Langmuir 1989,5, 1217-1222 emulsion systems is that it represents the minimum concentration of surfactant required to achieve the lowest possible oil-water interfacial tension under given conditions. Hence, it is a measure of surfactant effectiveness in tension lowering, therefore lowering the free energy required for emulsion formation. The importance of the spontaneous curvature of the monolayer in emulsions is demonstrated by the observation that the type of emulsion produced (i.e., O/W or W/O) when a two-phase oil/water system is shaken is generally the same as the microemulsion type produced upon phase separation a t equil i b r i ~ m . ' ~Hence, ~ ~ ~ the emulsification of a Winsor I system gives a water-continuous emulsion whereas a Winsor I1 system gives an oil-continuous dispersion. It is clear from this that emulsion type is not determined solely by the surfactant but is also affected by a range of variables which involve the nature and state of the oil and water phases, e.g., temperature, salt concentration,and oil type." The roles of the rigidity modulii ( K and K ) in emulsion stability are less certain. The coalescence of droplets may proceed via a dumbell-shaped transition state which involves a highly curved interface in the region where the droplets initially connect. The energy difference between this postulated transition state and the two initial droplets (i.e., the energy barrier to droplet coalescence) should be related to the energy required to bend the interface (against its natural curvature) and hence the film rigidity. It should be noted that these considerations may only be relevant in the case of emulsions stabilized by low molar mass surfactants. Other factors, which might include surface rheological properties, are likely to be important (42) Salager, J. L.;Maldonaldo, I. L.; Perez, M. M.; Silva, F. J.Dispers.
Sci. Technol. 1982, 3, 279.
1217
in many commercial emulsions, which are commonly stabilized by polymer surfactants. The above discussion indicates the relevance of the parameters estimated from microemulsion studies to important emulsion properties. Hence, these parameters may form a useful basis by which to classify surfactants and give a clearer understanding of the underlying principles of the HLB concept. It should be remembered, however, that these equilibrium properties cannot in the end completely account for emulsion stability, which is a kinetic property. With this in mind, we are currently extending these equilibrium studies to include dynamic measurements of the rates of microemulsion and macroemulsion droplet coalescence. Microemulsion droplet coalescence has been shown previously to be slowest a t the solubilization boundary corresponding to the lower temperature phase limit for W/O microemulsions stabilized by an ionic ~ u r f a c t a n t . ~Our ~ measurements for the C12E5/heptane water droplet coalescence indicate the rate is again slowest a t the solubilization boundary.44 In this case, however, this corresponds to the upper temperature single-phase limit. Macroemulsion stability has been shown to pass through a minimum at conditions corresponding to phase i n v e r s i ~ n .Estimation ~ ~ ~ ~ ~ of the monolayer properties in this type of study, together with measurements of the system dynamics, offers a promising approach in understanding emulsion stability. Registry No. CI2E5,3055-95-6;heptane, 142-82-5;tetradecane, 629-59-4. (43) Fletcher, P. D. I.; Howe, A. M.; Robinson, B. H. J.Chem. Soc., Faraday Trans. 1 1987,83, 985. (44) Clark. S.: Fletcher. P. D. I.: Ye. X.. unuublished results. (45) Bourrel, M.;Gracid, A.; Schechkr, R. S.; Wade, W. H. J. Colloid Interface Sci. 1979, 72, 161.
Sorption Studies of the Microstructure of Hydrated Cement F. G. R. Gimblett,* C. D. Lawrence,+and K. S. W. Sing Department of Chemistry, Brunel University, Kingston Lane, Uxbridge, Middlesex UB3 8PH, England Received August 9, 1988. In Final Form: January 20, 1989 Nitrogen and butane sorption data have been obtained on a wide range of hardened cement pastes. In general, it is believed that data from the two adsorptives provide a consistent and reliable description of the pore systems developed. When dried rapidly, hardened pastes contain slit-shaped pores whose widths lie between 2 and 4 nm for a range of paste densities. The hydration products approximate to platelike aggregations with overall average thicknesses between 30 and 1000 nm, depending on the waterlcement ratio and the degree of hydration. Slowly dried pastes have much reduced surface areas and contain wider, more symmetrical pores.
