EQUILIBRIA IN THE SYSTEM CALCIUM HYDROXIDECALCIUM SULPHATE-ALKALI SULPHATE-WATER AT 5-30°C, AND THEIR RELEVANCE TO PORTLAND CEMENT HYDRATION IN THE PRESENCE OF SUPERPLASTICIZERS by Ellis M. Gartner1, Serge Sabio1 and Jean-Philippe Perez2 1
Lafarge Laboratoire Central de Recherche, 38291 France. E-mail:
[email protected] 2 LRSS, University de Bourgogne, Dijon, France. E-mail:
[email protected]
ABSTRACT: We report new experimental data on the system calcium hydroxide-calcium sulphate-alkali sulphate –water over the temperature range 5° – 30°C. Experiments have been performed using either potassium or sodium as the alkali metal, and results for the pure system at 30°C are in very good accord with those of Hansen and Pressler (1947). The principal solid phases present are portlandite, gypsum and syngenite. Conductimetric measurements show that, in the “admixture-free” system, equilibrium is usually obtained within only a few tens of minutes. The solubility product of syngenite decreases with decreasing temperature, while those of portlandite and of gypsum increase, displacing the intersections between the solubility isotherms. In the presence of organic admixtures of the type generally used as “superplasticizers” (SP) in portland cement concretes, some minor changes in apparent solubility products are observed that can be attributed to weak complexation of calcium ions and/or the neutralisation of hydroxide ions by the SP molecules . 1.0 INTRODUCTION There is great interest in understanding the mechanisms by which high-range water-reducing admixtures, commonly known as superplasticizers (SP), are able to “improve concrete rheology” (i.e., to reduce the effective viscosity of fresh concrete at constant mix composition.) It is now well-established that SPs function by adsorption onto the surfaces of the particles in concrete, producing repulsive forces, which can be either electrostatic or steric in nature [1]. These repulsive forces can greatly reduce the tendency of cement grains and other fine particles in the concrete to flocculate, giving a suspension which is much better dispersed than the SP-free system. However, we still do not know exactly how typical SPs are adsorbed on the surfaces of cement grains, and how the performance of a given SP is related to its chemical composition. There are also important unanswered questions regarding the effectiveness of specific cement-SP combinations, since no two cements or SPs are identical: some couples work better than others for certain types of concrete. In order to better understand cement-SP interactions, we must be able to model the chemistry of the aqueous phase in fresh concrete, or at least the equivalent liquid extracted from cement pastes or mortars which are believed to simulate the chemistry of fresh concrete. It has been known for many years that, in the admixture-free cement-water system, the aqueous phase contains principally alkali metal (potassium and sodium) ions, calcium ions, sulphate ions and hydroxide ions [2]. Other species, such as silicate, aluminate, iron compounds, etc., are present at much lower concentrations, except in very unusual circumstances. In many respects, the system can be regarded as a perturbed lime-gypsum solution, that is, it is close to equilibrium with crystalline portlandite and gypsum. In practice, it is usually slightly supersaturated relative to one or both of these phases [3]. The main independent variable is thus the amount of soluble alkali in the cement. Portland cement clinkers
