- Email: [email protected]
Investigation by atomic force microscopy of forces at the origin of cement cohesion
Ultramicroscopy 86 (2001) 11–21
Investigation by atomic force microscopy of forces at the origin of cement cohesion Samuel Lesko, Eric Lesniewskaa,*, Andre´ Nonatb, Jean-Claude Mutinb, Jean-Pierre Goudonneta a
Physics Laboratory L.P.U.B. UMR CNRS 5027, University of Bourgogne, B.P. 47870, F- 21078 Dijon Cedex, France b Chemistry Division L.R.R.S. UMR CNRS 5613, University of Bourgogne, F- 21078 Dijon Cedex, France Received 13 July 2000
Abstract In cement paste, the cohesion results of the interactions between calcium silicate hydrate (CSH) surfaces in an interstitial ionic solution. (N, V, T) Monte Carlo simulations show that the interactions are due to the ion correlation forces influenced by the surface charge density, the ionic concentration and the ion valence. This paper deals with the direct measurement in solutions by atomic force microscopy (AFM) of the forces and the interaction ranges between a probe and an atomically smooth substrate covered by CSH nanoparticles. Different electrolytic solutions (Ca(OH)2, CaCl2, NaCl, NaOH) have been used in order to determine influent parameters permitting to identify the nature of acting forces. Investigations have been rendered possible by selecting appropriate experimental setup and solutions. The selected probe and substrate on which CSH nanoparticles have previously grown are neutral regarding the reactivity during experiments permitting the exchange of solutions. Results show that a force originates from electrostatic nature and differs from Derjaguin–Landau–Verwey–Overbeek (DLVO) theory. Agreement is found between experiments and (N,V,T) Monte Carlo simulations of ionic correlation forces. These forces are at the origin of the cohesion of cement paste. # 2001 Elsevier Science B.V. All rights reserved. PACS: 07.79.L; 07.10.P; 82.70.-Y; 81.05.Ys Keywords: Atomic force microscopy; Measurement of forces; Colloids; Nanophase materials
1. Introduction Although ordinary portland cement (OPC) is used in huge amount worldwide, their mechanical properties are debated at fundamental levels to know which forces may act to ensure such a *Corresponding author. Tel.: +33-380-39-60-34; fax: +33-380-39-60-24. E-mail address: [email protected] (E. Lesniewska).
cohesive behaviour. Early cement paste can be schematically presented (Fig. 1) as multicompound grains that are highly charged at their surface and develop interactions through an interstitial electrolytic solution composed in the major part of calcium and hydroxide ions. Chemical reactions between cement grains and electrolytic solution will mainly give rise to precipitation of calcium silicate hydrates (CSH) on the surface of grains [1,2]. Thus, interactions between cement grains are
0304-3991/01/$ - see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 3 9 9 1 ( 0 0 ) 0 0 0 9 1 - 7
12
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 1. Schematic representation of the reaction between cement powder and water. Once cement powder (A) contacts water, it partially dissolves (B) before CSH precipitates (C) onto surface of powder. In set solid (D), the contact area increase between grains is responsible for the mechanical strength increase, the acting force remaining the same.
progressively replaced by interactions between CSH. Identifying the nature of acting force between CSH becomes consequently the striking point to understand cement cohesion. At macroscopic scale, rheology performed on cement paste at an early age has been successfully applied to understand which kind of interactions may occur along the cohesive phenomenon. Results showed that such interactions are off colloidal short range type and remain unchanged from paste to set solid, the strength increase being explained by the such increase of interaction area between cement grains [3]. At microscopic scale, (N,V,T) Monte Carlo simulations of the ion distribution between two charged disks immersed in a large simulation cell [4] have demonstrated that DLVO theory cannot be applied anymore for CSH interactions since ionic correlation has to be accounted for. In this way, CSH interactions have been modelled in the scope of a primitive model where surfaces are infinite planes, interstitial solutions a continuum media of dielectric constant e and counterion hard spheres bearing their valences. Results showed the
sensitivity of CSH interactions to both the charge density and the type of exchangeable ion (radius and valence) present in interstitial solution. For monovalent counterions, the electrostatic contribution is always negligible and the force is mainly repulsive. For divalent counterions, the electrostatic attraction becomes significant, leading to attracto/repulsive behavior. At mesoscopic scale, Atomic Force Microscopy (AFM) has already permitted to measure interactions forces [5–9] and to understand cohesion force involved in plaster, another mineral binder [10,11]. During plaster setting, a cohesion process is caused by the precipitation of gypsum micrometric crystals and the associated interactions. Experimental procedure consisted thus to measure forces in calcium sulphate solutions between two gypsum micrometric crystals. One gypsum crystal (probe) associated to an AFM cantilever is facing a second crystal (substrate) for probing the interactions at the origin of the setting of plaster. To the best of our knowledge, no direct force measurement between CSH or cement grains in electrolytic solutions has been conducted. The
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
purpose of this study is to present such AFM measurements between a CSH probe and a CSH substrate in several electrolytic solutions in order to clearly identify the acting force in cement. To reach this goal, several steps have to be over passed to ensure measurement of CSH interaction free of external perturbations. Firstly, it is interesting to notice that the CHS morphology has been previously identified as assembly of oriented nanoparticles with 60 30 5 nm3 characteristic dimensions [12]. The main point is that the CSH particle size differs enormously of the gypsum crystal size easy to manipulate. This implies to uniformly cover both the probe and substrate with CSH nanoparticles. Then, the CSH coverage being rough by nature, experimental set up has to be optimised to prevent influence of roughness onto measurements. In the same manner, CSH coverage has to be in equilibrium with electrolytic solutions imposed to ensure constancy of ionic concentrations. Secondly, identification of the acting forces between CSH nanoparticles should be possible by selecting appropriate electrolytic solution such as: Ca(OH)2, CaCl2, NaCl and NaOH solutions. Ionic concentrations will vary for each one to check whether the force originates from electrostatic nature. This set of solutions is also driven to verify the validity of theoretical predictions from Monte Carlo simulations. Notably, those solution permit to compare the effects of counterion valence (Na+ or Ca2+). Furthermore, counterions are used either in chloride or hydroxide solutions in order to test counterions influence without or with according to pH variations. Actually, varying the pH means varying the surface charge density since CSH surface is composed of disrupted silicate chains ended by silanol groups. Unfortunately, these basic solutions bring experimental problems. Means have to be found to prevent carbonation and ensure constancy of ionic concentrations. Other consequence is the reactivity of mostly commercial AFM cantilevers with respect to basic solutions. CSH growth is observed by reaction between silica or silicon nitride substrates and hydroxide solutions. This triggers not only change of ionic concentrations but also dissolution and precipitation processes onto the
13
AFM cantilever during the experiments. Influence of this reactivity should be evaluated by monitoring the modification of the cantilever spring constant. It is clear that an AFM study of the acting forces between CSH nanoparticles should provide quantitative and qualitative information on the appropriate probe/substrate system. The substrate should be atomically flat, uniformly covered by CSH nanoparticles, and reach equilibrium for performing force measurements. Two systems are presented: silica/silica and silica/calcite. To this end, interaction range and force obtained on various samples, taken in different electrolytic solutions, are compared and discussed. Discussion about results will point out the possibility to effectively measure the interaction between CSH nanoparticles. The importance of validating the data interpretation using (N,V,T) Monte Carlo simulation results is pointed out to clearly identify the forces at the origin of the cohesion of cement paste.
