Cement Energy and Environment

that the amount of heat produced per mole of water consumed (transformed to IW) was constant. Deviation from this proportionality occurred only during the deceleration phase in the absence of retardant where the heat flow was higher than suggested by the water consumption . Two reactions have been suggested to dominate the production of heat prior to sulphate depletion (39]: 1) silicate reaction (Eq. (1 )) where C-S-H formation and portlandite precipitation occur synchronously with C3S dissolution, and 2) ettringite precipitation (Eq. (2)) from ions resulting from initial dissolution of C3A and calcium sulphates. I (CaOIJ.SI~ + 3.9H10- (Ca0) 17 .SI~ .(H10h 0 + IJCl(OHh (I ) 6Ca 2 ' + 2AI 31 + 3504 1- + 120W + 26H)O (2) - (CaOJ,.AI10l.(C.lSO•h·(Hz0)n (Ettrlnglte) Ettringite precipitation has been shown to be relatively constant during the main peak of hydration (39], therefore heat production and water consumption are expected to be proportional to the rate of the silicate reaction. Following sulphate depletion, C3A dissolution also contributes to heat production, however, water will not be transferred to the IW fraction during this reaction since no solid phase is formed. This decoupling of heat production from IW production during the initial burst of C3A dissolution could account for higher heat flow per mole of water consumed during this period. The lack of this decoupling where retardant was present likely indicates a change in the aluminate- sulphate balance resulting from effects such as the reactivity of cement phases, the dispersion and wetting of cement grains, and the dissolution rate of calcium sulphates (40]. Based on the rate plots of Fig. 7, the proportionality between water consumption and heat production was calculated as -17,000 joules per mole of water. This value sits favourably between the calculated values of -21,000 J mol-1 for silicate reaction and - 10,000 J mol-1 for ettringite precipitation, using enthalpies of -561 Jg-1 and -214 Jg-1 for the reactions shown in Eqs. (1 ) and (2) respectively (39], along with molar weights of 228 and 1255 for C3S and ettringite respectively. These values imply that the silicate reaction accounted for approximately 76% of the heat production, a value less than that estimated (-80 to 85%) from study of OPC hydration reported by Jansen et al. [39]. It should be noted however that the cement used by Jansen et al. was higher in CaO and Si02, lower in Al203, and reacted more rapidly such that the peak heat flow was NS mW g-1 , suggesting that the silicate reaction may have formed a greater part of the total heat flow than was the case in our study. 3.6 Initial T2 decline The ITO forms a distinctive feature of the T2 distributions (Fig. 4) and appears in all cases , with and without retardant, to immediately precede the start of gel pore formation (Figs. 5 and 6) . Importantly, while commencing just after the onset of the acceleration period of hydration, the completion of the ITO coincides exactly with the end of the acceleration period and the hydration peak. Therefore, the profound changes in microstructure occurring during the ITO appear to be closely linked with the processes controlling the rate of hydration. A large decrease in the T2 (and T1 ) values, usually an order of magnitude, during early hydration is a ubiquitous feature 1 H NMR relaxometry studies of cement. A very early study using Portland cement measured both T1 and T2 changes during hydration and observed T2 decline commencing at 10 h but which continued till about 50 h [41 ], much longer than is the case here. A number of investigators of white cement hydration have reported large declines in the T2 value of the capillary pores for the initial 100 h [11] and for N100 h (30]. More recent studies have reported initial order of magnitude declines in the T2 values of the capillary pores over 24 h during the initial hydration of OPC derived endodontic cements (28,29] and of white cement [21] . In the latter study, however, gel pore formation commenced after 2 h of hydration, suggesting differences in the relationship between hydration reactions and microstructural development compared to OPC. Wang et al. [37] showed that the initial decline in the T1 value of the capillary pore peak correlated well with penetration depth of a Vicat need le. The end of this decline matched exactly the end of the setting period, while the start of the setting period matched the point of upturn in a plot of 1/T1 . These findings provide strong support for the idea that the processes involved in the ITO are linked to the percolation characteristics of the solid network. While the links between hydration and attainment of the percolation threshold are complex and difficult to resolve [42], the final 61