In 1980, ASTM Committee C01.23 subcommittee on Compositional Analysis established a task group to develop a XRD method for phase analysis of clinker and hydraulic cements. Initial work on measurement of the interstitial phases of clinker (periclase, aluminates and ferrite) [7] using selective extractions, resulted in adoption of a standard test method in 1998 [8]. This method adopted the qualification approach after ASTM C 114 (chemical analysis), where the analyst qualifies by demonstrating acceptable estimates of a set of SRM cements based upon limits set in the standard. For the XRD method, the NIST SRM clinkers are used because they have certified phase composition and were initially produced to aid in the development and testing of methods for phase abundance analysis [9].
Earlier round robin testing applied a standardization using pure clinker phases and either peak area or whole-pattern measurement techniques. This approach limited the number of willing participants due to the difficulty and time requirements in standardization and peak area measurements. The Rietveld method approach overcomes a number of these difficulties; 1) selection of reference standards, 2) development of standardization curves, and 3) measurement of intensity from patterns where substantial peak overlap poses difficulty for single peak-measurement approaches. An earlier report on a clinker round-robin [6] demonstrated that a significant improvement in precision was seen and was attributed to the improved means for measuring pattern intensities. A new round-robin study was initiated for hydraulic cements using four mixtures of SRM clinker-calcium sulfate blends. The materials were ground to <10 µm and interblended with known amounts of laboratory-prepared calcium sulfate phases: gypsum, bassanite, and anhydrite, and for one sample, calcite. The round robin procedure required each specimen to be scanned three times, with sample re-packing for each scan. Each of 11 participants was requested to report the scan conditions, mass fractions without microabsorption corrections, and to submit raw data. Participants needed to perform qualitative phase identifications, which are required for selecting appropriate structure models for their Rietveld input files.
Although the standard does not specify the details of the analysis, scans that range from 20º to 77º two-theta (Cu Kα) encompass the interval of strong diffraction peak intensities. Lower angles may be problematic as, depending upon slits and specimen size, the irradiated area is likely to fall beyond the powder onto the sample holder. A suggested configuration based upon our system was provided but not required of the participants: 20º − 77º, 4 s count time, 1º divergent slit, 2.3º Soller slits on incident and diffracted beam sides, graphite monochromater, a 0.2 mm antiscatter and 0.6 mm detector slits.