Phase Composition Analysis of the NIST Reference Clinkers by Optical Microscopy and X-Ray Powder Diffraction.
Phase Composition Analysis of the NIST Reference
Clinkers by Optical Microscopy and X-Ray Powder
Stutzman, P. E.; Leigh, S.
NIST TN 1441; 49 p. September 2002.
Available from: Superintendent of Documents, U.S.
Government Printing Office, Mail Stop SSOP, Washington,
cements; clinker; composition; optical microscopy;
reference material; Rietveld method; x-ray powder
Certification of the phase compositions of the three
NIST Reference Clinkers will be based upon more than one
independent method. The current reference values were
established using an optical microscope examination,
with additional optical microscope data taken from an
ASTM C 1356 round robin. The present X-ray powder
diffraction (XRD) study provides a second, independent
estimate of the phase abundance. Reitveld refinement of
the powder diffraction data allowed calculation of a set
of best-fit reference patterns and their scale factors.
Because of significant contrast in the linear absorption
coefficients of ferrite and periclase relative to the
estimated mean matrix linear absorption coefficient, the
scale factors were adjusted for microabsorption effects.
The XRD data generally agree with the optical data with
the exception of aluminate. This disagreement may
reflect the difficulty in resolving this fine-sized
phase using the optical microscope. The XRD data show
greater precision than replicate measurements by
microscopy. Measurements from different sources,
laboratories, instruments, and methods can exhibit
significant between-method variability, as well as
distinct within-method variances. The data sets were
analyzed using both unweighted and weighted schemes to
establish optimal consensus values and to provide
meaningful consensus uncertainties. While the consensus
mean values of individual phase abundance do not vary
significantly across methods of combining data sources,
the associated uncertainty values do. The
Mandel-Paule-Vangel-Rukhin maximum likelihood method of
combining the data sets is favored as this method
produces a weighted mean whose weighting scheme does not
necessarily skew the consensus value in the direction of
the large number of XRD values, and it takes between- as
well as within-method variation explicitly into account.