AI-Driven Discovery of Critical-Element-Free Magnetic Materials
A team of scientists from the Ames National Laboratory has unveiled a groundbreaking machine learning model designed to identify critical-element-free permanent magnet materials.
This novel model is a predictive tool for assessing the Curie temperature of fresh material combinations, marking a pivotal advancement in the application of artificial intelligence to forecast new permanent magnet materials. This development follows the team’s recent achievement in uncovering thermodynamically stable rare earth materials.
The Significance of High-Performance Magnet
High-performance magnets are indispensable for various technologies, including wind energy, data storage, electric vehicles, and magnetic refrigeration.
These magnets often contain critical materials like cobalt, Neodymium, and Dysprosium—scarce resources in high demand. This scarcity has spurred researchers to seek innovative ways of designing magnetic materials that reduce reliance on critical elements.
Harnessing the Potential of Artificial Intelligence
Machine learning (ML), a facet of artificial intelligence, relies on data and iterative algorithms to enhance predictive capabilities continually.
The research team leveraged experimental data on Curie temperatures and theoretical modeling to train their ML algorithm. The Curie temperature signifies the maximum temperature at which a material retains its magnetic properties.
The Role of ML in Material Discovery
Yaroslav Mudryk, a scientist at Ames Lab and the research team’s senior leader, emphasized the importance of identifying compounds with high Curie temperatures.
These materials can sustain magnetic properties at elevated temperatures, making them crucial for designing permanent magnets and other functional magnetic materials.
Traditionally, the search for such materials relied on expensive and time-consuming experimentation. However, ML offers a more efficient and resource-saving alternative.
Building the ML Model on Scientific Foundations
Prashant Singh, a scientist at Ames Lab and a research team member, underscored the project’s focus on developing an ML model rooted in fundamental scientific principles.
The model was trained using known magnetic materials, establishing connections between various electronic and atomic structure attributes and Curie temperature. These patterns provide the ML model with a foundation for identifying potential candidate materials.
Putting the Model to the Test
To validate their model, the team experimented with compounds based on Cerium, Zirconium, and Iron—an idea proposed by Andriy Palasyuk, another scientist on the team. Their goal was to explore unknown magnet materials derived from readily available elements.
The success of the ML model in predicting the Curie temperature of these material candidates represents a significant stride toward creating a high-throughput method for designing future permanent magnets.
In the words of Prashant Singh, “We are writing physics-informed machine learning for a sustainable future.” This innovative approach could revolutionize how we discover and develop materials for sustainable and advanced technological applications.
Read the original article on Science Daily.
Read more: New Algorithm Aces University Math Course Questions.
