0 
0 
Ames National Laboratory scientists have devised a machine learning model to predict new magnet materials without using scarce elements. This innovative approach, focusing on a material’s Curie temperature, offers a more sustainable path for future technological applications.
Scientists use AI to find new magnetic materials without critical elements.
A team of researchers from the U.S. Department of Energy’s Ames National Laboratory developed a new machine learning model for discovering critical-element-free permanent magnet materials. The model predicts the Curie temperature of new material combinations. It is an important first step in using artificial intelligence to predict new permanent magnet materials. This model adds to the team’s recently developed capability for discovering thermodynamically stable rare earth materials.
Importance of High-Performance Magnets
High-performance magnets are essential for technologies such as wind energy, data storage, electric vehicles, and magnetic refrigeration. These magnets contain critical materials such as cobalt and rare earth elements like Neodymium and Dysprosium. These materials are in high demand but have limited availability. This situation is motivating researchers to find ways to design new magnetic materials with reduced critical materials.
Photo of a magnet. Credit: U.S. Department of Energy Ames National Laboratory
The Role of Machine Learning
Machine learning (ML) is a form of artificial intelligence. It is driven by computer algorithms that use data and trial-and-error algorithms to continually improve its predictions. The team used experimental data on Curie temperatures and theoretical modeling to train the ML algorithm. Curie temperature is the maximum temperature at which a material maintains its magnetism.
“Finding compounds with the high Curie temperature is an important first step in the discovery of materials that can sustain magnetic properties at elevated temperatures,” said Yaroslav Mudryk, a scientist at Ames Lab and senior leader of the research team. “This aspect is critical for the design of not only permanent magnets but other functional magnetic materials.”
According to Mudryk, discovering new materials is a challenging activity because the search is traditionally based on experimentation, which is expensive and time-consuming. However, using a ML method can save time and resources.
Developing the Model
Prashant Singh, a scientist at Ames Lab and member of the research team, explained that a major part of this effort was to develop an ML model using fundamental science. The team trained their ML model using experimentally known magnetic materials. The information about these materials establishes a relationship between several electronic and atomic structure features and Curie temperature. These patterns give the computer a basis for finding potential candidate materials.
Model Testing and Validation
To validate the model, the team used compounds based on Cerium, Zirconium, and Iron. This idea was proposed by Andriy Palasyuk, a scientist at Ames Lab and member of the research team. He wanted to focus on unknown magnet materials based on earth-abundant elements. “The next super magnet must not only be superb in performance, but also rely on abundant domestic components,” said Palasyuk.
Palasyuk worked with Tyler Del Rose, another scientist at Ames Lab and member of the research team, to synthesize and characterize the alloys. They found that the ML model was successful in predicting the Curie temperature of material candidates. This success is an important first step in creating a high-throughput way of designing new permanent magnets for future technological applications.
“We are writing physics-informed machine learning for a sustainable future,” said Singh.
Reference: “Physics-Informed Machine-Learning Prediction of Curie Temperatures and Its Promise for Guiding the Discovery of Functional Magnetic Materials” by Prashant Singh, Tyler Del Rose, Andriy Palasyuk and Yaroslav Mudryk, 2 August 2023, Chemistry of Materials.
DOI: 10.1021/acs.chemmater.3c00892
