Nanoscale Experimental Investigation of Particle Interactions at the

Although a fully hydrated cement paste can exhibit a high compressive strength .... and the pH was measured with a high alkalinity electrode (radiomet...
0 downloads 0 Views 412KB Size
Download PDF
Langmuir 2005, 21, 7263-7270

7263

Nanoscale Experimental Investigation of Particle Interactions at the Origin of the Cohesion of Cement Ce´dric Plassard,† Eric Lesniewska,† Isabelle Pochard,‡ and Andre´ Nonat*,‡ Physics Laboratory LPUB CNRS UMR 5027, University of Bourgogne, Dijon F-21078, France, and Chemistry Laboratory LRRS CNRS UMR 5613, University of Bourgogne, Dijon F-21078, France Received February 18, 2005. In Final Form: May 31, 2005 Atomic force microscopy has been used to investigate the force at the origin of the cohesion of cement. The cohesion of cement grains is caused by surface forces acting between calcium silicate hydrate nanoparticles in interstitial electrolytic solution. Direct measurement of the interaction between two calcium silicate hydrate surfaces is performed in air and different aqueous solutions. In dry air, starting with the van der Waals forces, the interaction area between calcium silicate hydrate nanoparticles can be estimated. In electrolytic solution, the evolution of these forces is extensively dependent on both surface and solution chemistry. The roles of the calcium hydroxide concentration, pH, and ionic strength are investigated. The force measurements allow us to confirm the pre-eminence of ionic correlation forces in the cohesion of cement.

I. Introduction C-S-H is a hydrated calcium silicate phase. It is the major constituent of hydrated Portland cement paste and is responsible for its main physical and chemical properties (in cement chemistry notation C, S, and H refer to CaO, SiO2, and H2O, respectively).1 It is formed by a dissolutionprecipitation process during the so-called hydration of cement. When Portland cement is mixed with water, it begins to dissolve and produces a solution containing calcium, silicate, and hydroxide ions, which is quickly supersaturated with respect to C-S-H.2 C-S-H then precipitates on the surface of cement grains in the form of small sheets (typically 60 × 30 × 5 nm3).3,4 Simultaneously to this chemical evolution, cement particles form a network established a few minutes after mixing.5-7 Setting of the cement paste is due to the strengthening of the cement grains network by C-S-H nanoparticles precipitating at the contact points between the cement grains. The strength of the cement paste increases during the hydration process because of the augmentation of the number of contact points between the cement grains by the multiplication of C-S-H particles. Although a fully hydrated cement paste can exhibit a high compressive strength (more than 100 MPa), its tensile strength is quite low (about 2 MPa).8 It is probably due to the fact that the elastic limit of the material is small: the critical strain * To whom correspondence should be addressed. E-mail: [email protected]. Telephone: +33-3-80-39-61-66. Fax: +33-3-80-39-61-32. † Physics Laboratory LPUB CNRS UMR 5027. ‡ Chemistry Laboratory LRRS CNRS UMR 5613. (1) Taylor, H. W. Cement Chemistry, 2nd ed.; Academic Press: London, U.K., 1990; chapter 1. (2) Barret, P.; Bertrandie, D. J. Chim. Phys. 1986, 83, 765-775. (3) Gauffinet, S.; Finot, E. et al. Compte Rendu de l’Acade´ mie des Sciences de Paris 1998, 327, 231-236. (4) Garrault, S.; Nonat, A. Langmuir 2001, 17, 8131-8138. (5) Jiang, S. P.; Mutin, J. C. et al. Cem. Concr. Res. 1995, 25, 779789. (6) Jiang, S. P.; Mutin, J. C. et al. Cem. Concr. Res. 1996, 26, 491500. (7) Nachbaur, L.; Mutin, J. C. et al. Cem. Concr. Res. 2001, 31, 183192. (8) Maso, J. C. Le Be´ ton Hydraulique, Presses de l’Ecole Nationale des Ponts et Chausse´es: Paris, France, 1982; 275-293.

