Quantitative polytype-composition analyses of SiC using X-ray diffraction: a critical comparison between the polymorphic and the Rietveld methods

Abstract

Abstract X-ray powder diffraction (XRD), in conjunction with an analysis routine, is the method of choice for performing quantitative phase-composition measurements in multi-phase mixtures. However, accurate determination of phase compositions in SiC-based ceramics, which can contain a multitude of SiC polytypes with overlapping Bragg reflections, is a difficult task. In order to determine the best analysis method for the quantitative phase-composition measurement in SiC-based ceramics, a critical comparison between two commonly used analysis methods — polymorphic method (Ruska) and whole pattern method (Rietveld) — was performed. Owing to the difficulty in obtaining high-purity standard mixtures of SiC polytypes experimentally, we have simulated a set of XRD patterns corresponding to different SiC polytype mixtures of wide ranging compositions as standards. Within the polymorpic method, the following three different approaches were used: (i) raw XRD peak heights were obtained and corrected for mean background level; (ii) a non-linear Levenberg-Marquardt least-squares fit was used to obtain peak heights; and (iii) using the same fitting procedure as in (ii) integrated intensities were obtained. We demonstrate that the Rietveld method yields the most accurate phase-composition measurements, with mean and maximum errors of 0.5 wt.% and 0.9 wt.%, respectively. In the polymorphic method, we show that a fitting procedure is essential for the improvement in the accuracy of the analysis. Furthermore, we find that the use of integrated intensities, obtained from the fit, for quantitative XRD analyses gives results that are less accurate compared with when corrected raw peak heights are used.