Exploring and anticipating supramolecular synthons: from fundamental science to practical applications

Abstract

Four different methods; molecular electrostatic potentials (MEPs), hydrogen-bond energies (HBE), hydrogen-bond propensities (HBP) and hydrogen-bond coordination (HBC) were used for mapping out the structural landscape of twelve pyrazole and twelve thiazole based molecules. In seven out of eight crystal structures obtained in pyrazoles, a combination of HBE and HBP predicted the experimentally observed synthons correctly. In all eight crystal structures obtained in thiazoles, the synthons were predicted correctly using all four methods. A series of co-crystallizations between twelve pyrazole with twenty carboxylic acids (240 experiments), and twelve thiazole with twenty carboxylic acids (240 experiments) were carried out to build an experimental library that could be used for evaluating the ability of electrostatics, energies, propensities and molecular complementarity methods to rationalize the observed intermolecular interactions. The results suggested that a combination of electrostatics and molecular complementarity are essential for identifying the predominant molecular recognition events in the pyrazole based study, and methods such as MEPs, HBE, and HBP all predicted the observed synthons in co-crystals of the thiazole-based molecules. In order to examine competition between hydrogen and halogen bonds, and to synthesize ternary co-crystals, four thiazole based molecules were co-crystallized with 15 hydrogen-bond donors and one halogen bond donor resulting in new co-crystals in 44 out of 60 experiments, and the crystal structures of two ternary co-crystals were obtained. A series of eight unactivated and activated amide functionalized molecules were synthesized to establish a supramolecular halogen-bond hierarchy. The positive electrostatic potential on the halogen atoms was enhanced through an sp-hybridized carbon and electron-withdrawing fluoro group(s) next to amide group. Tetraflourinated and iodoethynyl based molecules were identified as the most effective halogen-bond donors and were therefore least successful for co-crystal synthesis. In order to predict crystallizability of 83 drug-like molecules a molecule, logistic regression approach was employed using molecular descriptors such as molecular weight, rotatable bond, surface area, heteroatom, melting temperature, glass transition temperature, and molecular shape/volume. Four different models were developed, and the success rate was above 85% (using experimental DSC data for the crystallization classification). Finally, the solid-form landscape of urea was explored using full interaction maps (FIMs), and data from the CSD to develop optimum protocols for synthesizing co-crystals of this compound. As a result, 49 of 60 attempted reactions produced new co-crystals. Moreover, the goal of reducing solubility and lowering hygroscopicity of the parent compound was achieved, which, in turn, offers new opportunities for a slow-release fertilizer with limited hygroscopicity thereby reducing many current problems of transport, handling, and storage of urea.