Using low temperature calorimetry and moisture fixation method to study the pore structure of cement based materials

Pore structure is one of the most important characteristics of cement based materials. This PhD study focused on using a low (micro-)temperature calorimetry (LTC) and a moisture fixation method to study the pore structure of cement based materials, especially at the nanometric level. Special attention was devoted to investigating important factors influencing the analysis of measured LTC data and using LTC to characterize the pore structure of cement based materials. Besides, the moisture fixation method was selected as a comparison and complementary method to the LTC. Attempts have been made to correlate the porosity characteristics determined by the two different techniques.
The experimental investigations were mainly conducted on a mono-sized model material MCM-41 and samples of two types of well hydrated cement pastes made from CEM I 32.5 R and CEM III/B 42.5 N, respectively. The primary consideration of including the model material in this investigation was to validate the applicability of the chosen methods in the context of pore size determination. In addition, data from literature were used.
LTC investigations conducted in this PhD study include the ice content determination from measured data, the impact of sample saturation on the detected porosity, the effect of frost damage on the pore size distribution determination by LTC, the effect of preconditioning the cement paste samples on the freezing and melting behavior of the pore solution, the impact of sample crushing and a preliminary exploration of the influence of ions present in cement pore solution during measurements on the porosity determination by LTC.
Important results of the LTC investigations can be summarized as follows: (1) Two factors are of paramount importance in the analysis of measured LTC data, i.e., the baseline determination of the measured heat flow and the values adopted for the heat of fusion for the water/ice confined in pores. (2) Sample saturation has an impact on the porosity determination by LTC. The freezing/melting point of the water/ice in a non-fully saturated pore system is lower compared with the case when the system is fully saturated. Using non-fully saturated samples, some pores will be misinterpreted in the sense that they are estimated to be too small. (3) Frost damage may take place in the used cylinder samples of the studied cement pastes. It was concluded that frost damage potentially changes the pore connectivity while it has limited effect on changing the interior size distribution of the meso-pores. (4) Hardened cement paste samples were preconditioned in a big amount or in a small amount of saturated limewater for a relatively long time. The results indicate that the two preconditioning cases have very limited influence on the freezing and melting behaviors of the pore solution in the studied cement paste samples. (5) Two types of samples, i.e., in the form of powders and cylinders, of two types of cement pastes (CEM I and CEM III) were used to study the impact of sample crushing on the porosity determination by LTC. For the CEM I samples, the results indicate that sample crushing only changes the pore connectivity while it has limited effect on the pore (interior) size distribution and the total pore volume. For the CEM III samples, however, sample crushing not only changes the pore connectivity but also the pore (interior) size distribution and the total pore volume. (6) Thermodynamic modeling using the program PHREEQC was performed on relevant cement paste samples. The results suggest that for the studied paste samples, the temperature depression caused by the ions present in the pore solution only affects the determination of the pore size distribution by LTC to a limited extent.
Moisture fixation behaviors of the materials were studied by the “dynamic (water) vapor sorption” (DVS) measurements at the hygroscopic range and pressure plate measurements at the over-hygroscopic range.
Key points of the sorption studies can be summarized as follows: (1) A resaturation study was used to investigate the possible pore structure changes at low relative humidities (RHs). The results showed that the drying at low RHs does not cause any microstructure changes or, alternatively, that the absorption process must have been able to fully restore the pore structure of hardened cement paste samples, indicating the possible microstructure changes are reversible. (2) A study on sorption isotherms suggests that the differences between the sorption isotherms measured at different temperatures are mainly a consequence of the temperature dependent properties of water. The pronounced impact of temperature on desorption isotherms of cement based materials as reported in some references in literature were not found in this study. (3) The calculated specific surface areas were very much dependent on the type of equations used for describing multilayer adsorption, indicating that the calculated specific surface area may not represent the “real” geometrical surface area. (4) The important factors influencing the analyzed pore size distribution (PSD) results using sorption data were reviewed. For the studied hardened cement pastes CEM I and CEM III, three characteristic peaks with the radii of about 1.4, 1.8 and 3.0 nm were found in the calculated PSD curves from the desorption isotherms. The peak at 1.4 nm was missing in the PSD curves calculated from the absorption isotherm. The network theory, suggesting desorption is controlled by the pore entry sizes while absorption is controlled by the interior pore sizes, tends to be of great relevance in explaining the results.
The PSD results determined by the LTC and the water vapor sorption method were compared and only a certain degree of agreement has been found. Due to the uncertainties and many unsolved factors involved in the data analysis, it is concluded that probably none of these two methods can deliver the “true” (actual) pore size distribution information. That is, before the uncertainties and unsolved factors being solved, both methods are only semi-quantitative and meaningful for comparison purposes.