2. Materials and methods 2.1. Experimental set-up All experiments were performed in a glovebox free of carbon dioxide to prevent carbonation of hydroxide solutions. Inside, a commercial AFM (Nanoscope IIIa, Veeco Inst., Digital Inst., Santa Barbara, CA) was used. The microscope was kept above 80% relative humidity by passing a decarbonated flux of dry nitrogen or helium gas through lime solution. Measurement of relative humidity was carried out by keeping a hygrometer (Quick Novo 74880; EBRO GmbH, Ingoldstadt, Germany) and the head of the microscope under a glass bell-jar. By this method, solution evaporation from the liquid cell was limited during the experiment, which never exceeded 1 h. Concerning the surface force measurements, the tip–sample contact force was controlled before each measurement. The zero force was defined as the force when the tip retracted from the surface. The force applied was minimized (i.e. 0.5 nN) to prevent the cantilever from perturbing the surface.
14
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Interaction measurements were performed with usual calibration process to transform experimental cantilever deflection curves as a function of the vertical scanner displacement Dz to force-distance curves [13]. Using the slope of the retraction deflection curves in the contact region, the cantilever deflection is then converted into a force using the Hooke’s law F ¼ kDz where k represents the cantilever spring constant. Since AFM makes no absolute measurement of the probe–substrate separation distance, the latter may be calculated by correcting the scanner position. For each given electrolytic solution, the experiment consisted on recording force curves on 10 different locations for 10 different probe/substrate couple. In order to check CSH coverage, surfaces were imaged before and after force measurement in contact mode AFM in solution. All systems presented have been the result of preliminary investigations onto various systems on which CSH precipitation was possible. Those systems have been held back due to their efficiency and their relevance with respect to aims fixed onto introduction. 2.2. Probe preparation
manufacturer, E the Young’s modulus, and r the density of the material of the cantilever [19]. To obtain CSH coverage, the silicon nitride tip was immerged in large volume (V=30 ml) of saturated calcium hydroxide solution during 48 h in order to consume all silica layers and to precipitate CSH. The probe is rendered non-reactive due to the difficulty to dissolve bulk silicone nitride. The complete consumption of silica layer (2 nm thick) leads to less than 5% of the spring constant variation. 2.3. Substrate preparation As already mentioned, silica is reactive against calcium hydroxide solutions. Consequently, the obvious choice for the substrate was pure silica sheet (Goodfellow, mean roughness: 1 nm for 1 1 mm2 area). CSH coverage was obtained by immerging it in saturated calcium hydroxide solution (Ca(OH)2 , [Ca2+]=22 mmol lÿ1 at 258C, pH=12.6) via the following chemical reaction where x represents the calcium–silica ratio (C/S): SiO2 þ2H2 O ! H4 SiO4 ;
ð2Þ
xCa2þ þH4 SiO4 þ2xOHÿ ! C2S2H:
ð3Þ
Silicon nitride tip attached to commercial rectangular cantilever with a nominal spring constant of 0.3–0.1 N mÿ1 (Park Scientific Instruments, Sunnyvale, CA) was used for two reasons. Firstly, the tip can react with calcium hydroxide solution to form CSH coverage since it is naturally covered by a thin layer of silica (2 nm thick) [14,15]. Secondly, interactions between CSH surfaces (60 30 nm2) can be reached as the resultant interaction area is in order of 10 10 nm2 [16]. Conversely, cantilever spring constant was chosen less than 1 N mÿ1 in order to be sufficiently sensitive to acting forces in liquid [17]. Concerning the evolution of cantilever spring constant, the resonant frequency n of the different cantilevers was recorded during reasonable time (48 h). For the estimation of the associated spring constant k, we utilized the expression [18]
In fact, for silica substrate, CSH coverage was made in situ before to perform force measurements in the resulting solution. This was done by introducing saturated calcium hydroxide (V=500 ml) in the AFM liquid cell. After several hours, CSH coverage effectively appeared on silica substrate (Fig. 2A) as imaged by AFM. Silica substrate continuously reacts as long as diffusion of silicate ions does not become limitative [12]. The system is thus not suitable for variation of electrolytic solutions. Much more efficient choice was the {1 0 1 4} cleavage plane from optical quality calcite (Iceland spar). CSH coverage was obtained by immersing in concentrate sodium silicate solution (Rm=[SiO2]/ [Na2O]=0.33, [SiO2]=0.5 mol lÿ1, pH=14.2). The following chemical reaction occurs:
k¼ 2ðpLnÞ3 wðr3 =EÞ1=2 ;
CaCO3 ! Ca2þ þCO2ÿ 3 ;
ð4Þ
xCa2þ þH4 SiO4 þ 2xOHÿ ! C2S2H:
ð5Þ
ð1Þ
where L and w represent the length and width, respectively, of the cantilevers as supplied by the
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
15
Fig. 2. Two different CSH coverages imaged in contact mode AFM in electrolytic solution. (A) Nanoparticles of CSH on silica substrate after 4 h on saturated calcium hydroxyde solution ([Ca2+]=22 mmol lÿ1, pH=12.6 at 258C). Scan size: 1 mm. Scan force: 80 pN. Scan rate: 3 Hz. Relative height: 10 nm. (B) Micrometric atomically smooths domains of CSH on calcite substrate after several days of equilibrium where atomic resolution is allowed. Scan size: 15 mm. Scan force: 70 pN. Scan rate: 2 Hz. Relative height: 500 nm.
Silicate concentration was chosen in such way to shift equilibrium to CSH precipitation. There are two advantages: initial surface is atomically smooth and non-reactive against further electrolytic solutions used (Ca(OH)2, CaCl2, NaCl, NaOH) for force measurements. This kind of substrate cannot disturb the ionic concentrations of solutions. Concerning the calcite substrate, CSH coverage was made ex situ. Freshly cleaved crystal was immerged in large volume (V=100 ml) of sodium silicate solution during 24 h. CSH coverage appeared to be similar to those onto silica substrate (Fig. 2A). Long time of equilibrium for CSH coverage onto calcite substrate also triggered partial recrystallization. Micrometric atomically smooth domains (Fig. 2B) appeared rolling so out the influence of roughness onto measurement. It must be mentioned that CSH nanoparticles are still present in the lowest part of the sample. On these microdomains, atomic resolution can be obtained (not shown). The calcite substrate was then quickly rinsed with bidistilled water (milliQ) before immersion in a given electrolytic solution (V100 ml) to maintain equilibrium for 4 h. Force measurements were performed onto these microdomains in a droplet of the electrolytic solution (V=500 ml) in which samples were equilibrated. Actually, CSH stoichiometry depends onto ionic
concentrations of imposed solutions [1]. This is the reason why CSH surfaces undergo equilibrium time, especially long for change of solution type to ensure a system at complete equilibrium. 2.4. Solution preparation Electrolytic solutions used for interaction measurements are summed up into Table 1. All solutions were prepared with target ionic concentrations before to be saturated against CSH by dissolving CSH powder (x=C/S=0.66, issued from chemical synthesis) in excess. Solutions undergo equilibrium period of 1-week prior to be filtered through 0.1 mm Millipore filter. The saturation against CSH ensures the keeping of CSH coverage both onto probe and substrate regardless of solution used. In the same time, ionic concentrations are also preserved.
3. Results 3.1. Silica probe/silica substrate Typical force curve between silicone nitride tip and silica substrate showed pure attractive interaction in saturated calcium hydroxide solution.
