Details of the X-ray environmental chamber are provided in Refs.  and . The X-ray system consists of an X-ray source, a detector, and a positioning table to move the source and detector relative to the specimen being evaluated. The detector uses an NaI crystal and measures the photon count for each of 256 discrete energy channels. The entire system is computer controlled, so that the user may set up a grid of specimen points to be evaluated at periodic intervals.
For this study, three different cements were used: 1) an ASTM Type I/II ordinary portland cement, issued as Cement 133 in June 1999 by the Cement and Concrete Reference Laboratory at NIST , 2) a coarse low C3 A content cement ground to a fineness of 254 m 2/kg (median particle diameter of about 25 µm) , and 3) an ultra-fine cement ground to a fineness of 654 m2/kg (median particle diameter of 5 µm). Cement 133 has a potential Bogue composition of 58.6 % C3 S, 14.8 % C2 S, 10.6 % C3A, and 7.5 % C4AF based on oxide analysis, with a Blaine fineness of about 350 m2 /kg . The low C3 A cement has a Bogue composition of 59.0 % C3S, 25.9 % C 2S, 0.6 % C3 A, and 14.2 % C4AF, as determined by quantitative microscopy, with hemihydrate added at a mass fraction of 0.05. For the ultra-fine cement, the Bogue composition (via quantitative microscopy) is 73.5 % C3S, 16.5 % C2S, 7.1 % C 3A, and 2.5 % C 4AF, once again with a 5.0 % by mass hemihydrate addition. Small (50 g to 100 g) samples of cement paste and mortars were prepared in glass beakers and mixed by hand using a spatula for two to three minutes. In general, cement pastes and mortars of w/c=0.3 and 0.45 were prepared.
The fresh cement pastes were placed in either small inverted Lego2 blocks (a common children's toy building block) or larger parallelepiped cuvettes, which were either left open or sealed with a cap. The Lego blocks and plexiglass cuvettes were chosen as sample holders due to: 1) their low absorption of X-rays , 2) their inherent stackability (allows repeatable placement of the block within the X-ray chamber), 3) the ease with which they can be sealed by adding a cap, and 4) the ease with which they can be filled with a level volume of the viscous cement paste. The cuvettes have the additional advantage of being transparent, so that the drying behavior may be directly inferred from changes in the "brightness" of the sample. The basic experimental setup, illustrated for the composite (layered) cement paste specimens to be described below in the Results and Discussion section, is shown in Fig. 1.
Each block was labelled and weighed (to ± 0.01 g) before the cement paste was added. The masses of the cement paste-filled blocks were determined initially and periodically throughout the exposure period. The blocks were located sequentially on a holder (an inverted Lego base element) placed at a fixed location within the X-ray chamber. After specific periods of exposure, the X-ray system was used to scan vertically (in 0.2 mm increments) a distance of either 8 mm (Legos) or 20 mm (cuvettes) along the central y-axis of each block. A five second count time was used at each location to improve the signal to noise ratio. Assuming a Poisson process, the relative standard uncertainty in the sum of the counts obtained for channels 50 to 150 should be on the order of 300 counts or 0.4 % (for a sum of 70,000 counts, a typical value for the cement paste specimens).
Examples of spectra measured for two of the fresh cement pastes are provided in Fig. 2. The small peak at channel 200 is from an internal Cobalt source used for system calibration. The denser w/c=0.3 cement paste is seen to absorb more of the X-rays as indicated by its lower count values at all channels up to 150. Based on the pattern observed in Fig. 2, the sum of counts between channels 50 and 150 was selected as the dependent variable to be analyzed as being representative of the density (and water content) of the cement paste. This value was normalized by dividing it by the ratio of the counts achieved in free air at each measuring time to the counts achieved in free air for the first measuring time (to account for inherent variability in the X-ray source). In the graphs which follow, these normalized counts have been divided by one thousand producing values generally between 70 and 140. In some cases, differential density (counts) profiles have been produced by subtracting the 3 h absorption values at each spatial location from all subsequent readings at the same location, to highlight the changes occuring after the initial setting/settling of the cement paste is complete.