Introduction research is the prediction of long-term behavior in different environments, and in this context the development of resistance to diffusion for various molecular or ionic species is of particular importance to concrete durability.12 In practice, the number of significantly different composition used is very large; also, performance is known to depend on environ-
mental history. It is clear, therefore, that traditional laboratory tests are unable to provide sufficient information to characterize COI-Icrete field performance. In principle, investigation of the microstructure of hardened pastes is fundamental to the understanding of long-term Performance, and the influence of water vapor and the structural modifications introduced by temperature change, drying, mechanical stress, etc., are important
'Materials Science Department, British Cement Association,
(1) Tuutti, K. Corrosion of Steel in Concrete; Swedish Cement & Concrete Research Institute: Stockholm, 1982; p 468. (2) Lawrence, C. D. Tech. Rep. 544-Cem. Concr. Assoc. 1981, 24.
onegoal for
Wexham Springs, Slough, Berkshire SL3 6PL, England.
0743-7463/89/2405-1217$01.50/0 0 1989 American Chemical Society
1218 Langmuir, Vol. 5, No. 5, 1989
aspects which require study. All reported investigations agree that hardened cement is a porous material which contains a high proportion of calcium silicate hydrate gel. The latter is initially amorphous, although the increasing tendency of the material toward crystallization with aging or thermal treatment has a considerable influence on the overall p ~ r o s i t y . ~ Convincing methods for investigating the porosity of hardened cement pastes have been difficult to establish, since each method employed seems to provide a somewhat different picture of the microstructural details! Whereas mercury porosimetry has been used extensively in recent years, the applicability of this method is generally limited to systems in which the pores are greater than ca. 4 nm in diameter.6 For smaller pore sizes, sorption techniques have been favored for the study of internal surfaces and pore structures and applied to a wide range of materials. Even with hardened cement pastes, much work has been reported on the sorption of water vapor;6 however, it is possible that chemical interactions may have obscured the correct interpretation of the results in these cases. The sorption of nitrogen is a very well-established procedure for porous materials generally,' and the literature indicates that the shape of nitrogen isotherms on hardened cement pastes is responsive to changes in the conditions of formation and storage? By the same token, precautions must be taken to ensure that the necessary drying, outgassing, and cooling procedures do not produce misleading results.g The interactions which occur during sorption between nitrogen molecules and most solid surfaces are assumed to be purely physical in nature. However, detailed investigations of hardened cements reveal that, over the range of relative pressures 0 < p / p o < 0.4, several different shapes of isotherm may be obtained quite reproducibly.1° Surfaces in hydrated cements have some ionic character, and since the electrical quadrupole in the nitrogen molecule allows a degree of specificity in its interaction with such surfaces, it is conceivable that these changes in nitrogen isotherm shape may be attributed to alterations in the chemical nature of the internal surfaces.l' It has been suggested recently that the nitrogen moelcule may not be the most suitable sorptive for characterizing all microporous solids and that a number of carefully selected molecules should be used as probes for microstructure evaluation.12 The same argument may be applied to hardened cement pastes, especially those containing pores whose sizes are only marginally greater than micropores (ca. 2-nm diameter). In these cases, other sorptives should be studied in order to confirm structure evaluations based on nitrogen as a probe molecule. One such example is butane; its molecule contains no perma(3) See, for example: Gimblett, F. G. R.; Qui, M. U. Z. J. Colloid Interface Sci. 1988, 123, 148; 1988,125, 534. (4) Diamond, S. Hydraulic Cement Pastes: Their Structure and Properties, Proc. Sheffield Uniu. Conf., 1976;Cement & Concrete Association: Slough, 1976; p. 2. (5) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area & Porosity, 2nd. ed.; Academic Press: London, 1982; p 176. (6) Hagymassy, J., Jr.; Odler, I.; Yudenfreund, M.; Skalny, J.; Brunauer, S. J. Colloid Interface Sci. 1972, 38, 20. (7) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309. (8) Lawrence, C. D. Tech. Rep. 530-Cem. Concr. Assoc. 1980, 28. (9) Litvan, G. G. Cem. Concr. Res. 1976,6, 139. (10) Lawrence, C. D.; Gimblett, F. G. R.; Sing, K. S. W. R o c . 7th Int. Congr. Chem. Cement, Paris, 1980; Editions Septima: Paris, 1981;Vol. 111, Section VI, p 141. (11)Carruthers, J. D.; Payne, D. A.; Sing, K. S. W.; Stryker, L. J. J. Colloid Interface Sci. 1971, 36, 205. (12) Carrott, P. J. M.; Sing, K. S. W. In Characterization of Porous Solids; Unger, K. K., Rouquerol, J., Sing, K. S. W., Kral, H., Eds.; Elsevier: Amsterdam, 1988 p 86.
