All cement pastes were prepared using Cement and Concrete Reference Laboratory (CCRL) cement proficiency sample 152 [16]. Specimens with w/c of 0.3 and 0.4 were prepared. Water and cement were mixed for several minutes in a temperature-controlled high-speed blender. After mixing, small disks of cement paste (about 5 mm to 6 mm thick and 47 mm in diameter) were cast in pre-weighed plastic Petri dishes and “consolidated” by tapping the dish numerous times on a laboratory counter top. The specimens were covered, weighed, and then placed in an environmental chamber at 20 oC where they were cured under either sealed or “saturated” (small volume of distilled water on top) conditions. At ages of 8 h and (1, 2, 3, 8, 15, and 28) d twin specimens of similar mass were selected and removed from their dishes. The specimens cured under sealed conditions were easily removed, due to their ongoing autogenous shrinkage, while for those cured under saturated conditions (which had actually expanded slightly during hydration due to water imbibition), it was usually necessary to break open the bottom Petri dish in order to remove the disk specimen. The mass of each specimen was measured upon removal. Each specimen was then inverted and placed back in a bottom Petri dish, but now being placed on top of a rubber o-ring spacer so that the top of the specimen was slightly higher than the top rim of the Petri dish. These specimens were then placed directly into the Hot Disk Thermal Constants Analyzer[1] for evaluation of their thermal conductivity.
Fig. 1 – Schematic of experimental configuration for measuring the thermal properties of a pair of (twin) cement paste specimens.
The basic experimental configuration for measuring the thermal properties of cement paste specimens using the thermal constants analyzer is provided in Figure 1. The thermal constants analyzer consists of a variety of transient plane source probes connected to a computerized control unit. The transient plane source measurement technique has been previously described in detail by Gustafsson and Log [17, 18] and theoretical considerations have been summarized recently by He [19]. For the current study, a 6.403 mm radius probe (Ni foil encased in Kapton) was selected. The probe was sandwiched horizontally between the cast sides of the twin hardened cement paste specimens in a holding frame and the entire setup was placed in a small closed chamber (to minimize the influence of air currents, etc.). After an equilibration time of at least 45 min in a laboratory nominally maintained at 23 oC, measurements were obtained with a power of 0.3 W applied for a measurement time of 10 s. The measured response of the probe/sensor was analyzed using the built-in software to determine the thermal conductivity of the specimens, after providing the heat capacity value that was determined in a separate experiment (described below). The analyzer samples 200 points during the 10 s measurement time and points 75 to 200 were used in the quantitative analysis. Fresh cement pastes were also analyzed separately using a 6.631 mm radius Ni/Kapton probe (0.3 W for 10 s) that was placed vertically into a 35 mm inner diameter plastic cup holding about 50 g of fresh cement paste (height of about 25 mm). In this case, both the thermal conductivity and the heat capacity of the fresh cement pastes were estimated in a single probe measurement. According to the manufacturer, the thermal conductivity measurements made in this way are reproducible within ± 2 %.
For the separate heat capacity experiments on the hydrated cement pastes, after their thermal conductivity measurements were completed, a small piece of one of the disks of known mass (typically 1.5 g to 3.0 g) was broken off and placed in the Hot Disk heat capacity unit, consisting of a special probe attached to the base of a gold pan/lid. For the measurements, the gold pan with its lid is surrounded on all sides by polystyrene insulation, in an attempt to minimize energy loss. First, a reference measurement is made with an empty pan, followed by the measurement with the specimen placed in the pan. For these heat capacity measurements, a power of 0.1 W was applied for a measurement time of 160 s. In this case, points 100 to 200 (of the total 200 sampled in the 160 s) were used in the quantitative analysis. Knowing the mass of the specimen, its heat capacity in units of J/(g·K) can be easily determined. For the analysis employed in this paper, any heat generation due to the ongoing cement hydration reactions during the 160 s of measurement time was ignored, as even for the earliest 8 h specimen, the maximum (heat) power generation due to hydration was estimated to be 0.005 W. The heat capacity measurements of the hydrating cement paste were facilitated by placing the (flat) cast surface of the specimen in the bottom of the pan to produce a high quality (thermal) contact between pan and sample. Additionally, the thickness of the specimens was such that the top of the specimen did not contact the bottom surface of the gold lid, as recommended in the Hot Disk User’s Manual. According to the manufacturer, the heat capacity measurements made in this way are reproducible within ± 2 %.
After breaking one of the disks to obtain a specimen for the heat capacity measurement, the remainder of the (broken) disk was crushed to a fine powder for measurement of its degree of hydration via loss-on-ignition (LOI) analysis. For this purpose, the non-evaporable water content (wn) of each sample was determined as the mass loss between 105 oC and 1000 oC divided by the mass of the ignited sample, corrected for the LOI of the unhydrated cement powder, determined in a separate LOI measurement. Previously, the expanded uncertainty in the calculated wn has been estimated to be 0.001 g/g cement, assuming a coverage factor of 2 [20]. The values of wn were then converted to estimated degrees of hydration based on the phase composition of the cement and published coefficients for the non-evaporable water contents of the various cement clinker phases [21]. Based on a propagation of error analysis, the estimated uncertainty in the calculated degree of hydration is 0.004.