A model for reactions that occur in hydrating portland cement is now generally well developed. Incorporation of various by-products to form blended cements modifies both the hydration reactions and the physical properties of the resulting pastes. A review of recent progress in understanding the effects of blending agents on these reactions is presented. The blending agents considered are low-calcium (Class F) fly ash, high calcium (Class C) fly ash, blast furnace slag, silica fume, biosilica and natural pozzolans. Effects of the blending agents on physical properties such as rheology are also considered. Particular attention is given to the essential role of alkalies in pore solutions and the beneficial reactions that occur with high silica content blending agents.

The hydration of a blended cement through hydraulic or pozzolanic reactions results in heterogeneous polyphase materials. Because portland cement clinker is the major component in most cement blends, the microstructural development of portland cement hydrates, including C-S-H and pore structures, is first discussed. Slag, fly ash, silica fume and limestone filler cements are then compared to portland cement with regards to C-S-H morphology and composition, aluminate crystallization, cement paste interfaces and pore size distribution.

Pore structure development in portland/fly ash blends was investigated using mercury porosimetry and methanol exchange techniques. The progress of hydration was monitored using compressive strength tests. The specimens were made using four water-cement ratios and were hydrated over a one-year period in lime-saturated water. Mercury porosimetry results indicated that the blended cement pastes generally had higher total porosity than plain cement pastes. The major contribution to this increase in porosity was in the form of smaller pore sizes. With reactive fly ash at 20% replacement, the pore structure of mature paste consists mainly of pores nominally smaller than 0.05 μm in diameter. Diffusion parameters obtained from the methanol exchange results were found to be inversely related to the volume of large pores (nominally <0.05 μm) and also to the volume of small pores (nominally <0.05 μm). The effects of the physical and chemical properties of cements and fly ashes on pore structure development are discussed.

Pastes of a portland cement (60%) blended with a granulated blastfurnace slag (40%) were examined, principally by SEM with EDX analysis. Reaction rims around slag particles and relicts of fully reacted slag particles had compositions compatible with mixtures, in varying proportions, of C-S-H having Si/Ca -0.62 and a phase of the hydrotalcite family having Al/Mg -0.38. Calculations taking into account relevant densities and water contents indicated that replacement of the slag by its in situ hydration products entails little or no change in the numbers of Mg and 0 atoms per unit volume, but that substantial proportions of the Ca, Si and Al are released and an equivalent amount of H gained. In other respects, the microstructures qualitatively resembled those of pure portland cement pastes of similar ages, but less CH was formed and the C-S-H not formed in situ from the slag had a Si/Ca ratio of 0.56, higher than that of 0.50 to 0.53 found in the absence of slag. None of the individual phases in the slag cement pastes showed significant compositional variation with time in the 28 day to 14 month period studied. The relative amounts of Ca, Si and Al expelled from the slag are such that, in order to form C-S-H and AFm phase, more Ca is required. It is obtained partly at the expense of CH formation, and partly through increase in the Si/Ca ratio of the C-S-H formed from the clinker phases. Mass balance, volume composition and bound water content were calculated for the 14 month old paste and compared with corresponding results for the pure portland cement.

Fly ashes having three distinctly different levels of calcium, designated low-calcium (Class F), intermediate-calcium (Class F/C), and high-calcium (Class C), comprised the basic material for the present study. Pastes and mortars were made using cement and one of three types of fly ashes at various levels of replacement and water-solid ratios. Chloride ion diffusion was measured by applying an electrical potential across cured cylindrical samples and measuring the amount of current passed in a certain period of time (proportional to amount of CE” passed in this time). Other supportive measurements were made, e.g. porosity, pore size distribution, water permeability and surface area. The Cl− ion diffusivity was correlated with the chemical composition of fly ash, mix proportion, and water permeability of the hardened paste or mortar.

Cement paste and concrete specimens containing three different mixtures of portland cement (PC) and high-lime fly ash (FA) were subjected to various curing conditions, and the strengths, non-evaporable water, and porosities compared with control mixes containing portland cement only. Strength and porosity data indicated that the cementing action of the mixtures was in all cases inferior to portland cement. For each mixture the strength loss resulting from imperfect curing, i.e. sealed or exposed to 50% relative humidity, was greater than for portland cement. In the case of concrete drying from one surface, strength differences were found, but these were not as clearly defined as those obtained in the paste specimens.

Proton pulse nuclear magnetic resonance can be used to follow the hydration of C3S and C3A, individually and mixed, with and without gypsum and superplasticizers. The variation of the spin-lattice relaxation time from early ages up to the times of setting and hardening is related first to the mobility of water molecules in rapid exchange, then to the progress of solid-water interfaces.

Plasticizers act as retarders to lengthen the induction period. They behave differently with pure phases taken separately than with mixtures, in which a mutual interaction of C3S, C3A, gypsum and admixture has been characterized. Among the three superplasticizers studied, the sulfonated melanine formaldehyde was the strongest retarder, while the sodium sulfonated naphthalene was the weakest.

