Ranking intermolecular interactions offers insights into why molecules crystallise

A researcher in the UK has shed new light on which interactions are important in the packing of crystal structures.1

Robin Taylor from the Cambridge Crystallographic Data Centre (CCDC) analysed the line-of-sight interactions between the most common elements found in organic crystal structures. He found that the probability of an interaction taking place boils down to the exposed surface area an atom presents. With 137,560 appropriate crystals available from the Cambridge Structural Database, and the inclusion of several statistical considerations, Taylor was able to keep potential sources of uncertainty to a minimum.

Throughout the chemical crystallography literature, a host of interactions have been deemed significant in stabilising crystal structures. Back in 2009, Jack Dunitz, of ETH Zurich, Switzerland, and Angelo Gavezzotti, of the University of Milan, Italy, urged caution in assigning significance to interactions as important contributors to intermolecular stabilisation.2 Taylor, inspired by Dunitz and Gavezzotti’s work, set out to determine whether an interaction occurs more often than expected by chance, and therefore whether it does in fact play an important role within the crystal.

Interactions may be longer than the sum of Van der Waals radii yet still play a significant stabilising role

He says that whilst the underlying physics of molecular interactions is well understood, this is a valid way of ranking interactions, and will hopefully lead to more objectivity in crystal analysis. ‘This method enables a statistically rigorous way of ranking the importance of interactions in controlling the crystal packing, rather than chemical intuition.’

The highest-ranked interactions were hydrogen and halogen bonds, which are well known for their importance and strength. The most striking result was the high ranking of H[C]–F and H[C]–Cl interactions. These are known to be energetically weak, but were surprisingly common relative to their random expectations. This could be due to an entropic effect. As H[C]–F and H[C]–Cl have weak directional preferences, it is just easier to pack a molecule with less rigid directional constraints.

Santiago Alvarez, a theoretical bonding expert at the University of Barcelona, Spain, and a member of the European Academy of Sciences, says that if this process can be automated, then it should be possible to routinely analyse primary interactions. ‘This will facilitate a better understanding of molecular packing principles and new crystal design strategies’.

As with any statistical study, the main limitation is data. Results become more relevant by using more specific groups of atoms depending on their electronic environments. However the more sub-categories, and the more distinct the classification, the less data there is available.

Robin hopes to extend this work to include group–group interactions, and to try to understand why stable crystals which contain a higher proportion than expected of less favourable interactions have formed stable crystals.