A method for quantitative pyrite abundance in mine rock piles by powder X-ray diffraction and Rietveld refinement

The abundance of pyrite and other sulfide minerals in mine rock piles is a potentially significant if not a determinative factor in terms of the geochemical and geomechanical evolution of the dumps as oxidation produces acid solutions that drive hydrolysis reactions. A technique is presented here that supports the quantitative determination of pyrite abundance in mine rock dumps by heavy liquid mineral separation to concentrate pyrite for powder X-ray diffraction and then Rietveld method refinement of the diffraction data on a large number of samples using commonly available laboratory equipment. In order to improve and constrain the accuracy of XRD results, binary (pyrite-quartz) and 6-part mineral mixtures (pyrite and rock-forming andesite minerals) spanning a wide range of pyrite concentrations were prepared gravimetrically and run as standards. These standards were then used to minimize errors in pyrite abundance data by constraining key input parameters in the Rietveld refinement. A new polynomial relationship was derived between diffracting crystallite size and the Brindley microabsorption correction input size. This method provides a means to determine uncertainties in pyrite abundance, whereas conventional Rietveld refinement techniques done without the use of standards yield only statistical measures of the least-squares fit, rather than absolute uncertainties in mineral constituent weight percentages. The technique was applied to a number of mine rock pile samples and the uncertainty in the results determined by applying the relationship derived from the 6-part gravimetric standards to the results of the Brindley corrected Rietveld refinements. Uncertainties determined by this method are found to be on the order of ±10% for samples with pyrite content greater than ∼10 wt% and ±30% for samples with pyrite content less than 10 wt%. In order to evaluate the technique's improvement upon traditional visual mineral abundance estimation the quantitative results are compared to manual volumetric estimates