H.J.H.

Brouwers

Chemical Reactions in hydrated Ordinary Portland Cement based on the Work by Powers and Brownyard
In a pioneering work, Powers and Brownyard (1948) were the first to systematically investigate the reaction of cement and water and the formation of cement paste. In the late 1940s they presented a thorough model of the cement paste, in which unreacted water and cement, the hydration product, and (gel and capillary) porosity were considered. Via extensive and carefully executed experiments major paste properties were determined such as the amount of retained water and the chemical shrinkage associated with the hydration reaction. These properties were furthermore related to the content of the four most important clinker phases, viz. alite, belite, aluminate and ferrite, in the cement. Locher (1975), Hansen (1986), Taylor (1997), Neville (2000) and Brouwers (2003) summarise the most important features of their work. Here, it will be demonstrated that their results enable the study of the reactions of the four clinker phases and their reaction products, which is a principal innovation. In the past, the water binding of OPC as predicted by their model has been compared to pure C3S hydration by Locher (1966) and Young and Hansen (1987). This approach is permitted, as C3S is the major constituent of OPC. The model of Powers and Brownyard (1948), however, contains valuable specific information in regard to the reaction of each individual clinker phase, such as C3S. This aspect of their model and experiments has been overlooked hitherto. This will be presented here, and will result in a new reaction model. Water retention measured by Powers and Brownyard (1948) Powers and Brownyard (1948) executed and reported numerous experiments with cements of different compositions, with neat cement and with mortars, and at various water/cement ratios (w0/c0) and various hardening times at ambient temperature. Their cements had a wide variation in composition and the specific surface was 1610-2045 cm2/g. The water retention of the hardened cement was measured both at saturated state and at P-dried state. The water that can be removed by P-drying (using vacuum and magnesium perchlorate hydrate at 23 oC) was named evaporable water, the remaining water non-evaporable or chemically bound water. The evaporable water comprises both absorbed water (gel) and capillary water. Assuming that all cement particles smaller than 50 m have reacted after a long time (e.g. 1 year or even longer) and at high w0/c0 (more than 0.44), Powers and Brownyard (1948) fitted the following relation between wn/c and the clinker composition: wn/c = 0.187 x C3S + 0.158 x C2S + 0.665 x C3A + 0.213 x C4AF (1)

This equation (mass of water retained in P-dried state) can be transferred in moles of retained water per mole of reacted clinker phase via

1-0553

mC3S = x C3S c; mC2 S = xC2S c; etc.

(2) (3)

w = M H n H ; mC3S = M C3 S nC3S ; mC2 S = M C3 S nC2 S ; etc. yielding nH,n = 2.37 nC3S + 1.51 nC2 S + 9.97 nC3 A + 5.74 nC4 AF

(4)

whereby the molecular masses of water and the clinker phases (and most hydration products) can among others be found in Brouwers (2003). Powers and Brownyard (1948) used the P-dried samples to perform sorption experiments using water vapour. They found that at relative water vapour pressures below 45%, the amount of water held is proportional to the amount of cement reacted and hence, was assigned absorbed or gel water. Above this relative vapour pressure, the water also condenses in the larger capillary pores. Furthermore, using the B.E.T. theory, a quantity Vm was measured and seen as the mass of water to cover the P-dried hydration product with one layer of water. It was observed that Vm is linearly proportional to the amount of non-evaporable water. This was explained with the fact that the internal surface of the hydration product is proportional to the cement reacted. In later work Brunauer and Kantro (1964) experimentally confirmed for C2S and C3S hydration that the surface development closely follows the degree of hydration indeed. Accordingly, the amount of gel water was related to the non-evaporable water. Furthermore, it was recognised that the amount of internal surface depends on the composition of the cement as each clinker phase produces a hydration product with its specific internal surface. From the experiments it was furthermore concluded that the maximum amount of water that can be retained by the hydration product, the i.e. gel water, corresponds to 4Vm per mass of reacted cement. Water in surplus of 4Vm is capillary water. This result was explained with the concept that Vm is the mass of water to cover the hydrated cement in the hydration product with one layer of water, and that with 4 layers the hydration product (gel space) is saturated. Powers and Brownyard (1948) therefore recommended the following empirical fit: wg/wn = 4Vm = 4 (0.230 x C3S + 0.320 x C2S + 0.317 x C3A + 0.368 x C4AF ) (5)

The total water retained by the reaction products (non-evaporable + gel water) thus reads:
wd/c = (wn + wg)/c = (1 + 4Vm) wn/c

(6)

The total retained water could also depend on the C S content, so this mass fraction should be included in a fit as well. Accordingly, a least squares method is employed to the experimental data of Powers and Brownyard (1948). Details of this approach can be found in Brouwers (2003). The fitting yields
wd/c = 0.334 x C3S + 0.374 x C 2S + 1.410 x C3A + 0.471 x C 4AF + 0.261 x C S (7)

