281180 Computational Crystal Structure Prediction of Organic Molecules

Monday, October 29, 2012: 4:35 PM
Crawford West (Westin )
Erik E. Santiso1, Andrei V. Kazantsev1, Manolis Vasileiadis1, Claire S. Adjiman2 and Constantinos C. Pantelides1, (1)Department of Chemical Engineering, Imperial College London,Center for Process Systems Engineering, London, United Kingdom, (2)Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London, United Kingdom

Crystallization is an extremely important unit operation the food, chemical, and pharmaceutical industries. The experimental determination of all possible crystal forms of a given material is, however, a very challenging problem due to the large experimental parameter space that needs to be explored. Predicting crystal structures in silico, however, is also challenging: for many materials there are multiple crystal configurations very close in energy to each other, and there are no general empirical potentials available that can predict their energies with the required accuracy. This means that accurate ab initio calculations are needed to predict crystal structures

In this work, we present our most recent work in a systematic methodology to identify crystal structures for small to intermediate size organic molecules. Our methodology involves an initial global search for candidate polymorphs with the CrystalPredictor1 algorithm, followed by a more accurate minimization of the lowest-energy structures with the CrystalOptimizer2 algorithm. Both algorithms use quantum mechanical calculations augmented with an empirical dispersion correction to estimate the energies of different crystalline conformations. In CrystalOptimizer, we have found that using Local Approximate Models (LAMs) is an effective way to reduce the computational cost associated with the ab initio calculations. In this work, we show how this methodology can be extended to the global search in CrystalPredictor to produce a set of candidate structures that is improved in terms of both the geometries and the energy rankings. We illustrate this methodology with several examples.

[1] P.G. Karamertzanis, C.C. Pantelides, "Ab Initio Crystal Structure Prediction - I. Rigid Molecules", Journal of Computational Chemistry, 2005, 26, 304-324.
[2] A.V. Kazantsev, P.G. Karamertzanis, C.S. Adjiman, C.C. Pantelides, "Efficient Handling of Molecular Flexibility in Lattice Energy Minimization of Organic Crystals", Journal of Chemical Theory and Computation, 2011, 7, 1998-2016.

Extended Abstract: File Not Uploaded