16
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Table 1 Cationic concentrations in mmol lÿ1 of electrolytic solutions in which interactions have been measured. All solutions have been saturated with respect to CSH by adding CSH powder into solutions and filtering them once equilibrium is reacheda. Ca(OH)2 CaCl2 NaCl pH : 11.1 ! 12.6 pH=9 pH=9 [Ca2+] 0.7 4.5 9.8 19.4
[Ca2+] 1.4 5.3 10.3 20.2 30.7
[Na+] 9.5 16.7 48.3 97.8
NaOH pH:12.0 ! 13.0 [Ca2+]a 0.4 0.5 0.3 0.6
[Na+] 9.8 18.0 50.2 92.8
[Ca2+]a 0.3 0.2 0.2 0.2
a
Calcium ions are issued from saturation of chloride and hydroxide sodium solutions versus CSH.
Due to the small value of spring constant, the force at small separation was not studied, surfaces jumping into contact when force gradient exceeds the value of the spring constant. Nevertheless, the range of interaction was determined as the separation at which cantilever begins to sense a force. In this case, the attractive force varied between 2 and 4 nN (Fig. 3A). Likewise, adhesion force was measured in the lowest part of hysteresis loop before snapping out the surface. Adhesion varied from 0.6 to 1.8 nN (Fig. 3B). By mapping the forces for different probe/substrate couples, the lack of adhesion is due to discontinuous assembly of CSH nanoparticles onto the silica substrate. The surface roughness obviously played a significant role. It is demonstrated [20–22] and experimentally observed by AFM [23–26] that roughness reduces magnitude of surface force at both the short and long ranges affecting significantly the interaction range. From another point of view, roughness enhances the probability for the probe to contact several particles and increases the adhesion. This trend was obvious for adhesion measurements originated from the same location where all values are in proportional ratio with the minimum. Furthermore, the relative orientation between opposing CSH nanoparticles enlarges the data distribution [17]. The next striking point lies in the incapability to control the layout of CSH nanoparticles over the apex of the tip. SEM description of the apex was particularly difficult since it is known that such a
Fig. 3. Interaction range (A) and adhesion (B) measured between silica probe and a silica substrate both covered by CSH nanoparticles. Each graph represents the repartition (left axis) and the cumulative (right axis) of experimental measurements expressed in percentage.
delicate assembly disjoins and shrivels up under vacuum procedure [27] rendering the conclusion uncertain. Furthermore, the consequence of high curvature on the morphology of CSH is unknown. Notably, comparison between usual estimated radius of 50 nm for the tip apex and the lateral dimensions of the elementary CSH particle is meaningful. Regardless of the uncertainties, the goal to reach interaction between particles can be however approached by estimation of interaction radius. Energy involved in adhesion is estimated by the formula (rangeadhesion)/2 leaving to the order of 10ÿ18 J. To extrapolate a mean radius with the assumption of a circular shape for contact area,
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
17
Fig. 4. Deflection curves (raw data) measured with respect to electrolytic solutions Ca(OH)2, CaCl2, NaOH, NaCl, respectively, marked by (A), (B), (C) and (D). Scan rate=0.1 Hz.
interfacial energy of CSH in solution has to be known. Indeed this value has already been extracted from kinetic experiment [12] and is concerned with 10 mJ mÿ2. This gives an interaction radius of about 10 nm, which is lower than characteristic size of CSH particle (60 30 nm2). Consequently, the use of AFM tip as the interacting probe gives us the opportunity to probe on the case interaction on a single CSH nanoparticle free of roughness perturbations better. It should be highlighted however that interaction between two single particles remains unsure due to uncertainty concerning the CSH layout onto the apex of tip. 3.2. Silica probe/calcite substrate As a consequence of the experimental set up, silicone nitride tip and calcite substrate permits to measure interactions free of concentration change and of roughness perturbations even by varying electrolytic solutions. Fig. 4 sums up general trends for each type of electrolytic solution. Interactions clearly depend on the solution imposed. The presence of calcium ions rises up purely attractive interaction (Figs. 4A and B) whereas the presence of sodium ions gets repulsive interactions at the approach at least (Figs. 4C and D). In fine
approach, interactions in sodium hydroxide solution (Fig. 4C) appears purely repulsive contrary to interactions in sodium chloride in which attractive force is pointed out either by jump in to contact at the approach or adhesion to the retract part of deflection curve (Fig. 4D). Concerning the respective evolution of CSH interaction in accordance with ionic concentration, the mean values measured for the interaction range of force (attractive or repulsive) and the adhesion as function of concentration of either sodium or calcium ion are reported in Fig. 5. Cumulative experimental data of interaction range (Fig. 6) and adhesion (Fig. 7) concern at least 100 force–distance curve measurements experienced for each electrolytic solution and for each ionic strength. Three types of behavior are observed. The first concerns the different solutions of calcium hydroxide in which interaction tends to increase (both in range and adhesion), respectively, to the calcium concentration and the pH. The short range (2–5 nm, Fig. 5A) of attractive force (0.1–1.8 nN, Fig. 5B) experienced on atomically smooth domains in calcium hydroxide varied with the calcium concentration (1–20 mM). The second concerns the sodium solutions (NaOH and NaCl) where range of repulsive interaction decays from
18
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 5. Evolution of the interaction range (A) and the adhesion force (B) versus the concentration of the cations (Ca2+ or Na+) labeled in mmol lÿ1 (mM). For the range, the upper part of the graphic is concerned with purely attractive interaction, e.g. range of attractive force whereas the lower part deals with repulsive interaction at approach, e.g. range of repulsive force.
14–10 to 4–3 nm as ionic strength increases (Fig. 5A). Fine details render possible the distinction between the two solutions. In case of sodium chloride solutions, interaction tends to become purely attractive even if this trend is solely expressed by a larger number of attractive force curves, always in minor place against repulsive ones (30%/70%, see Fig. 6D). However, a decrease of adhesion (from 1.2 to 0.2 nN) is remarked in the same time (Fig. 7D). Moreover, repulsive force at the approach remains unchanged along concentration changes (not shown). This is not the case for sodium hydroxide solution. No trend to attractive potential has been observed (Fig. 6C) while repulsive force decreases with respect to ionic concentration, adhesion remaining tiny in all cases (Fig. 5D). The third behavior concerns the calcium chloride solutions. No significant evolution is noted with respect to concentration both in interaction range (2–2.5 nm) and in adhesion (0.4–0.9 nN), particularly seen onto cumulative of experimental data in Figs. 6B and 7B.
4. Discussion The electrolytic solutions used (Ca(OH)2, CaCl2, NaCl, NaOH) have been chosen in order to check the sensitivity of interaction forces on ionic concentrations, valence of ion, and pH, to identify the nature of acting force between CSH surfaces. The major point before to begin with discussion about interaction is to mention the mechanism of charge for CSH surface. CSH exhibit at their surface disrupted silicate chains ended by silanol groups [1]. Thus, the pH would normally control the negative charge density of the CSH surface as no specific adsorption of ion is supposed according to Si2OH ! Si2Oÿ þHþ :
ð6Þ
Although this approximation is suitable for monovalent sodium ion, it cannot be further applied for divalent calcium ion that chemically adsorb onto silanol groups [3]: Si2OH þ Ca2þ ! Si2O2Caþ þHþ :
ð7Þ
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
19
Fig. 6. Cumulative of experimental range in respective solutions: Ca(OH)2 (A), CaCl2 (B), NaOH (C) and NaCl (D). Horizontal axis and vertical axis correspond respectively to the interaction range (nm) and to the percentage of measurement. Interaction ranges of attractive force are shown in (A) and (B) diagrams while ranges of repulsive force in (D) or attracto-repulsive force in (C) are represented. Negative ranges in (C) indicate the range of purely attractive force.
