Abstract
Bulk oxide determinations from two pairs of ASTM C150/C150M-12 portland cements are used to calculate precision and accuracy values for X-ray fluorescence (XRF) by both the fused glass bead and the pressed powder sample preparation methods. The first pair of cements are ordinary portland cement (OPC) with 45 participants, while the second pair contains ca. 3 % limestone, with 24 labs participating. Each lab provided results from six replicates analyzed in duplicate, covering fifteen analytes, CaO, SiO2, Al2O3, Fe2O3, SO3, MgO, Na2O, K2O, TiO2, P2O5, Mn2O3, SrO, ZnO, Cr2O3, and Cl, with the laboratories roughly split between the two different sample preparations. Chemical data using traditional chemical analyses (the Reference Methods per ASTM C150) from the Cement and Concrete Reference Laboratory (CCRL) proficiency test program were included for comparison to the XRF results. Precision measures for within- and between-laboratory performance are presented as 1σ and 95 % limits (ASTM d2s). Accuracy criteria are based upon a two-sided 95 % prediction interval for the mean of two separate test results, defining the range of values one might expect for each analyte relative to a certified value of a reference material. Comparing to current C114 limits, based upon classical wet chemical analysis, within-lab precision for CaO is larger, wheras SiO2precision is similar to existing limits. Al2O3 and Fe2O3 precisions are substantially better, whereas for the remaining analytes, the precision is generally similar to ASTM C114 limits. A comparison of the ASTM accuracy criteria shows that CaO accuracy is poorer than the current limit, SO3 is larger, SiO2 and Na2O are similar, while the remaining accuracy limits for XRF are generally better than the ASTM C114 criteria. The powder method shows a positive relative bias for CaO for the cements containing less than 5 % limestone, suggesting a need for more limestone-containing reference materials for powder method calibration for these cements.