Heat evolution and electrical conductivity curves plotted against hydration time for C 3 S hydration (a) in water at a w/c = 18 and (b) in a 22 mM lime solution at w/ 

Context in source publication

Context 1

... discussed later. Owing to variations in the size of the Burgers vector b and in the solid – liquid interfacial energy, different crystal faces will behave differently during etch pit formation and the number of etch pits will vary according to the types and number of dislocations present [51]. The principal features discussed above are illustrated in Fig. 5 for surfaces of quartz (100) under different undersaturation conditions [46]. Experimental results for the net dissolution rate of various minerals as a function of undersaturation (adapted from [2]) are shown in Fig. 6. Note that the dissolution rates are de fi ned as negative value as opposed to growth rates. It can be observed that there are two regimes of dissolution, corresponding to the regime in which the undersaturation of the solution is low enough ( σ ≪ Δ G crit * ) to allow the formation of etch pits at dislocations, followed by a sharp decrease in the rate of dissolution at undersaturations, still quite far from equilibrium, where only step retreat can occur. (The regime of homogeneous formation of etch pits is not observed in these plots). Most of these minerals show large values of Δ G * crit − from 2 to tens of kJ/mol. Therefore, slow dissolution rates are observed not only close to equilibrium, but also quite far from it. For example in the case of albite the slowdown in dissolution, due to the fact that etch pits may no longer form at dislocations occurs at an undersaturation of around 11 ( ∼ 33 kJ/mol) (Figs. 4 and 6). Phenomenologically this drastic slowdown in dissolution rate is similar to what is seen for alite. This observation led us to carry out some new experiments and re-examine published data on the early stages of alite and C 3 S reaction. In order to investigate the relevance of the above theories to the early dissolution process of alite, studies were made on alite synthesised by the authors, quenched in air [52] and then fractured to give fresh surfaces. The theory of dissolution presented above indicates that the state of saturation of the solution should be of great importance. The samples were immersed in two different solutions: deionised water and saturated lime solution for 2 and 30 min. The water to cement ratio was in the range of 1000. The samples were not coated for examination in the SEM; the pictures (Fig. 7) were taken in the secondary electron mode at an accelerating voltage of 3 kV. After only 2 min samples immersed in water show extensive surface pitting. There is also a marked difference in Fig. 7(a) between the crystal in the top right of the picture and the one in the bottom left. This is a consequence of the crystallographic orientation of the grains, which would have different interfacial energies, different dislocations densities and therefore different energy barriers for the nucleation of pits. It is also pertinent to note that the clear de fi nition of these etch pits suggests that there is no hydrate layer covering the surface. Furthermore, these reactive sites of dissolution cover the surface fairly evenly, so the formation of a protective phase only at reactive sites unlikely. After 30 min of hydration (Fig. 7(c)) one can observe that the surface has been severely corroded with step heights reaching several hundreds of nanometers. In the case where the alite was hydrated in saturated lime solution, the surface does not undergo such extensive dissolution. After 2 min portlandite and other hydrates have precipitated, maybe during specimen drying or at places where supersaturation is locally reached (Fig. 7(b)). After 30 min (Fig. 7(d)), C – S – H with a “ sheaf of wheat ” morphology [53,54], is observed. However, at both times most of the surface appears smooth and unattacked. This strongly suggests that under saturated lime conditions, the predominant regime of dissolution is step retreat since the formation of etch pits is not observed for these samples. These observations con fi rm that the concentration of the solution plays an important role in the dissolution process of alite as seen in studies of other minerals and explained in the previous section. Fig. 8 from a normally hydrated alite paste shows that even at much later ages dissolution from certain areas is favoured to give a very rough surface, while other crystal facets show a much lower degree of reaction. In fact, examination of the literature indicates other studies in which the effect of solution concentration on dissolution rate is apparent. For example, Barret and Ménétrier [6] analysed fi ltrates obtained by passing a limited amount of solvent (distilled water or lime solution at various lime concentrations) through C 3 S spread on a millipore fi lter. They showed that the lime concentration increase was always smaller when the lime concentration in the solvent was high. Several studies on dilute suspensions with