that cement paste can support without being destroyed is smaller than 10-4. It has been suggested that this is because the cohesion of the network, which is due to shortrange surface forces between C-S-H particles.9 The aim of this experimental work is to investigate these surface forces. C-S-H presents a lamellar structure close to the one of rare minerals, tobermorite and/or jennite.10-12 Its actual structure is still debated, but the tobermorite model seems to be the most probable.13 Both C-S-H and tobermorite have a layered structure based on a calcium plane flanked on each side by linear silicate “dreierketten” chains,14 as demonstrated by 29Si MAS NMR measurements.15-18 Therefore, the most developed surfaces of the C-S-H particles, which are supposed to interact together in a cement paste, consist of parallel linear chains of dimeric silicates eventually bridged together by another silicate tetrahedron. On average, each silicate has a silanol group that gives, according to the crystal cell of C-S-H deriving from the tobermorite structure, a density of 4.8 silanol/ nm2 on the surface.19 During the hydration process, the C-S-H surface is in contact with a calcium hydroxide solution saturated with respect to portlandite. In the absence of other ions, the solubility of portlandite is 20 mmol L-1 at 20 °C corresponding to pH 12.5. In cement paste, because of the presence of alkali ions, the pH may be greater than 13. In such conditions, all or part of the surface silanols may (9) Rueb, C. J.; Zukoski, C. F. Proceedings of the Materials Research Society Symposium; Pittsburgh, PA, 1991; 279-286. (10) Taylor, H. F. W. J. Chem. Soc. 1950, 3682-3690. (11) Gard, J. A.; Taylor, H. F. W. Cem. Concr. Res. 1976, 6, 667-677. (12) Taylor, H. F. W. J. Am. Ceram. Soc. 1986, 69, 464-467. (13) Nonat, A. Cem. Concr. Res. 2004, 34, 1521-1528. (14) Hamid, S. A. Z. Kristallogr. 1981, 154, 189-198. (15) Wieker, W.; Grimmer, A. R. et al. Cem. Concr. Res. 1982, 12, 333-339. (16) Lippmaa, E.; Gi, M. M. et al. Cem. Concr. Res. 1982, 12, 597602. (17) Cong, X.; Kirkpatrick, R. J. Adv. Cem. Bas. Mater. 1996, 3, 144156. (18) Klur, I.; Pollet, B. et al. Nuclear Magnetic Resonance Spectroscopy of Cement-Based Materials; Springer: Berlin, Germany, 1998; 119141. (19) Viallis-Terrisse, H.; Nonat, A. et al. J. Colloid Interface Sci. 2001, 244, 58-65.

10.1021/la050440+ CCC: $30.25 © 2005 American Chemical Society Published on Web 07/06/2005

7264

Langmuir, Vol. 21, No. 16, 2005

Plassard et al.

be ionized according to

≡SiOH + OH- / ≡SiO- + H2O

(1)

It is difficult to estimate the actual degree of ionization. According to a simple phenomenological model based on surface reactions, the degree of ionization of the surface silanol groups varies from about 10 to 90% when the calcium hydroxide concentration increases from 1 mM (≈pH 10) to 20 mM (≈pH 12.5).13,20 These values correspond to charge densities of respectively 0.08 and 0.69 C/m2. Although the highest value is probably overestimated, C-S-H particles in aqueous suspensions are charged, as demonstrated by electrophoresis measurements.19,21 An apparent charge reversal is observed when elecrophoretic measurements are made in calcium hydroxide solutions. The ζ potential is negative for low Ca(OH)2 concentrations and becomes positive when the concentration is above 2 mmol L-1. Particle interactions in cement paste result then from surface forces between charged particles in electrolytic solutions. According to the classical DLVO theory, the attractive part of the surface forces is due to van der Waals forces and the repulsive part is due to the pressure of the counterion cloud neutralizing the surface charges. However, because of the potentially high surface charge and the divalence of the calcium counterions, an attractive contribution because of the ionic correlations may occur in the osmotic pressure, which is not taken into account in the meanfield Poisson-Boltzmann equation. This was demonstrated by both anisotropic hypernetted chain (HNC) calculations and Monte Carlo simulations in a primitive model.22-26 Thus, this work aims to experimentally investigate if such non-DLVO forces are at the origin of the cohesion of cement pastes. The surface forces between C-S-H particles are measured when immersing in different electrolytic solutions similar to cement paste pore solutions. Varying the pH, the ionic strength, and the nature and concentration of the counterions in the bulk solution in equilibrium with the charged surfaces may be key to demonstrating the origin of the intervening forces. Force measurements between surfaces immersed in concentrated electrolytic solutions are not common. Kekicheff et al.27 demonstrated that it is possible to use an atomic force microscope (AFM) to measure forces between a mica surface and a silicon nitride tip immersed in concentrated calcium chloride solution. From this way, they experimentally provided evidence for the first time of ionic correlation attractions in the presence of divalent counterions. Finot et al.28 proved that AFM allows force measurements in symmetric systems such as gypsum/ gypsum immersed in calcium sulfate solutions by sticking a microcrystal on the tip. Preliminary studies on the C-SH/C-S-H system showed that these measurements were (20) Nonat, A.; Courault, A. C. et al. Cem.-Wapno-Beton 2001, 5, 184-191. (21) Nachbaur, L.; Nkinamubanzi, P. C. et al. J. Colloid Interface Sci. 1998, 202, 261-268. (22) Guldbrand, L.; Jo¨nsson, B. et al. J. Chem. Phys. 1984, 80, 22212228. (23) Kjellander, R.; Marcelja, S. J. Chem. Phys. 1985, 82, 21222135. (24) Valleau, J. P.; Ivkov, R. et al. J. Chem. Phys. 1991, 95, 520-532. (25) Delville, A. Langmuir 1994, 10, 395-402. (26) Pellenq, R.; Crespin, M. et al. Second Rilem Workshop on Hydration and Setting; Rilem Editions: Dijon, France, 1997; 63-86. (27) Ke´kicheff, P.; Marcelja, S. et al. J. Chem. Phys. 1993, 99, 60986113. (28) Finot, E.; Lesniewska, E. et al. Langmuir 2000, 16, 4237-4244.