Gimblett et al. Table I. Chemical Analyses of Cements Employed comDosition. mass % cement SiOz CaO SO3 FeZ03 A1203 21.04 OPCl 2.36 5.48 64.37 2.35 21.02 OPC2 2.10 5.63 63.55 2.47 OPC3 19.37 5.74 3.98 63.33 2.76 OPC4 21.60 4.88 1.70 65.48 2.75 30.18 c2s1 3.50 1.10 62.20 0.14 26.01 0.19 C3S1 0.01 73.78 24.04 1.22 72.06 0.15 C3S2 0.50 23.70 2.10 2.08 70.10 0.05 c3s3 24.92 c3s4 2.87 0.06 72.07 composition, mass % cement C,S (alite) C,S (belite) C,A C,AF, SRPC1" 64.8 19.7 1.4 10.3 71.9 11.9 1.5 13.5 SRPC2" ~
Estimated by quantitative X-ray diffraction analyses.
nent electrical charge separation, and ita physical interactions with solid surfaces should simply involve nonspecific van der Waals' forces. For this reason, the use of a combination of butane and nitrogen sorption procedures as in the present study was felt to be a promising approach to the investigation of hardened cement paste microstructure, providing advantages associated both with a well-establishedtechnique and the use of accepted methods of interpretation. If it is possible to obtain consistent evidence for the presence of pores with diameters approaching the micropore range in this way, this may be of importance in the calculation of permeability and the prediction of concrete behavior. Experimental Section Materials. Cement paste samples have been prepared over a wide range of water/cement ratios and curing conditions. In the results presented below, their identity is contained within a code; thus, cement paste CEM1/0.4/200 was prepared from cement CEMl at a water/cement ratio of 0.4 and cured for 200 days a t 298 K under water. More complex storage conditions were applied to some specimens-for example, additional curing at 333 K for 2 days after 200 days a t 298 K is indicated by CEM1/ 0.4/200-2, and autoclave curing for 16 h a t 463 K is indicated by the addition of the letter A, i.e., CEM1/0.4/200A. Pastes containing AC and CR in their designation were taken from a research program investigating creep. Analyses of the cements are given in Table I; the laboratory cements were free from alkali-metal ions and generally free from calcium sulfate. Water used in the preparation of the cement pastes and in their subsequent cure was doubly distilled, while the BDH AnalaR methanol employed in solvent-displacement studies was dried by conventional methods.13 Nitrogen gas used throughout this work was of the "oxygen-free" variety (nominally 99.9% pure) produced by the British Oxygen Co. In thermogravimetric studies, any moisture present in the gas was removed prior to use by flowing through a cold trap cooled in an acetone/solid COz bath. In sorption studies, moisture was removed by condensing in a cold trap cooled with liquid nitrogen and slowly evaporating the gas into the volumetric apparatus. The n-butane was instrument grade (British Oxygen, 99.5% purity) and used without further purification. Sample Drying. Before sorption isotherms could be determined on the samples, which had been previously crushed and sieved, various drying and outgaeaing procedures were undertaken. Isotherms have been obtained on D-dried samples (samples evacuated for several weeks through a vapor trap cooled to 194 K with solid carbon dioxide), on samples subjected to rapid drying a t 373 K, and on freshly crushed systems placed in the sorption apparatus while still saturated with water as well as on samples (13) See, for example: Vogel, A. I. A Text-book of Practical Organic Chemistry, 3rd ed.; Longman: London, 1956; p 169.