can contain alkali sulphates at levels of up to about 2% by mass, and these highly soluble salts dissolve very rapidly when the cement is mixed with water. Some additional alkali may also be released into solution at early ages (as hydroxide) by the hydration of the aluminate phases. The amount of soluble alkali, and its potassium/sodium ratio, is a characteristic of a particular cement. In order to determine exactly how an organic additive, such as a SP, perturbs the composition of the aqueous phase in cement systems, it is not sufficient simply to run a control mix with no admixture. This is because such admixtures can influence both the equilibrium parameters and the kinetics of cement hydration; and, in studying fresh concrete, we are dealing with a reacting system far from equilibrium, in which many reactions occur simultaneously, several of which may be perturbed by the presence of the additive. Ideally, we should first study reference systems in which a true chemical equilibrium can be obtained, in order to obtain the influence of the additive on the equilibrium system, before studying the reacting system. We have chosen to do this by starting with the simple portlandite-gypsum system. The compositions of alkali solutions in equilibrium with portlandite and gypsum at 25°C and 30°C were established by Hansen and Pressler (H&P) over 55 years ago [4], but, as far as we are aware, no study has yet been made on how this system responds to perturbation by water-soluble organic polymers. 2.0 EXPERIMENTAL In the present study, we chose to work at 30°C, 20° and 5°C. 30°C was chosen so as to have data in common with those of H&P, 20°C was chosen because it is commonly used as a standard in lab experiments on cement and concrete, and 5° was chosen a “practical” lower limit. We chose to use solid alkali sulphates as the source of alkalis, since they are easier to weigh in air than the hydroxides used by H&P; thus, we approached equilibrium from a different direction. We also chose a set of commercial superplasticizers that span the range of types commonly used in concrete. For each equilibrium assemblage, we generally began by adding 6.75g of reagent-grade calcium hydroxide (portlandite) and 0.75g of reagent grade calcium sulphate dihydrate (gypsum) to 75ml of either de-ionised water or prepared solution in a 100ml Pyrex flask (which minimised the dead volume once the flask was sealed with a glass stopper). The prepared solutions contained dissolved superplasticizers, or alkali sulphates, or both, in order to give the desired initial concentrations of alkali metal ions and superplasticizers in the liquid phase. In no case did the amount of alkali sulphate or SP added approach the titration capacity of the 6.75 g of portlandite, thus ensuring that there was always some solid portlandite and gypsum present. [Note: since the alkali was provided as a sulphate, the ratio of solid portlandite/gypsum in the equilibrium mixtures decreased with increasing alkali concentration, but in cases where significant amounts of syngenite formed gypsum too was consumed, and in these cases more than 0.75g was used.] The sealed flasks were immersed in water baths controlled to within ± 0.1°C and agitated gently for at least 48 hours. In order to be able to follow the approach to equilibrium, the conductivities of the solutions were measured frequently, and were observed to approach steady-state values very rapidly (within a few minutes) in most cases, although some SPs appeared to retard this process slightly. Nevertheless, in order to be absolutely sure of equilibrium, we held all of the suspensions at the desired temperature for at least 48 hours before filtering and analysing any of the mixtures. The suspensions were allowed to settle for a short period before samples of the supernatant were taken and filtered rapidly under an elevated pressure of nitrogen (to avoid carbonation). The resulting filtrate samples were weighed and immediately titrated with 0.1N HCl to pH 7.0, to stabilise them against carbonation and to obtain an estimate of the hydroxide concentration. The stabilised solutions were analysed for sodium and potassium by flame photometry, for calcium by