1-0554

Using eqs. (2) and (3) the water retention can now be written in moles as: nH,d = 4.23 nC3S + 3.58 nC2 S + 21.1 nC3 A + 12.7 nC4 AF + 1.97 n C S (8)

Eqs. (7) and (8) have as advantage that they provide the mass and moles of totally retained water as a function of each clinker phase and that it includes calcium-sulphate. Here, the information on water retention in P-dried and saturated paste will be used to investigate the reactions of the four clinker phases. Reactions of calcium silicate phases Powers and Brownyard (1948) presented a literature review of the reaction products and were aware that the products of the clinker phases C3S and C2S were microcrystalline CH and a colloidal gel, also named colloidal hydrous silicate and calcium silicate hydrate (p. 106-132, p. 260, p. 488). In later work this product was called tobermorite gel (Brunauer et al. (1958), Brunauer and Greenberg (1960), Kantro et al. (1960), Brunauer and Kantro (1964), Kantro et al. (1966)) and nowadays it is generally called C-SH (Taylor (1997)). It is known to be a poorly crystalline to almost amorphous material. Both clinker phases and said reaction products are most abundant in cement paste, and can react independently from the other clinker phases. Accordingly, the following relations are put forward: C2S + (2 x + y) H CxSHy + (2 x) CH C3S + (3 x + y) H CxSHy + (3 x) CH (9) (10)

Note that at that time x, i.e. the C/S ratio, was unknown, though it was known that the C2S reaction hardly produces CH (from Powers and Brownyard (1948), p. 488). In other words, x should be close to 2, being in agreement with the present knowledge of x being in the range 1.5 to 2. The C/S ratio depends among others on the hydration conditions (bottle, paste), w0/c0, particle size, age and the analytical method employed (Taylor (1997)). Eq. (10) was proposed by Locher (1966), who was the first to couple the overall non-evaporable water of Portland cement (thus not C3S in particular) as determined by Powers and Brownyard (1948), to the hydration of the single clinker phase C3S. Note that (2 x + y) and (3 x + y) correspond to n H , and that nC2 S = 1 and nC3S = 1 in eqs. (9) and (10), respectively. They depend on the drying conditions: see eqs. (4) and (8) for their values in P-dried and saturated state, respectively. P-dried state Based on the findings of Brunauer et al. (1958), Brunauer and Greenberg (1960) and Kantro et al. (1966), P-drying only removes water from the C-S-H (and not from the crystalline CH) and it does not affect the C/S (x) ratio of the C-S-H. X-ray analysis on P-dried C2S and C3S/alite by Kantro et al. (1966) revealed that (x y) amounts to 0.5 and 0.4 in C-S-H formed by C2S and C3S/alite, respectively. These values are stable in time (up to 1700 days) and are found for water-cement ratios ranging

1-0555

from 0.45 to 0.7. Using nC2 S = 1 in eq. (4) yields n H ,n = 1.51 , and using nC3S = 1 in eq. (4) yields n H ,n = 2.37 . So it follows that (x y) equals 0.49 and 0.63 for the C2S and C3S/alite reactions, respectively. For C2S the agreement with the aforesaid values of Kantro et al. (1966) is excellent, for C3S/alite it is very good as well. Note that Locher (1966) experimentally found n H ,n = 2.66 (i.e. the coefficient of C3S in eq. (1) being 0.21 instead of 0.187). This value yields (x y) = 0.34, this figure is in better agreement with the aforementioned value in Kantro et al. (1966). To quantify the two reaction products, a value for the C/S ratio, x in eqs. (9) and (10), now needs to be specified. An average value of 1.7 is often found and generally accepted (Brunauer et al. (1958) and Brunauer and Greenberg (1960), Copeland and Kantro (1964), Locher (1966), Young and Hansen (1987), Taylor (1997)) and also taken here, yielding the following approximate reactions C2S + 1.51 H C1.7SH1.21 + 0.3 CH C3S + 2.37 H C1.7SH1.07 + 1.3 CH (11) (12)

Saturated state The next step is to study the hydration of C2S and C3S in saturated conditions. From eq. (8) it readily follows that (2 x + y = 3.58) and (3 x y = 4.23) for the C2S and C3S reactions, respectively. Again invoking x = 1.7, eqs. (9) and (10) yield the following approximate reactions C2S + 3.58 H C1.7SH3.28 + 0.3 CH C3S + 4.23 H C1.7SH2.93 + 1.3 CH (13) (14)