Langmuir, Vol. 5, No. 5, 1989 1219
Sorption Studies of Hydrated Cement
........ ........methano! ........ .................. .... D-dried .:.
' i , Group A
L
..a.
I
r
o
u
D B
C
-Group
y //
..I.**
G
-Group
D
Group E Group F Group G Group H Group
0
200
400
600 Furnace
800
I
1000
temperature/'C
F i g u r e 1. Mass loss from sample OPC4/0.6/7 with increasing temperature for specimens outgassed at room temperature after D-drying and after soaking in methanol for 21 days. subjected to solvent replacement procedures using methano1.l' Thermogravimetric investigations employing a Stanton-Redcroft TG-750 instrument (sample size, 10 mg; heating rate, 10 "C min-'; nitrogen flow rate, 10 cm3 mi&) indicated that samples dried by different procedures contained different quantities of strongly bound water,15 with methanol-dried materials possessing greater amounts than simply outgassed systems (Figure 1). Measurement of Sorption Isotherms. An automatic gravimetric apparatus has been designed and constructed16to allow the determination of butane sorption/desorption isotherms at ca. 274 K. Nitrogen isotherms were measured at 77 K by using a conventional volumetric apparatus and techniques which have been described previously." The calibration of this apparatus was checked by means of a sample of colloidal silica TK800 from the SCI/IUPAC/NPL set of surface area standards,'8the isotherm on this nonporous material being used subsequently as a reference for the interpretation of the isotherms obtained on the wide range of hydrated cement specimens studied in this work.
0.1
0.3 0.4 Relative pressure
0.2
Figure 2. Summary of the nitrogen isotherm shapes observed on hardened cement pastes and on silica TK800, respectively. e 1 0
f
I Butane
I
isotherms
Group A
Results Isotherm Shapes. All nitrogen and butane isotherms measured were T y p e I1 in t h e BDDT c l a s s i f i ~ a t i o n , ~ ~ usually exhibiting hysteresis between the adsorption and desorption branches which disappeared (or almost disappeared) at p / p o values of 0.45. Closure of the hysteresis loops indicated that sorption was reversible a n d purely physical and that processes such as flexing of the pore walls of the adsorbent were unlikely during the sorption of either butane or nitrogen. Most of the b u t a n e isotherms were measured at 274 K; however, three were carried o u t at 270 K and three at 266 K. The reduced isotherms obtained o n t h e same sample at 266,270, and 274 K had exactly t h e
Group
Group C Group D _2___
1 0
(14) Swanson, J. W. In Characterization of Porow Solids; Gregg, S. J., Sing, K. 9. W., Stoeckli, H. F., Eds.; Society of Chemical Industry: London, 1979; p 339. (15) Dollimore, D.; Gamlen, G. A.; Mangabhai, R. J. In 2nd European Symposium on Thermal Analysis; Dollimore, D., Ed.;Heyden: London, 1981; p 485. (16) Lawrence, C. D. Tech. Rep. 554-Cem. Concr. Assoc. 1982, 26. (17) Gimblett, F. G. R.; Rahman, A. A.; Sing, K. S. W. J. Colloid Interface Sci. 1981, 84, 337. (18) Eyerett, D. H.;Parfitt, G. D.; Sing, K. S. W.; Wilson, R. J.Appl. Chem. Bwtechnol. 1974,24, 199. (19) Brunauer, S.;Deming, L. S.;Deming, W. S.;Teller, E. J. Am. 1940,62, 1723. Chem. SOC.
