Advancements in sustainable magnetic materials could significantly decrease reliance on rare earth elements, reduce the costs of electric vehicles, and bolster the manufacturing sector in the U.S., according to Suman Itani, the lead researcher and a doctoral student in physics.
A Comprehensive Magnetic Materials Resource
The newly introduced Northeast Materials Database facilitates the exploration of materials that play a vital role in contemporary technology. Magnets are essential in devices like smartphones, medical equipment, power generators, and electric vehicles. However, the strongest magnets available today rely on rare earth elements that are expensive, predominantly imported, and increasingly challenging to obtain. Despite the vast array of known magnetic compounds, no new permanent magnet has emerged from this existing pool.
Published in Nature Communications, the study details the development of an AI system capable of analyzing scientific literature and extracting crucial experimental data. This information was utilized to train models that assess the magnetic properties of materials and determine the temperature at which they lose magnetism, culminating in a single, searchable database.
Minimizing Dependence on Rare Earth Elements
Researchers have long recognized that numerous magnetic materials may still be undiscovered. However, testing every potential combination of elements, which could reach millions, would require substantial time and financial resources in a laboratory environment.
"We are addressing one of the most significant challenges in materials science -- finding sustainable alternatives to permanent magnets -- and we are hopeful that our experimental database, along with advancing AI technologies, will help us achieve this objective," stated Jiadong Zang, a physics professor and co-author of the study.
Enhancing AI's Role in Scientific Research and Education
The research team also includes Yibo Zhang, a postdoctoral researcher specializing in both physics and chemistry. The scientists envision that the large language model used in this project could extend its applications beyond just building this database, particularly in higher education. For instance, this technology could transform images into modern rich text formats, aiding in the updating and preservation of library collections.
This project has received backing from the Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, U.S. Department of Energy.