In Brouwers (2003) the model of Powers and Brownyard and the reactions (11)-(14) are furthermore used to derive the volume fraction of C-S-H in the hydrated product, the density of C1.7SH3.2 (2.27 g/cm3), and the porosity of C-S-H (33%). Reactions of aluminate and sulphate phases In contrast to the calcium silicate phases, which basically form CH and C-S-H, the aluminate phase can react in several ways and to more different hydration products. In some reactions calcium sulphate and carbon dioxide may also be involved. In contrast to the C-S-H gel, which is poorly crystalline to amorphous, the water content of most conceivable hydration products, which are (quasi-)crystalline, are known, as well as their specific volumes and other data. The retained water of these crystalline reactions products is therefore of other nature than the gel water of the C-S-H. Likewise the CH, these crystalline hydration products are impermeable and in the paste their gel water (they dehydrate upon drying) can, unlike the C-S-H gel space, not be considered as gel space. In this Section the reaction of aluminate and sulphate phases is discussed, and it will be assume a priori that the ferrite phase does not react with the sulphate phase. In the following Section the reason of this major assumption is motivated. The results of this Section also support this hypothesis.

1-0556

In Taylor (1997) a comprehensive review of the hydration of C3A is presented, which is briefly summarised here. In water, the pure compound will mainly hydrate to hydrogarnet (C3AH6). In a real water-cement system at room temperature, in the presence of calcium sulphate and calcium hydroxide, the aluminate hydrate C4AH13-19 (formed from C3A, CH and H), and aluminate sulphate phases are formed, such as C4A S H12-14 (monosulphate) and C6A S 3 H32-36 (ettringite). Firstly, ettringite is formed, which is then partly converted into mono-sulphate. This mono-sulphate is very susceptible to carbonation, resulting in the formation of ettringite and C4A C 0.5 H12 (hemi-carbonate). Only a few tenths of mass percent CO2 on cement mass is already sufficient to prevent the formation of mono-sulphate. When sufficient CO2 is present, hemi-carbonate can be replaced by the CO2 richer mono-carbonate (C4A C H11). For the present analysis is not important whether hemi-carbonate or mono-carbonate are formed, as the water retention by both substances is almost identical, so that here attention is restricted to C4A C 0.5 H12. In this paper, only the reaction (products) in saturated state is discussed, in P-dried state can be find elsewhere (Brouwers (2003)). The major conclusion of the analysis of Pdried products is that the sulphate in the cements used by Powers and Brownyard most likely has been present as anhydrite and/or in the clinker. Saturated state In saturated state, the amount of retained water in relation to the moles of C3A and C S is provided by eq. (8) as (approximately)
n H ,d = 21nC3 A + 2nC S

(15)

In the absence of C S ( nC S = 0) and in the presence of CH, only the aluminate hydrate and hemi-carbonate can be formed. The high water retention renders the formation of hemicarbonate unlikely, so from eq. (15) it readily follows that in such case C3A + CH + 21 H C4AH22 (16)

This high water retention is higher than given by Taylor (1997), but in line with the C4AH21 found by Le Chatelier in the 19th century (quoted by Schwiete and Ludwig (1969)). Fischer and Kuzel (1982) synthesised aluminate hydrates and with XRD, IR and DTA, also measured the presence of C4AH19 and an indication of hydrates with even more (interlayer) water. At a relative humidity lower than 65%-88%, the hydrate is readily dehydrated to C4AH13 (Schwiete and Ludwig (1969), Fischer and Kuzel (1982)). In case sulphate is present too, ettringite and mono-sulphate will be formed as well (Taylor (1997)). In case of carbon dioxide present, carbonation will occur, whereby monosulphate is the most unstable and will first react to hemi-carbonate (C4A C 0.5 H12), and possibly also to mono-carbonate (C4A C H11) in case sufficient C is present. Sources of C are mixing water, ambient air and C C in the cement (Kuzel (1996), Taylor (1997)). The Pdried analysis has revealed that the sulphate phase does not carry water (Brouwers (2003)).

1-0557

Furthermore, in saturated conditions, most likely C4A C H11, C4A S H14 and C6A S 3 H36 are formed (Dosch et al. (1969), Pllmann et al. (1989), Kuzel (1996), Pllmann (2003)). This information is used to formulate the following additional reactions: C3A + CH + C C + 11.5 H C4A C 0.5 H12 C3A + C S + 14 H C4A S H14 C3A + 3 C S + 36 H C6A S 3 H36 (17) (18) (19)

The formation of hemi-carbonate has been imposed, but with the formation of monocarbonate almost the same amount of water is involved (11 moles of H instead of 11.5). Implicitly we assume that the initial mono-sulphate is able to consume all CO2 and hence, that the ettringite and aluminate hydrate (and portlandite) do not carbonate. Now it is possible and interesting to compute which part of the C3A is converted into C4AH22, C4A C 0.5 H12, C4A S H14 and into C6A S 3 H36. Mole balances of C3A, S and H yield
nC4 AH 22 + nC nC
4 A S H 14 4 A C 0.5 H 12