Depending onto the pH and the calcium concentration, the surface will be either negatively or positively charged. Notably, it has been already established by Zeta potential measurement that the isoelectric point in calcium hydroxide solution was near 2 mM [3]. Different evolutions observed may so arise from such e competition between the pH effect and calcium activity. Those preliminary items being underlined, it is worthy of note to remark that interactions are indeed sensitive to all solutions, behavior being different for each case on regard with ionic concentration. This major point clearly demonstrates the electrostatic origin for CSH interactions. Indeed, interactions between symmetric bodies in liquid are usually viewed in the framework of DLVO theory comprising a competition
between van der Waals and electrostatic doublelayer forces. This competition gives rise to attracto-repulsive forces function of ionic concentration and valence of ion, similarly to what is experimentally observed. If comparisons are made in framework of DLVO theory, conclusions are not so easy to arrive at since sometimes experiment results are in accordance with DLVO theory whereas others are contradictory. Accordance is for example found in case of chloride sodium solutions where decay of repulsive range agrees well with power law of Debye length. The theory of DLVO applies well with this case due to the low density of negative charge induced by low pH [4]. Moreover, for sodium chloride solution, the trend to pass from repulsive to attractive force is slightly marked in
20
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 7. Cumulative of experimental adhesion force in respective solutions: Ca(OH)2 (A), CaCl2 (B), NaOH (C) and NaCl (D). Horizontal axis and vertical axis correspond, respectively, to the adhesion (nN) and to the percentage of measurement.
qualitative agreement with DLVO theory. However, one must keep in mind that the major part of interaction measured (70%, Fig. 6D) remains repulsive in contradiction with DLVO prediction. This remark applies even further to sodium hydroxide solutions since interactions are always repulsive regardless to cationic concentration. Concerning the calcium solutions, evolution of force pattern is triggered by simultaneous action of concentration of calcium ion and of the pH. Increase of calcium concentration at low pH has apparently no effect onto interaction. The fact that attractive force depends on both pH and ionic concentration turns down the action of van der Waals force. Even if one might suppose its action, it must be remembered that ranges of attractive force lie for the major part up to 2 nm (Fig. 5A and 5B) which exceed theoretical limit for van der Waals force in geometry used, assumption being made on the Hamaker constant of the CSH to be
in order of 10ÿ20 J and on radius for interacting sphere to be roughly 50 nm. In such a case, this points out the action of an attractive force of electrostatic nature to reach such a longer range. Instead, forces issue from divalent ion–ion correlation effect ranges from 2 to 5 nm regardless of geometry factors [17,28]. This is indeed in good agreement with experiments. A clue of the action of ionic correlation force is brought by Monte Carlo simulations performed for cement and clay systems [4]. Simulations have shown repulsive interaction for monovalent cations whereas presence of divalent cations exhibit attractive ones. Furthermore, adhesion is found to increase until a certain range with pH. Both of these results agree well with experiments letting appear the probable effect of ionic correlation force onto CSH interactions in calcium hydroxide solutions. Consequently, CSH interactions in real cement system are governed by those non-DLVO forces.
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
5. Conclusion The problematic of directly measuring the interactions between CSH nanoparticles by AFM has been presented in order to be aware of limitations. Solutions to tackle problems encountered have pointed out the system where a silica probe is opposed to a cleavage plane of calcite both of them being covered ex situ by CSH. The microdomains recrystallized of CSH nanoparticles allow atomic resolution and structural characterization of the CSH surface depending on the surrounded electrolytic solution. This system has permitted us to measure CSH interactions free of the concentration change and roughness effect, in several appropriate electrolytic solutions. Regarding the respective sensitivity of CSH interaction to ionic parameters, it can be concluded that ionic correlation force acts. The ionic correlation forces are at the origin of cement cohesion.
Acknowledgements We deeply thank Drs Delville, Pellenq (CRMD, Orleans) and Dr. Baron (ATILH) for fruitful discussions. This project is a part of the Research Program Contract ‘‘Mechanic and Chemistry of Cement Materials’’ financially supported by CNRS (France) and Association Technique des Industries des Liants Hydrauliques (France). Samuel Lesko was supported by a scholarship from the Re´gion Bourgogne.
References [1] H.F.W. Taylor, Cement Chemistry, Academic Press, London, 1990. [2] A. Nonat, J.-C. Mutin, From hydration to setting, Proceedings of the International RILEM Worshop, E&FN SPON, London, 1991. [3] L. Nachbaur, P.-C. Nkinamubanzi, A. Nonat, J.-C. Mutin, J. Colloid. Interface Sci. 202 (1998) 261.
21
[4] R. Pellenq, J. Caillot, A. Delville, J. Phys. Chem. B 101 (1997) 8584. [5] H.W. Hao, A.M. Baro, J.J. Saenz, J. Vac. Sci. Technol. B 9 (1991) 1323. [6] A.L. Weisenhorn, P. Maivald, H.-J. Butt, P.K. Hansma, Phys. Rev. B 45 (1992) 11–226. [7] N.A. Burnham, R.J. Colton, H.M. Pollock, Nanotechnology 4 (1993) 64. [8] T. Thundat, X.-Y. Zheng, G.Y. Chen, R.J. Warmack, Surf. Sci. Let. 294 (1993) 2150. [9] M. Radmacher, J.P. Cleveland, M. Fritz, H.G. Hansma, P.K. Hansma, Biophys. J. 66 (1994) 2159. [10] E. Finot, E. Lesniewska, J.-C. Mutin, J.-P. Goudonnet, Surf. Sci. 384 (1997) 201. [11] E. Finot, E. Lesniewska, J.-C. Mutin, J.-P. Goudonnet, J. Chem. Phys. 111 (1999) 6590. [12] S. Gauffinet, E. Finot, E. Lesniewska, A. Nonat, C. R. Acad. Sci. II 327 (1998) 231. [13] W.A. Ducker, T.J. Senden, R.M. Pashley, Nature 353 (1991) 239. [14] P.H. Holloway, H.J. Stein, J. Electrochem. Soc. 123 (1978) 723. [15] T. Matsuo, M. Esashi, Sensor Actuator 1 (1981) 77. [16] B. Cappella, G. Dietler, Surf. Sci. Rep. 34 (1999) 1. [17] J.N. Israelachvili, Intermolecular and Surfaces Forces, Academic Press, London, 1992. [18] J.P. Cleveland, S. Manne, D. Bocek, P.K. Hansma, J. Struct. Biol. 105 (1990) 54. [19] D. Sarid, Scanning Probe Microscopy, Oxford, New York, 1991. [20] M. Kostoglou, A.J. Karabelas, J. Colloid Interface Sci. 171 (1995) 187. [21] D. Maugis, J. Adhes. Sci. Technol. 10 (1996) l6l. [22] M.C. Herman, K.D. Papadopoulos, J. Colloid Interface Sci. 142 (1991) 331. [23] D. Atkins, P. Ke´kicheff, O. Spalla, J. Colloid Interface Sci. 188 (1997) 234. [24] D.M. Schaefer, M. Carpenter, B. Gady, R. Reifenberger, J. Adhesion Sci. Technol. 9 (1995) 1049. [25] G. Toikka, R.A. Hayes, J. Ralston, Langmuir 12 (1996) 3783. [26] C. Larson, C.J. Drummond, D.Y.C. Chan, J. Phys. Chem. 99 (1995) 2114. [27] S. Gauffinet, E. Finot, E. Lesniewska, S. Collin, A. Nonat, AFM and SEM studies of CSH growth on C3S surface during its early hydration, Proceedings of the 20th International Conference on Cement Microscopy Mexico, 1998. [28] P. Ke´kicheff, S. Marcelja, T.J. Senden, V.E. Schubin, J. Chem. Phys. 99 (1993) 6098.
