Alibaba's AI Agent Discovers Four New Superconducting Materials

Alibaba's Damo Academy, in collaboration with Renmin University of China and the University of Chinese Academy of Sciences, has announced the development of an AI agent named ElementsClaw. This agent has successfully predicted and identified four novel superconducting materials, which have subsequently been synthesized and experimentally validated.

The ElementsClaw agent leverages AI for Science technologies to accelerate material discovery. It is designed not only to predict superconductivity but also to review literature, assess synthesis feasibility, and propose experimental designs, significantly enhancing efficiency. The AI screened 2.4 million crystal structures, identifying 68,000 potential candidates, from which the four new materials were derived.

Among the verified materials are Hf₂₁Re₂₅, Zr₄VRe₇, HfZrRe₄, and Zr₃ScRe₈, with the latter showing a critical temperature as high as 6.5K. This achievement marks the first instance of AI agents discovering and having their findings experimentally confirmed in the field of superconductors. The Damo Academy has made the database of all predicted 2.4 million stable crystals publicly available for researchers.

This development is expected to broaden the scope of AI applications in materials science, potentially aiding in the discovery of electrolytes for solid-state batteries, multiphase catalysts, and thermoelectric materials.