+ nC

4 A S H 14

+ nC

6 A S 3 H 36

= n C3 A = n H ,d = 21nC3 A + 2nC S

(20) (21) (22)

+ 3nC

6 A S 3 H 36

= nC S + 14nC
4 A S H 14

21nC4 AH 22 + 11.5nC

4 AC 0.5 H 12

+ 36nC

6 A S 3 H 36

respectively, whereby eq. (15) has been inserted. This set of three equations contains 4 unknowns. Most likely, as discussed in the foregoing and as in most real pastes, full carbonation of mono-sulphate has occurred. Accordingly, henceforth it is assumed that the data of Powers and Brownyard (1948) are based on carbonated pastes. According to Kuzel (1996) this situation is most likely in practice as CO2-free conditions are very difficult to achieve and only a little CO2 is needed to prevent the formation of mono-sulphate. The amounts formed products in that case follow from eqs. (20)-(22) as
nC4 AH 22 = nC3 A 37 2 nC S nC3 A nC S 57 3

(23) (24) (25) (26)

nC

4 A S H14

=0

1 n C A S 3 H = nC S 6 36 3 18 1 nC AC 0.5 H = nC S n C S 4 12 57 3

Note that the sum of all these hydration products formed from C3A, times their water retention (eqs. (16)-(19)), indeed complies with eq. (15). For the full carbonation of mono-sulphate, nC C / nC S apparently needs to be 1/6 (see eq. (26)), or mC C / mC S = 12%. So, related to the mass of calcium sulpate, only 12% of calcite is required for full carbonation. As the sulphate constitutes only a few percent of the

1-0558

cement mass, a few tenths of a percent of calcite in relation to cement mass is required, which is in line with findings in literature (Kuzel (1996), Taylor (1997)). Ettringite and hemi-carbonate are stable (in contrast to mono-sulphate) and delayed ettringite formation is not likely (Kuzel (1996)) when mono-sulphate is absent. From eqs. (23)-(26) the following quantitative conclusions follow. The moles of hemi-carbonate (plus possible mono-carbonate) and ettringite formed only depend on nC S , but their ratio being constant
nC
4 A C 0.5 H12

nC A S 3 H
6

36

18 1 19

(27)

For full carbonation of the mono-sulphate, as assumed to prevail during the experiments of Powers and Brownyard (1948) and in most practical cases, the molar ratio of hemicarbonate and ettringite is about unity. The hemi-carbonate formed can be expressed in amount of ettringite formed, as this latter product is always present (in case sulphate is present). The ratios appear not to depend on nC3 A nor nC S . It is interesting now to determine the ratio of tetra calcium aluminate hydrate formed to ettringite formed. Combining this eq. (23) with eq. (25) yields:
nC4 AH 22 nC A S 3 H
6 36

=3

nC3 A nC S

nC A 2 37 3 3 57 nC S 3

(28)

Apparently, to bind all sulphate (i.e. so that no aluminate hydrate is formed), the number of moles C3A ( nC3 A ) needs to be 2/3 the number of moles of C S ( nC S ). This minimum value for nC3 A / nC S (= 2/3) is larger than the value that one would expect if all C3A would be converted into ettringite, namely 1/3. This is due to the fact that mono-carbonate is formed simultaneously in a carbonating environment, which is also binding C3A. If no (or partial) carbonation takes place in the cement paste, then another line of reasoning needs to be followed and alternative expressions are obtained for the quantities of the four products formed. The quantities then depend on the degree of carbonation (Brouwers (2003)). Reaction of ferrite phase The clinker phases C3S, C2S and C3A were already known to Le Chatelier in 1887. The ferrite phase, on the other hand, was discovered much later, in 1928 (Steinour (1961)). The reaction of this phase, which is the most impure one, in a cement-water system still raises questions. First of all, distinction needs to be made between the reaction of the pure phase at one hand, and the reaction of the impure phase as found in cement clinker, reacting in the presence of other clinker phases (Taylor (1997)). The first situation is prevailing in a

