Robustness of the quantitative phase analysis of X-ray diffraction data by the Rietveld method

Abstract

The quality of X-ray powder diffraction data and the number and type of refinable parameters has been examined with respect to their effect on quantitative phase analysis (QPA) by the Rietveld method using data collected from two samples from the QPA round robin [Madsen et al. J. Appl. Cryst. (2001), 34, 409--26]. For specimens where the diffracted intensity is split between all phases approximately equally accurate results could be obtained with a maximum observed intensity in the range of 1000--200000 counts. The best refinement model was one that did not refine atomic displacement parameters, but did allow other parameters to refine. For specimens where there existed minor or trace phases, this intensity range changed to 5000--1000000 counts. The refinement model with the most accurate results was one that refined a minimum of parameters, especially for the minor/trace phases. Given that all phases were quite crystalline, step sizes for both types of specimen could range between 0.01--0.04 deg 2Th, and still yield acceptable results. Data should be collected over a 2Th range that captures the lowest angle peak, and continues at least until (i) there is a constant increase in cumulative intensity with angle, (ii) a point where peaks no longer appear, or (iii) the upper 2Th limit of the goniometer. The wide range of these values show that QPA by the Rietveld method is quite robust with regards to data quality. As these are ideal specimens, these values indicate a best-case scenario for the collection of diffraction data for QPA by the Rietveld method, but does show that the analysis can be quite forgiving of lower quality data.