5.3      The Hydration Reactions

     Each of the four main cement minerals reacts at a different rate and tends to form different solid phases when it hydrates. The behavior of each of these minerals has been studied by synthesizing it in its pure form and hydrating it under controlled conditions, and these reactions are discussed in this section. It should be noted that during the actual cement hydration process all the minerals dissolve into the same pore solution, and thus the solid hydration products are associated with the pore solution as a whole rather than a particular cement mineral. However, the individual reactions provide a good approximation of the overall hydration behavior of cement.

Hydration of the calcium silicate minerals (C3S and C2S)
     Tricalcium silicate (C3S) is the most abundant and important cement mineral in Portland cements, contributing most of the early strength development.  The hydration of C3S can be written as:


where 1.7C-S-Hx is the calcium silicate hydrate (C-S-H) gel phase and CH is calcium hydroxide, which has the mineral name portlandite. The variable x in eq. 5.1 represents the amount of water associated with the C-S-H gel, which varies from about 1.4 to 4 depending on the relative humidity inside the paste and on how much of the water associated with the C-S-H is considered to be part of its actual composition. The state of water in cement paste and in C-S-H gel is discussed further in Section 5.6. The kinetics of hydration of C3S are substantially similar to those of Portland cement as a whole (see Figure 5-1). Much of the reaction occurs during the first few days, leading to substantial strength gains and reduction in capillary porosity.
     The dicalcium silicate phase (C2S) reacts according to:


The hydration products are the same as those of C3S, but the relative amount of CH formed is less.  C2S is much less soluble than C3S, so the rate of hydration is much slower.  C2S hydration contributes little to the early strength of cement, but makes substantial contributions to the strength of mature cement paste and concrete. 

Hydration of the calcium aluminate/ferrite minerals  (C3A and C4AF)
     The hydration of the aluminate and ferrite minerals is somewhat more complex than that of the calcium silicate minerals, and the reactions that take place depend on whether sulfate ions are present in the pore solution.  C3A is highly soluble, even more so than C3S.  If C3A is hydrated in pure water, calcium aluminate hydrates form.  The reaction sequence is:


where the first reaction is very rapid and the second reaction occurs more slowly.  The final reaction product, C3AH6, is called hydrogarnet.  The initial reaction is so rapid that if it is allowed to occur in a portland cement paste it would release large amounts of heat and could cause the paste to set within a few minutes after mixing, an undesirable condition known as flash set.  The purpose of adding gypsum (C H2) to Portland cement is to prevent this from happening.  The gypsum is also highly soluble, rapidly releasing calcium and sulfate ions into the pore solution.  The presence of the sulfate ions causes the C3A to undergo a different hydration reaction.  The reaction of C3A and gypsum together is:


where C6A 3H32 is the mineral ettringite.  Since all portland cements contain gypsum, eq. 5.4 is the main hydration reaction for C3A.  Small amounts of hydrogarnet formed by eq. 5.3 can sometimes be found in cement pastes, however.  If the gypsum in the cement reacts completely before the C3A, then the concentration of sulfate ions in the pore solution decreases drastically and the ettringite becomes unstable and converts to a different solid phase with less sulfate:


where the new reaction product, C4A H12, is called monosulfoaluminate.  Most cements do not contain enough gypsum to react with all of the C3A, and as a result most or all of the ettringite is converted to monosulfoaluminate within the first day or two of hydration via the reaction of eq. 5.5. 

     Reactions 5.4 and 5.5 are both exothermic and contribute to the heat of hydration of cement.  The early hydration of C3A to form ettringite via eq. 5.4 is quite rapid; this is a major contributor to the stage 1 kinetics shown in Fig 5.1.  As with the early C-S-H formation, this separates the C3A particles from the pore solution, slowing their dissolution.  After a period of several hours the early-forming ettringite converts to monosulfoaluminate via eq. 5.5, allowing the C3A to undergo renewed rapid hydration, with kinetics that are roughly similar to the main hydration peak of C3S (stages 3 and 4).  When the C3A is hydrating within a cement paste, this second period of reaction often creates a shoulder in the decelerating (Stage 4) rate period just after the main peak (not shown in Fig. 5.1).
     The ferrite phase (C4AF) reacts in a similar fashion to the C3A (eqs. 5.1-5.5), but more slowly.  One important difference is that some of the aluminum in the reaction products is substituted for iron.  The amount of substitution depends on many factors including the composition of the C4AF and the local conditions in the paste.  A convenient way to represent these reactions is [1]:



where (A,F) indicates aluminum with variable substitution of iron, and (F,A) indicates iron with variable substitution of aluminum.  The (F,A)H3 is an amorphous phase that forms in small amounts to maintain the correct reaction stoichiometry.  Because of the substituted iron, the main reaction products are not pure ettringite and monosulfoaluminate, although they have the same crystal structure.  Instead, cement chemists have given them the names AFt and AFm, respectively, where the m indicates monosulfate (one sulfate ion) and the t indicates trisulfate.  In a portland cement paste where the C3A and C4AF are intimately mixed together, it can be safely assumed that the aluminum-bearing reaction products are never completely free of iron, and so the terms AFm and AFt are more correct.  However, as with the terms alite (impure C3S) and belite (impure C2S), this is a bit more confusion than many people are willing to deal with, and thus the terms ettringite and monosulfoaluminate are commonly used to refer to these phases in cement pastes.

Reaction with additional sulfate ions
    As noted above, most portland cements do not contain enough added gypsum to fully hydrate the C3A and C4AF via reactions 5.4 and 5.6 to form ettringite. Once the gypsum is consumed, the ettringite reacts with the remaining C3A and C4AF to form a new lower-sulfate phase called monosulfate (reactions 5.5 and 5.7). Thus in a mature portland cement paste it is normal to find monosulfate and little or no ettringite.
    However, if a new source of sulfate ions becomes available in the pore solution, then it becomes thermodynamically favorable to form ettringite again, just as it was initially. This will occur at the expense of the existing monosulfate:


where it is understood that the A sites will contain some F. The gypsum on the left side of eq. 5.8 represents the equivalent amount of dissolved ion, as no solid gypsum need form in the paste.
    Reaction 5.8 is more than just a theoretical point: in fact, it is all too common for sulfate ions present in ground water, sea water, and soil to diffuse into concrete, allowing the reformation of ettringite to proceed. This occurs primarily, but not exclusively, in concrete below ground level, such as building foundations. The problem with this phenomenon is that reaction 5.8 is expansive, meaning that the ettringite occupies a larger volume than the monosulfate it replaces. Thus expansive stresses are created that can cause cracking and other deterioration. Unfortunately, this is actually only the first step in the sulfate attack process, as once all of the monosulfate is consumed other chemical reactions can occur that further weaken the cement paste (assuming a continued ingress of sulfate ions). We will discuss sulfate attack further in Chapter 12, and conclude this discussion by mentioning that the best way to minimize the damaging effects of sulfate attack and similar chemical attack processes is to use a low w/c to keep the permeability of the concrete low.

<-previous    next->

Home Pictures Site Map About the Authors Links