A generalized method for refining and selecting random crystal structures using graph theory.

Area of Science:

• Materials Science
• Crystallography
• Computational Chemistry

Background:

• Predicting unknown crystal structures is crucial for materials discovery.
• Current methods often require extensive computational resources or prior knowledge.
• Efficient generation of diverse and stable initial structures is a key challenge.

Purpose of the Study:

• To develop a general, minimally-informed method for refining and selecting random crystal structures.
• To improve the efficiency and success rate of crystal structure prediction algorithms.

Main Methods:

• A novel approach using quotient graphs derived from near-neighbor analysis.
• Refinement of initial random structures guided by graph-based topological information.
• Validation across nine diverse chemical systems.

Main Results:

• The method successfully generated a large number of low-energy crystal structures.
• Demonstrated effectiveness in refining random structures across various systems.
• The approach requires minimal prior information for structure generation.

Conclusions:

• The developed method offers a robust way to generate high-quality initial structures for crystal structure prediction.
• Integration into existing algorithms can significantly expedite the discovery of ground-state crystal structures.
• This technique enhances the efficiency of exploring the materials design space.