Image Courtesy: Esa Kapila
Artificial intelligence is transforming the search for superconductors, giving scientists a faster way to identify materials that can conduct electricity with zero resistance. Researchers say the AI-powered approach could significantly accelerate the hunt for a room-temperature superconductor, a breakthrough that would revolutionize energy, computing, and transportation.
The research comes from the international SuperC consortium, led by Professor Päivi Törmä of Aalto University. By combining machine learning with advanced quantum physics calculations, the team has demonstrated a new method for rapidly identifying promising superconducting materials from an almost limitless number of chemical combinations. The findings, recently published in Physical Review Research, mark an important proof of concept.
Superconductors allow electricity to flow without losing energy, but today’s materials only work at extremely low temperatures, often requiring expensive cooling systems close to absolute zero. They already play a critical role in technologies such as quantum computers, MRI scanners, fusion reactors, and magnetic levitation trains. A practical superconductor that functions at room temperature could dramatically reduce energy losses across power grids, computers, and data centers.
The SuperC consortium was founded in 2023 with the ambitious goal of discovering a room-temperature superconductor by 2033. In its latest study, researchers used machine learning to screen enormous numbers of possible materials before applying detailed quantum mechanical calculations to the most promising candidates. The approach led to the discovery of two new superconductors, YRu?B? and LuRu?B?, whose properties arise from electrons moving through flat bands in a kagome lattice, a geometric structure inspired by traditional Japanese basket weaving.
After the AI identified the materials, scientists at Rice University synthesized the compounds and experimentally confirmed that both exhibit superconductivity, validating the predictions.
Researchers say the new workflow addresses one of the biggest challenges in the field. More than 7,000 superconductors have been discovered over the decades, mostly through trial and error, while only about 20 have been theoretically predicted because of the immense computational effort involved. By using AI to eliminate unlikely candidates before performing expensive calculations, scientists believe they could eventually evaluate billions of potential materials.
If the approach continues to deliver results, it could bring researchers significantly closer to developing a room-temperature superconductor, one of the most sought-after breakthroughs in modern physics.
