An adaptive genetic algorithm for crystal structure prediction.

Area of Science:

• Materials Science
• Computational Chemistry
• Solid State Physics

Background:

• Discovering novel materials with desired properties requires efficient structure prediction methods.
• Traditional methods often face limitations in exploring large and complex chemical spaces.
• First-principles calculations, like DFT, offer high accuracy but are computationally expensive for exhaustive searches.

Purpose of the Study:

• To develop a highly efficient genetic algorithm (GA) for accelerated materials structure prediction.
• To combine the speed of classical potentials with the accuracy of DFT for enhanced computational materials discovery.
• To enable the investigation of larger and more complex material systems using first-principles calculations.

Main Methods:

• An adaptive and iterative approach integrating classical potentials for rapid structure exploration.
• Density functional theory (DFT) calculations for accurate energy and structure refinement.
• A genetic algorithm (GA) framework to guide the search process efficiently.

Main Results:

• Achieved several orders of magnitude increase in efficiency for DFT-based GA structural searches.
• Successfully identified complex binary and ternary intermetallic compound structures (36 and 54 atoms/cell).
• Discovered a novel multi-TPa Mg-silicate phase with up to 56 atoms per unit cell.

Conclusions:

• The developed hybrid GA significantly enhances the efficiency of first-principles materials discovery.
• This approach allows for the study of significantly larger and more complex material systems.
• The discovered Mg-silicate phase has potential implications for understanding exoplanetary mantle composition.