complexometry, and for sulphate by turbidimetric titration against a solution of BaCl2. All three techniques were conducted following the procedures of EN 196, and the typical coefficients of variation (in-lab repeatability) were conservatively assumed to be 1.3% for the alkali metals, 1.2% for calcium and 2% for sulphate (by taking the higher of either the typical values quoted in the European norm or those obtained in our own calibrations). We also conservatively assumed a coefficient of variation of 5% for hydroxide. (Note: A second series of solutions, including those prepared at 5°C, was analysed slightly differently, and no hydroxide titrations were done on them.) Four different commercial SPs were chosen for this study; one was of the MSFC class, (a modified melamine sulfonate formaldehyde condensate containing sulphanilic groups,) and the other three were members of the “PCP” class (comb polymers based on polycarboxylic acid polyoxyethylene esters.) The dosages used were based on the concentrations that were found to be necessary to give about the same initial slump in typical superplasticised concretes. In a few cases, the liquid phase was analysed for total SP by the total organic carbon method (TOC), conducted by an outside laboratory following the EN 1484 procedure, with a typical coefficient of variation of 2%. 3.0 RESULTS 3.1 Equilibrium in the absence of superplasticizers All of the results obtained at 20°C are shown in Figure 1, and the results obtained at 5°C and 30°C are shown in Figure 2. The ‘2nd series’ results in these figures were obtained by a different operator than the first series, but there is reasonably good agreement, as can be seen in the graphs. (Note that the results shown in figure 2 are all 2nd series except for those with Na at 30°C.) 800 Ca x 10, (R=K), (2nd series) SO3, (R=K), (2nd series)
700
Ca x 10, without portlandite, (R=K) SO3 without portlandite, (R=K)
Calcium (x10) or Sulphate, mM
600
Ca x 10, (R=K) SO3, (R=K) Ca x 10, (R=Na)
500
SO3, (R=Na) Ca x 10, (R=Na), (2nd series) 400
SO3, (R=Na), (2nd series)
300
200
100
0 0
100
200
300
400
500
Total dissolved alkali as oxide [R2O], mM
Figure 1. Results at 20°C
600
700
800
Also included in Figure 1 are a set of data obtained in the absence of portlandite, i.e. for the simple equilibrium between gypsum and potassium sulphate at 20°C. As expected, they show lower calcium concentrations and higher sulphate concentrations than the solutions in simultaneous equilibrium with portlandite at equivalent alkali concentrations. 800 Ca x 10, R=Na, 30°C 700
SO3, R=Na, 30°C Ca x 10, R=Na, 5°C
Calcium (x10) or Sulphate, mM
600
SO3, R=Na, 5°C Ca x 10, R=K, 5°C
500
SO3, R=K, 5°C 400
300
200
100
0 0
100
200
300
400
500
600
700
800
Total dissolved alkali as oxide, [R2O], mM
Figure 2. Results at 5° and 30° C Table 1 gives the concentrations measured in the first series of experiments, without organic admixtures. (Note: In this case, Na+ concentrations were not measured but rather assumed constant in the runs with added K2SO4, and vice-versa for K+ in the runs with added Na2SO4.) The data at the highest sodium concentrations show a rather poor anion/cation balance, but are still within the probable range of error quoted for our analytical methods. Below [R2O] = 200mM, they generally show good agreement (at 30°C) with the data obtained by H&P [4], and there is also relatively little difference between our data obtained at 20°C and their data obtained at 25°C. However, at higher alkali concentrations there are some differences, which can be ascribed to two different effects: I.
The formation of syngenite, (K2SO4•CaSO4•H2O), in the system with R=K. This was also observed in H&P’s work. The onset of syngenite formation leads to decreases in potassium, calcium and sulphate concentrations below the portlandite-gypsum equilibrium curves, and this onset occurs at lower K2O concentrations at 20°C than at 30°C. The effect of syngenite formation at 20°C can clearly be seen in Figure 1 for the second series of data with potassium; similar results were obtained in the first series but we found that the amount of gypsum used in the initial mixtures had been set to low to compensate for syngenite formation, and it was increased in the 2nd series to ensure that there was always some residual gypsum at equilibrium. (Note: We also observed that the onset of syngenite formation occurred at lower alkali concentrations as the temperature decreased, but did not attempt to locate the invariant point precisely in the experiments reported here.)
II.
