Paul Verwer

Crystal structure prediction of small organic molecules: a second blind test.

The first collaborative workshop on crystal structure prediction (CSP1999) has been followed by a second workshop (CSP2001) held at the Cambridge Crystallographic Data Centre. The 17 participants were given only the chemical diagram for three organic molecules and were invited to test their prediction programs within a range of named common space groups. Several different computer programs were used, using the methodology wherein a molecular model is used to construct theoretical crystal structures in given space groups, and prediction is usually based on the minimum calculated lattice energy. A maximum of three predictions were allowed per molecule. The results showed two correct predictions for the first molecule, four for the second molecule and none for the third molecule (which had torsional flexibility). The correct structure was often present in the sorted low-energy lists from the participants but at a ranking position greater than three. The use of non-indexed powder diffraction data was investigated in a secondary test, after completion of the ab initio submissions. Although no one method can be said to be completely reliable, this workshop gives an objective measure of the success and failure of current methodologies.

Crystal structure predictions for acetic acid

Possible crystal structures of acetic acid were generated, considering eight space groups and assuming one molecule in the asymmetric unit. Our grid‐search method was compared with a Monte Carlo approach as implemented in the Biosym/MSI Polymorph Predictor. This revealed no sampling deficiencies. A large number of possible crystal structures were found (∼100 within only 5 kJ/mol), including the experimental structure. Energy minimizations were done with a united‐atoms force field (GROMOS), an all‐atoms force field (AMBER), and a potential that describes the electrostatic interactions with distributed multipoles (DMA). In all cases, the experimental structure had a low lattice energy. The number of hypothetical crystal structures was reduced considerably by removing space‐group symmetry constraints, or by a primitive molecular dynamics shake‐up. Nevertheless, sufficient structures of equal or lower energy compared with the experimental structure remained to suggest that other factors need to be considered for genuine structure prediction. © 1998 John Wiley & Sons, Inc. J Comput Chem 19: 459–474, 1998

Method for the computational comparison of crystal structures

A new method for assessing the similarity of crystal structures is described. A similarity measure is important in classification and clustering problems in which the crystal structures are the source of information. Classification is particularly important for the understanding of properties of crystals, while clustering can be used as a data reduction step in polymorph prediction. The method described uses a radial distribution function that combines atomic coordinates with partial atomic charges. The descriptor is validated using experimental data from a classification study of clathrate structures of cephalosporins and data from a polymorph prediction run. In both cases, excellent results were obtained.

Computational Approaches to Crystal Structure and Polymorph Prediction

Detailed knowledge of crystal structure at the atomic level is a prerequisite for rational control of crystallisation processes, polymorphism and solid state properties. Recent advances in computer hardware and software have enabled the prediction of crystal structure and polymorphism for simple organic compounds. Here, we present an overview of crystal structure prediction, and we discuss a number of pitfalls frequently encountered. A concise review of various approaches is given, followed by a description of recent improvements in polymorph prediction methodology developed at Molecular Simulations.

On the irrelevance of electrostatics for the crystal structures and polymorphism of long even n-alkanes

It is known that the experimental triclinic crystal structures of even n‐alkanes are not well reproduced upon energy minimization with current force fields. The inclusion of electrostatics does not solve this, and, moreover, some charge schemes show unphysical features such as positively charged carbon atoms or charge alternation. The effect of the electrostatics on the energies of the crystal structures of the even n‐alkanes, and thereby on their polymorphism, has never been established. A new charge scheme is introduced that yields physically sensible charges without constraints. It will also be shown, however, that electrostatics are relevant neither for the structures of the crystals, nor for their energies.

On the influence of thermal motion on the crystal structures and polymorphism of even n-alkanes

Discrepancies between the crystal structures of short n-alkanes as obtained from experiment and as obtained from molecular mechanics tended to worsen at longer chain lengths. The same holds for the relative stabilities of the two experimentally observed polymorphs. In this paper it is argued that the discrepancies are caused by thermal effects, and that the triclinic polymorph is the most stable polymorph for all chain lengths at 0 K. A phase transition is predicted but has yet to be found experimentally. Current force fields cannot reproduce the experimental observations without explicit introduction of temperature by means of molecular dynamics.

Crystal structure prediction of organic pigments

The present work investigates the feasibility of predicting crystal structures of industrially important pigments when only low quality XRPD patterns are available. Despite the commercial significance of organic pigments only few crystal structures of these materials have been reported. Pigments are practically insoluble in most solvents and, therefore, it is difficult to grow a single crystal of good enough quality for the structure determination. In industry, pigments are produced by precipitation reactions leading to very fine powders. The crystallites are often so small that they cause a considerable line broadening in X-ray powder diffraction. Therefore, a powder pattern of limited quality is usually the only information available. The structures of organic pigments PV19, PV23 and PR202 were successfully predicted using the Polymorph Predictor of Cerius in combination with XRPD patterns of limited quality. After generation and energyminimization of the possible structures, their powder patterns were compared to the experimental ones. On this basis, structures of pigments were chosen from the list of all the structures. Rietveld refinement was used to validate the right choice of the structures. m29.o04