Professor Richard I. Cooper, Department of Chemistry, University of Oxford.
Crystallisation of Molecular Materials: Prediction, Measurement, Analysis.
Crystallization of single- and multi-component crystals is key for controlling physicochemical properties, including bioavailability, dissolution rate and stability of active pharmaceutical ingredients (APIs) . There has been great progress in ab initio prediction of molecular crystal structure in recent times, but the interaction of many factors that cause a molecule to be crystallizable at all remain poorly understood. Statistical models of crystallization propensity built from the CSD  and databases of materials using a machine learning method can predict crystallization propensity with surprising accuracy, using descriptors of molecules calculated from the 2D molecular diagram[3,4]. These models have been validated computationally and experimentally, and feature extraction allows identification of constituent molecular properties which have most influence on the crystallizability, thereby providing guidance on small changes which could be made to induce (or avoid) crystallisable molecules.
A similar approach has been applied to the formation of co-crystal materials, using in-house results of a co-crystal screen of 34 API-like molecules with 20 common co-formers. A machine learning model produces a ranked list of co-formers which increases successful co-crystallization rates in experiments with previously unseen APIs . External data has been used to validate the model and demonstrates that it performs well for APIs in the same chemical neighbourhood as the training data. Both of these approaches use descriptors of the molecular components derived solely from 2D atomic and connectivity information, avoiding structure prediction, lattice energy and even any specific intermolecular interactions.
 Aakeröy, C. B.; Forbes, S.; Desper, J. J. Am. Chem. Soc. 131, 17048–17049 (2009).
 Groom, C. R.; Bruno, I. J.; Lightfoot M. P.; Ward, S. C. Acta Cryst. B 72, 171-179 (2016).
 Wicker, J. G. P.; Cooper, R. I. CrystEngComm 17, 1927-1934 (2015).
 Wicker, J. G. P.; Cooper, R. I. J. Chem. Inf. Model., 56, 2347–2352 (2016).
 Wicker, J. G. P.; Crowley, L. M.; Robshaw, O., Little, E.J.; Stokes, S. P.; Cooper R. I.; Lawrence, S. E. CrystEngComm 19, 5336 – 5340 (2017).
ABOUT THE PRESENTER
Richard Cooper received his Master’s degree in Chemistry from the University of Oxford in 1996 and D.Phil degree in 2000. His doctoral thesis was undertaken in the Chemical Crystallography Laboratory in Oxford and included studies of infrequently observed crystal packing motifs and methods of structure refinement. Throughout his post-doctoral research and short stints in industry he has maintained an interest in the use of crystallographic results for designing new materials and predicting their properties. He is an author and current maintainer of the crystallographic refinement software, CRYSTALS, which is used throughout the world for structure analysis and in particular when investigating unusual diffraction data. His recent research has focussed on analysis of the hundreds of thousands of experimental crystal structures that are available in the literature via the CSD to try to predict properties of materials (e.g. propensity to crystallize) and the likelihood of co-crystal formation for pairs of molecules.
Tea/coffee will be available at 10h45.
For further information, please contact: Michael.firstname.lastname@example.org