B
I
I
01
02
Group E
I
I
I
03 0 4 Relative pressure
Figure 3. Summary of the butane isotherm shapes observed on hardened cement pastes and on silica TKSOO, respectively. same shape and produced very similar BET surface areas. This was regarded as evidence t h a t sorption equilibrium h a d been attained. It should be noted t h a t imperfections in t h e initial apparatus, which allowed ingress of water vapor to t h e hygroscopic cement paste samples, resulted in t h e failure of some b u t a n e isotherms to a t t a i n reversibility below p / p o values of 0.45 during preliminary runs.
1220 Langmuir, Vol. 5, No. 5, 1989
A summary of nitrogen isotherm shapes over the relative pressure range 0 < p / p o < 0.4 is illustrated in Figure 2. In this figure, the isotherm measured on silica TK800 is labeled group A. It was found that the isotherm shape was determined partly by the method of drying adopted. Thus, when samples were outgassed directly from the watersaturated state, the isotherm shapes a t low relative pressures belong to groups B, D, F, G, or H. If an organic solvent was used to displace water from the samples, the isotherms tended to belong to groups C, E, or I. A corresponding summary of butane isotherm shapes is illustrated in Figure 3 (groups A-E, with the isotherm on silica TK800 now being incorporated in group E). A wide range of isotherm shapes was again obtained quite reproducibly, although no clear correlation existed in this case between the shape and the drying procedure adopted. The absence of such a correlation suggests that any changes in surface chemistry occurring during the various drying procedures can only be identified by nitrogen sorption studies. Physical effects such as the occurrence of a relationship between the micropore volume retained and the drying procedure, which should be revealed by both butane and nitrogen isotherm shapes, are apparently not operating. BET Surface Areas and Analyses of Isotherms. Nitrogen BET plots' for isotherms on hydrated cements were linear over the relative pressure range 0.035 < p / p o < 0.28. Apparent surface areas, based on the outgassed sample masses, lay between 0.4 and 183.3 m2 g-l. As established previously for calcium silicate hydrate gels,2o the particular value for a given sample was found to be influenced considerably by the drying procedures employed. The mean BET "C" constant for hydrated cements was 82 with a standard deviation for the population of 42; C constants were significantly reduced when samples were dried by organic solvent replacement procedures. Butane BET analyses were carried out over the relative pressure range 0.1 < p / p o < 0.4. McClellan and Harnsberger21have quoted that the surface area assumed for the adsorbed butane molecule which produces data most consistent with the nitrogen BET surface areas is 0.444 nm2, and this value has been used to calculate specific surface areas in this work. Initial experiments indicated that surface areas tended to decrease when samples were held in the apparatus for extended time periods; however, reproducible data were achieved after modifications to apparatus design had been undertaken. Such surface areas had values which lay between 4.0 and 140.9 m2 g-l, being influenced yet again by the manner in which the samples had been dried. A shallow "knee" was exhibited by all the isotherms measured, indicating that interaction energies between butane molecules and the surfaces of hydrated cements are small. As expected, the resulting C values were low and extended over the range 3.6-30, averaging 11.6 with a standard deviation for the population of 4.8. These low values must raise some doubts concerning the reality of BET surface areas as derived from butane isotherms. Despite this reservation, however, all the butane and nitrogen BET surface areas measured were quite reproducible (to within f2.5 m2 g-l). Comparison of these surface areas indicated that, in general, those determined with butane were approximately three-quarters those for (20) Gimblett, F. G. R.; Sing, I(.S. W.; Mohd. Amin, Z . Proc. 7th Int. Congr. Chem. Cement, Paris, 1980, Editions Septima: Paris, 1981; Vol. 11, Section 11, p 225. (21) McClellan, A. L.; Harnsberger, H. F. J. Colloid Interface Sci. 1967,23, 511.
Gimblett et al.
