Microabsorption Corrections

Microabsorption results in biased phase fraction estimates. Where weakly-absorbing phases exhibit greater intensities than expected, while strongly-absorbing phases exhibit lower intensities. Calculation of the linear attenuation coefficients for the cement phases (Table 1) and for a mixture, RM 8486, show that ferrite, periclase, and free lime (when present) may be expected to have the greatest estimate errors unless compensated for. This effect is not problematic in studies utilizing standardization mixtures, as the error is inherent in the standardization curve; however, in Rietveld analyses this effect may be significant.

Fine-grinding reduces microabsorption effects, but trials in our laboratory using binary mixtures of ferrite and periclase show it may still be problematic. Use of the Brindley absorption correction changes the scale factors based upon the differences between the phase and mixture linear attenuation coefficients, with an adjustment made for particle size [13]. Using the actual median particle size of 2µm resulted in an over correction of the phase fractions of the known mixture. An example showing true values vs raw data and Brindly-corrected values based upon a 1µm median particle size is provided in Figure 4. As expected, ferrite and periclase values better reflect the true values based upon the laboratory-prepared mixture. The low orthorhombic aluminate value may partly result from impurities in the laboratory material.

Figure 5: True vs calculated mass fraction values for a laboratory-prepared mixture of clinker phases showing both raw data (diamonds) and data corrected for microabsorption (circles) using the Brindley microabsorption correction and a median particle size of 1 µm.