also possible.29 In the present study, the validity of the measurements was first checked in dry air, in which case, the adhesion is essentially due to van der Waals forces. Then, analyses of the forces between C-S-H surfaces were carried out in different electrolytic solutions. II. Materials and Methods The experimental procedure consists of measuring the interaction forces acting between a probe and a substrate with an atomic force microscope in air or in aqueous solution. A judicious choice of the nature of probe and substrate are necessary, especially when the measurements are made in solutions. Indeed, certain experimental conditions prove to be essential: the probe and substrate should not react with the solution to ensure the stability and reproducibility of force measurements. Moreover, the surface of the substrate must be atomically flat to avoid roughness effects. A. Measurements Environment. All experiments were performed in a CO2-free glovebox to prevent carbonation of hydroxide solutions. Inside, a commercial AFM (Nanoscope IIIa, Veeco Institute, Santa Barbara, CA) was used. For studies in aqueous solutions, an adapted commercial fluid cell was used. The temperature of the surrounding wall was maintained at 25 °C and humidity controlled to avoid evaporation of the solution. In air, force measurements were also conducted under controlled atmosphere: low humidity (1%) was ensured by a decarbonated flux of dry helium gas. Measurement of the relative humidity (RH) was performed with a precision of 1% and a range of 0-100% (Quick 74880 hygrometer; EBRO GmbH, Ingoldstadt, Germany). B. Force Measurements. The interaction forces were directly measured between modified or nonmodified tips and atomically smooth surfaces, both prepared as described in the next section. Interaction measurements were performed with the usual calibration process to transform experimental cantilever deflection curves as a function of the vertical scanner displacement ∆z into force-distance curves.30 Using the slope of the retraction deflection curves in the contact region, the cantilever deflection is then converted into a force using Hooke’s law

F ) -k∆z

(2)

where k is the stiffness constant of the AFM cantilever used, determined by the resonant frequency method or thermal noise analysis. Force curves obtained give the force F (nN) against the tip-sample separation (nm). For each experiment, statistics of over 100 force measurements were established by recording 5 force curves on 20 different locations from each sample. C. Substrate Preparation. 1. Measurements in Air. The experiments in a controlled atmosphere have been performed using four different substrates and their most developed face (in brakets), namely, muscovite green mica (New York Mica Co., New York) face {0001}, natural gypsum crystal CaSO4‚2H2O (Lafarge Co., Mazan, France) face {010}, calcite CaCO3 (Lafarge Co.) face {10Ih4}, and C-S-H face {001}. The three first substrates were obtained by cleaving a single crystal just before the experiment and did not undergo any other transformation. The C-S-H substrate was prepared following the same procedure as for the measurements in solution as described below, rinsed in pure alcohol, and helium dry. Obviously, these samples will not use anymore for studies in solution. 2. C-S-H-Modified Substrates. C-S-H is obtained by hydrating tricalcium silicate, which is what occurs with Portland cement, or from silica in saturated calcium hydroxide solution, which is what occurs with Roman cements. However, tricalcium silicate and silica substrates continuously react with the solution to form C-S-H, and it is then not possible to get atomically flat surfaces in chemical equilibrium with the solution. A more effective choice for preparing C-S-H substrates was the {101 h 4} cleavage face of calcite CaCO3 (Iceland spar). The advantages consist of an atomically flat and nonreactive surface with respect to a Ca(OH)2 solution, which is actually the pore solution of (29) Lesko, S.; Lesniewska, E. et al. Ultramicroscopy 2001, 86, 1121. (30) Ducker, W. A.; Senden, T. J. et al. Nature 1991, 353, 239-241.

Nanoscale Experimental Investigation of Particle Interactions

Langmuir, Vol. 21, No. 16, 2005 7265

Table 1. Electrolytic Solutions Concentrations in mmol L-1 in Which Interactions between C-S-H Surfaces Were Measureda series 1 Ca(OH)2

series 2 CaCl2

series 3 Ca(OH)2 + NaNO3

[Ca2+]

pH

[Ca2+]

pH

[Ca2+]

