C3S the induction period

The causes of the induction period and of its termination have been the subject of much debate. Hypotheses have been reviewed in a collaborative paper (T25). The main ones are as follows  

The induction period is shortened by adding prehydrated CjS (013), but additions of lime or CH, including that formed from CjS, are variously reported to be ineffective (013,B67) or to lengthen it, though shortening it with cement (Ull). In cement mixes, additions of pfa or some other finely divided materials accelerate hydration after the first day, apparently by acting as nucleation sites for C-S-H (Section 9.3.3). Additions of reactive silica markedly accelerate hydration (S53). Most of this evidence supports hypothesis 3 and tells against hypothesis 4. Hypothesis 3 does not exclude hypothesis 1, as the breakdown of a protective layer could be associated with formation of a new product. 

The recent microstructural evidence (Section 5.3.1) gives no indication that a membrane or other product distinct from that formed later is formed during the initial reaction in C3S pastes, though a gelatinous coating is formed in cement pastes, which show an induction period similar to that observed with C3S (Section 7.5.1). For C3S pastes, this evidence excludes hypothesis 2, and gives no positive support to hypothesis 1. It does not exclude the formation of an altered layer on the C3S surface, no more than a few nanometres thick. Tadros et al. (T28) postulated the formation of a SiOj-rich layer with chemisorbed Ca , and Barret et al. (B63) that of a superficially hydroxylated C3S, formed by protonation of the 0 and SiO ions, balanced by loss of Ca .  

The calculated solubility of C3S is about 1 molal (S50). In the absence of an altered or protective layer, the observed low rate of reaction during the induction period could be explained only if the concentrations close to the C3S surface were of this order of magnitude. It is doubtful whether the very high concentration gradients that this implies could exist (J15,G49). 

The balance of the evidence favours a combination of hypotheses 1 and 3. The following model is essentially that of Gartner and Gaidis (G49), but has features in common with those of Barret and Bertrandie (B57) and of Grutzeck and Ramachandran (G51). 

---