different lime concentrations in the aqueous phase have been undertaken during the past decades [36,42,55,56]. In Fig. 9, it can be seen that the fi rst peak of dissolution monitored by isothermal calorimetry is completely different depending on whether the aqueous phase is pure water or a saturated lime solution. This clearly shows the effect of the initial saturation state of the early dissolution. In order to study the in fl uence of crystallographic defects on the early reactions of alite with water, narrow particle size distributions of quenched alite were thermally-treated at 650 °C for 6 h. The particle size distributions were centred around values of d v50 of 38, 61 and 82 μ m the particle size distribution is shown in (Fig. 10) [52]. The annealing temperature of 650 °C is one at which recovery – annealing of defects and a decrease in the number of dislocations – is expected to occur. The treated samples changed from polymorph MIII to TI as demonstrated by the X-ray diffraction patterns in Fig. 11, which is consistent with the decrease in defect density. The particle size distribution was however not affected by this heat treatment. Narrow particle size distributions were chosen to avoid dispersion of the reactivity due to different particle sizes. The heat of hydration was followed by isothermal calorimetry at 20 °C, with a water to cement ratio of 0.4. Fig. 12 shows the heat evolution of the quenched and thermally-treated samples. Fig. 12(a), in which the samples were mixed inside the calorimeter, shows that the fi rst heat peak is much smaller for the thermally-treated sample than for the control. In Fig. 12(b) the samples were mixed externally by hand before insertion into the calorimeter. The sharp initial heat peak is followed by a period of low chemical activity which is prolonged for several hours for the thermally-treated specimens, with a de fi nite, almost fl at, induction period. However, the acceleration and deceleration parts of the curves are very similar for both treated and untreated samples, which suggest that the same mechanisms and rate laws apply to both the quenched and the heat treated samples at this stage. These experiments show that the early dissolution behaviour is dramatically affected by the defect density of the crystals and that this affects the time needed to reach the acceleration period. According to the theory of dissolution, a low density of crystallographic defects will lead to a low number of etch pits. Therefore when dissolution becomes limited to the step retreat process (as the solution concentration increases) the surface is less rough due to the low number of etch pits so there are less steps. Therefore, dissolution proceeds at a lower rate compared to the quenched alite. In fact, the role of defects in the early hydration processes of cementitious materials was already noted by previous researchers [31 – 34]. Maycock and co-workers [33] and Odler and Schüppstuhl [31] studied the effect of quenching rate on the reactions of alite and found that faster quenching, likely to induce more crystal defects, resulted in shorter induction periods. Fierens and Verhaegen [34] cooled tricalcium silicate at different rates from 1600 °C to 1300 °C before quenching. Besides calorimetry (Fig. 13), they used thermoluminescence to follow the changes during the early reaction [57,58]. Thermoluminescence gives an indication about the presence of crystallographic defects in the structure by excitation with plasma, γ ray, or UV sources followed by relaxation upon heating, leading to the emission of light [59]. Two populations of defects were identi fi ed, whose amount increased with higher cooling rates. One of these populations of defect progressively disappeared during early hydration. Furthermore, the length of the induction period was found to be inversely proportional to the original magnitude of this thermoluminescence peak. (The changes in the calorimetry curve are different to those reported above due to the fact that a broad range of particle sizes is present.). Dislocations and planar defects (such as stacking faults) were observed by TEM in alite thermally-treated at (700 °C) by Hudson and Groves [60]. The density of dissociations in these annealed samples, was too low to be measured by TEM (the limit of resolution of TEM method for determining dislocation densities dislocation densities is around 10 6 cm − 2 [61]). It is however not excessive to assume that surfaces of quenched alite should be intersected by numerous defects. Dislocations and stacking faults in alite would arise from the growth and cooling process, possibly affected by the presence of impurities, and during the grinding processes. Sakurai and co-workers [32] etched alite with a solution of 0.4% HF with 0.6% HNO 3 in ethyl alcohol in order to reveal defects, grain boundaries, etc. Their SEM images reproduced in Fig. 14 show that etching begins at grain boundaries and etch pits are formed that they associated with the emergence of dislocations. Ménétrier and co-workers [45] observed by SEM a non-uniform ...