The hydration reaction of C3A with gypsum (molar ratio = 1:1.5) has been studied in water and in sodium hydroxide solutions at 30°C. The hydration product were analyzed microscopically and by means of X-ray diffraction. The concentration of sulfate and aluminum in solution was determined spectrophotometrically. In the absence of alkali, the amount of ettringite formed passes through a maximum and is accompanied by almost complete consumption of sulphate ion in solution. Alkali depresses the formation of ettringite and accelerates that of calcium hydroxide. The monosulfate hydrate appears immediately in alkaline solutions and water then reappears after a day as a solid solution with the tetracalcium aluminate hydrate. The formation of the monophase solid solution is accompanied by the consumption of C3A, ettringite, and calcium hydroxide and by an increase in the amount of the dissolved aluminum. The presence of alkali markedly increases the initial sulfate ion concentration from solution.

Very high strength concretes with water-cement ratios ranging from 0.21 to 0.27, having compressive strengths varying between 73 and 118 MPa, were prepared. One series was made with only high early strength cement (Type III), and the other series contained 6% to 11% silica fume.

In general, the microstructure of very high strength concrete is very dense and is composed mainly of C-S-H in the gel and crystalline phases. Mg, Al, S, Cl, K and Fe were detected in a number of C-S-H locales. The Ca/Si ratio was variable. In concretes without silica fume, the CH content is much lower than in normal concrete, and in the silica fume concretes it is still lower and not well crystallized. A few large, partly reacted and unreacted silica fume particles with surface cracks were present.

Strong cement-aggregate bonding is seen in concretes with silica fume containing limestone aggregates, whereas the gravel concretes show microcracks and a weaker bonding.

Effective particle dispersion is believed to play a significant role in hydration reactivity of slag/cement blends and consequently on the development of physical and mechanical properties of the hardened pastes. The effects of various factors (superplasticizer and mixing time) on the particle dispersion were evaluated using a computer controlled scanning electron microscope. Backscattered electron signals and elemental maps were resolved for polished mounts of slag and cement particle mixtures. The adequacy of dispersion of slag and cement particles was determined by analyzing about 1000 particles per analysis with appropriate calculations and statistical treatment. These results were related to the hydration behavior of the slag/cement blends. The results obtained can be used to evaluate some of the physical phenomena and the kinetics of slag-cement hydration.

A salt-saturated and sulfate-expansive grout was formulated with four levels of sodium citrate to determine effects on workability and setting time. Linear expansion of restrained grout prisms increased with amount of citrate used, apparently a function of extent of retardation of the grouts. This relationship was observed both for grouts cured for at least 53 days, and for a concrete formulated from the same cementitious materials cured for at least 180 days age.

The microstructures of the grouts at 28 days age showed notable increases in extent of reaction of the sulfate phases and in formation of ettringite with increasing citrate content, and thus with increasing retardation of set. XRD also revealed an increase in well-crystallized ettringite and decrease in gypsum with greater retardation. These observations provide a microstructural and compositional link between retardation and expansion of cementitious materials formulated to be expansive.

Computer-based models represent an important development that should provide a scientific basis for in-depth predictions of the behavior of materials. Aspects of microstructure models for cement-based systems are described. A computer model is described that can incorporate many assumptions in the form of sub-models. Specific algorithms are selected to solve specific parts of the problem. The algorithm chosen depends on specific problem requirements at each stage of calculation, and these problem requirements may not be anticipated in advance. An important part of these computer models is that they incorporate a file of numbers, representing the microstructure, which can be operated on to effect an incremental change in the system.

A general method has been developed for determining the effects of particle size distribution (PSD) on the kinetics of hydration. Given the PSD and the kinetic rate for particles having a single size, expressions can be derived for the variation of the total degree of hydration with time. The method was applied to PSD and kinetic functions typical of cement systems that account for reaction, diffusion and expansion of spherical hydrate layers. The PSD was found to critically affect the kinetics. A key feature of the theory follows with time those particles that have become totally hydrated, so-called “dead” particles. This produces a dead-particle hydration curve having approximately the same shape and location as the overall hydration curve. It can be represented analytically in all cases and gives a measure of the disguise provided by the distribution.

Hardened cement paste can be regarded as a highly dispersed system of solid particles, air voids and water filled pores ranging in size from a submicroscopic to a macroscopic scale. Using a statistical model, the elastic moduli of solid particles, air voids and pore water can be combined appropriately to find a correlation between the modulus of hardened cement paste and the moduli, as well as respective volume fractions, of its constituents. Ice formation and the addition of aggregates in a mortar can easily be taken into account. On this basis the measurement of the dynamic elastic modulus of hardened cement paste and mortar at different temperatures and its evaluation provides much interesting data. The interaction of particles and pores is better understood. The ice formation can be studied. Since the freezing temperature of pore water is lowered in small gel pores, the pore size distribution can be calculated. Frost damage is observed directly. Therefore, this method is a valuable tool to improve the Munich model of hardened cement paste.

Volume changes can be calculated from the hydration stoichiometry of C3S if the composition of C-S-H is taken as C1.7SH4.0 with a density of 1.85 g/cm3. The results are in general agreement with the volume changes determined by Powers. However, the calculated evaporable water content is higher and the space-limiting water-cement ratio is calculated to be 0.42. The calculations can be applied also to the pozzolanic reaction and predict a marked increase in solid volume. In that case the composition of C-S-H is modified to C1.5SH3.8.