The determination of crystal structures from powder diffraction data is a field that is rapidly expanding due to a range of computational and experimental developments. A major driving force of this expansion is the continuing development of direct space methods of structure solution. This work will show the development of the program POSSUM as a suite of direct space structure determination methods and its subsequent successful application to a number of molecular organic and inorganic materials whose crystal structures were previously unknown. Direct space methods utilise global optimisation algorithms to locate the crystal structure. This work describes the successful application of the differential evolution optimisation algorithm to structure determination from powder diffraction. Differential evolution is shown to be a robust and efficient optimisation technique with the limited number of control parameters associated with the method ensuring that optimisation of the searching is easily achieved. Investigation into improving the computer performance of the method also focused on reduction of the time taken to evaluate agreement between experimental and calculated patterns through the application of the discrete wavelet transform. This effectively reduces the number of points in the powder pattern, yet retains the same level of information as the original data set and is shown to illustrate good discrimination between solutions generated in a direct space structure solution.

This PhD project had two parts. The first part concerned the development of a number of specific instrumentation and software approach of general use for X-ray and neutron crystallography. Central to this was a proof of principle study in which novel 3D computational methods were developed to allow an accurate model to be built of cryo-frozen crystals and supporting loop and liquid. It is demonstrated that this type of approach may be used to optimise crystallographic absorption corrections that offer particular advantages for situations where data redundancy is low.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This thesis concerns the development of software techniques for polycrystalline fibre diffraction analysis and the application of the techniques to the determination of the structures of target molecules. Fibre diffraction analysis is one of a powerful method for determining the structure of polycrystalline fibres. However, one of the main shortcomings of this method is the paucity of intensity data; the observed intensities need to be modelled using suitable methods. This particular problem is further complicated by the fact that no direct Fourier synthesis methods can be used because of the cylindrical averaging inherent in fibre diffraction patterns. An attractive strategy is to use software programs for performing semi-automated data extraction from fibre diffraction patterns using refined method of data stripping and modelling. Even though there are some software packages available to do part of this, process the best one, the CCP13 suite, required a great deal of improvement and development. In this thesis, the design, development and applicability of such a software package, FibreFix, are described. Using several practical examples, we show how this software package has aided in solving practical structure-determination problems. Covered in the thesis are: (i) Development,of several techniques as part of the CCP13 software for performing data stripping and structure analysis; (ii) Design and development of a new integrated CCP13 software package, FibreFix, incorporating these techniques; (iii) Evaluation of FibreFix to show that the analysis and modelling processes in data reduction for fibre diffraction can be simplified substantially using the FibreFix software. To show this two different, but related, case studies are described: crystalline natural rubber and E-DNA. Our results show that the structures modelled by the software package are well above our threshold for satisfactory modelling. This claim is supported by very satisfactory X-ray amplitude fitness and in reduction in the R-factor of the models. We have analysed the structure of oriented natural rubber, obtaining a very good fit with the observed X-ray amplitudes and a best R factor of 0.18. We have also analysed different possible models for E-DNA. The R-factors indicate that 172 D-DNA-like models are better than 152 models. We have a found a satisfactory structure for E-DNA, which gives a good fit with the X-ray amplitudes and a best R-factor 0f 28%. This project has substantially, improved the CCP13 fibre analysis software, now all incorporated within FibreFix, which can be used for data stripping of many forms of fibre diffraction pattern. The structures of rubber and E-DNA have also been used as targets to utilize and enhance the capability of the LALS (linked atom least-squares) modelling and refinement program.