Quantitative mineralogical analysis using the Rietveld full-pattern fitting method

Abstract

Ansrnlcr Quantitative phase analysis of a number of multicomponent standard and natural mineral mixtures has been done using an adaptation of the Rietveld method. Binary mixtures (most 50:50 by weight) of corundum with quartz, hematite, ilmenite, magnetite, biotite, analcime, mordenite, or clinoptilolite were analyzed,using digital powder X-ray diffraction (XRD) data. In addition, a suite of standard mixtures of hematite and corundum, a natural feldspar mixture, the G-l standard granite, two natural bauxite samples, and a mixture of biogenic carbonate minerals were also analyzed. Quantitative information was extracted from refined individual scale factors and unit-cell voiumes (derived from refined unit-cell parameters), obtained with a Rietveld refinement program modified to analyze up to ten phases. The quantitative results for standard mixtures were within 2.50/o (absolute) of the true values, with the exception of the hematite, ilmenite, and magnetite mixtures. Results for the latter mixtures using CuKa data were severely afected by microabsorption, but analysis of the hematite mixtures using FeKa radiation gave results with absolute errors <2010. Results for the G-l granite and the natural feldspar agreed well with optically determined modes, and the method facilitated separation of the significant overlaps in the pattern of the carbonate minerals. Quantitative mineralogical analysis by the Rietveld method has several significant advantages over conventional methods of quantitative analysis. The method uses all intensity data in a pattern rather than a few of the most intense reflections, partially compensating for preferred orientation and extinction. In addition, standard dala are calculated for each phase during analysis, overcoming the troublesome requirement of obtaining standards representative of the materials in an unknown. It is also possible to gain a wealth of information from each sample in addition to amounts of phases. Because some of the most troublesome systematic errors, including sample displacement and zero-point shift, can be refined, the method yields unit-cell parameters of an accuracy comparable with that obtained when using an internal d-value standard. The method should find a wide application in geology, including in modal analysis and compositional determinations of individual mineral components using unit-cell parameter systematics.