[Na+]

pH

[Ca2+]

series 4 Ca(OH)2 + NaOH [Na+]

pH

series 5 Ca(OH)2 + CaCl2 [Ca2+]

pH

0.16 0.82 1.84 3.15 4.53 10.31 14.69 19.13

10.9 11.1 11.3 11.6 11.9 12.1 12.3 12.6

0.06 1.38 2.45 5.14 7.11 14.72 17.37

11.3 11.3 11.3 11.3 11.3 11.3 11.3

0.71 1.62 3.10 4.40 9.89 14.69

26.01 24.88 22.64 20.98 8.71 0.00

9.8 10.1 10.8 11.0 11.6 12.3

0.03 0.36 3.20 8.00 11.30 19.13

41.50 33.78 31.43 20.53 9.21 0.00

12.6 12.6 12.6 12.6 12.6 12.6

26.75 36.94 48.34 60.01 67.19 74.97

12.5 12.5 12.4 12.4 12.2 12.2

a Before force measurements, both the C-S-H substrate and tip are equilibrated in these solutions during 1 month. All solutions were saturated with respect to C-S-H by adding C-S-H powder into solutions and filtering them once equilibrium is reached. The different solutions were analyzed by inductively coupled plasma optical emission spectrometry (Vista pro, Varian).

cement. The freshly cleaved face of calcite was immersed in a sodium silicate solution ([SiO2]/[Na2O] ) 0.33 and [SiO2] ) 0.5 mol L-1 at pH 14.2). The following chemical reactions then occur: partial dissolution of calcite substrate

CaCO3 f Ca2+ + CO32-

(3)

Precipitation of C-S-H on the calcite cleavage plane by heterogeneous nucleation from calcium ions provided by the calcite dissolution and silicate and hydroxide ions contained in the solution according to

2Ca2+ + 3H2SiO42- / Ca2H2Si3O9‚xH2O + 2OH-

(4)

Figure 1. Scanning electron microscopic image of a silicon nitride tip after immersion during 48 h in a saturated calcium hydroxide solution. C-S-H nanocrystals are mainy precipitated on the edge and top of the pyramid.

The C-S-H growth on calcite was materialized by the aggregation of C-S-H nanoparticles of approximately 5 nm high and a developed face of 60 by 30 nm.29 Then, the calcite surface covered by C-S-H was immersed in various electrolytic solutions as listed in Table 1. These solutions were previously saturated with respect to C-S-H by adding C-S-H powder, previously synthesized as described elsewhere. Solutions undergo equilibrium for 1 week prior to being filtered through 0.1 µm Milliporesystem filtration. In such solutions, the silica concentration is too low (a few mmol L-1 or less) to displace the solubility equilibrium of calcite toward the formation of C-S-H. C-SH-covered calcite substrates were stored in these C-S-H equilibrium solutions for 1 month at 25 °C in closed containers to avoid carbonation. Calcium and silicate concentrations of each solution were analyzed by ICP-OE spectrometry (Vista pro, Varian), and the pH was measured with a high alkalinity electrode (radiometer analytical high alkalinity combined pH electrode). Calcium concentrations and the corresponding pH of the solutions are given in Table 1. After 1 month, the samples are studied by AFM in their equilibrium solution for topographical and force investigations. The calcite surface is covered by micrometric atomically smooth C-S-H domains formed by Ostwald ripening, thus smoothing out the influence of roughness on force measurements. It should be mentioned that C-S-H nanoparticles are still present in the troughs. Atomic resolution imagings were made on the C-S-H microdomains. For Ca(OH)2 concentrations ranging between 0.1 and 20 mmol L-1, we noticed a structural evolution of the C-S-H surface as a consequence of the change of C-S-H stoichiometries with the lime concentration in the equilibrium solution.31 D. Probe Preparation. For measurement in air, silicon cantilevers (Veeco Co., Santa Barbara, CA) of a high spring constant ranging from 2 to 30 N m-1 were used. For solution investigations, pyramidal silicon nitride Si3N4 tips fixed to commercial triangular cantilevers, double-side Au coated with a measured spring constant ranging from 0.06 to 0.6 N m-1 (Veeco Co.), were used. To obtain C-S-H coverage, the silicon nitride tips, which are naturally covered by a thin layer of silica (∼5 nm thick), were immerged in a large volume (V ) 50 mL) of a saturated calcium hydroxide solution during 48 h. Under these conditions, C-S-H precipitates on the extremity of the tip from

A. Force Measurements in Air. 1. Validation of the Methodology by Force Measurements between Model Surfaces. To check the validity of the experimental set up, measurements of adhesion forces between well-defined flat mineral surfaces and a silicone tip were first performed in dry air. In such conditions, the interaction forces between the mineral surface and the tip are dominated

(31) Plassard, C.; Lesniewska, E. et al. Ultramicroscopy 2004, 100, 331-338.