1-0559

laboratory setting and the results reported (Collepardi et al. (1979), Fukuhura et al. (1981)), but the latter situation is obviously of more relevance to real cement pastes. As said, in many publications it is reported that similar products can be formed as formed by the C3A reactions, such as C4AH13-19, C4A S H12-14 and C6A S 3 H32-36, whereby A can be partly substituted by F (Schwiete and Ludwig (1969), Taylor (1997)). The partial- replacement of A by F in ettringite, yielding approximately C6A0.75F0.25 S 3 H31, however, has only been found in pure ferrite and C S hydration experiments (e.g. Fukuhara et al. (1981)). The replacement has not been found in mono-sulphate yet. The formation of FH3 has not been detected either in real cement pastes, though it could be amorphous and therefore difficult to detect. These findings render the similarity between C3A and C4AF reactions questionable (Taylor (1997)). In real cement pastes, on the other hand, F has been found in hydrogarnets. Using XRD, Kantro et al. (1960) and Copeland et al. (1960) suspected a phase with a composition in the vicinity C6AFS2H8 in hydrated cement. In a pure system of C3S and C4AF, Schwiete and Iwai (1964) found that with increasing C3S/C4AF, the S/F ratio (x) in the formed C6AFSxH12-2x also increased, but x not exceeding a value of 2 (at room temperature). For x > 1.5 the hydrogarnet was furthermore stable to sulphate attack (practically no transition to ettringite). Using XRD and EMPA, Taylor and Newbury (1984) confirmed the presence of a hydrogarnet of composition near C6A1.2F0.8S2H18 in hydrated OPC. Based on SEM and TEM, Rodger and Groves (1989) suggested a composition of C6A0.6F0.6S2Hx in OPC and OPC-fly ash blends. Paul and Glasser (2000) investigated OPC pastes that were prolonged cured (8.4 years) at 85 oC, which has limited predictability towards cement hydration at ambient temperature, of course. But using XRD and DTA/TGA, they estimated that the observed hydrogarnet in their paste is having a composition near C6A1.26F0.51M0.46S2.8 S 0.58H5.59. By Brouwers (2003) it is demonstrated that the water retention results of Powers and Brownyard (1948), both in P-dried and saturated state, render the occurrence of C3A analogue reactions (to form F substituted calcium aluminate/ferrite hydrate, mono-sulphate and ettringite) not very likely, being in line with the literature summarized above. Consequently, here the possible formation of the F and S containing hydrogarnets will be discussed only.
P-dried state The formation of hydrogarnet can be modelled as follows

C4AF + x C2S + (2x + y 2) H C6AFSxHy + (2x 2) CH C4AF + x C3S + (3x + y 2) H C6AFSxHy + (3x 2) CH In dried state, for the hydrogarnet furthermore holds 2x + y = 12 And from eq. (4) follows

(29) (30)

(31)

1-0560

n H ,n = 1.51nC2 S + 2.37 nC3 S + 5.74nC4 AF

(32)

Applying eq. (31) to eq. (29) yields n H ,n = 10 . Furthermore, nC2 S = x , nC3S = 0 and
nC4 AF = 1 . Substitution of these values and n H ,n = 10 into eq. (32) yields x = 2.8, and eq.

= 1 , and substitution into eq. (32), now yields x = 3.1. with nC2 S = 0 , nC3S = x and n With eq. (31) then y = 5.8 is obtained. The computed compositions of the hydrogarnets are very close (an S/F ratio of about 3), irrespective if they are formed from C2S or C3S, and their composition comes close to composition that has been measured by the authors mentioned above. The analysis here supports the idea that C4AF reacts with C2S and/or C3S to form a hydrogarnet. The absence of F and S containing phases could be due to the fact that C3A reacts more rapidly than C4AF, and might consume all sulphate in the cement system. As previous authors found an S/F ratio of 2 at ambient temperatures, the C4AF and calcium silicates most likely react to C6AFS2H8. Note that if the coefficient 5.74 appearing in eqs. (4) and (32) would only have been 25 % larger (i.e. a water retention of about 7.14 moles of water per mole C4AF instead of 5.74), x 2 would have been obtained from the present analysis indeed.
C4 AF

(31) then yields y = 6.4. Applying eq. (31) to eq. (30) yields n H ,n = 10 + x , combining this

Saturated state Using the values for x (2 or 3) as obtained from the analysis of the P-dried state, the hydrogarnet reactions in saturated state can modelled by equations (29) and (30), whereby y is yet unknown. Similarly as for the aluminate hydrate in the previous Section, which appeared to retain extra interlayer water, the hydrogarnet will be allowed to retain more water than prescribed by eq. (31). From eq. (8) follows the following water retention is saturated state
n H ,d = 3.58nC2 S + 4.23nC3S + 12.74nC4 AF

(33)

