A team of scientists at the University of New Hampshire has leveraged artificial intelligence to tackle one of the most stubborn bottlenecks in clean energy technology: our dependence on rare earth magnets. Their AI system has identified 25 previously unknown magnetic materials capable of maintaining their properties at high temperatures — a discovery that could dramatically reduce the cost of electric vehicles and renewable energy systems.
The research, published in Nature Communications, describes the creation of the Northeast Materials Database, a searchable resource containing 67,573 magnetic compounds that is now freely available to researchers worldwide.
Why Magnets Matter
Permanent magnets are everywhere in modern life. They are critical components in smartphones, medical imaging devices, wind turbines, power generators, and — crucially — the electric motors that drive EVs. The most powerful magnets available today rely on rare earth elements like neodymium and dysprosium.
Despite their name, rare earth elements are not particularly rare in the Earth's crust. The problem is that they are difficult and environmentally damaging to extract, and the global supply chain is heavily concentrated. China controls approximately 60 percent of rare earth mining and nearly 90 percent of processing capacity, creating significant supply chain vulnerabilities for other nations.
As the world electrifies its transportation fleet, demand for these materials is expected to surge. Finding alternatives is not just an academic exercise — it is an economic and strategic imperative.
Teaching AI to Read Science Papers
The breakthrough came from an unconventional approach. Rather than testing materials one by one in a laboratory — a process that could take decades given the millions of possible element combinations — the team built an AI system capable of reading scientific papers and extracting experimental data about magnetic properties.
"By accelerating the discovery of sustainable magnetic materials, we can reduce dependence on rare earth elements, lower the cost of electric vehicles and renewable-energy systems, and strengthen the U.S. manufacturing base," said Suman Itani, the study's lead author and a doctoral student in physics.
The AI processed thousands of published studies, pulling out information about magnetic compounds and their behavior at different temperatures. This data was then used to train models that could predict whether a material would be magnetic and calculate the temperature at which it would lose its magnetism — a critical property for practical applications.
25 New Candidates
Among the database's 67,573 compounds, the team identified 25 materials that had never previously been recognized as high-temperature magnets. These candidates could potentially serve as alternatives to rare earth magnets in motors, generators, and other applications where maintaining magnetic strength at elevated temperatures is essential.
The discovery is particularly significant because no entirely new permanent magnet has been identified from the existing pool of known magnetic compounds in recent memory. The field has been stuck, and AI may have just provided the key to getting unstuck.
Beyond the Database
Professor Jiadong Zang, a co-author of the study, is optimistic about the future: "We are tackling one of the most difficult challenges in materials science — discovering sustainable alternatives to permanent magnets — and we are optimistic that our experimental database and growing AI technologies will make this goal achievable."
The project, supported by the U.S. Department of Energy, represents a growing trend of using AI not just to analyze data but to accelerate fundamental scientific discovery — bringing us closer to a cleaner, more affordable energy future.
