When the specimens were crushed for the degree of hydration evaluation, small pieces (typically 2 mm to 3 mm in size) were removed and stored in ethanol, to replace the water remaining in the specimens and halt hydration. The pieces were subsequently placed in a low viscosity epoxy to replace the ethanol and cured in an oven at 60 ºC. After curing, the specimens were prepared for viewing at a magnification of 500X in the SEM as described previously.17 Four representative images, each with 1024 x 768 pixels, were acquired for the microstructures achieved after 92 d of curing, for both of the w/c and the three curing conditions investigated in this study. At this magnification, each pixel is approximately 0.5 µm by 0.5 µm, with an area of 0.25 µm2. Each image was analyzed to estimate the area fractions of four components of the hydrated cement paste microstructure: capillary porosity, calcium hydroxide (CH), calcium silicate hydrate gel (C-S-H) along with other hydration products, and unhydrated cement. Based on stereological principles, the two-dimensional area fractions measured in this manner should be equivalent to the three-dimensional volume fractions of the four phases.
To apply a consistent analysis technique to all of the images, the following procedure was employed. A greylevel histogram, such as that shown in Fig. 2, was obtained for each image. This histogram indicates the number of pixels in the image having each possible brightness value (between 0 and 255). For hydrated cement paste, capillary pores are darkest, C-S-H and other (aluminate) hydration products are dark grey, CH is light grey, and unhydrated cement is brightest. The greylevel histogram in Fig. 2 clearly contains several distinct peaks, corresponding to these components of the microstructure. Our goal is to segment (separate) the image into these four components, so that the area (volume) fractions can be determined by simple pixel counting. For separating out the unhydrated cement and the CH, threshold values were selected to be equal to the local minimum between the third and fourth peaks (about 180 for the histogram in Fig. 2) and between the second and third peaks (110 for the histogram in Fig. 2) in the greylevel histogram, respectively. To obtain a consistent separation between the C-S-H and the capillary porosity, an alternative procedure to using the local minimum was employed. For a given cement, the volumetric ratio of C-S-H and other hydration products to CH should approach a constant value at later ages regardless of curing conditions. Microstructural modeling using the CEMHYD3D hydration model 18, 20 has indicated that this value should be about 3.7 for CCRL cement 152. Given the expected image-to-image variation in this quantity for the finite size sample area, a threshold greylevel to separate the capillary porosity from the C-S-H and other hydration products was chosen such that the computed ratio of this quantity was always between 3.58 and 3.82 for the selected threshold.
Figure 2- Greylevel histogram for a microstructure image for CCRL cement 152 with w/c=0.35, cured under sealed/saturated conditions for 92 d.
The degree of hydration of the cement was then estimated according to the equation:7, 15
 | (1) |
where
| α | = | the computed degree of hydration of the cement; |
| Vcem(t) | = | the volume (area) fraction of unhydrated cement at time t; and |
| VRS | = | the initial volume fraction of readily soluble (non clinker) phases in the cement. |
For cement 152, VRS≈ 0.08, with 6 % calcium sulfates and the remaining 2 % corresponding to the alkali sulfates, free lime, etc. Vcem(0) was calculated based on the final w/c of the prepared pastes, assuming a density of 3200 kg/m3 for the powder. These SEM-determined degrees of hydration will be compared to those determined using loss on ignition analysis.