Applying x = 2 to eq. (29) yields n H ,d = 2 + y , nC2 S = 2 , nC3S = 0 and nC4 AF = 1 . Substitution of these values into eq. (33) yields 2 + y = 19.92 and hence, y = 17.92 (i.e. C6AFS2H17.92 is formed). Applying x = 2 to eq. (30) yields n H ,d = 4 + y , nC2 S = 0 ,
nC3S = 2 and nC4 AF = 1 . Substitution of these values into eq. (33) yields 4 + y = 21.2 and

hence, y = 17.2 (C6AFS3H17.2 is formed). In comparison to dried hydrogarnet (C6AFS2H8), saturated hydrogarnet per mole apparently retains 9 to 10 moles of water more. When the same procedure is also carried out with x = 3, then C6AFS3H19.48 and C6AFS3H18.43 follows from the reactions with C2S or C3S, respectively. Also in these cases the saturated hydrogarnet holds 12 to 13 more moles of water than in dried state (C6AFS3H6). This high water retention could be caused by the formation of very small crystals and/or imperfectly ordered structures (Taylor (2002)). Another explanation could be that the hydrogarnet is formed by the reaction of an unstable F-substituted gehlenite hydrate (C2A0.5F0.5SH8) and CH, resulting in aforesaid super-saturated hydrogarnets. The analogue

1-0561

reaction of C2ASH8 and CH is explained by Damidot and Glasser (1995) using a phase diagram, and it was observed by Locher (1960) as well. A conceivable reaction would be: C4AFS2H16 + 2 CH C6AFS2H18 (34)

In other words, the formation of the gehlenite hydrate would be an intermediate step in overall reactions (29) and (30), and that the involved water is retained in a super saturated hydrogarnet. However, it should be noted that this water retention has never been measured, nor the presence of F-substituted gehlenite hydrate, and therefore remains speculative. But the water retention experiments by Powers and Brownyard (1948) on real cement pastes and the analysis there from here, confirm the formation of the afore-mentioned hydrogarnet with high water retention. That this water retention has not been measured yet might be caused by the prevailing experimental conditions: mostly the paste is dried to some extent, so being not saturated anymore. For the same reason usually also C4AH13, C4A S H12 and C6A S 3 H32 are detected, and not C4AH19-22, C4A S H14 and C6A S 3 H36, which can persist only in saturated state (Dosch et al. (1969), Fischer and Kuzel (1982), Pllmann (1989), Kuzel (1996), Pllmann (2003)). Note that C4AF consumes C2S and/or C3S to form the crystalline hydrogarnet, and consequently, less C2S and/or C3S is available for reactions (9) and (10), and less C-S-H gel is formed. On the other hand, with the formation of hydrogarnet both C2S and C3S produce per mole more CH than with the formation of C-S-H (no C4AF involved). This can easily be verified by considering eqs. (29) and (30) with x = 2, and eqs. (9) and (10) with x = 1.7. In the case of C2S, per mole of this clinker phase, 1 mole of CH is produced instead of 0.3, in the case of C3S, per mole of this clinker phase, 2 moles of CH are produced instead of 1.3. Sufficient CH needs to be produced for enabling reaction (16). Summary In the present article the water retention results as measured by Powers and Brownyard (1948) have been used to investigate the reactions of the clinker phases alite, belite, aluminate, calcium sulphate and ferrite in real cement pastes. Most likely, the ferrite reacts with the alite and belite to form a hydrogarnet. The remaining calcium silicates react to C1.7SH3.2 (when saturated) and CH. The sulphate seems to react solely with the aluminate. In case of carbonation, which is mostly the case in real cement paste, hemi-carbonate, ettringite and tetra calcium aluminate hydrate is formed. This insight leads to the following modified retention nH,d = 4.5 nC3S + 3.5 nC2 S + 21 nC3 A + 13 nC4 AF + 2 nC S (35)

This equation provides the moles of retained water in saturated state as a function of the moles of the individual clinker phases present in the cement. Using eqs. (2) and (3) this equation can be written as

1-0562

wd/c = 0.355 x C3S + 0.366 x C2S + 1.401 x C3A + 0.482 x C4AF + 0.265 xC S

(36)

