The application of transmission electron microscopy to the study of the early stages of hydration C3S and the microstructure of very mature C3 S paste is reported. In the former case the effect of water-cement ratio appears to be very important. In the latter, a detailed evaluation of the microstructures of inner C-S-H, outer C-S-H and CH is possible and the relationship of these to models of cement hydration is discussed.

The hydration of a tricalcium silicate paste has been examined from very early stages, i.e. during the induction period, up to two and half years using both scanning and transmission electron microscopy as well as solid state nuclear magnetic resonance. This latter technique can be used to reveal the degree of hydration of the sample and the nature of the silicate units which make up the C-S-H gel. Solid state NMR shows that during the induction period a small amount of material containing hydrated monomeric silicate units is formed, although this is difficult to identify using SEM. However during the acceleratory period SEM shows the development of the typical fibrillar C-S-H material. Backscattered electron imaging shows the cores of anhydrous material surrounded by the denser inner product.

Thermogravimetric analysis has been used to find the C/S ratio of this material, which is approximately 1.8, whilst solid state NMR shows that the hydrated silicate units are largely dimeric at this stage. As hydration proceeds, the fracture surfaces become increasingly dominated by large calcium hydroxide crystals, although some fibrillar material can be identified even after over two years. The C-S-H contains progressively more middle units in silicate chains at the expense of the end units. A sample of tricalcium silicate hydrated for 26 years has also been examined.

An understanding of the microstructure of cement paste in actual field concrete is considered to be the primary objective of cement microstructural studies. Cement paste in concrete is shown to consist of narrow ribbons of the order of 80 to 150 μm separating sand and coarse aggregate grains, and interrupted by frequent air voids. Individual cement grains form shells around themselves early in the hydration process. In subsequent hydration they either deposit dense layers of internal hydration product or empty out in a hollow-shell mode and precipitate hydration products in the empty spaces between nearby shells. Analogous shells are similarly formed around fly ash particles, which then react by progressive etching from within the shells. Interfacial zones around aggregates are pictured as having a complex characteristic structure including at least four different forms of calcium hydroxide particles with various preferred orientations. It is suggested that the conventional picture of the “aureole de transition” derived from successive x-ray diffraction measurements may have been misinterpreted.

The suspension hydration of tricalcium silicate (water-solid 500) has been studied under both floating and fixed pH (8 to 12) conditions. In addition to solution analysis, the microstructure and chemical composition of the hydrate products was also studied. The results are interpreted as suggesting the existence of two C-S-H phases in the system CaO-SiO2-H2O, the stabilization of each being dependent on the pH of the solution phase.

The hydration and microstructural development of cement are affected by the arrangement of anhydrous phases within the cement grains. Backscattered electron images and X-ray microanalysis can be used to characterize the microstructure of anhydrous cement. An image analyzer has been used to measure the overall proportions of silicate and interstitial phases in several cements and to measure the proportions of these phases that are exposed on the surface. These results are compared with the rates of heat evolution during the hydration of the cements.

The problem of defining bound water in a cement paste is discussed; a reasonable definition is one that includes interlayer water in C-S-H and AFm phases, structural water in ettringite, and adsorbed water, but not water in micropores or in larger pores. On this basis, structural considerations indicate a value of around 32% on the ignited weight for a fully hydrated paste. ‘Non-evaporable’ water, typically around 22% on the ignited weight at full hydration, cannot be identified with bound water, because dehydration to the state in which only non-evaporable water remains causes major loss of interlayer water and destruction of ettringite. In the interpretation of pore solution data, the definition of bound water, and the value assumed for this quantity, are important, because the ionic concentrations in the pore solution are greatly affected by the volume of free water available to dissolve them. If cement is partially replaced by low calcium fly ash, the quantity of bound water at any given age is substantially reduced. This effect contributes to the relatively low concentrations of alkali metal and hydroxyl ions that are observed in the pore solutions of many portland-fly ash cement pastes.

