Obtaining models suitable for Rietveld refinement: the EXPO2011 techniques

Rietveld refinement succeeds if the model is sufficiently close to the target structure. Severe distortions in the structural parameters or relatively high percentage of missed atoms do not allow the refinement to converge. Ab initio techniques like Direct or Patterson methods may rely only on the reflection intensities provided by full pattern decomposition algorithms. Owing to the relevant errors in the intensity estimates, the structural models provided by such phasing techniques are poor and require supplementary improvements before being submitted to Rietveld refinement. The situation is more favorable for non-ab initio techniques, when they can exploit the prior information on the full molecular geometry: their models can be easily submitted to Rietveld refinement, but often the accuracy of the refined model relies on the geometrical information rather than on the experiment. In this paper ab natio and non-ab initio phasing tools of EXPO2011, the updated and more powerful version of EXPO2009 [1], are described together with ancillary techniques for improving and completing the structural models.