Next: Freezing of water
Up: Main
Previous: Introduction
Cement paste was prepared by adding deionized water to ordinary Portland cement and mixing with a spatula for 1 minute. Samples of dimensions 30 mm x 15 mm x 2.5 mm were cast into plastic molds. Stainless steel electrodes of dimensions 7.5 mm x 7.5 mm were placed into the wet cement paste approximately 20 mm apart. All samples were stored in an airtight chamber above a pool of lime-saturated water to prevent drying.
The freezing apparatus consisted of an insulated thermoelectric stainless steel plate controlled by a variable power source. The cold chamber consisted of a stainless steel cylinder open at both ends and a stainless steel lid. A thermistor (10 kΩ ± 1% at 25 ºC) connected to a multimeter was used to monitor the temperature inside the chamber. Two 160 mm long 24 gauge wires were run through the top of the cold chamber and attached to the electrodes in the sample. The opposite ends were connected to the impedance analyzer via a two-electrode connecting jack.
Two impedance analyzers, a Hewlett-Packard Model 4192A and a Schlumberger Model 1260* [16], with frequency ranges of 13 MHz to 5 Hz and 32 MHz to 10 µHz, respectively, were used. Impedance data were collected within a frequency range of 10 MHz to 10 Hz using a personal computer interfaced to the impedance analyzer, and were analyzed using "Equivalent Circuit" [17]. The resistance and capacitance of each sample were determined by fitting the experimental impedance curves. Variations in the fitting procedure resulted in only about a 20% difference in the calculated dielectric constant. This amount of precision was considered to be acceptable for this experiment. Data for frequencies above 3 MHz were not used because of the known unreliability of impedance data in this frequency range [4].
High frequency measurements were corrected to account for the residual inductive and capacitive effects of the wires and the internal resistance of the impedance analyzer. Since these corrections for the frozen pastes resulted in a change of less than 1% for the resistance, and less than 5% for the capacitance, the corrections were not actually required to achieve an acceptable level of accuracy. More details of the measurement process can be found elsewhere [2,4,10,11].