Magnets quietly power the modern world, serving as critical components in everything from smartphones and medical devices to electric vehicles and power generators. Despite their essential role in daily life and clean energy technologies, developing stronger and more sustainable magnetic materials has historically progressed at a painstaking pace. Today’s most powerful permanent magnets depend heavily on rare-earth elements. These specific elements are notoriously costly, largely imported, and increasingly difficult to secure. Furthermore, traditional laboratory testing of every possible elemental combination is prohibitively expensive and time-consuming, leaving countless potential magnetic compounds undiscovered.
To overcome this massive bottleneck, researchers at the University of New Hampshire have developed a powerful artificial intelligence system to accelerate the discovery of advanced magnetic materials. By shifting the search from the physical laboratory to the digital realm, the team has successfully identified new functional materials that could transform renewable energy systems and electric vehicles.
Building the Northeast Materials Database
The University of New Hampshire research team, led by physics doctoral student Suman Itani, recently published their findings in the journal Nature Communications. Together with physics professor Jiadong Zang and postdoctoral researcher Yibo Zhang, the team built a modern large language model designed specifically to read existing scientific papers. This artificial intelligence system automatically extracts crucial experimental details and data from the text.
Once the information is extracted, the data feeds directly into advanced computer models. These machine learning models are trained to classify materials, predict key magnetic properties, and estimate the specific temperatures at which compounds lose their magnetism. The automated extraction and prediction process eliminates the enormous effort previously required to collect such information by hand. The result is the newly created Northeast Materials Database, a massive and highly searchable resource containing 67,573 magnetic materials entries.
Discovering High-Temperature Alternatives
Within that extensive collection of over 67,000 compounds, the machine learning models highlighted dozens of high-probability candidates for future laboratory validation. Most notably, the AI system successfully identified 25 previously unrecognized compounds capable of remaining magnetic even at elevated temperatures. These discoveries are particularly significant because many materials in the dataset do not contain any rare-earth elements.
Finding materials that can maintain their magnetic behavior at temperatures exceeding practical operating thresholds has long been one of the most difficult challenges in materials science. Until this project, no entirely new permanent magnet had been discovered from the vast pool of known magnetic compounds. The identification of these 25 new high-temperature materials provides scientists with a targeted, highly accurate shortlist. This curated list suggests promising new pathways for developing affordable, sustainable magnets without relying on imported rare-earth elements.
Economic and Environmental Implications
The transition to rare-earth-free options carries major implications for both the economy and the environment. “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,” Itani noted.
Professor Zang echoed this optimism regarding the search for sustainable alternatives to permanent magnets. He emphasized that the combination of the new experimental database and rapidly growing artificial intelligence technologies makes this previously daunting goal achievable. This breakthrough could ultimately influence a wide range of industries that rely on advanced electronics and motors.
Beyond Magnets: Expanding the AI Framework
The impact of this research extends far beyond the discovery of sustainable magnetic materials. The underlying large language model developed for this project demonstrates significant potential for widespread use in other fields, particularly within higher education. For example, the researchers suggest the artificial intelligence framework could be utilized to convert legacy images into modern rich text formats, providing a streamlined method for modernizing and preserving library holdings.
Additionally, the artificial intelligence framework could easily be adapted to explore other complex areas of materials science. This adaptability opens the door for faster innovation across the energy sector, advanced electronics, and quantum technologies. The pioneering materials database project was formally supported by the United States Department of Energy through its Office of Basic Energy Sciences, Division of Materials Sciences and Engineering.