Nitrogen bet surface
Figure 4. Comparison of the butane and nitrogen BET surface areas (both in m2 g-') evaluated for a range of hardened cement paste samples. Table 11. BET Analyses on the Same Sample but with Different Sorptive8 at Different TemDeratures BET analyses, m2 g-' 1st: butane, 2nd: nitrogen, 3rd: butane, 274 K 77 K 274 K sample" C3S1/0.47/CR 137.1 168.8 140.9 C3S1/0.47/AC-7 127.3 121.9 121.9 OPC1/0.50/1081 67.1 70.9 68.2 C3S4/0.18/16h 51.5 56.7 50.9 C2S1/0.50/2267A 37.6 42.4 35.0 14.2 15.1 12.9 C3S4/0.18/4h
" All samples dried by solvent replacement procedures using methanol. nitrogen. Indeed, a rough correlation existed between the values as illustrated in Figure 4, suggesting that both types of molecule were exploring the same pore system. To check the effect of cooling on samples in determining their nitrogen isotherms at 77 K, butane sorption experiments were conducted both before and after nitrogen adsorption. The results obtained demonstrate that butane surface areas appear to be substantially unchanged by intermediate cooling of the samples to 77 K (Table 11). Analyses of the isotherms have been carried out by using the a, method,z in which measurements on cements pastes were compared with those on silica TK800. The nitrogen isotherm data show that capillary condensation occurred during adsorption for the following samples: autoclaved specimens, specimens dried a t 373 K, dense pastes subjected to D-drying, and some porous pastes subjected to D-drying. A downward deviation of the a,plot arose with nitrogen isotherms associated with the following: roomtemperature-cured specimens which had been outgassed directly from the saturated state in the sorption apparatus, room-temperature-cured specimens pretreated with methanol, and most porous pastes subjected to D-drying. (22) Gregg, S. J.; Sing, K. S. W. In Surface & Colloid Science; Matijevic, E., Ed.; Wiley: New York, 1976; Vol. 9, p 231.
Langmuir, Vol. 5, No. 5, 1989 1221
Sorption Studies of Hydrated Cement
Table 111. Microstructure Development during Hydration As Estimated from Butane Sorption Experiments hydraulic apparent surface area, uptake at p J p o = 0.995, radius of pores, FHH plot sample' m2g-l (10-5) g i g nm asplot shapeb %'' value 2.29 C3S4/0.18/4h 14.2 3528 4.1 SIP 4.57 C3S4/0.18/16h 51.5 4077 1.3 P P 6.78 1.1 3009 45.7 c3s4j0.18ji 5.83 1.1 P 891 13.7 C3S4/0.18/ 550 1.36 5.7 C 2874 7.5 C3S1/0.47/1 5.17 1.4 P 5085 59.9 c3slJ0.47J7 4.92 P 1.3 7084 91.5 C351/0.47/28 4.29 P 1.4 5578 67.3 C3S1/0.41 f 42 4.08 P 1.5 6367 69.1 c3slJ0.47 180-I 1.65 3.9 C 3825 6.5 SRPC2/0.6/17h 2.94 P 2.2 8612 66.2 SRCP210.612.5 3.37 P 1.9 11433 SRCP210.6f 11 98.4 3.05 2.2 P 10871 83.1 SRCP2/0.6/ 16 2.55 P 12232 2.0 SRCP2f 0.6135 103.5
'All samples dried by solvent replacement procedures using methanol. * C = concave; S = straight; P
0
0
0
0
0
= convex.
2
8 8 Reduced I o r m o n ,or
0
r,andara
Figure 5. an plots derived from butane isotherms on sample C3S1/0.47 after different times of hydration.
Figure 6. agplots derived from butane isotherms on sample SRPC210.6 after different times of hydration.
