Geographic-style maps with a local novelty distance help navigate in the materials space

Area of Science:

• Materials Science
• Computational Chemistry
• Crystallography

Background:

• The rise of self-driving labs necessitates automated methods for identifying novel materials.
• Existing structural databases are vast, requiring efficient tools to prevent duplication of synthesized compounds.

Purpose of the Study:

• To develop and validate a computational approach for real-time novelty checking of periodic materials.
• To define a rigorous metric for material novelty based on structural similarity.

Main Methods:

• Defining material novelty as the minimum distance to the nearest neighbor in known structural databases.
• Utilizing ultra-fast structural invariants to rapidly identify nearest neighbors, even with transformations (e.g., unit cell changes, atom perturbations, element substitution).
• Demonstrating the method by checking for near-duplicates of newly synthesized materials within large inorganic crystal structure databases.

Main Results:

• The developed method can identify nearest neighbors within seconds on standard computing hardware.
• Near-duplicates were successfully identified for 43 materials synthesized by Berkeley's A-lab in the Inorganic Crystal Structure Database and Materials Project.
• The approach enables real-time novelty assessment crucial for automated materials synthesis.

Conclusions:

• A fast and accurate method for checking material novelty has been established.
• The proposed approach, using invariant descriptors, facilitates navigation of materials space for efficient discovery.
• This work supports the advancement of self-driving laboratories in their quest for novel materials.