(32) Cleveland, J. P.; Manne, S. et al. Rev. Sci. Instrum. 1993, 64, 403.

the silicate ions provided by the dissolution of silica in the alkaline medium and from the calcium and hydroxide ions from the solution (Figure 1). After complete consumption of the silica layer, the probe was also rendered nonreactive, as long as the silicon nitride bulk was preserved from oxidation. The cantilever spring constants were determined by measuring the resonant frequency and using the following expression:32

k ) 2(πLf)3wxF3/E where L and w represent the length and width, respectively, of the cantilevers as supplied by the manufacturer, f represents the resonant frequency, E represents the Young’s modulus, and F represents the density of the material of the cantilever. The spring constant of cantilevers with a C-S-H-modified-probe was determined before making the C-S-H coverage. Because of the double-side Au coated, the cantilevers do not undergo any alteration during the immersion in saturated calcium hydroxide solution. Therefore, their spring constant remain unchanged with or without C-S-H coverage on the tip. This was checked by measuring the resonant frequency after tip immersion in lime solutions. The complete consumption of the silica layer leads to less than 5% of the resonant frequency variation. This variation is mainly due to the slight change in the mass of the tip caused by the C-S-H precipitation according to:

ω0 ∝ xk/m*

with m* ) mC + 0.24mTip (from ref 32)

III. Results and Discussion

7266

Langmuir, Vol. 21, No. 16, 2005

Plassard et al.

Table 2. Adhesion Forces Measured between the Silica Tip and Atomically Smooth Mineral Surfaces and Hamaker Constants of Each Component and Each Couple34,35 substrate/tip

mica/silica

calcite/silica

gypse/silica

C-S-H/silica

C-S-H/C-S-H

adhesion force measured (nN) Hamaker constant of substrate (10-20 J) Hamaker constant of substrate/tip system (10-20 J)

24.1 7.034 6.8

36.8 10.135 8.2

22.9 5.434 6.0

39.4 14.0a 9.6a

60.0 14.0a 14.0a

a

Values calculated starting from the curve in Figure 2. The value of the Hamaker constant of silica is 6.6 × 10-20 J.34

Figure 2. Characteristic of the adhesion force measured in air between three couples substrate/silica according to the Hamaker constant. The couples are mica/silica, gypsum/silica, and calcite/ silica. Hamaker constants are given in Table 2.

by van der Waals forces. The van der Waals forces may be estimated from the Hamaker constants of the materials according to the geometry of the setup and then compared to the measured ones. The measurements were made using three types of surfaces consisting of freshly cleaved mica {0001}, calcite {101 h 4}, and gypsum {010}, respectively, and a silicon AFM tip, which is actually naturally covered by silica. The tip may be approximated by a silica sphere, and the measurements correspond to interactions between a sphere and a plane. In these conditions, according to the Derjaguin approximation, the van der Waals force is given by

F(D) ) -

AR 6D2

(5)

where R is the tip radius, D is the tip-plane distance in contact, and A is the Hamaker constant corresponding to the tip-plane couple. The adhesion forces in each case are reported in Table 2 with the Hamaker constants of each component and of each couple. These last have been approximated according to33

A12 ) xA11A22

(6)

These data make it possible to represent the characteristic of the adhesion force according to the Hamaker constant for each couple substrate/silica (Figure 2). The characteristic obtained is linear, and according to eq 5 the slope is R/6D2 ) 4.10 × 10+11 m-1. This value is in the order of magnitude of what it is expected considering typical values of 20 nm for the curvature radius of the tip and 0.2 nm for the contact separation. The validity of the measurements gives the possibility to estimate the Hamaker constant of C-S-H for which, in our knowledge, there is no experimental determination. 2. Validation on the C-S-H/Silica System. The same kind of measurements were then made between the same silicon tip and a calcite single crystal covered by a (33) Israelachvili, J. N. Proc. R. Soc. London, Ser. A 1972, 331, 3955.

Figure 3. Deflection of cantilever against piezoelectric displacement. Substrate and tip are covered with C-S-H. Measurements in saturated calcium hydroxide solution.

crystallized layer of C-S-H to estimate the Hamaker constant of C-S-H. Knowing the value of R/6D2 and by using the eq 5, it is possible to deduce the value of the Hamaker constant for the C-S-H/silica couple. We find a Hamaker constant of 9.6 × 10-20 J for the C-S-H/silica couple corresponding to a Hamaker constant of 14.0 × 10-20 J for the C-S-H. As already said, no experimental determination of the C-S-H Hamaker constant has been found in the literature. Nevertheless, the value found in the present work is in accordance with the values suggested for condensed phases in the literature (4-40 × 10-20 J). 3. Estimation of the Interaction Surface in the C-SH/C-S-H System. Finally, force measurements were made between the calcite single crystal covered by a crystallized layer of C-S-H and a silicon tip covered by C-S-H. It was thus possible to estimate the area of the surface interaction in the configuration used in the experiments in solution described below. In this case, the measurements correspond to interactions between two planes of C-S-H. The van der Waals force is given by

F/area unit ) -

A 6πD3

(7)

where D is the tip-plane distance in contact and A is the Hamaker constant corresponding to the couple tip-plane. If we consider that the tip is in atomic contact with the surface at a separation of D ) 0.2 nm, the force per area unit is equal to F/area unit ) 9.3 × 108 N/m2. The interaction surface between C-S-H can then be estimated at

S)

Fadhesion C-S-H/C-S-H ) 64 nm2 F/area unit

(8)

B. Force Measurements between Two C-S-H Surfaces Immersed in a Chemically Equilibrated Solution. The force measurements, with our experimental setup being validated in air, were undertaken between two C-S-H surfaces equilibrated in solutions. As previously stated, the sample preparation procedure ensures an absence of artifacts because of chemical evolution and surface roughness.