Investigation by atomic force microscopy of forces at the origin of cement cohesion Samuel Lesko, Eric Lesniewskaa,*, Andre´ Nonatb, Jean-Claude Mutinb, Jean-Pierre Goudonneta a
Physics Laboratory L.P.U.B. UMR CNRS 5027, University of Bourgogne, B.P. 47870, F- 21078 Dijon Cedex, France b Chemistry Division L.R.R.S. UMR CNRS 5613, University of Bourgogne, F- 21078 Dijon Cedex, France Received 13 July 2000
Abstract In cement paste, the cohesion results of the interactions between calcium silicate hydrate (CSH) surfaces in an interstitial ionic solution. (N, V, T) Monte Carlo simulations show that the interactions are due to the ion correlation forces influenced by the surface charge density, the ionic concentration and the ion valence. This paper deals with the direct measurement in solutions by atomic force microscopy (AFM) of the forces and the interaction ranges between a probe and an atomically smooth substrate covered by CSH nanoparticles. Different electrolytic solutions (Ca(OH)2, CaCl2, NaCl, NaOH) have been used in order to determine influent parameters permitting to identify the nature of acting forces. Investigations have been rendered possible by selecting appropriate experimental setup and solutions. The selected probe and substrate on which CSH nanoparticles have previously grown are neutral regarding the reactivity during experiments permitting the exchange of solutions. Results show that a force originates from electrostatic nature and differs from Derjaguin–Landau–Verwey–Overbeek (DLVO) theory. Agreement is found between experiments and (N,V,T) Monte Carlo simulations of ionic correlation forces. These forces are at the origin of the cohesion of cement paste. # 2001 Elsevier Science B.V. All rights reserved. PACS: 07.79.L; 07.10.P; 82.70.-Y; 81.05.Ys Keywords: Atomic force microscopy; Measurement of forces; Colloids; Nanophase materials
1. Introduction Although ordinary portland cement (OPC) is used in huge amount worldwide, their mechanical properties are debated at fundamental levels to know which forces may act to ensure such a *Corresponding author. Tel.: +33-380-39-60-34; fax: +33-380-39-60-24. E-mail address: [email protected] (E. Lesniewska).
cohesive behaviour. Early cement paste can be schematically presented (Fig. 1) as multicompound grains that are highly charged at their surface and develop interactions through an interstitial electrolytic solution composed in the major part of calcium and hydroxide ions. Chemical reactions between cement grains and electrolytic solution will mainly give rise to precipitation of calcium silicate hydrates (CSH) on the surface of grains [1,2]. Thus, interactions between cement grains are
0304-3991/01/$ - see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 3 9 9 1 ( 0 0 ) 0 0 0 9 1 - 7
12
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 1. Schematic representation of the reaction between cement powder and water. Once cement powder (A) contacts water, it partially dissolves (B) before CSH precipitates (C) onto surface of powder. In set solid (D), the contact area increase between grains is responsible for the mechanical strength increase, the acting force remaining the same.
progressively replaced by interactions between CSH. Identifying the nature of acting force between CSH becomes consequently the striking point to understand cement cohesion. At macroscopic scale, rheology performed on cement paste at an early age has been successfully applied to understand which kind of interactions may occur along the cohesive phenomenon. Results showed that such interactions are off colloidal short range type and remain unchanged from paste to set solid, the strength increase being explained by the such increase of interaction area between cement grains [3]. At microscopic scale, (N,V,T) Monte Carlo simulations of the ion distribution between two charged disks immersed in a large simulation cell [4] have demonstrated that DLVO theory cannot be applied anymore for CSH interactions since ionic correlation has to be accounted for. In this way, CSH interactions have been modelled in the scope of a primitive model where surfaces are infinite planes, interstitial solutions a continuum media of dielectric constant e and counterion hard spheres bearing their valences. Results showed the
sensitivity of CSH interactions to both the charge density and the type of exchangeable ion (radius and valence) present in interstitial solution. For monovalent counterions, the electrostatic contribution is always negligible and the force is mainly repulsive. For divalent counterions, the electrostatic attraction becomes significant, leading to attracto/repulsive behavior. At mesoscopic scale, Atomic Force Microscopy (AFM) has already permitted to measure interactions forces [5–9] and to understand cohesion force involved in plaster, another mineral binder [10,11]. During plaster setting, a cohesion process is caused by the precipitation of gypsum micrometric crystals and the associated interactions. Experimental procedure consisted thus to measure forces in calcium sulphate solutions between two gypsum micrometric crystals. One gypsum crystal (probe) associated to an AFM cantilever is facing a second crystal (substrate) for probing the interactions at the origin of the setting of plaster. To the best of our knowledge, no direct force measurement between CSH or cement grains in electrolytic solutions has been conducted. The
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
purpose of this study is to present such AFM measurements between a CSH probe and a CSH substrate in several electrolytic solutions in order to clearly identify the acting force in cement. To reach this goal, several steps have to be over passed to ensure measurement of CSH interaction free of external perturbations. Firstly, it is interesting to notice that the CHS morphology has been previously identified as assembly of oriented nanoparticles with 60 30 5 nm3 characteristic dimensions [12]. The main point is that the CSH particle size differs enormously of the gypsum crystal size easy to manipulate. This implies to uniformly cover both the probe and substrate with CSH nanoparticles. Then, the CSH coverage being rough by nature, experimental set up has to be optimised to prevent influence of roughness onto measurements. In the same manner, CSH coverage has to be in equilibrium with electrolytic solutions imposed to ensure constancy of ionic concentrations. Secondly, identification of the acting forces between CSH nanoparticles should be possible by selecting appropriate electrolytic solution such as: Ca(OH)2, CaCl2, NaCl and NaOH solutions. Ionic concentrations will vary for each one to check whether the force originates from electrostatic nature. This set of solutions is also driven to verify the validity of theoretical predictions from Monte Carlo simulations. Notably, those solution permit to compare the effects of counterion valence (Na+ or Ca2+). Furthermore, counterions are used either in chloride or hydroxide solutions in order to test counterions influence without or with according to pH variations. Actually, varying the pH means varying the surface charge density since CSH surface is composed of disrupted silicate chains ended by silanol groups. Unfortunately, these basic solutions bring experimental problems. Means have to be found to prevent carbonation and ensure constancy of ionic concentrations. Other consequence is the reactivity of mostly commercial AFM cantilevers with respect to basic solutions. CSH growth is observed by reaction between silica or silicon nitride substrates and hydroxide solutions. This triggers not only change of ionic concentrations but also dissolution and precipitation processes onto the
13
AFM cantilever during the experiments. Influence of this reactivity should be evaluated by monitoring the modification of the cantilever spring constant. It is clear that an AFM study of the acting forces between CSH nanoparticles should provide quantitative and qualitative information on the appropriate probe/substrate system. The substrate should be atomically flat, uniformly covered by CSH nanoparticles, and reach equilibrium for performing force measurements. Two systems are presented: silica/silica and silica/calcite. To this end, interaction range and force obtained on various samples, taken in different electrolytic solutions, are compared and discussed. Discussion about results will point out the possibility to effectively measure the interaction between CSH nanoparticles. The importance of validating the data interpretation using (N,V,T) Monte Carlo simulation results is pointed out to clearly identify the forces at the origin of the cohesion of cement paste.