Pastes pretreated with methanol exhibited increased downward deviations in their nitrogen a, plots, in agreement with the view that methanol pretreatment allows retention of a greater volume of micropores. Examples of the asplots derived from butane sorption isotherms are illustrated in Figures 5 and 6, where the data refer to samples whose microstructure developed during the early stages of hydration. The ability to condense liquid butane in very young pastes appears to be lost as the intergrain voids fill with platelike hydrate particles and the samples develop interparticle slit-shaped pores. The BET surface areas reach a maximum during early hydration in the case of pastes with a low waterlcement ratio but continue to increase (at least over the first 2 months) when pastes with a waterlcement ratio of 0.6 are considered (Table 111). The ability to condense butane during adsorption was enhanced by prolonged storage of samples in the sorption apparatus and by slow drying. Sometimes, capillary condensation was also seen to develop in dried samples during repetition of the sorption procedures. This suggests that structures where capillary condensation does not occur are probably the closest to those originally existing in the water-saturated pastes. Application of the Frenkel-Halsey-Hill (FHH) method of isotherm analysisz3often provides information on pore texture. Since the method is comparative, a reference isotherm on a nonporous sample is not required. Selections from the nitrogen adsorption data have been converted to FHH plots by fitting data over the relative pressure range 0.5 < p / p o < 0.99 to the equation In ( p / p o )= -K/(W/ W0,4)', where W and Wo.4 are the uptakes a t relative
pressures of p / p o and 0.4, respectively, while K is an empirical parameter linked to the adsorption energy of the first layer of adsorbate. Nitrogen isotherms on silica TK800 exhibit linear FHH plots, with s = 2.8. The data for hydrated cements which have been dried rapidly give s values which are larger than that expected theoretically (s = 3.0), indicating that sorption has been restricted at high relative pressures as would be found in a system containing slit-shaped micropores.24 In systems which show capillary condensation during adsorption (as estimated from a, analyses), the FHH plots give s values less than 3, suggesting the development of a system containing wider pore sizes. The FHH equation has also been applied to the data arising from the butane sorption isotherms. The range of linearity in these plots was more restricted than for the nitrogen data, although linear regions could usually be identified. The value of s for butane sorption on silica TK800 was only 1.69, being in approximate agreement with data quoted by PierceB for sorption of CFzClzon the same adsorbate. As for the nitrogen data, the value of s for rapidly dried systems was again found to be very high (within the range 4-7), and again the simplest explanation appears to be that restricted sorption occurred a t high relative pressures.
(23) Pierce, C. J. Phys. Chem. 1960, 64, 1184.
Discussion It is clear that the interactions of nitrogen and butane molecules with dried hardened cement pastes a t 77 and 274 K, respectively, occur via purely physical processes. (24) Carrott, P. J. M.; McLeod, A. I.; Sing, K. S. W. In Adsorption at the Gas-Solid and Liquid-Solid Interface; Rouquerol, J., Sing,K. S. W., Eds.; Elsevier Scientific Publishing Company: Amsterdam, 1982; p 403.
1222
Langmuir 1989,5, 1222-1225
The general similarity of the data arising from the sorption of these two adsorptives supports the view that similar pore spaces are being explored and that a satisfacotry analysis of the microstructure of the pastes is thereby possible. In a previous preliminary analysis,l0we have suggested that the simplest interpretation of the nitrogen data on rapidly dried pastes is that slit-shaped pores are present with widths that range from micro to macro. This model is supported by the more detailed examination presented in this report. However, despite the use of two different physical adsorptives, it is not possible to assess precise values for the total mesopore volumes from the type I1 isotherms obtained, although structural developments during early hydration seem to produce an increasing volume of mesopores in the samples. For pastes with a low water/cement ratio, a peak in the BET surface area occurred a t intermediate ages. More sophisticated pore size analyses of these paste samples have been reported elsewhere by one of us;25these show that for rapidly dried pastes the cumulative pore surfaces are much smaller than the BET surface areas. This supports the view that a considerable