Eq. (35) supports the formation of C1.7SH3.2 by both C3S and C2S, insofar these clinker minerals are not involved with C4AF in the formation of hydrogarnet of composition C6AFS2H18, and this is matched by the factor 13 in the equation. In practical application, e.g. in concrete, the bulk of the cement paste is isolated and can effectively be considered as sealed. Hydration stops at a relative humidity of about 80% (Powers (1947)), and water retention under this sealed condition is interesting to determine as well. As long as water is present the reactions proceed to form the saturated reaction products. When this water is consumed, it is likely that the saturated hydration products partly dehydrate with falling relative humidity. This enables the cement hydration to proceed till a final relative humidity in the entire paste of about 80% is achieved. So, it is necessary to assess the water retention of all hydration products formed at this relative humidity to assess the minimum required water for full hydration in a sealed system. It can be expected that the saturated hydrogarnet (C6AFS2H18), aluminate hydrate (C4AH22) and ettringite (C6A S 3 H36) lose their most loosely bound water and are readily dehydrated to C6AFS2H8, C4AH13 and C6A S 3 H32, respectively. By Brouwers (2003) it is derived that at 80% relative humidity the C-S-H is dehydrated to C1.7SH2.5. The portlandite (CH) and hemi-carbonate (C4A C 0.5 H12) retain their water. To assess the quantity (moles) of aluminate hydrate, hemi-carbonate and ettringite formed, eqs. (23)-(26) can be used. All information condenses in the following water retention relation at 80% relative humidity: nH,d80 = 3.8 nC3S + 2.8 nC2 S + 12 nC3 A + 4.4 nC4 AF + 6.5 nC S Using eqs. (2) and (3) this equation can be written as wd80/c = 0.300 x C3S + 0.293 x C2S + 0.800 x C3A + 0.163 x C4AF + 0.860 xC S (38) (37)

In eq. (37) the factor 4.4 pertaining to C4AF is introduced to match the factors pertaining to C3S/C2S, this former factor is required to stochiometrically permit the formation of C6AFS2H8 and 4CH/2CH. The factors pertaining to C3A and C S follow from eqs. (23)(26) whereby the proper water retention is used (eqs. (16)-(19)). To compare the water required for full hydration in saturated and in sealed condition, in eqs. (36) and (38) the composition of a typical CEM I (ASTM Type III) are substituted: x C3S = 0.61, x C2S = 0.15, x C3A = 0.06, x C4AF = 0.10 and xC S = 0.04, yielding wd/c = 0.40 and wd80/c = 0.33. In other words, a relatively small drop in humidity (from 100% to 80%) leads to a substantial reduction in water required for complete hydration. The moles of formed hydration products in the paste have been quantified, and are related to the mass (moles) of the clinker phases originally present in the cement. As seen, this new reaction model enables relations between clinker composition and water retention at different humidity. Furthermore, each product formed affects the macroscopic properties such as permeability, strength, creep, chemical shrinkage and binding capacity of the paste.

1-0563

Quantification of these products in the paste facilitates the future engineering with OPC and application of OPC-based materials. Acknowledgements The author is grateful for the advice given by late Dr. H.F.W. Taylor, Emeritus Professor of Inorganic Chemistry, University of Aberdeen, U.K. and by Dr. Dr. H. Pllmann, Professor of Mineralogy/Geochemistry, Martin-Luther University of Halle-Wittenberg, Germany. He furthermore wishes to thank the following persons and institutions for providing copies of references: Mr. W.J. Burns from the Portland Cement Association (PCA), Skokie, Il, U.S., and Dr. M. Schneider and Mrs. B. Bumer from the Verein Deutscher Zementwerke e.V. (VDZ), Dsseldorf, Germany. Literature Brouwers, H.J.H. (2003), The work of Powers and Brownyard revisited: microstructure and reactions of hydrating Portland cement paste, Research Report, University of Twente, to be published. Brunauer, S., Kantro, D.L. and Copeland, L.E. (1958), The stoichiometry of the hydration of beta-dicalcium silicate and tricalcium silicate at room temperature, Bull. 86, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from J. Am. Chem. Soc., Vol. 80 (1958), p. 761-776. Brunauer, S., and Greenberg, L.E. (1960), The hydration of tricalcium silicate and dicalcium silicate at room temperature, Bull. 152, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from Proc. 4th ISCC, Vol. 1, p. 135-165. Brunauer, S. and Kantro, D.L. (1964), Chapter 7 in The chemistry of cements, Vol. 1, edited by H.F.W. Taylor, Academic Press, London, U.K.. Collepardi, M., Monosi, S., Moriconi, G. and Corradi, M. (1979), Tetracalcium aluminoferrite hydration in the presence of lime and gypsum, Cem. Concr. Res., Vol. 9, p. 431-437. Copeland, L.E., Kantro, D.L. and Verbeck, G. (1960), Chemistry of hydration of Portland cement, Bull. 153, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from Proc. 4th ISCC, Vol. 1, p. 429-465. Copeland, L.E. and Kantro, D.L. (1964), Chapter 8 in The chemistry of cements, Vol. 1, edited by H.F.W. Taylor, Academic Press, London, U.K.. Damidot, D. and Glasser, F.P. (1995), Investigations of the CaO-Al2O3-SiO2-H2O system at 25 oC by thermodynamic calculations, Cem. Concr. Res., Vol. 25, p. 22-28.

