The crystallographic characterization of materials has become a key step in understanding molecular behavior and predicting physical properties. The contributions of crystallography range from providing insights into life’s basic processes, as illustrated by the recent crystal structure determinations of the ribosome, to the prediction of optical and magnetic properties of new materials solely from crystal symmetry. In spite of the proven utility of this approach, many compounds fail to form large crystals suitable for conventional diffraction methods. This talk explores the development of NMR and modeling methods designed to provide crystal structures for these challenging materials. One area of focus is a highly promising non-diffraction alternative known as crystal structure prediction (CSP). The aim of CSP is the accurate computational prediction of solid-state structure solely from the knowledge of atomic connectivity. The promise of CSP arises from its complete independence from diffraction data, eliminating difficulties associated with diffraction (e.g. poor x-ray diffraction from lighter elements). Successful theoretical prediction of crystal structure has, however, remained an elusive goal, despite over three decades of intense research. Recent work in our laboratory demonstrates that when solid-state NMR data from powders are included in the CSP process, the correct structure can consistently be identified from among dozens of candidates. Presently, most CSP structures are of low quality and even when a correct structure is identified, a refinement step is needed to “clean up” the structure. Thus a second area of emphasis is the theoretical refinement of CSP structures using DFT methods that include lattice effects. Solid-state NMR chemical shift tensors (13C and 15N) are remarkably sensitive to these refinements and are used in our lab to monitor refinements. This NMR work suggests that such refinements have the potential to create ultra-high resolution structures from CSP data that rival or surpass the accuracy of single crystal neutron diffraction coordinates. Remarkably, we find such DFT refinements also improves data from most conventional diffraction methods (e.g. x-ray single crystal) and we have now refined structures from x-ray single crystal, neutron single crystal and x-ray powder data to create structures that appear to surpass the accuracy of single crystal neutron diffraction coordinates.
Read More