2. Materials and methods 2.1. Experimental set-up All experiments were performed in a glovebox free of carbon dioxide to prevent carbonation of hydroxide solutions. Inside, a commercial AFM (Nanoscope IIIa, Veeco Inst., Digital Inst., Santa Barbara, CA) was used. The microscope was kept above 80% relative humidity by passing a decarbonated flux of dry nitrogen or helium gas through lime solution. Measurement of relative humidity was carried out by keeping a hygrometer (Quick Novo 74880; EBRO GmbH, Ingoldstadt, Germany) and the head of the microscope under a glass bell-jar. By this method, solution evaporation from the liquid cell was limited during the experiment, which never exceeded 1 h. Concerning the surface force measurements, the tip–sample contact force was controlled before each measurement. The zero force was defined as the force when the tip retracted from the surface. The force applied was minimized (i.e. 0.5 nN) to prevent the cantilever from perturbing the surface.
14
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Interaction measurements were performed with usual calibration process to transform experimental cantilever deflection curves as a function of the vertical scanner displacement Dz to force-distance curves [13]. Using the slope of the retraction deflection curves in the contact region, the cantilever deflection is then converted into a force using the Hooke’s law F ¼ kDz where k represents the cantilever spring constant. Since AFM makes no absolute measurement of the probe–substrate separation distance, the latter may be calculated by correcting the scanner position. For each given electrolytic solution, the experiment consisted on recording force curves on 10 different locations for 10 different probe/substrate couple. In order to check CSH coverage, surfaces were imaged before and after force measurement in contact mode AFM in solution. All systems presented have been the result of preliminary investigations onto various systems on which CSH precipitation was possible. Those systems have been held back due to their efficiency and their relevance with respect to aims fixed onto introduction. 2.2. Probe preparation
manufacturer, E the Young’s modulus, and r the density of the material of the cantilever [19]. To obtain CSH coverage, the silicon nitride tip was immerged in large volume (V=30 ml) of saturated calcium hydroxide solution during 48 h in order to consume all silica layers and to precipitate CSH. The probe is rendered non-reactive due to the difficulty to dissolve bulk silicone nitride. The complete consumption of silica layer (2 nm thick) leads to less than 5% of the spring constant variation. 2.3. Substrate preparation As already mentioned, silica is reactive against calcium hydroxide solutions. Consequently, the obvious choice for the substrate was pure silica sheet (Goodfellow, mean roughness: 1 nm for 1 1 mm2 area). CSH coverage was obtained by immerging it in saturated calcium hydroxide solution (Ca(OH)2 , [Ca2+]=22 mmol lÿ1 at 258C, pH=12.6) via the following chemical reaction where x represents the calcium–silica ratio (C/S): SiO2 þ2H2 O ! H4 SiO4 ;
ð2Þ
xCa2þ þH4 SiO4 þ2xOHÿ ! C2S2H:
ð3Þ
Silicon nitride tip attached to commercial rectangular cantilever with a nominal spring constant of 0.3–0.1 N mÿ1 (Park Scientific Instruments, Sunnyvale, CA) was used for two reasons. Firstly, the tip can react with calcium hydroxide solution to form CSH coverage since it is naturally covered by a thin layer of silica (2 nm thick) [14,15]. Secondly, interactions between CSH surfaces (60 30 nm2) can be reached as the resultant interaction area is in order of 10 10 nm2 [16]. Conversely, cantilever spring constant was chosen less than 1 N mÿ1 in order to be sufficiently sensitive to acting forces in liquid [17]. Concerning the evolution of cantilever spring constant, the resonant frequency n of the different cantilevers was recorded during reasonable time (48 h). For the estimation of the associated spring constant k, we utilized the expression [18]
In fact, for silica substrate, CSH coverage was made in situ before to perform force measurements in the resulting solution. This was done by introducing saturated calcium hydroxide (V=500 ml) in the AFM liquid cell. After several hours, CSH coverage effectively appeared on silica substrate (Fig. 2A) as imaged by AFM. Silica substrate continuously reacts as long as diffusion of silicate ions does not become limitative [12]. The system is thus not suitable for variation of electrolytic solutions. Much more efficient choice was the {1 0 1 4} cleavage plane from optical quality calcite (Iceland spar). CSH coverage was obtained by immersing in concentrate sodium silicate solution (Rm=[SiO2]/ [Na2O]=0.33, [SiO2]=0.5 mol lÿ1, pH=14.2). The following chemical reaction occurs:
k¼ 2ðpLnÞ3 wðr3 =EÞ1=2 ;
CaCO3 ! Ca2þ þCO2ÿ 3 ;
ð4Þ
xCa2þ þH4 SiO4 þ 2xOHÿ ! C2S2H:
ð5Þ
ð1Þ
where L and w represent the length and width, respectively, of the cantilevers as supplied by the
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
15
Fig. 2. Two different CSH coverages imaged in contact mode AFM in electrolytic solution. (A) Nanoparticles of CSH on silica substrate after 4 h on saturated calcium hydroxyde solution ([Ca2+]=22 mmol lÿ1, pH=12.6 at 258C). Scan size: 1 mm. Scan force: 80 pN. Scan rate: 3 Hz. Relative height: 10 nm. (B) Micrometric atomically smooths domains of CSH on calcite substrate after several days of equilibrium where atomic resolution is allowed. Scan size: 15 mm. Scan force: 70 pN. Scan rate: 2 Hz. Relative height: 500 nm.
Silicate concentration was chosen in such way to shift equilibrium to CSH precipitation. There are two advantages: initial surface is atomically smooth and non-reactive against further electrolytic solutions used (Ca(OH)2, CaCl2, NaCl, NaOH) for force measurements. This kind of substrate cannot disturb the ionic concentrations of solutions. Concerning the calcite substrate, CSH coverage was made ex situ. Freshly cleaved crystal was immerged in large volume (V=100 ml) of sodium silicate solution during 24 h. CSH coverage appeared to be similar to those onto silica substrate (Fig. 2A). Long time of equilibrium for CSH coverage onto calcite substrate also triggered partial recrystallization. Micrometric atomically smooth domains (Fig. 2B) appeared rolling so out the influence of roughness onto measurement. It must be mentioned that CSH nanoparticles are still present in the lowest part of the sample. On these microdomains, atomic resolution can be obtained (not shown). The calcite substrate was then quickly rinsed with bidistilled water (milliQ) before immersion in a given electrolytic solution (V100 ml) to maintain equilibrium for 4 h. Force measurements were performed onto these microdomains in a droplet of the electrolytic solution (V=500 ml) in which samples were equilibrated. Actually, CSH stoichiometry depends onto ionic
concentrations of imposed solutions [1]. This is the reason why CSH surfaces undergo equilibrium time, especially long for change of solution type to ensure a system at complete equilibrium. 2.4. Solution preparation Electrolytic solutions used for interaction measurements are summed up into Table 1. All solutions were prepared with target ionic concentrations before to be saturated against CSH by dissolving CSH powder (x=C/S=0.66, issued from chemical synthesis) in excess. Solutions undergo equilibrium period of 1-week prior to be filtered through 0.1 mm Millipore filter. The saturation against CSH ensures the keeping of CSH coverage both onto probe and substrate regardless of solution used. In the same time, ionic concentrations are also preserved.
3. Results 3.1. Silica probe/silica substrate Typical force curve between silicone nitride tip and silica substrate showed pure attractive interaction in saturated calcium hydroxide solution.
16
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Table 1 Cationic concentrations in mmol lÿ1 of electrolytic solutions in which interactions have been measured. All solutions have been saturated with respect to CSH by adding CSH powder into solutions and filtering them once equilibrium is reacheda. Ca(OH)2 CaCl2 NaCl pH : 11.1 ! 12.6 pH=9 pH=9 [Ca2+] 0.7 4.5 9.8 19.4
[Ca2+] 1.4 5.3 10.3 20.2 30.7
[Na+] 9.5 16.7 48.3 97.8
NaOH pH:12.0 ! 13.0 [Ca2+]a 0.4 0.5 0.3 0.6
[Na+] 9.8 18.0 50.2 92.8
[Ca2+]a 0.3 0.2 0.2 0.2
a
Calcium ions are issued from saturation of chloride and hydroxide sodium solutions versus CSH.
