The analysis of powder diffraction data

A comparison has been carried out between the results of analyses of several sets of neutron powder diffraction data using three different methods: the Rietveld method [Rietveld (1967). Acta Cryst. 22, 151-152; (1969) J. Appl. Cryst. 2, 65-71], a modification of the Rietveld method to include off-diagonal terms in the weight matrix [Clarke & Rollett (1982). Acta Cryst. Submitted] and the SCRAP method, which involves the estimation of observed Bragg intensities [Cooper, Rouse & Sakata (1981). Z. Kristallogr. 157, 101-117]. Two simulations have also been carried out to demonstrate the way in which the results can differ in more extreme cases. This study has confirmed that the values of the estimated standard deviations given by the Rietveld method are not reliable and that, of the methods considered, only the SCRAP method will in general give reliable values for the estimated standard deviations of the structural parameters.