A symmetry-oriented crystal structure prediction method for crystals with rigid bodies

Area of Science:

• Materials Science
• Computational Chemistry
• Solid-State Chemistry

Background:

• Crystal structure prediction (CSP) is crucial for understanding material properties.
• Existing CSP methods may face challenges with compounds containing recurring molecular units or rigid bodies.

Purpose of the Study:

• To develop an efficient CSP method tailored for compounds with recurring molecules or rigid bodies.
• To validate the method's performance on metal chalcogenides Li3PS4 and Na6Ge2Se6.

Main Methods:

• Developed a novel CSP approach treating specific chemical groups as rigid bodies (e.g., PS4 as tetrahedral, Ge2Se6 as ethane-like dimer).
• Applied density functional theory (DFT) calculations for energy evaluations.
• Compared results with the USPEX package, a genetic algorithm-based CSP tool.

Main Results:

• Successfully predicted experimentally observed crystal structures for Li3PS4 and Na6Ge2Se6.
• Discovered several novel, low-energy phases, including a new stannite-type Li3PS4 structure.
• Identified a potentially more stable structure for Na6Ge2Se6 than the experimentally observed one.
• The developed method showed superior performance in predicting metastable structures compared to USPEX.

Conclusions:

• The new CSP method is efficient and effective for predicting structures of compounds with rigid molecular units.
• The discovery of novel phases highlights the potential of this method for materials discovery.
• The Python implementation is publicly available for broader research use.