It was observed that standard ASTM C 109 sand (St. Peter Sandstone) from working quarries in the Ottawa area (northeastern Illinois) is nearly always sub-angular to rounded, frosted in appearance, and occasionally covered with localized, relatively large, terminated quartz overgrowths. At higher magnifications the frosted surface of the quartz sand was found to consist of a thin layer of micron-sized, coprecipitated quartz crystals and occasional kaolinite flakes. Upon reaction of the sand with either a saturated Ca(OH)2 solution or as part of a sand-cement paste, some of these features were preserved, whereas others were lost. Micrographs of sand cured in cement paste at 90°C are presented showing that the kaolinite flakes react and become “puffy” and rounded in appearance. Also, the smooth surfaces of the large, terminated quartz overgrowths are severely etched as a result of dissolution. Sand reacted at 38°C in saturated Ca(OH)2 solution demonstrated more subtle effects, perhaps obscured by precipitated CaCO3 and C-S-H on the surfaces of the quartz overgrowths. These results suggest that ASTM C 109 sand is not entirely inert, but exhibits a limited amount of temperature-dependent pozzolanic activity.

Pore structures of portland and blended cement mortars prepared with sea water were assessed by mercury intrusion porosimetry. Comparison is made with similar mortars made with fresh water. Mortars were made using cement containing 0%, 25% or 50% (by volume) of one of two Alberta fly ashes. They were tested after 90 days of continuous immersion in sea water at 5°C or 20° C. Results show that the mortars made with sea water generally contained a much higher volume of fine pores although the porosity was, in some cases, greater than that of mortars made with fresh water. Nevertheless, the volume of larger pores was lower in the mortars made with sea water. The pore structure of the various mortars is discussed in relation to potential durability.

In backscattered electron (bse) images of polished cement sections, anhydrous material, calcium hydroxide, other hydration products (mainly C-S-H) and porosity can be distinguished on the basis of their grey level in the image. Using an image analyzer connected directly to the SEM, it is possible to resolve these four components and so measure their relative proportions and distributions. The effects of magnification and the number of fields measured on the accuracy of bse image analysis are examined. The volume fractions of anhydrous material, porosity and calcium hydroxide derived from bse image analysis are compared with those obtained by other techniques and good correlation was found for the measurement of anhydrous material and porosity.

Two concretes were made in the Canadian Arctic at sub-freezing temperature using an unusually high amount of calcium chloride, about 9% of the cement mass. Calcium chloride in such a high dosage was mainly used as an antifreezing admixture, because it was impossible to heat the cement and the aggregates prior to the batching and cure the concrete properly after its pouring.

One of these two concretes was cast inside a mine, where the temperature is about −10°C all year round. After one year of service, the strength of this concrete was determined to be 39 MPa at room temperature. The second one was cast outside the mine at −30°C during winter and was exposed to temperatures above 0°C the following summer. After one winter of service the strength of this was 5 MPa.

These two concretes were examined under the SEM and show quite a dense microstructures containing chloroaluminates.

A model has been developed to describe a granule of unground clinker, to simulate the grinding process by the removal of particles according to a user-defined particle size distribution, and to characterize the exposed surface areas and volumes of each clinker phase present. The composition of the clinker granule and the distribution of phases within the granule are also user defined.

Preliminary models for predicting the performance of exposed concrete show that the porosity sub-model plays a pivotal role in linking properties with the developing gradients of cement paste microstructure that arise from drying and carbonation. The difficulties of porosity characterization are reviewed and some alternatives to traditional methods are considered. Property/porosity relationships are considered with particular reference to the volume, size-distribution and continuity of the larger pores. The factors influencing porosity under normal exposure conditions, such as cement hydration, drying and carbonation, are reviewed and the possibilities of modeling their effects are considered. Some comparisons between predicted and measured values of porosity are given. It is concluded that progress in modeling the performance of exposed concrete is likely to be linked with improvements in methods of pore structure characterization.

The pore structure (i.e. surface area, pore size distribution and pore volume) of well-hydrated portland cement pastes of water-cement ratios 0.4, 0.6, and 0.75 were investigated by the nitrogen sorption and mercury intrusion porosimetry (MIP) techniques. The effect of solvent replacement by methanol on the pore structure was studied as well. It was concluded that the solvent replacement drying procedure preserves the original pore structure of hydrated cement because the calculated and measured bulk densities of the different water-cement ratio systems investigated were in excellent agreement. Capillary condensation analysis was used to estimate the volume of capillary pores smaller than 4 nm in pore diameter for the 0.6 and 0.75 water-cement ratio pastes. The 0.4 water-cement ratio paste has pores smaller than can be determined from capillary condensation analysis. The volume of pores smaller than 4 nm was estimated from volume-thickness (V-t) analysis. For the three systems investigated, the volume of pores greater than 4 nm was obtained by MIP. For solvent-replaced pastes that showed capillary condensation according to V-t analysis, excellent agreement was obtained between the nitrogen sorption and MIP techniques in the pore diameter range of 4 nm to 30 nm.