Nanoscale Experimental Investigation of Particle Interactions

Langmuir, Vol. 21, No. 16, 2005 7267

Figure 4. Typical experimental force curves deduced from the approach-retraction cycle of the tip. (a) Attracto-repulsive interaction. (b) Purely attractive interaction.

Cantilevers with low spring constant values have been chosen to be sufficiently sensitive to forces in liquid. In solution, the rate of the vertical motion performed during the approach-retract cycles was lowered to 50 nm s-1 to avoid changing the viscosity of the medium. The force versus separation curves presented in this paper are limited to the range of 0-40 nm to increase the resolution of the measurements. However, in each case, first measurements are performed with great displacements (1 µm). In these conditions, the probe and the substrate are kept in contact until 2 s. The maximum force applied by the cantilever (k ) 0.6 N m-1) reaches 10-20 nN. In the compliance regime, the deflexion was always equal to the displacement without any instability, proving that there is no other interaction regime at a shorter distance (see Figure 3). Two examples of the force curves directly calculated from the cantilever deflection and scanner displacement according to eq 2 are presented in Figure 4; in each case, the upper curve relates to the approach of the tip from the substrate, whereas the lower one corresponds to the retraction. A hysteresis appears between the two curves because of the sticking of the tip onto the substrate. Negative force values correspond to attraction and positive ones to repulsion between C-S-H surfaces. The vertical slope at small separation distances corresponds to the contact between the tip and the substrate. Several things can be deduced from these force curves. First, the nature of the interaction between C-S-H particles, i.e., attractive (only negative forces occurring), repulsive (only positive forces occurring), or attracto-repulsive (both negative and positive forces occurring) can be determined along the tipsubstrate distance. Second, the values of the forces can be measured. Adhesion force is measured at the lowest point of the hysteresis loop before the probe snaps away from the surface; likewise, repulsive force is measured in the upper part of the hysteresis loop while approaching the surface. 1. Interaction in Calcium Hydroxide Solutions. The simplest solution simulating the pore solution of a hydrated cement paste is a calcium hydroxide solution. That is why it is important to first investigate the surface forces in lime solutions. Details of typical force curves measured between C-S-H surfaces in equilibrium with solutions of different calcium hydroxide concentrations ranging from 0.2 to 19.1 mmol L-1 are presented in Figure 5. The force curves reveal different behaviors with the increase of the lime concentration in solution: for the lowest calcium hydroxide concentration, the interaction is purely repulsive; for concentrations less than or equal to about 3.15 mmol L-1, attracto-repulsive interactions are observed; and the interactions are purely attractive

Figure 5. Interaction forces measured between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip immersed in Ca(OH)2 solutions of different concentrations (series 1 in Table 1). Lines are only guides for eyes.

Figure 6. Evolution of the adhesion between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip according to the Ca(OH)2 concentration in the solution in which they are immersed. The error bars are twice the standard deviation on 100 different measurements.

for higher concentrations. Furthermore, the intensity of the attractive force increases with the calcium hydroxide concentration. The evolution of the adhesion force with the calcium hydroxide concentration is plotted in Figure 6. As previously mentioned, the measurement error is estimated from the standard deviation of the values obtained from 100 different measurements. These dispersions in the case of the lowest and highest lime concentrations are shown in Figure 7. The widest dispersion observed with the most concentrated lime solution may be due to higher roughness, as revealed by AFM imaging. Although the experimental error on the measurement increases with the lime concentration, the adhesion force significantly increases and is about 5 times greater at the highest concentration than at the lowest. The change from an attracto-repulsive to a purely attractive regime corresponds to the jump in the evolution of the adhesion force between 2 and 4 mmol L-1 of calcium hydroxide in solution. In the mean field approach, according to the DLVO theory, the interaction forces between two charged surfaces

7268

Langmuir, Vol. 21, No. 16, 2005

Plassard et al.

Figure 7. Top of the figure represents the force value distributions obtained on 100 measurements. (a) In the case of the lowest calcium hydroxide concentration in solution. (b) In the case of the highest calcium hydroxide concentration in solution. The AFM images (size, 2 × 2 µm2; relative height, 100 nm) of the surfaces on which the measurements were made are shown at the bottom of the figure. The dispersion of measurements may be correlated with the roughness of the substrates.

