Automated Method For Crystal Structure Refinement In Powder Diffraction Based On Parameters Sorting Space

In recent years, crystal structure refinement in powder diffraction has been studied more and more widely. Rietveld refinement method, which is a classical method in this field, used nonlinear least squares method to make full spectrum fitting for the experimental and calculated intensity. It got a very brilliant achievements in the field. With the development of computer science, some software based on Rietveld refinement method have been developed, and it brought convenient for crystal structure refinement. However, when we use the existing refinement software, there exist some outstanding issues including a great number of manual interventions(manual refinement), time-consuming, low efficiency and low accuracy because of that the researchers need to develop a refinement strategy in advance and operate software manually to refine. The refinement strategy means that formulating of the refinement order of more than 100 parameters and settings of convergence degree.On the basis of the existing Rietveld refinement method and Fullprof software, this paper proposes an automated method for crystal structure refinement in powder diffraction based on parameters sorting space to solve the above issues. Firstly, this paper introduces the related theory of Rietveld refinement method and the basic working mechanism of Fullprof software. Then, that the refinement order of crystal parameters(such as the crystal cell parameters, the atomic coordinates, etc.) is the core of refinement has been found. So, this paper has defined the "crystal parameter space" i.e. nine groups of parameters of different meaning and "crystal parameters sorting space" i.e. full permutation of nine groups of parameters. It provides the crystallography theoretical basis for the automated method proposed by this paper.The automated method for crystal structure refinement proposed by this paper mainly includes the automated refinement for a refinement order between crystal parameter groups and optimal Rwp search algorithm. Among them, The former is based on Fullprof black-box model and "crystal parameter space", and the latter is implemented by "crystal parameters sorting space " in combination with the former. Finally, an automated auxiliary system has been developed by Python to implement the application of the automated method to experimental data. The application results of the automated refinement for Y2O3, LaMnO3 and PbSO4 show that the automated method is effective with no refinement strategy and manual operation. Meanwhile, it can save time and reduce the error rate caused by a lot of repeated manual refinement experiments, and improve the refinement efficiency.