1-0564

Dosch, W., Keller, H. and Zur Strassen, H. (1969), Written discussion, Proc. 5th ISCC, Vol. 2, p. 72-77. Fischer, R. and Kuzel, H.J. (1982), Reinvestigation of the system C4A.nH2OC4A.CO2.nH2O, Cem. Concr. Res., Vol. 12, p. 517-526. Fukuhura, M., Goto, S., Asaga, K., Daimon, M. and Kondo, R. (1981), Mechanism and kinetics of C4AF hydration with gypsum, Cem. Concr. Res., Vol. 11, p. 407-414. Hansen, T.C. (1986), Physical structure of hardened cement paste, a classical approach, Materials and Structures, Vol. 19, p. 423-436. Kantro, D.L., Copeland, L.E. and Anderson, E.R. (1960), An X-ray diffraction investigation of hydrated Portland cement paste, Bull. 128, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from Proc. ASTM, Vol. 60, Philadelphia, 1960, p. 1020-1035. Kantro, D.L., Weise, C.H. and Brunauer, S. (1966), Paste hydration of beta-dicalcium silicate, tricalcium silicate, and alite, Bull. 209, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from Symp. on Structure of Portland Cement Paste and Concrete, Highway Res. Board Spec. Rep. No. 90, Highway Research Board, Washington, 1966, p. 309-327. Kuzel, H.J. (1996), Initial hydration reactions and mechanisms of delayed ettringite formation in Portland cements, Cem. Concr. Composites, Vol. 18, p. 195-203. Locher, F.W. (1960), Hydraulic properties and hydration of glasses of the system CaOAl2O3-SiO2, Proc. 4th ISCC, Vol. 1, p. 267-276. Locher, F.W. (1966), Stoichiometry of tricalcium silicate hydration, Symp. on Structure of Portland Cement Paste and Concrete, Highway Res. Board Spec. Rep. No. 90, Highway Research Board, Washington, 1966, p. 300-308. Locher, F.W. (1975), Volumennderungen bei der Zementerhrtung, Sonderheft aus Zement und Beton, Heft 85/86, p. 1-4 (in German). Neville, A.M. (2000), Properties of Concrete (4th ed.), Prentice Hall/Pearson, Harlow, U.K.. Paul, M. and Glasser, F.P. (2000), Impact of prolonged warm (85oC) moist cure on Portland cement paste, Cem. Concr. Res., Vol. 30, p. 1869-1877. Pllmann, H., Kuzel, H.J. and Wenda, R. (1989), Compounds with ettringite structure, Neues Jahrbuch Miner. Abh., Vol. 160, p. 133-158. Pllmann, H. (2003), Private communications.

1-0565

Powers, T.C. (1947), A discussion of cement hydration in relation to the curing of concrete, Bull. 25, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from Proc. Highway Res. Board, Vol. 27 (1947), p. 178-188. Powers, T.C. and Brownyard, T.L. (1948), Studies of the physical properties of hardened portland cement paste, Bull. 22, Res. Lab. of Portland Cement Association, Skokie, IL, U.S, reprinted from J. Am. Concrete Inst. (Proc.), Vol. 43 (1947), p. 101-132, p. 249-336, p. 469-505, p. 549-602, p. 669-712, p. 845-880, p. 933-992. Rodger, S.A. and Groves, G.W. (1989), Electron microscopy study of ordinary Portland cement and ordinary Portland cement-pulverized fuel ash blended pastes, J. Am. Ceram. Soc., Vol. 72, p. 1037-1039. Schwiete, H.E. and Iwai, T. (1964), ber das Verhalten der ferritischen Phase im Zement whrend der Hydratation, Zement-Kalk-Gips, Vol. 17, p. 379-386 (in German). Schwiete, H.E. and Ludwig, U. (1969), Crystal structures and properties of cement hydration products (hydrated calcium aluminates and ferrites), Proc. 5th ISCC, Vol. 2, p. 37-69. Steinour, H.H. (1961), Progress in the chemistry of cement, l887-1960, Bull. 130, Res. Lab. of Portland Cement Association, Skokie, IL, U.S., reprinted from J. PCA Res. and Dev. Lab., Vol. 3, p. 2-11. Taylor, H.F.W. and Newbury, D.E. (1984), An electron microprobe study of mature cement paste, Cem. Concr. Res., Vol. 14, p. 565-573. Taylor, H.F.W. (1997), Cement chemistry (2nd ed.), Thomas Telford, London, U.K.. Taylor, H.F.W. (2002), Private communications. Young, J.F. and Hansen, W. (1987), Volume relationships for C-S-H formation based on hydration stoichiometries, Mat. Res. Soc. Symp. Proc., Vol. 85, p. 313-322. Author Dr. ir. H.J.H. Brouwers Department of Construction Technology (Bt) Faculty of Engineering Technology (CTW) University of Twente P.O. Box 217 NL 7500 AE Enschede

1-0566