Figure 8. Interaction forces measured between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip immersed in solutions of an increasing Ca(OH)2 concentration in which the ionic strength is controlled by the addition of NaNO3 (I ≈ 3 × 10-2 mol L-1) (series 3 in Table 1). (a) Typical evolutions of surface force with the interaction distance. Lines are only guides for eyes. (b) Evolution of adhesion according to the Ca(OH)2 concentration.

immersed in an aqueous solution result from the balance between the van der Waals attractive contribution and the repulsive contribution because of the pressure of the counterions cloud. If this theory applies here, the increase in the adhesion force should be due to the decrease of the repulsive contribution because the van der Waals contribution would remain practically constant for a given system. In fact, the main consequences of the increase of the lime concentration in solution are - Increase of the calcium concentration, - Increase of the pH from 10 to 12.5, - Increase of the ionic strength from 0.05 × 10-2 to 5.35 × 10-2 mol L-1. The main consequence of an increase of the pH is the increase of the surface charge density of C-S-H, whereas the increase of the ionic strength will lead to a decrease of the diffuse layer range, i.e., a decrease of the Debye length. In the framework of the DLVO theory, an increase of the charge densities of the interacting surfaces would lead to an increase in the repulsive force contribution because the counterion clouds will be more

consequent. In doing so, the overall adhesion force should decrease. Oppositely, an increase of the ionic strength would lead to a decrease in the repulsive force contribution because the counterion clouds will be more condensed. Therefore, the overall adhesion force should increase. To verify if this theory would apply here, it is essential to separate the effect of the pertinent parameters (pH and ionic strength) on the adhesion force between C-S-H. This is what is done on the two next paragraphs. The role of calcium ions will be discussed later. 2. Role of the pH. To check the role of the pH on surface forces, force measurements were performed in the same range of lime concentrations as in Figures 5 and 6 but while maintaining constant with the ionic strength to about 3 × 10-2 mol L-1 by adding an appropriate sodium nitrate amount (series 3 in Table 1). The results are shown in Figure 8. Contrary to what it is expected according to the DLVO theory, the overall adhesion force increases with the calcium hydroxide concentration, i.e., with the pH.

Nanoscale Experimental Investigation of Particle Interactions

Figure 9. Interaction between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip immersed in solutions of an increasing Ca(OH)2 concentration (series 1) and in solution of an increasing Ca(OH)2 concentration in which the ionic strength is adjusted to 3 × 10-2 mol L-1 by the addition of NaNO3 (series 3). Evolution of the adhesion is according to the pH.

2. Role of the Ionic Strength. When the results obtained with series 1 (ionic strength varying between 0.07 × 10-2 and 4 × 10-2 mol L-1) are compared with series 3 (ionic strength constant and equal to 3 × 10-2 mol L-1), it is noticeable in Figure 9 that the adhesion force does not depend on the ionic strength in the bulk solution. Contrary to what it is expected according to the DLVO theory, the overall adhesion force follows the same evolution whatever the ionic strength, proving that the increase of the intensity of the attractive force is not due to the condensation of the counterion cloud. According to Figure 9, the pH appears to be a determining factor for the forces interacting between C-S-H surfaces. Therefore, the origin of the attractive forces is not only the van der Waals forces. Other forces must also be considered. 3. Role of the pH and Calcium Concentration. In the experiments described above, the calcium concentration increases in the same way as the pH. To separate the effect of these two parameters, three different series of experiments were performed: - In series 2 in Table 1, the calcium concentration was increased at low pH by adding calcium chloride to a low concentrated lime solution, ∼pH 11.3, [Ca] < 20 mmol L-1 - In series 5, the calcium concentration was increased at high pH by adding calcium chloride to a saturated lime solution, 12.1 < pH < 12.5, 26 < [Ca] < 80 mmol L-1. - In series 4, the pH was adjusted to a high value (pH 12.6) by adding sodium hydroxide to the different lime solutions. In this case, for the same pH, the calcium concentration decreases as the sodium hydroxide is increased and a part of the divalent Ca counterions are replaced by Na monovalent counterions. The respective role of calcium and pH is well-revealed in Figure 10 in which the evolution of the adhesion force is plotted versus the calcium concentration for low and high pH. At low (series 2) and high (series 4 and 5) pH, the adhesion force increases with the calcium concentration and reaches a plateau but only for 5 mmol L-1 at low pH and for more than 20 mmol L-1 at high pH. Moreover, the value of the adhesion force on the plateau at high pH is more than twice that at low pH. Also, it is noticeable that, when the pH is high in the presence of sodium cations, a greater calcium concentration in solution is required to observe attractive forces (series 4, [Ca] ) 8 mmol L-1 at pH 12.5 against [Ca] < 1 mmol L-1 at pH 10). Concerning the repulsive contribution, it is only seen at low calcium concentrations (Figure 11). Measurements at low pH show a different behavior. The force measured

Langmuir, Vol. 21, No. 16, 2005 7269

Figure 10. Interaction between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip two C-S-H immersed in solutions of an increasing calcium concentration at low (series 2 in Table 1) and high pH (series 4 and 5 in Table 1). Evolution of adhesion is according to the Ca concentration.

