules. The materials might then be tested by the Rice team to verify their superconductivity.
Why Does It Matter?
Finding novel superconductors is a difficult endeavor because of the complexity of the quantum mechanical theory of superconductivity.
Over the decades researchers have recognized over 7,000 superconductors, but mostly serendipitously. The process of identifying possible materials is so computationally heavy that, in fact, researchers have only been able to theoretically predict the viability of about 20 of these.
Päivi Törmä, Professor, Aalto University
Even if one uncovers what appears to be a promising combination, the majority of them are absolutely worthless. According to Törmä, they are challenging to synthesize and scale. As a result, finding feasible superconductors needs massive computer capacity to screen materials. SuperC's machine-learning methodology debunks that notion.
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. With machine learning, we may be able to push the number of materials we can process into the billions. This will take us a critical step closer to finding a room-temperature superconductor.
Päivi Törmä, Professor, Aalto University
Journal Reference:
Mustaf, A. R., et al. (2026) Machine-learning-guided discovery of kagome superconductors YRu 3 B 2 and LuRu 3 B 2. Physical Review Research. DOI:10.1103/lpqj-7hyg. https://journals.aps.org/prresearch/abstract/10.1103/lpqj-7hyg.