AI-accelerated discovery of altermagnetic materials

Area of Science:

• Condensed Matter Physics
• Materials Science
• Computational Materials Science

Background:

• Altermagnetism is a distinct magnetic phase with unique properties, but its study is limited by the scarcity of known materials.
• Discovering new altermagnetic materials is crucial for understanding altermagnetism and developing next-generation information technologies.

Purpose of the Study:

• To develop an automated approach for discovering novel altermagnetic materials.
• To expand the library of known altermagnetic materials with diverse electronic properties.

Main Methods:

• An AI search engine utilizing a graph neural network was developed to learn material crystal structure features.
• A classifier was fine-tuned with limited positive samples to predict altermagnetism probability.
• First-principles electronic structure calculations were used for confirmation.

Main Results:

• 50 new altermagnetic materials were discovered, including metals, semiconductors, and insulators.
• The discovered materials exhibit diverse electronic structures, enabling properties like the anomalous Hall and Kerr effects, and topological properties.
• Four novel i-wave altermagnetic materials were identified for the first time.

Conclusions:

• The AI search engine significantly outperforms human experts in discovering altermagnetic materials.
• The newly discovered materials offer a wide range of properties for potential applications in advanced technologies.
• This AI-driven approach accelerates the discovery of materials with targeted properties.