Figure 6: Final processed 2-D SEM/X-ray images for a portland cement ground to
different fineness: top- 540.7 m2/kg, and
bottom- 197.0 m2/kg.
Each image is 256 µm by 200 µm. Colors correspond to the following
phases: red- C3S, aqua- C2S,
green- C3A,
orange- C4AF, pale
green- calcium sulfates, yellow- K2SO
4
, and white- free lime.
In Table I, the quantitative phase analysis results from the final segmented 2-D SEM/X-ray images are compared to one another and to the phase fractions estimated using optical microscopy point counting on the raw clinker. In general, the agreement between the different cement powder images and the raw clinker image is quite reasonable. The volume fractions of the C 4AF phase are slightly greater in the SEM images than in the clinker analysis, perhaps illustrating that an accurate determination of the ferrite phase via conventional optical microscopy is somewhat difficult, perhaps due to its small grain size. When comparing the results for the different finenesses, there are not any significant differences between the surface area phase fractions (relative to the corresponding volume fractions) to indicate that one particular phase is being preferentially exposed by the grinding. In general, the surface area fractions for the C3 A are higher than its corresponding volume fractions, and vice versa for the C3 S. This indicates that the cement particle surfaces are enriched in C3 A and deficient in C3S relative to the cement's bulk composition. This is at least partially due to the fact that the C3S tends to be located in the larger particles, which have a reduced surface area to volume ratio relative to the smaller particles often containing the C3 A phase.
One important application for the images and the quantitative phase analysis results presented in this paper is to provide input for cement hydration and microstructural development models [11,12]. Based on the quantitative phase fractions and the two-point correlation functions measured for the phases, a 3-D cement particle image can be reconstructed which matches the phase volume fractions, phase surface area fractions, and correlation structure of any cement of interest [8]. It is only after the starting cement powder is adequately characterized that accurate modelling of the hydration behavior and physical properties of cement-based materials is possible [13].
Cement powder | ||||||
Blaine fineness (m2/kg) | ||||||
Phase | Clinker | 654.0 | 540.7 | 370.0 | 279.0 | 197.0 |
Volume fractions | ||||||
C3S | 0.740 | 0.770 | 0.752 | 0.746 | 0.643 | 0.763 |
C2S | 0.163 | 0.142 | 0.134 | 0.130 | 0.212 | 0.089 |
C3A | 0.076 | 0.045 | 0.077 | 0.084 | 0.077 | 0.083 |
C4AF | 0.022 | 0.043 | 0.037 | 0.041 | 0.068 | 0.065 |
Surface area fractions | ||||||
C3S | 0.713 | 0.712 | 0.692 | 0.599 | 0.713 | |
C2S | 0.176 | 0.131 | 0.122 | 0.233 | 0.098 | |
C3A | 0.061 | 0.110 | 0.123 | 0.098 | 0.116 | |
C4AF | 0.050 | 0.048 | 0.064 | 0.070 | 0.073 |
Figure 7: Image analysis surface area to volume ratio for four major clinker
phases vs. measured Blaine fineness for a portland cement ground to five
different finenesses. Error bars indicate one standard deviation for cases
where two images were analyzed at a given fineness