Both microscopical and QXRD methods have been used to estimate phase abundance of clinker and are considered direct methods of analysis. Error in microscopy due to incorrect identification of the constituent phases is considered to exceed the error due to counting statistics [15] as fineness of the constituents may preclude their identification. For the finer-grained clinkers this may be especially problematic for the interstitial phases periclase, free lime, and the alkali sulfates.
X-ray powder diffraction analyses are not limited by crystal size and so are suitable for both clinker and cements. The accuracy of XRD, given careful experimental procedure, is about 2-5 % absolute for alite and belite and 1-2 % for aluminate and ferrite [15]. Generally, one should see good agreement between XRD and microscopy for the silicates and the total interstitial phases.
Limitations of the accuracy derive from the suitability of the reference standards (the structure models), and the ability to identify and control correlations between variables. For example, because of a high degree of overlap of the strong diffraction peaks from cubic and orthorhombic aluminate, their scale (intensity) factors are correlated. This can be controlled through determination of the peak profile shape characteristics using a salicylic acid / methanol extraction residue enriched in aluminate. In fitting the whole-clinker pattern, these profile shape variables served as initial values and, for both aluminate forms, were constrained to be the same. Box plots of the initial data set showed an anti-correlation between belite and ferrite. In this case, the peak shape of ferrite would broaden at the expense of the intensity factor of belite. This was controlled by determining their respective peak profile shape parameters using a potassium hydroxide / sucrose and nitric acid / methanol extraction residues and holding these fixed throughout the fitting of the whole-clinker patterns.