Search for Stable and Low-Energy Ce-Co-Cu Ternary Compounds Using Machine Learning

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Materials Science

Background:

• Cerium-based intermetallics are promising candidates for advanced permanent magnets.
• Developing new magnetic materials requires efficient exploration of complex compositional spaces.

Purpose of the Study:

• To investigate the Ce-Co-Cu ternary system for novel permanent magnet materials.
• To accelerate the discovery of stable and high-performance magnetic compounds using computational methods.

Main Methods:

• Utilized a machine learning framework, specifically a crystal graph convolutional neural network (CGCNN).
• Integrated CGCNN with first-principles calculations for efficient material screening.
• Predicted thermodynamic stability and dynamic properties of candidate compounds.

Main Results:

• Identified five stable Ce-Co-Cu compounds with formation energies below the convex hull.
• Discovered hundreds of low-energy, potentially metastable Ce-Co-Cu ternary compounds.
• Predicted high magnetization in two cobalt-rich compounds: Ce4Co33Cu and Ce4Co31Cu3.

Conclusions:

• The ML-guided approach significantly accelerates the discovery of new magnetic materials.
• The Ce-Co-Cu system offers promising candidates for next-generation permanent magnets.
• Computational methods are crucial for navigating complex material landscapes and identifying high-performance compounds.