Anhydrite formation. There is also a divergence in the results at the highest sodium concentrations; H&P’s final data point at 30°C and [Na2O] = 375 mM gives much lower concentrations of both calcium and sulphate ions than our data, and there is little reason to believe that any double salt (sodium-calcium-sulphate) forms at this sodium concentration. This result supports the hypothesis of Gartner et al [6], who suggested that anhydrite may have formed in H&P’s experiments at high sodium concentrations, because the alkali would decrease the activity of water in the system relative to the alkali-free system, and thus should lower the temperature of the gypsum-anhydrite equilibrium point, which is close to 40°C in the absence of other salts. There was no evidence for anhydrite formation (checked by IR analysis of the solid residue) in any of our experiments; however, the rates of nucleation, growth and dissolution of anhydrite are known to be very slow relative to other calcium sulphates, and they may thus, under these conditions, be sensitive to the presence of anhydrite nuclei in some sources of gypsum [7]. Table 1. ‘First Series’ of results obtained at 20° and 30°C in the absence of admixtures Initial concentration of Alkali sulfate used, mM Ca++ None added 20°C 66.5 30°C 63.5
Analyses of equilibrated solutions (meq/L) Na+ K+ SO4= OH-
Sum
2.9 2.9
0 0
20.2 24.5
49.0 39.5
0 2
Na2SO4 @ 20°C 49.4 98.4 203.9 404.7 798.2
39.0 32.4 30.7 31.0 36.7
98.2 204 409 804 1512
0* 0* 0* 0* 0*
72.2 141 314 677 1385
70.6 104 136 144 160
-6 -9 -10 13 4
K2SO4 @ 20°C 49.6 100.0 202.1 401.0 595.4
39.2 33.8 32.5 19.7 8.8
2.9* 2.9* 2.9* 2.9* 2.9*
98.2 208 407 552 948
75.6 146 308 421 666
65.6 100 126 157 283
-1 -1 8 -3 11
Na2SO4 @ 30°C 49.4 98.4 203.9 404.7 798.2
38.4 33.1 30.0 31.6 35.9
86.4 191 391 816 1566
0* 0* 0* 0* 0*
73.6 151 311 667 1410
61.8 81.6 109 134 147
-11 -8 2 47 45
The data for the SP-free solutions at 20°C were used to calculate simple ionic solubility products for the compositions Ca(OH)2, CaSO4 and K2Ca(SO4)2, based on the simplifying assumption that the solutions contained only the simple ions K+, Na+, Ca++, OH and SO4=. Results are shown in Figure 3. The first two of these three simple ionic solubility products represent the saturation values for portlandite and for gypsum, respectively, given that we had deliberately made sure that excess of these two solids was always present. The third solubility product represents the saturation values for syngenite, which is known to form at potassium sulphate concentrations of above about
165 mM at 20°C in the simple K2SO4-CaSO4 system [5]. This product is only shown for the three solutions with values of [K2O] well above 200mM, and formation of syngenite in these three mixes was confirmed by IR and DTA analyses of the filtered solids. Simple ionic products are, of course, not expected to be constant with varying ionic strength. All of the observed simple ionic solubility products increase with increasing alkali concentration, as would be expected, since the activity coefficients of all of the ions decrease with increasing ionic strength over the range of interest. At first sight it seems surprising that the portlandite ionic product shows a smaller dependence than that of gypsum on alkali concentration. However, this is probably due to the two moles of water included in each mole of gypsum, because the activity of water decreases as alkali concentration increases, (which should then require an increase in the simple ionic solubility product to compensate). The syngenite solubility product shows by far the greatest alkali concentration dependence, which is not surprising since it is a quintuple ionic product, and the simple molecular formula of the solid phase also contains one molecule of water.
Simple ionic solubility products .
1000
100
10
Ca(OH)2 mM^3/10^3 CaSO4 mM^2/10^2 K2Ca(SO4)2 mM^5/10^10
1 1
10
100
1000
10000
[R+], mM
Figure 3. Simple ionic solubility products at 20°C as a function of alkali metal concentration. 3.2 Equilibrium in the presence of superplasticizers The compositions of Na2SO4 solutions equilibrated with portlandite and gypsum in the presence of the four SPs are summarised in Table 2. Apart from the small increases in [Na+] observed with added SPs, due to the presence of sodium as the principal cation in all of the SPs, there is clearly relatively little change in the overall solution compositions due to the presence of SPs at the dosages used here. This result can be seen more clearly in the effects of SP addition on the simple ionic solubility products for portlandite and gypsum, shown in Figures 4 and 5, respectively. [Note: In the presence of SPs, hydroxide ion concentrations were estimated based on a calculation for electrical neutrality, deliberately neglecting the anionic nature of the SP. The results were generally close to those obtained by the titration of the solutions to pH 7 but presumably contains some error due to the influence of the SP on the charge balance. Estimated errors in the ionic products are shown on the graphs.]
