Routine organic crystal structure determination from powder diffraction data alone is an important goal for both the chemical industry and chemists in general. Recent developments in global optimisation methods, coupled with innovative processing of powder diffraction data, have made this goal realisable in an increasing number of cases. The global optimisation methods rely upon one having an isolated molecular description of the compound under study. Once the unit cell and space group have been determined, trial crystal structures can be postulated and optimised using a technique such as simulated annealing to give the best agreement between the measured and calculated diffraction data.

With organic compounds, it is often the case that the chemist is interested mainly in the packing and conformation of the molecule under study. In these circumstances, it is normally sufficient to solve the crystal structure to limited (perhaps 1.5 - 2.0 Å) resolution. In contrast to direct methods, global optimisation methods have little problem coping with this relatively low resolution, as prior information in the form of known molecular connectivity has already been imposed. With this low resolution comes the possibility of rapid data collection and thus the possibility of high-throughput powder diffraction structure solution.

In a set of powder diffraction experiments performed over a period of 9 shifts on beamline BM16, data for structure solution and refinement were collected from a series of organic and organometallic compounds. Data were typically collected to ca. 1.1 Å using an optimised count time scheme in which only about 3 hours of data were collected per sample. Where appropriate, the samples were also cooled to around 130 K using a Cryostream in order to improve the diffracted intensities at higher angles. By processing each data set whilst the next sample was being mounted and collected, six crystal structures were solved during the duration of the experiment; specifically, the anti-ulcer drug famotidine (forms A and B), the anti-convulsant remacemide (and its nitrate and acetate salts), and a palladium organometallic catalyst (Figure 24). Famotidine is an excellent example of a high-value pharmaceutical compound that exhibits polymorphism; whilst the form A structure was known from a single crystal study, the crystal structure of Form B was unknown. None of the remacemide structures were known, the salt forms being typical of ones evaluated in the early stages of drug development by way of improving aqueous solubility and dissolution. The actual structure solutions were obtained using data to typically ca. 1.5 Å resolution, with subsequent constrained Rietveld refinements carried out on the full range of collected data. Taking due account of the count time scheme, the important conclusion is that each structure was solved from less than one hour of diffraction data. Five more samples were successfully indexed and their structures are currently being analysed.

By matching data collection and analysis strategies to the capabilities of the powder diffractometer on BM16, the expectations of what can be achieved in terms of crystal structure solution from powder diffraction data during a period of synchrotron beam time have been redefined. Whilst powder diffraction cannot, as yet, be considered to be as straightforward as single-crystal diffraction for structure solution purposes, it need no longer be considered a method of last resort.

Principal Publications and Authors
K. Shankland (a), W.I.F. David (a), T. Csoka (a), Z. Krist, 212, 550-552 (1997).
W.I.F. David, K. Shankland and N. Shankland (b), Chem. Commun., 931 (1998).

(a) ISIS Facility, Rutherford Appleton Laboratory (UK)
(b) Department of Pharmaceutical Sciences, University of Strathclyde (UK)