Statistical validation of standardless and standard-based analysis by X-ray fluorescence spectrometry in iron ores characterisation

An X-ray fluorescence (XRF) study comparing the performances of standardless or semiquantitative analysis in relation to the traditional standard-based or quantitative analysis technique is presented. This comparison using precision lines and simultaneous confidence intervals provides reliable information about the precision and trueness that each technique can achieve. This study enables the comparison of the results obtained with two different analytical approaches for iron ore samples in terms of precision and accuracy of results, sample preparation and availability of standards. A new technique for assessing the accuracy of an analytical method using linear regression is presented, wherein the results of both analyses are regressed against certified reference materials. The statistical test performed is based on the joint confidence interval for the slope and the intercept of the regression line calculated considering the uncertainties in both axes or in both analytical methods. The slope, intercept and variances associated with the regression coefficients are calculated with bivariate least-squares regression. Some chosen standard-samples were analysed for statistical validation of the methods. These samples had sufficient replicates to perform the calculations and later statistical comparison between the accuracy (precision and trueness) of each analytical method and evaluation of its precision lines.