Chemical shrinkage, the volume reduction associated with the reaction between cement and water in hydrating cement paste, was assessed using the method described by Geiker [11], which is similar to that recently employed by Tazawa et al. [12]. While the latter authors concluded that chemical shrinkage is directly proportional to degree of hydration, they further stated that chemical shrinkage is not directly related to autogenous or self-desiccation shrinkage. Conversely, Hua et al. [13,14] have recently established a model which successfully directly relates autogenous shrinkage to the capillary pressures induced by chemical shrinkage. Thus, measurements of chemical shrinkage may serve a dual purpose, quantification of hydration rates and indication of system susceptibility to self-desiccation shrinkage.
To assess chemical shrinkage, a known mass of cement paste (typically 10 g) was placed in the bottom of a small glass jar, with a diameter of 2.5 cm and a height of about 6 cm. After covering the cement paste with about 1 ml of water, the remainder of the jar was filled with an hydraulic oil. The jar was then sealed with a rubber stopper encasing a pipette graduated in 0.01 ml increments. The jar was then placed in a constant temperature water bath ( T = 25 ºC ) and the oil level monitored to the nearest 0.0025 ml over time. A control sample using only cement powder and oil (no water) was used to correct for minor room temperature fluctuations. By normalizing the change in volume by the mass of cement in the sample, the chemical shrinkage per gram of initial cement (ml/g cement) could be determined. In all cases, two specimens were run for each w/c ratio and cement, with the average result being reported.