Comparison of profile and integrated-intensity methods in powder refinement

Abstract

If a powder pattern consists of well resolved peaks, and if it is adequately described by a crystallographic model, as determined by statistical tests, the estimates of the parameters and their standard deviations will not be significantly different, in a statistical sense, whether the method of refinement uses peak profiles or integrated intensities. If the model does not fit adequately, the standard deviations cannot be calculated by statistical methods. If peaks overlap, there are correlations between the intensities of the individual peaks that are handled automatically in the profile method, but must be included explicitly in the integrated-intensity method. A background function should be included in the model, and its parameters refined. The background parameters have correlations with other parameters that become important if the pattern is complex, and therefore poorly resolved.