Due to the small value of spring constant, the force at small separation was not studied, surfaces jumping into contact when force gradient exceeds the value of the spring constant. Nevertheless, the range of interaction was determined as the separation at which cantilever begins to sense a force. In this case, the attractive force varied between 2 and 4 nN (Fig. 3A). Likewise, adhesion force was measured in the lowest part of hysteresis loop before snapping out the surface. Adhesion varied from 0.6 to 1.8 nN (Fig. 3B). By mapping the forces for different probe/substrate couples, the lack of adhesion is due to discontinuous assembly of CSH nanoparticles onto the silica substrate. The surface roughness obviously played a significant role. It is demonstrated [20–22] and experimentally observed by AFM [23–26] that roughness reduces magnitude of surface force at both the short and long ranges affecting significantly the interaction range. From another point of view, roughness enhances the probability for the probe to contact several particles and increases the adhesion. This trend was obvious for adhesion measurements originated from the same location where all values are in proportional ratio with the minimum. Furthermore, the relative orientation between opposing CSH nanoparticles enlarges the data distribution [17]. The next striking point lies in the incapability to control the layout of CSH nanoparticles over the apex of the tip. SEM description of the apex was particularly difficult since it is known that such a
Fig. 3. Interaction range (A) and adhesion (B) measured between silica probe and a silica substrate both covered by CSH nanoparticles. Each graph represents the repartition (left axis) and the cumulative (right axis) of experimental measurements expressed in percentage.
delicate assembly disjoins and shrivels up under vacuum procedure [27] rendering the conclusion uncertain. Furthermore, the consequence of high curvature on the morphology of CSH is unknown. Notably, comparison between usual estimated radius of 50 nm for the tip apex and the lateral dimensions of the elementary CSH particle is meaningful. Regardless of the uncertainties, the goal to reach interaction between particles can be however approached by estimation of interaction radius. Energy involved in adhesion is estimated by the formula (rangeadhesion)/2 leaving to the order of 10ÿ18 J. To extrapolate a mean radius with the assumption of a circular shape for contact area,
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
17
Fig. 4. Deflection curves (raw data) measured with respect to electrolytic solutions Ca(OH)2, CaCl2, NaOH, NaCl, respectively, marked by (A), (B), (C) and (D). Scan rate=0.1 Hz.
interfacial energy of CSH in solution has to be known. Indeed this value has already been extracted from kinetic experiment [12] and is concerned with 10 mJ mÿ2. This gives an interaction radius of about 10 nm, which is lower than characteristic size of CSH particle (60 30 nm2). Consequently, the use of AFM tip as the interacting probe gives us the opportunity to probe on the case interaction on a single CSH nanoparticle free of roughness perturbations better. It should be highlighted however that interaction between two single particles remains unsure due to uncertainty concerning the CSH layout onto the apex of tip. 3.2. Silica probe/calcite substrate As a consequence of the experimental set up, silicone nitride tip and calcite substrate permits to measure interactions free of concentration change and of roughness perturbations even by varying electrolytic solutions. Fig. 4 sums up general trends for each type of electrolytic solution. Interactions clearly depend on the solution imposed. The presence of calcium ions rises up purely attractive interaction (Figs. 4A and B) whereas the presence of sodium ions gets repulsive interactions at the approach at least (Figs. 4C and D). In fine
approach, interactions in sodium hydroxide solution (Fig. 4C) appears purely repulsive contrary to interactions in sodium chloride in which attractive force is pointed out either by jump in to contact at the approach or adhesion to the retract part of deflection curve (Fig. 4D). Concerning the respective evolution of CSH interaction in accordance with ionic concentration, the mean values measured for the interaction range of force (attractive or repulsive) and the adhesion as function of concentration of either sodium or calcium ion are reported in Fig. 5. Cumulative experimental data of interaction range (Fig. 6) and adhesion (Fig. 7) concern at least 100 force–distance curve measurements experienced for each electrolytic solution and for each ionic strength. Three types of behavior are observed. The first concerns the different solutions of calcium hydroxide in which interaction tends to increase (both in range and adhesion), respectively, to the calcium concentration and the pH. The short range (2–5 nm, Fig. 5A) of attractive force (0.1–1.8 nN, Fig. 5B) experienced on atomically smooth domains in calcium hydroxide varied with the calcium concentration (1–20 mM). The second concerns the sodium solutions (NaOH and NaCl) where range of repulsive interaction decays from
18
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 5. Evolution of the interaction range (A) and the adhesion force (B) versus the concentration of the cations (Ca2+ or Na+) labeled in mmol lÿ1 (mM). For the range, the upper part of the graphic is concerned with purely attractive interaction, e.g. range of attractive force whereas the lower part deals with repulsive interaction at approach, e.g. range of repulsive force.
14–10 to 4–3 nm as ionic strength increases (Fig. 5A). Fine details render possible the distinction between the two solutions. In case of sodium chloride solutions, interaction tends to become purely attractive even if this trend is solely expressed by a larger number of attractive force curves, always in minor place against repulsive ones (30%/70%, see Fig. 6D). However, a decrease of adhesion (from 1.2 to 0.2 nN) is remarked in the same time (Fig. 7D). Moreover, repulsive force at the approach remains unchanged along concentration changes (not shown). This is not the case for sodium hydroxide solution. No trend to attractive potential has been observed (Fig. 6C) while repulsive force decreases with respect to ionic concentration, adhesion remaining tiny in all cases (Fig. 5D). The third behavior concerns the calcium chloride solutions. No significant evolution is noted with respect to concentration both in interaction range (2–2.5 nm) and in adhesion (0.4–0.9 nN), particularly seen onto cumulative of experimental data in Figs. 6B and 7B.
4. Discussion The electrolytic solutions used (Ca(OH)2, CaCl2, NaCl, NaOH) have been chosen in order to check the sensitivity of interaction forces on ionic concentrations, valence of ion, and pH, to identify the nature of acting force between CSH surfaces. The major point before to begin with discussion about interaction is to mention the mechanism of charge for CSH surface. CSH exhibit at their surface disrupted silicate chains ended by silanol groups [1]. Thus, the pH would normally control the negative charge density of the CSH surface as no specific adsorption of ion is supposed according to Si2OH ! Si2Oÿ þHþ :
ð6Þ
Although this approximation is suitable for monovalent sodium ion, it cannot be further applied for divalent calcium ion that chemically adsorb onto silanol groups [3]: Si2OH þ Ca2þ ! Si2O2Caþ þHþ :
ð7Þ
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
19
Fig. 6. Cumulative of experimental range in respective solutions: Ca(OH)2 (A), CaCl2 (B), NaOH (C) and NaCl (D). Horizontal axis and vertical axis correspond respectively to the interaction range (nm) and to the percentage of measurement. Interaction ranges of attractive force are shown in (A) and (B) diagrams while ranges of repulsive force in (D) or attracto-repulsive force in (C) are represented. Negative ranges in (C) indicate the range of purely attractive force.