Knowledge of the evolution of the permeability of cement throughout the course of hydration provides a suitable means to evaluate the evolution of the pore structure. The main difficulty is to measure permeability without disturbing the tenuous structure of the material at the beginning of the hydration. We have developed a differential permeability technique in which the applied flow is sufficiently weak that the structure of the medium is not disturbed. As an example of application of this technique we present measurement of the evolution of the critical permeability during a sol-gel transition.

Different series of cement paste specimens were prepared with ordinary portland cement, with portland, blast furnace slag cements having slag contents of 30, 50 and 75% by mass, with commercial fly ash cement and with portland cement containing fly ash additions of 10, 20, 30 and 50% by mass. Moist curing of the specimens varied between 3 and 28 days before the pore size distribution and characteristics of the phase composition were analyzed. Subsequent to curing, the specimens were subjected to drying in air of 65% RH with a controlled CO2 content of 0, 0.03 and 2% CO2 by volume. Depth of carbonation, pore size distribution of the carbonated paste, and the phase composition were investigated after 28 days and 6 months of drying, respectively. The results show that carbonation alters the prevailing pore structure of the hydrated paste. Important parameters are the type of cement used and the duration of curing.

Gas flow through concrete may be modelled by a relationship:

where D* is the specific permeability, N* is the number of unit pores per m2, and α, β and γ are experimental constants.

A unit pore is defined by the hydraulic mean radius a* equal to the total porosity divided by the BET surface area. The value of a* is estimated from porosity and hydration parameters to lie between 1.6 and 2.8 nm for water-cement ratios from 0.42 to 0.77 and initial porosities from 0.292 to 0.150. Estimation of pore radius a from gas flow yielded values of a/a* varying from 72 to 6. There appears to be fair correlation between N* and D*, but it appears that flow passes through microcracks rather than the natural pore system.

The effect of temperature on the progress of hydration and cement paste structure and properties was studied using four laboratory-synthesized portland cements, hydrated at temperatures between 5 and 95°C for up to 90 days. In addition to an accelerated progress of hydration, elevated temperatures altered the phase composition and pore structure of the resulting hydrated material. Below 75°C the effect of temperature on the intrinsic bond properties of the hydrates formed remained negligible.

The calculation of such macroscopic engineering properties as elastic modulus and compressive strength for cement pastes and concrete depends upon the establishment of a realistic model of the microstructure. Increasingly complex models are considered, which appear capable of predicting the elastic modulus in terms of a modified Rule of Mixtures. The same models are able to account for the broad features of the compressive strength, because strength is always scaled by the elastic modulus of the material. The actual value of the compressive or the bend strength is determined by the mechanics of crack initiation and crack propagation in the particular test used. Crack initiation is controlled by the defects present in the material and crack propagation by the fracture toughness of the different phases and the porosity in the microstructure. Thus the strength depends upon microstructure in a number of different but interrelated ways, determined by the fracture toughness of the material.

The resistance of a coherent granular material, such as mortar or concrete, to crack propagation cannot be appreciated without taking into account the heterogeneity of the material. Heterogeneity results from the presence of aggregates and from the local modifications of the composition and of the behavior introduced by these aggregates in the volumes of intergranular paste situated close to their interface, the “aureoles de transition.” It is possible to modify the characteristics of these zones to develop the global response of the material.

Stress corrosion phenomena, i.e. significant effects of mechanical stresses on the corrosion resistance of materials, have been reported for a large variety of materials, including metals, oxides and halides. Recently it has been shown [1,2] that cementitious materials are also sensitive to stress corrosion. The time dependent decrease in the flexural strength of chemically stressed concrete and mortar depends significantly on the mechanical stress acting simultaneously with the chemical attack.

For cementitious materials a comprehensive study of the stress corrosion phenomena has been started. In this paper the latest results from the current research program are presented and the interrelations between the microstructure of hardened cement mortar and the strength reduction caused by stress corrosion, are briefly discussed.