Scientists at the University of New Hampshire have used artificial intelligence to speed up the search for new magnetic materials. Their work led to the creation of an extensive database that could help reduce the dependence of electric vehicles and renewable energy on rare earth elements.
The researchers assembled the Northeast Materials Database — a resource with 67 573 magnetic compounds. Among them are 25 new materials that retain their magnetic properties at high temperatures. The database was the result of analyzing scientific papers using AI models.
"By accelerating the discovery of sustainable magnetic materials, we can reduce dependence on rare earth elements, make electric vehicles and renewable energy systems cheaper, and strengthen the U.S. manufacturing base," noted the lead author of the study, Suman Itani.
Modern strong magnets often require rare earth elements, which are expensive, mostly imported and difficult to extract. Although thousands of magnetic compounds are already known, no completely new permanent magnets have yet been found from this pool. Testing all possible combinations in the laboratory takes too much time and money.
The study was published in Nature Communications. The team used large language models to extract experimental data from papers, and then trained the models to predict magnetic properties and the Curie temperature. The project also included physics professor Jiadong Zang and postdoctoral researcher Yibo Zhang.
Support was provided by the Office of Basic Energy Sciences of the U.S. Department of Energy. The scientists hope that the database and growing AI technologies will make the search for sustainable alternatives a reality.
The work opens the way to cheaper and more environmentally friendly technologies for transport and energy. Magnets are used in smartphones, medical devices, generators and electric vehicles, so the impact of the discovery could be broad.