Depending onto the pH and the calcium concentration, the surface will be either negatively or positively charged. Notably, it has been already established by Zeta potential measurement that the isoelectric point in calcium hydroxide solution was near 2 mM [3]. Different evolutions observed may so arise from such e competition between the pH effect and calcium activity. Those preliminary items being underlined, it is worthy of note to remark that interactions are indeed sensitive to all solutions, behavior being different for each case on regard with ionic concentration. This major point clearly demonstrates the electrostatic origin for CSH interactions. Indeed, interactions between symmetric bodies in liquid are usually viewed in the framework of DLVO theory comprising a competition
between van der Waals and electrostatic doublelayer forces. This competition gives rise to attracto-repulsive forces function of ionic concentration and valence of ion, similarly to what is experimentally observed. If comparisons are made in framework of DLVO theory, conclusions are not so easy to arrive at since sometimes experiment results are in accordance with DLVO theory whereas others are contradictory. Accordance is for example found in case of chloride sodium solutions where decay of repulsive range agrees well with power law of Debye length. The theory of DLVO applies well with this case due to the low density of negative charge induced by low pH [4]. Moreover, for sodium chloride solution, the trend to pass from repulsive to attractive force is slightly marked in
20
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
Fig. 7. Cumulative of experimental adhesion force in respective solutions: Ca(OH)2 (A), CaCl2 (B), NaOH (C) and NaCl (D). Horizontal axis and vertical axis correspond, respectively, to the adhesion (nN) and to the percentage of measurement.
qualitative agreement with DLVO theory. However, one must keep in mind that the major part of interaction measured (70%, Fig. 6D) remains repulsive in contradiction with DLVO prediction. This remark applies even further to sodium hydroxide solutions since interactions are always repulsive regardless to cationic concentration. Concerning the calcium solutions, evolution of force pattern is triggered by simultaneous action of concentration of calcium ion and of the pH. Increase of calcium concentration at low pH has apparently no effect onto interaction. The fact that attractive force depends on both pH and ionic concentration turns down the action of van der Waals force. Even if one might suppose its action, it must be remembered that ranges of attractive force lie for the major part up to 2 nm (Fig. 5A and 5B) which exceed theoretical limit for van der Waals force in geometry used, assumption being made on the Hamaker constant of the CSH to be
in order of 10ÿ20 J and on radius for interacting sphere to be roughly 50 nm. In such a case, this points out the action of an attractive force of electrostatic nature to reach such a longer range. Instead, forces issue from divalent ion–ion correlation effect ranges from 2 to 5 nm regardless of geometry factors [17,28]. This is indeed in good agreement with experiments. A clue of the action of ionic correlation force is brought by Monte Carlo simulations performed for cement and clay systems [4]. Simulations have shown repulsive interaction for monovalent cations whereas presence of divalent cations exhibit attractive ones. Furthermore, adhesion is found to increase until a certain range with pH. Both of these results agree well with experiments letting appear the probable effect of ionic correlation force onto CSH interactions in calcium hydroxide solutions. Consequently, CSH interactions in real cement system are governed by those non-DLVO forces.
S. Lesko et al. / Ultramicroscopy 86 (2001) 11–21
5. Conclusion The problematic of directly measuring the interactions between CSH nanoparticles by AFM has been presented in order to be aware of limitations. Solutions to tackle problems encountered have pointed out the system where a silica probe is opposed to a cleavage plane of calcite both of them being covered ex situ by CSH. The microdomains recrystallized of CSH nanoparticles allow atomic resolution and structural characterization of the CSH surface depending on the surrounded electrolytic solution. This system has permitted us to measure CSH interactions free of the concentration change and roughness effect, in several appropriate electrolytic solutions. Regarding the respective sensitivity of CSH interaction to ionic parameters, it can be concluded that ionic correlation force acts. The ionic correlation forces are at the origin of cement cohesion.
Acknowledgements We deeply thank Drs Delville, Pellenq (CRMD, Orleans) and Dr. Baron (ATILH) for fruitful discussions. This project is a part of the Research Program Contract ‘‘Mechanic and Chemistry of Cement Materials’’ financially supported by CNRS (France) and Association Technique des Industries des Liants Hydrauliques (France). Samuel Lesko was supported by a scholarship from the Re´gion Bourgogne.
References [1] H.F.W. Taylor, Cement Chemistry, Academic Press, London, 1990. [2] A. Nonat, J.-C. Mutin, From hydration to setting, Proceedings of the International RILEM Worshop, E&FN SPON, London, 1991. [3] L. Nachbaur, P.-C. Nkinamubanzi, A. Nonat, J.-C. Mutin, J. Colloid. Interface Sci. 202 (1998) 261.
21
[4] R. Pellenq, J. Caillot, A. Delville, J. Phys. Chem. B 101 (1997) 8584. [5] H.W. Hao, A.M. Baro, J.J. Saenz, J. Vac. Sci. Technol. B 9 (1991) 1323. [6] A.L. Weisenhorn, P. Maivald, H.-J. Butt, P.K. Hansma, Phys. Rev. B 45 (1992) 11–226. [7] N.A. Burnham, R.J. Colton, H.M. Pollock, Nanotechnology 4 (1993) 64. [8] T. Thundat, X.-Y. Zheng, G.Y. Chen, R.J. Warmack, Surf. Sci. Let. 294 (1993) 2150. [9] M. Radmacher, J.P. Cleveland, M. Fritz, H.G. Hansma, P.K. Hansma, Biophys. J. 66 (1994) 2159. [10] E. Finot, E. Lesniewska, J.-C. Mutin, J.-P. Goudonnet, Surf. Sci. 384 (1997) 201. [11] E. Finot, E. Lesniewska, J.-C. Mutin, J.-P. Goudonnet, J. Chem. Phys. 111 (1999) 6590. [12] S. Gauffinet, E. Finot, E. Lesniewska, A. Nonat, C. R. Acad. Sci. II 327 (1998) 231. [13] W.A. Ducker, T.J. Senden, R.M. Pashley, Nature 353 (1991) 239. [14] P.H. Holloway, H.J. Stein, J. Electrochem. Soc. 123 (1978) 723. [15] T. Matsuo, M. Esashi, Sensor Actuator 1 (1981) 77. [16] B. Cappella, G. Dietler, Surf. Sci. Rep. 34 (1999) 1. [17] J.N. Israelachvili, Intermolecular and Surfaces Forces, Academic Press, London, 1992. [18] J.P. Cleveland, S. Manne, D. Bocek, P.K. Hansma, J. Struct. Biol. 105 (1990) 54. [19] D. Sarid, Scanning Probe Microscopy, Oxford, New York, 1991. [20] M. Kostoglou, A.J. Karabelas, J. Colloid Interface Sci. 171 (1995) 187. [21] D. Maugis, J. Adhes. Sci. Technol. 10 (1996) l6l. [22] M.C. Herman, K.D. Papadopoulos, J. Colloid Interface Sci. 142 (1991) 331. [23] D. Atkins, P. Ke´kicheff, O. Spalla, J. Colloid Interface Sci. 188 (1997) 234. [24] D.M. Schaefer, M. Carpenter, B. Gady, R. Reifenberger, J. Adhesion Sci. Technol. 9 (1995) 1049. [25] G. Toikka, R.A. Hayes, J. Ralston, Langmuir 12 (1996) 3783. [26] C. Larson, C.J. Drummond, D.Y.C. Chan, J. Phys. Chem. 99 (1995) 2114. [27] S. Gauffinet, E. Finot, E. Lesniewska, S. Collin, A. Nonat, AFM and SEM studies of CSH growth on C3S surface during its early hydration, Proceedings of the 20th International Conference on Cement Microscopy Mexico, 1998. [28] P. Ke´kicheff, S. Marcelja, T.J. Senden, V.E. Schubin, J. Chem. Phys. 99 (1993) 6098.