Table 2. Equilibrium concentrations at 20°C with and without superplasticizers (* Note: SP concentration is expressed as grams of dry extract per litre of solution.) Initial [Na2SO4]
Without SP
MSFC @7.39g/L
PCP1 @3.37g/L
PCP2 @3.22g/L
PCP3 @3.19g/L
11.1 37.9 68.0 153.2
10.5 37.7 71.1 156.4
36.7 21.9 18.4 16.4
35.1 21.4 20.3 15.9
3.8 112 210 425
3.2 107 212 429
1130 1140 1220 1260 1720
1260 1260 1260 1310 1610
[SO4=], (mM) 0 49.4 98.4 203.9
10.1 36.1 70.4 157.3
12.3 36.6 71.4 151.4
11.7 36.4 72.7 152.6
[Ca++], (mM) 0 49.4 98.4 203.9
33.3 19.5 16.2 15.4
34.9 22.6 18.8 16.6
35.0 21.2 18.2 16.3
[Na+], (mM) 0 49.4 98.4 203.9
2.9 98.1 203 409
9.6 119 221 419
5.9 111 222 429
Total organic carbon, (mg/L) 0 10 49.4 0 98.4 0 203.9 0 SP solution as introduced:
980 1630 1570 1690 2910
1020 980 1200 1250 1570
Generally, we observed that all four SPs, at the dosages used here, increased the simple portlandite solubility product by about 30-50% over the whole range of alkali concentrations used, but increased the simple gypsum solubility product by only 20-30%. There are several possible explanations for such increases, the most likely ones being (a) the decrease in activity coefficients due to the ionic strength increase caused by the large polyanionic SP molecules, (b) the partial neutralisation of hydroxide ions by the SP, and (c) weak complexation of calcium ions by the polymers. If calcium ion complexation were the whole explanation, the portlandite and gypsum solubility products should both have been influenced in exactly the same proportions. However, the possible errors in estimation of hydroxide ion concentration make it impossible to say if this was or was not truly the case; partial neutralisation of hydroxide by organic admixture could lead to significant errors in the calculated ionic products. Very similar results to those given above were also obtained with Na2SO4 at 30°C, and also with K2SO4 at 20°C at concentrations below the syngenite precipitation point. Thus, it can be assumed that, under conditions typical of fresh portland cement concrete, most common SPs have only a minor influence on the “effective solubility product” of gypsum and portlandite, and thus on the ionic equilibria in the liquid phase.
2500
[Ca][SO4], mM ^2
2000
1500 Without SP 1000
MSFC @ 7,39g/L PCP1 @ 3,37g/L
500
PCP2 @ 3,22g/L PCP3 @ 3,19g/L
0 0
100
200
300
400
500
[R+], mM Figure 4. Gypsum ionic product as function of alkali concentration and SP additions
4,0E + 05
.
