MS10-P05 Co-crystal or salt? A cautionary tale when inferring proton disorder solely from X-ray and computational data George Sackman (Department of Chemistry, University of Oxford, Oxford, United Kingdom) Richard Cooper (Department of Chemistry, University of Oxford, Oxford, United Kingdom) Alison Edwards (Australian Centre for Neutron Scattering, Sydney, Australia)email: pharmaceutical industry is constantly driven to enhance the physical properties of medicines and active pharmaceutical ingredients (APIs). These improvements can often be achieved by identifying novel solid forms of an API, such as co-crystals and salts.[1]

Co-crystals and salts are distinguished by the degree of proton transfer between an acid and a base in the structure. In a co-crystal, proton transfer does not occur and the proton remains on the acid. While in a salt, proton transfer to the base is complete. However, many materials can exist in an intermediate state along a continuum between co-crystals and salts, where proton transfer is partially complete.[1]

During a recent blind test of crystal structure prediction methods, one of the target structures was a 1:1 co-crystal of 3,5-dinitrobenzoic acid and Tröger's base (3,5-DNBA:TB). It was predicted, using density functional theory calculations, that 3,5-DNBA:TB may exhibit partial proton transfer of the carboxylic acid proton to an amine nitrogen atom of the base.[2]

Following this blind test, X-ray analysis of 3,5-DNBA:TB was performed. This provided supporting evidence for partial proton transfer, by indicating that the proton may be disordered over two sites (with occupancies of 0.58(3) on the carboxylic acid oxygen and 0.42(3) on the amine nitrogen).[3]

In order to study this potential hydrogen disorder, an accurate determination of the hydrogen atom positions was required. Therefore, neutron single crystal diffraction was employed, with Laue diffraction patterns collected at 150K on the KOALA Laue diffractometer at the Australian Centre for Neutron Scattering.

However, the neutron data shows that the proton is not disordered over two sites. The Slant Fourier map shows clear single site occupancy of the hydrogen closer to the carboxyl group (O-H 1.147(8)Å). This demonstrates the need for neutron data to confirm or deny whether a hydrogen is disordered or not.

Further neutron data was collected for a related material where the X-ray data indicated proton disorder: 1:1 3,5-bis(trifluoromethyl)benzoic acid and Tröger's base (3,5-BTFBA:TB). The neutron data again provided evidence that this proton was in fact not disordered over two sites. Therefore, we recommend that one should be cautious when inferring that a structure contains disordered protons, solely based on X-ray and computational data.

[1] Childs, S. L. et al. (2007). Mol. Pharmaceutics, 4(3), 323-338.

[2] Reilly, A. M. et al. (2016). Acta Cryst. B72, 439-459.

[3] Wheeler, K. A. & Breen, M. E. (2016b). CSD Communication: CCDC 1447528, doi:10.5517/cc1kl8gw.
Keywords: Neutron, Disorder, Hydrogen