ATTAINING 1% ACCURACY IN ABSOLUTE PHASE COMPOSITION LEVELS BY RIETVELD ANALYSIS, B.H. O'Connor and D.Y. Li, pp. 305-312

The paper addresses requirements for achieving Rietveld phase composition analysis (PCA) accuracy in the vicinity of 1% by weight with particular reference to scale factor precision. It is proposed from a statistical analysis that the target for scale factor estimated standard deviation (esd) should be in the vicinity of 0.2% relative, with additional targets being assumed for the ZMV factors, for the x-ray attenuation correction factors (when external standards are used) and for the phase composition of the calibration standard (either internal or external). Attaining such precision in the scale factors is very challenging in terms of diffraction data collection requirements and refmement strategies, particularly for the former. Results are presented for Rietveld refmements using difl?action data sets acquired with an o-Al203 powder, and also with mixtures made up with this material and a ZnO powder. INTRODUCTION The use of Rietveld refmement with diffraction data for PCA is now very popular with much of the work reported in the literature being so-called standardless PCA which is, in effect, relative KY. Absolute PCA is required in the ceramics development research of the authors and their colleagues, and clearly in many other applications. The authors have become increasingly concerned that the emergence of ‘standardless’ Rietveld PCA is inhibiting the development of reliable, truly quantitative Rietveld PCA procedures. It is of particular concern, based on our experience in XRF elemental analysis, that constraining phase composition totals to 100% may mask systematic effects in the results and also hide the presence of (i) undetected phases, (ii) crystalline components below the detection limit and (iii) amorphous material. Use of standardless analysis in effect ‘forces’ the sum of the reported concentrations to 100% and, in so doing, tends to mask systematic errors which bias the scale factors, eg due to microabsorption, preferred orientation and 20-dependent intensity errors. lpefs l-21 The question of accuracy in PCA is addressed only obliquely in Rietveld papers on refinement strategies references 3-6. This paper considers the scale factor precision requirements for achieving PCA accuracy in the vicinity of 1% by weight according to a 30 criterion. This accuracy target is very difficult to achieve as routine Rietveld analysis of Bragg-Brentano data typically delivers precision values of approximately fib relative for [o(si)/si] for major phases. The paper follows that in reference 7 on refinement strategies for Rietveld PCA Copyright(c)JCPDS-International Centre for Diffraction Data 2000,Advances in X-ray Analysis,Vol.43 305