Abstract
Uncertainty in ASTM C150-07 calculations of cement composition are assessed based upon the uncertainty in oxide determinations by different analytical methods. These methods include traditional chemical analyses (the “reference methods”), alternate chemical methods, X-ray fluorescence (XRF) using either a wavelength- or energy-dispersive detection system with either a fused glass or powdered specimen, atomic absorption, and inductively coupled plasma spectrometry. Within-laboratory and between-laboratory estimates for the CaO, SiO2, Al2O3, Fe2O3, and SO3 are determined for the different analytical methods using a paired sample approach used in cements proficiency testing. An analysis of variance test across these methods indicates that for CaO and SiO2, the methods do not return the same mean values; for Fe2O3, they do; and for Al2O3 and SO3, the results are mixed. XRF precision for CaO and SO3 is lower than the reference method for both glass and powder samples. Precision is lower for the powder compared to the glass, exception for Fe2O3. SiO2 XRF glass results exhibit similar precision to the reference methods, while the powder precisions are lower. Al2O3 precision for all X-ray methods and preparations exceed that of the reference method. All X-ray results are better than those from earlier studies indicating perhaps slight improvement in overall test performance within the industry. Box plots on Bogue-calculated values for each method indicate an overall agreement across methods, a consistency between the glass preparations for both wavelength-dispersive and energy-dispersive analyses, and a consistency between powder preparations for both XRF methods. An error propagation analysis is made to assess the effects of the uncertainty of the chemical analyses on the calculated phase compositions and to compare with the X-ray powder diffraction test method. The precision for the silicates and aluminates is similar, while ASTM C150-07 calculations for ferrite are more precise.