proportion of the pore volume is present in the micropore size region. For slowly dried material, agreement between cumulative mesopore surface areas and BET surface areas is good. If nitrogen and butane sorption techniques are capable of describing a representative fraction of the pore system in hardened cement pastes, then it is possible to construct a qualitative picture. The total uptake for pastes at relative pressures near 0.995 increases during early hydration even though the total pore volume decreases; thus pastes contain varying proportions of macropores which are not filled with liquid adsorbate under the experimental conditions. The average hydraulic radii of meso- and micropores (pore volume at p / p o = 0.995/BET surface area) for rapidly dried pastes of widely differing total mesopore volume are similar and lie between 1 and 2 nm; i.e., the
average pore slit widths lie between 2 and 4 nm. Since the surface areas for the majority of pastes lie between 5 and 150 m2 g-l, a simple calculation of the average thickness of the platelike hydrate can be made if a material density is assumed; for a hydrate density of 2500 kg m-3, average thicknesses are found to be between 1000 and 33 nm, depending on the waterlcement ratio and the degree of hydration. When slowly dried pastes are considered, the above hydraulic radius is increased considerably to about 7 nm. A picture of the hydration process also emerges from these studies. Thus, relatively large symmetrical voids between hydrating cement grains gradually fill with relatively well-dispersed filmy or platelike hydrate. At some point these form strongly bonded agglomerates, as continuing hydration increases the packing density; this leads to a reduction in the free surface and a corresponding increase in the strength of the material. Other evidence for the slow aging of the bonds between hydrate particles has been obtained from creep experiments.26 In addition, these systems do not behave as simple brittle solids, and it has been suggested that the sliding of adjacent particles relative to each other is possible under mechanical loading.27 Slow drying appears to cause coalescence of the dispersed filmy structure formed initially, leading to the reappearance of directly interconnected larger voids and the disappearance of finely dispersed slit-shaped pores. The relevance of this microstructural transformation to permeation processes when slow wetting and drying are possible is clear, and indeed evidence for the greatly increased permeability of cement pastes to liquids after initial drying is available in the literature.2s Quantitative estimates of permeability based on the above pore model are required, however, to enable the experimental values to be checked. Registry No. N2, 7727-37-9; butane, 106-97-8.
(25)Lawrence, C. D. In Principles and Applications of Pore Structural Characterization,Roc. Milan RILEMICNR Symp., 1983; Haynes, J . M., Rossi-Doria, P., Eds.; Arrowsmith: Bristol, 1985;p 339.
(26)Parrott, L.J. Ph.D. Thesis; University of London, 1973. (27)Hobbs, D. W.Ph.D. Thesis; University of Surrey, 1973. (28)Powers, T.C.;Copeland, L. E.; Hayes, J. C.; Mann, H. M. J . Am. Concr. Inst. 1954,51, 285.
Constant-Composition Study of Dicalcium Phosphate Dihydrate Crystal Growth in the Presence of Poly(acrylic acids) Zahid Amjad T h e BF Goodrich Company, Specialty Polymers and Chemicals Division, Avon Lake Technical Center, Avon Lake, Ohio 44012 Received September 20, 1988. I n Final Form: May 10, 1989
The constant-composition method has been used to study the influence of low (2100-20000) molecular weight poly(acry1ic acids) on the kinetics of crystal growth of dicalcium phosphate dihydrate (DCPD) on DCPD seed crystals at pH 6.00 and 37 O C . The results indicate that poly(acry1ic acid) concentration and polymer molecular weight markedly affect the growth rate. In addition, the kinetic data reveal that at a constant polymer concentration the growth rate increases with increasing molecular weight from 2100 to 20 000. The fit of the Langmuir adsorption model to the experimental data supports a mechanism of inhibition through molecular adsorption of inhibitor ions on the surface of growing crystals. Introduction The importance of calcium phosphates as the major constituents of teeth and bones and of pathological and industrial mineral deposits has stimulated extensive research on the precipitation of these comp~unds.l-~De0743-7463/89/2405-1222$01.50/0
pending upon the precipitation conditions, i.e., level of supersaturation, pH, and ionic medium, the following (1) Koutsoukos, P. G.; Amjad, Z.; Tomson, M. B.; Nancollas, G. H. J . Am. Chem. SOC.1980,102,1553.
0 1989 American Chemical Society