3,5E + 05
[Ca][OH][OH], mM^3
3,0E + 05 2,5E + 05 2,0E + 05 W ithout S P 1,5E + 05
MS FC @ 7,3 9g/L
1,0E + 05
P C P 1 @ 3 ,37 g/L P C P 2 @ 3 ,22 g/L
5,0E + 04
P C P 3 @ 3 ,19 g/L
0,0E + 00 0
1 00
2 00
3 00
4 00
[R+], m M
Figure 5. Portlandite ionic product as function of alkali concentration and SP additions
50 0
70%
[email protected]/L 60%
[email protected]/L
[email protected]/L
50%
[email protected]/L 40% 30% 20% 10% 0% 0
50
100 150 Na2SO4 concentration, mM
200
250
Figure 6. SP adsorption by portlandite-gypsum mix in Na2SO4 solutions at 20°C The amounts of SP adsorbed by the residual solid phases at equilibrium were estimated from their reduced TOC concentrations, as given in Table 2. The degrees of adsorption were quite significant, as shown in Figure 6, despite the relatively high water/solids ratio used. It is notable that all of the SPs showed partial adsorption on the solids in the suspensions, with the strongest adsorption (66%) occurring for MSFC in the absence of additional alkali sulphate. This suggests that MSFC is probably adsorbed significantly by portlandite. The other three SPs showed weaker adsorption but a similar trend towards less adsorption with increasing sulphate concentration. [Note: it should be remembered that the proportion of gypsum in the residual solid phases also increased with increasing sulphate concentration due to the way that the solutions were prepared]. 4.0 CONCLUSIONS •
We have extended the data of Hansen and Pressler [4] to 5° and 20°C, and obtained indirect evidence that anhydrite may have formed in their experiments with high sodium concentrations at 30°C. Our experimental techniques are also confirmed by good agreement with their results. There is indirect evidence that the invariant point syngenite-gypsum-portlandite occurs at a K2O concentration of about 200 mM at 20°C and at lower K2O concentrations as the temperature decreases, but we did not attempt to measure it precisely in these experiments.
•
Both MSFC-type and PCP-type superplasticizers, at concentrations typical of those found in the aqueous phase of superplasticised portland cement concretes, show only minor effects on the equilibrium solubility products of portlandite or gypsum in alkali solutions. The principal effect observed is a slight increase in the apparent solubility products of both solid phases, which can be interpreted as being due to either weak calcium ion chelation by the dissolved polymers, or to a decrease in ionic activity coefficients, or both.
•
Weak adsorption on the residual solid phases (a portlandite-gypsum mixture) was observed for all of the SPs tested. This was especially notable at low sulphate concentrations, from which it is deduced that the adsorption probably occurs mainly onto the surface of portlandite, which was the preponderant solid phase in the residue after equilibrium in those cases. This adsorption was significantly stronger for MSFC-type than for PCP-type SP polymers. However, the adsorptions observed for all of the SPs tested were relatively small compared to those typically observed in portland cement pastes at early ages, (when very little portlandite is present). This implies that portlandite is not the principal SP-adsorbing phase in hydrating portland cements.
5.0 ACKNOWLEDGEMENT We would like to give special thanks to Thierry Rouillon for his contribution to the experimental work. 6.0 REFERENCES [1] Uchikawa, H., Hanehara, S. and Sawaki, D. Effect of Electrostatic and Steric Repulsive Force of Organic Admixture in the Dispersion of Cement Particles in Fresh Cement Paste, 10th International Congress on the Chemistry of Cement, Gothenburg, 1997, vol. 3, paper 3iii001. [2] Lawrence, C.D. Symposium on the Structure of Cement Pastes and Concrete, Highway Research Board Special Report No. 90, Washington DC, 1966, pp. 378-91 [3] Gartner, E. and Skalny, J. Computation of Solubility Relationships for Hydrating Cement Systems, 8th International Congress on the Chemistry of Cement, Rio de Janeiro, 1986, Vol. III, pp. 244-50 [4] Hansen, W.C. and Pressler, E.E. Solubility of Ca(OH)2 and CaSO4•2H2O in Dilute Alkali Solutions, Industrial and Engineering Chemistry vol. 39, 1947, pp.1280-82 [5] Hill, A.E. Ternary Systems. XIX. Calcium Sulfate, Potassium Sulfate and Water, Journal of the American Chemical Society, vol. 56, 1934, pp.1071-78 [6] Gartner, E.M., Tang, F.J. and Weiss, S.J. Saturation Factors for Calcium Hydroxide and Calcium Sulfates in Fresh Portland Cement Pastes, Journal of the American Ceramic Society, vol. 68, 1985, pp. 667-73 [7] Ridge, M.J. and Beretka, J. Calcium Sulphate Hemihydrate and Its Hydration, Reviews of Pure and Applied Chemistry, vol. 19, 1969, pp.17-44
