A third blind test of crystal structure prediction.

Area of Science:

• Crystallography
• Computational Chemistry
• Materials Science

Background:

• The Cambridge Crystallographic Data Centre hosted the third collaborative crystal structure prediction (CSP2004) blind test, building on previous successful events (CSP1999, CSP2001).
• Eighteen research groups participated, employing diverse methods, primarily focused on global lattice energy minimization for crystal structure prediction.

Purpose of the Study:

• To assess current capabilities in crystal structure prediction through a collaborative blind test.
• To identify limitations in prediction methodologies, particularly concerning complex structures and flexible molecules.

Main Methods:

• Participants received molecular diagrams and submitted three predicted crystal structures for each molecule.
• Unlike previous tests, no restrictions were placed on space groups or the number of molecules per asymmetric unit (Z' = 2 allowed).
• A replacement molecule was provided when one test molecule's blind status was compromised.

Main Results:

• Success rates were lower than in prior tests, with only one successful prediction among the 'blind' molecules.
• The simplest rigid molecule's structure was not predicted, partly due to its crystallization with two molecules in the asymmetric unit.
• No successful predictions were made for the flexible molecule, mirroring outcomes from the 2001 blind test.

Conclusions:

• Improved energy models are crucial for accurately describing conformational and packing energies simultaneously.
• Enhanced search procedures are needed for crystals with multiple independent molecules and conformational flexibility.
• Accurate prediction of thermodynamically favored polymorphs requires addressing these limitations and understanding nucleation/growth processes.