How evolutionary crystal structure prediction works--and why.

Area of Science:

• Materials Science
• Computational Chemistry
• Crystallography

Background:

• Predicting material properties relies on knowing their crystal structure.
• Crystal structure prediction (CSP) is crucial for accelerating materials discovery and studying matter under extreme conditions.
• Both efficient search of energy landscapes and accurate energy calculations are critical for CSP.

Purpose of the Study:

• To review Crystal Structure Prediction (CSP) methods, focusing on the hybrid evolutionary algorithm USPEX (Universal Structure Predictor: Evolutionary Xtallography).
• To explain the effectiveness of evolutionary techniques for CSP by analyzing energy landscapes.
• To demonstrate the power of evolutionary CSP through diverse applications.

Main Methods:

• Utilizes a hybrid evolutionary algorithm, USPEX, for efficient exploration of multidimensional energy landscapes.
• Analyzes the intrinsic dimensionality and overall shape of chemical energy landscapes.
• Applies various search strategies including simulated annealing, metadynamics, and data mining.

Main Results:

• USPEX demonstrates significantly greater computational efficiency compared to random search methods for CSP.
• Evolutionary CSP has led to the discovery of novel materials, including a transparent phase of sodium, ionic boron, superconducting calcium, superhard carbon, and polymeric nitrogen.
• The method successfully predicted complex hydride superconductors like LiH(n) and CaLi(2).

Conclusions:

• Evolutionary algorithms like USPEX provide a powerful and efficient approach to solving the crystal structure prediction problem.
• These methods enable the discovery of new materials and phenomena, particularly under high-pressure conditions.
• Extensions of USPEX are being developed for molecular crystals, variable composition systems, and targeted property optimization.