Figure 11. Repulsive forces between a micrometric flat C-S-H surface and a C-S-H nanocrystal at the top of an AFM tip immersed in the different electrolytic solutions used in this study (series 1-4 in Table 1) versus the calcium concentration in each solution.

at the repulsion maximum changes a lot with a small variation of the calcium concentration in solution (or ionic strength), while the repulsive force measured in the other media is lower and does not change very much with the calcium concentration. This result is also difficult to interpret according to the mean field theory. Indeed, when the slight increase of the adhesion force with the calcium chloride concentration is explained by the screening of the surface charge, the same phenomenon cannot also explain the increase, at the same time, of the repulsive contribution. 4. Origin of the Forces. The data described above related to the surface forces between two C-S-H particles immersed in different electrolytic solutions clearly disagree with a mean-field approach of the interaction forces. The most important parameter seems to be the pH, and the second most important parameter seems to be the calcium concentration in solution. The main consequence of the pH augmentation is the evolution of the surface charge density, which is able to vary on a C-S-H surface between 0.08 and 0.68 C/m2. These finding are in agreement with ionic correlation attraction in strong correlated systems because of the high charge density of the surfaces, charge balanced by divalent counterions. These results are in close agreement with the Monte Carlo simulations recently performed by Jo¨nsson et al. using a primitive model.22,36 In this model, each surface is supposed to be an infinite plane with a homogeneous charge density. The solution confined between the two planes contains the (34) Ackler, D.; French, R. H. et al. J. Colloid Interface Sci. 1996, 2, 460. (35) Bergstro¨m L. Adv. Colloid Interface Sci. 1997, 70, 125. (36) Jo¨nsson, B.; Wennerstro¨m, H. et al. Langmuir 2004, 20, 67026709.

7270

Langmuir, Vol. 21, No. 16, 2005

divalent counterions, considered explicitly and needed to compensate the charge of the plane in a continuous medium characterized by its dielectric permittivity. The equilibrium distributions of the ions in the interstitial solution are then obtained, and the total osmotic pressure was calculated. The total osmotic pressure includes three terms: - The entropic pressure originates from the “gas” pressure of the counterions; its magnitude is proportional to the concentration of ions in the midplane of the cell defined by the two parallel charged surfaces. - The correlation pressure is caused by correlations between counterions on either side of the midplane; i.e., an excess of counterions on one side is correlated with a lack on the opposite side causing an electrostatic attraction. - The hard-core pressure describes the additional pressure because of the finite volume of the ions. The results show that, as the charge density rises, the number of counterions increases, but they become increasingly accumulated at the surfaces. In consequence, the entropic pressure decreases dramatically and the attractive correlation forces dominate the interactions. This simulation qualitatively agrees with the experimental data corresponding to series 1. In the present work, both the calcium counterion concentration and surface charge density rise when the lime concentration increases in solution. However, the experimental conditions imposed in series 2-5 are more complicated, and simulations in a grand canonical ensemble are required to equilibrate the solution confined between the two surfaces and the bulk solution. On the other hand, contrary to the case of mica in concentrated calcium chloride for which the specific adsorption of Ca ions promotes a change in the attractive regime at the reversal charge,27 a change in the attraction regime at the apparent charge reversal of C-S-H has not been observed in the present work. This is probably

Plassard et al.

because the charge reversal in the C-S-H case is not only due to specific adsorption but also results from ionic correlations as shown by Monte Carlo simulations.36 IV. Conclusion AFM measurements showed the existence of the medium range attraction between silicate surfaces in solutions similar to the pore solution of a hydrating cement paste. The attraction occurs on a relatively large range of pH and does not require a high calcium concentration. In conditions representative of the pore solution of a cement pastes, the pressure between C-S-H particles is about 30 MPa. This finding may explain the robustness of setting and hardening of cement. Furthermore, at a fixed pH, the adhesion force seems to be lower when part of the calcium counterions is replaced by sodium or potassium, that is, in agreement with what it is observed on the field. Pure repulsion can only occur for very low calcium concentrations in the bulk solution. It may be locally the case in concrete in which the alkali silica reaction occurs, promoting cracks in concrete. The attraction between C-S-H particles is a consequence of ion-ion correlations because of the very high negative charge density of the C-S-H particles and the presence of divalent counterions. A precise modeling of the interactions measured in the present work would require taking into account the acido-basic equilibrium of the surface and also equilibrating the solution confined between the two surfaces with the bulk solution. Acknowledgment. This project is part of the Research Program Contract “Mechanics and Chemistry of Cement Materials” financially supported by CNRS and Association Technique des Industries des Liants Hydrauliques (ATILH). C. P. was supported by a scholarship from the ministry of research MENRT. LA050440+