Cement Energy and Environment

setting point, which coincides with the peak in hydration rate, has been hypothesised to be due to percolation of hydration products within the pore spaces [43), based on a colloidal growth model. This model was able to match heat production (through scaling) very well and predicted a peak rate at lower and more realistic degrees of hydration compared to previous modelling. The mode of colloidal growth is also consistent with 1 H NMR relaxometry data reported by Bortolotti et al. [17) where clusters of constant composition , comprised of C-S-H gel globules and gel pores, were proposed to fill the capillary pore spaces during hydration. Concepts and studies regarding percolation and setting indicate that hydration progress during the ITO likely involved the penetration of the spaces between the cement clinker grains, responsible for the initial peak in the T:2 distribution at 20 ms, with diffuse and highly porous C-S-H as proposed by Bishnoi and Scrivener r44). A later review of hydration models concluded that in-filling of pores occurred via dendritic growth of low density hydrates [45] and TEM micronraphs have illustrated the fibrous manner of groWlh displayed by C-S-H during early hydration [46]. Such growth into the initial pore space will substantially increase the surface area within the pore volume and hence reduce relaxation times but without the formation of a population of pores producing shorter T2 times of hundreds of microseconds that characterize gel pores. Holly et al. [11] observed the appearance of a water population with a stable T2 value of 100 i..JS, assigned to interlayer water, after -10 h hydration. These authors proposed that this event signified a change in growth morphology from a fibrillar to a sheet-like structure. The cause of the rate peak has been hypothesised to be due to the impingement of hydration products where a low density form of C– S-H ramifies the pore space and latm densities [43-45] with this mechanism compatible with our observation that the rate peak coincides with a change in C-S-H growth morphology. -lowever, a recent paper [47] based on modelling of the hydration processes has proposed that the rate peak is due to a reduction in C3S dissolution rate due to surface coverage by C-S-H. Furthermore, this mechanism accounts for the insemsitivity of early hydration kinetics to the w/c ratio of the cement paste. A possible link between this proposed mechanism and the change in microstructural evolution observed here is unclear. Based on these observations, we propose that the space between cement grains is rapidly penetrated by very low density fibrillar growth during the ITO with this process providing sufficient additional surface area to decrease T2 relaxation times by an order of magnitude. After 5 to 7 h the ITO ends, marking a significant change in the system that apparently coincides with the point of final setting , a change in C-S-H growth morphology, and the start of formation of gel pores which are associated with layered C-S-H gel globules. The presence of retardants, while delaying the start of the ITO had only a small apparent effect on the measured characteristics of the ITO; the duration of the ITO was extended from -5 h with no retardant to -7 h in the presence of retardants. 3.7 Gel pore formation The start of gel pore formation coincides exactly with the end of the ITO and the peak in hydration rate (Fig. 6) and, based on the discussion in Section 3.6, with a change in C-S-H morphology from fibrillar to sheet-like structure. Considered within the context of the Jennings colloid model [15], the initial formation of aggregates of gel globules composed of a layered structure is associated with formation of small gel pore spaces and which together constitute the low density (LO) C-S-H. The mechanisms responsible for the deceleration in hydration rate likely are closely linked to the process of gel pore formation. Various hypotheses have been put forward for the onset of deceleration, and, recently, numerous (chiefly modelling) studies have concluded that the peak is a direct result of nucleation and growth mechanisms [2,47]. The activation energy of hydration has been shown to remain constant during the main hydration peak indicating that there is no transition in the rate controlling reaction at the hydration peak [48], yet it appears that there is some transition occurring in the growth mechanism of C-S-H. The distinct cessation of gel pore formation is in accord with a limitation, perhaps spatial, to the growth of LO C-S-H leading to the transition to the growth of high density (HO) C-S- H which contains no gel pore spaces [21]. The degree of hydration, as measured by the IW%, at the end of th is period is similar in all treatments (Fig. 8). 62