Realising the combined potential of solid-state NMR and structural databases

Databases of crystal structures are essential tools for researchers working in the solid state. Initially established as repositories of experimentally determined structures, the large data sets contained within databases, such as the Cambridge Structural Database (CSD), have become the subject of research in their own right through the development of "data mining". The usefulness of such databases is, however, highly dependent on the quality of the data they contain. In the vast majority of cases the structures were obtained via X-ray diffraction (XRD). While XRD is the pre-eminent tool for establishing the three-dimensional structure of crystalline materials, there are areas where XRD studies struggle and some art is required on the part of the crystallographer to establish a correct structure. For instance, hydrogen atoms scatter X-rays very weakly, and fragments with the same (OH vs F) or very similar (Si vs Al) numbers of electrons are very hard to distinguish. In addition, any disruption of the regular ordering of a crystal creates major challenges for structure solution; diffraction is not the natural tool for understanding such "disorder". Historically XRD experts have used measures such as "R factors" to assess how well a proposed structure fits to the experimental data, but ideally independent experimental evidence would be used to verify crystal structures.

We and other research groups have shown in recent years that solid-state nuclear magnetic resonance (SS-NMR) can now be used very effectively to distinguish between different possible crystalline structures. Developments in quantum chemistry (mostly notably through Density Functional Theory) allow NMR spectra to be calculated with excellent precision. Since the NMR spectrum is sensitive to very small changes in the local structure - deviations of the order of a picometre (10^-12 m) will change the spectrum measurably - even small imperfections in a crystal structure solution can be identified. Moreover different types of "disorder" e.g. due to the motion of atoms or irregular atomic positioning, have clear and distinct effects on the NMR spectrum.

This proposal seeks to develop systematic approaches to the validation of crystal structures via solid-state NMR and computational chemistry. We will establish which NMR experiments are required in order to distinguish crystal structure solutions and also to "validate" a structural solution. This will involve the creation of "NMR confidence parameters" which will measure the extent to which a structure is compatible with the NMR data available, and the effectiveness of these parameters will be verified against more traditional diffraction-based tools. By taking a systematic approach, we will be able to show how NMR can be used to resolve the different types of structural ambiguity and show the value of NMR as a complement to conventional diffraction-based studies.

The principal non-academic beneficiaries are industries for whom characterisation of solid forms is important. Most notably this includes the pharmaceutical industry, who need to understand the solid forms (and relationships between them) of drug substances, but also the agro-chemicals industry where the solid form also has an impact on performance. These industries generally rely on X-ray diffraction and databases of known structures (mostly notably the Cambridge Structural Database) for information on crystalline structures. Currently there is no independent means of verifying diffraction-derived structures and so developing techniques to validate structures will directly benefit these industries. Most pharmaceutical companies have access to solid-state NMR via in-house instrumentation or external services, and so will benefit immediately from the development of robust protocols for structure validation. The Durham research groups (PH and JAKH) work with most of the major pharmaceutical companies operating in the UK (GSK, AstraZeneca, Sanofi-Aventis)

Other direct beneficiaries of the project will be the curators of structural databases as they clearly have a long-term interest in demonstrating the integrity of data contained. The UK-based Cambridge Crystallographic Data Centre (CCDC) will be directly involved in the project and will be closely the monitoring this potentially disruptive methodology.

One route to achieving impact is via the recently formed Collaborative Computing Project in NMR Crystallography (CCP-NC, www.ccpnc.ac.uk) which has as one of its objectives to build bridges between different communities. Two of the investigators (PH and JRY) are members of the CCP-NC working group and actively involved in dissemination activities. Developments in this and other projects will be featured in the quarterly newsletters of CCP-NC. The CCP-NC will be organising the 4th SMARTER meeting (Structure elucidation by coMbining mAgnetic Resonance, compuTation modEling and diffRaction), an international conference on NMR crystallography, in 2014, with Durham as the likely hosts.

Reaching out to those currently unaware of the developing role of NMR in solid-state characterisation is naturally more challenging. Here the active involvement of the Cambridge Crystallographic Data Centre and other X-ray diffraction experts is key. Local diffraction experts (such as JAKH) will be involved in the day-to-day research, assisting with the practical problems of surveying structural databases and helping us understand the strengths and weaknesses of diffraction-based assessments of structure quality. This interaction is critical if the NMR-based metrics we will develop are to be taken seriously by the wider crystallographic community. The involvement of diffraction experts in the project Steering Group will ensure that we are asking the right questions and can present our results in a way and in venues that our target audience can appreciate.

In the longer term, raising the awareness of the usefulness of solid-state NMR for structural characterisation will benefit the suppliers of commercial NMR services, such as the Durham Solid-State NMR Research Service, and instrument manufacturers. The two principal manufacturers of solid-state NMR equipment, Bruker and Agilent (formerly Varian) have significant sites in the UK (Coventry and Yarnton respectively), and the superconducting magnets for the Agilent's machines are also manufactured in the UK (Yarnton).