- Email: [email protected]
Impact of Bulk Phase Transitions and Wetting Transitions on Adsorption Isotherms in Binary Systems in Contact with Solid
JOURNAL OF COLLOID AND INTERFACE SCIENCE ARTICLE NO.
207, 46 –53 (1998)
CS985542
Impact of Bulk Phase Transitions and Wetting Transitions on Adsorption Isotherms in Binary Systems in Contact with Solid A. Hamraoui and M. Privat1 UMR CNRS 6521, De´partement de Chimie UBO, 6 avenue Le Gorgeu, B.P. 809, 29285 Brest Cedex, France Received October 3, 1997; accepted March 26, 1998
wets the wall), for T . T W wetting is incomplete (there is a nonzero contact angle between the two liquid phases against the wall). The wetting transition occurs between class I (complete wetting) and IIA (partial wetting), according to Telo da Gama and Evans’ classification. The notion of wetting class was introduced by Dash (2) and developed by Sullivan (3) for one component system and extended to binary mixture by Telo da Gama and Evans (4). As a consequence of the one density character of their model, Telo da Gama and Evans found in such a case that the bulk wetting phase is the preferentially adsorbed species-rich phase; we have shown in our experiments that the 2,5DMP is preferentially adsorbed from a dilute phase (5) (from a water-rich phase), although this water-rich phase is the bulk wetting phase, which shows the incontestable two-density character of our system. When the wetting transition is first order, it is preceded by a transition called prewetting, whose existence in the phase diagram is represented by the prewetting line (PW) in Fig. 1. This line separates weak and strong adsorption zones. When passing the prewetting line, the adsorbed phase undergoes a phase change in surface. The surface phase transition is similar to the bulk phase transition (gas–liquid, liquid–liquid, solid– liquid), and it leads to the prewetting transition which is equivalent to a germination, in surface, of the wetting bulk phase. In our case, prewetting must be a liquid–liquid transition because the wetting transition occurs along a liquid–liquid coexistence. Understanding the mechanism of the prewetting transition, when it exists, for systems such as silica/water– 2,5DMP helps to make the connection between interfacial and macroscopic properties. The surface can undergo other phase transitions: a liquid– solid transition may appear beside a liquid–liquid one. Both can lead to layering, even if liquid–solid separations are preferentially observed in the case of layering. To distinguish between them may be the task of an experimentalist. He must also try to connect with the interactional properties of the system. In the results, some of them should be considered as classical, or as the generalization of classical results. Single wave isotherms (i.e. with a maximum and a minimum) have been known for a long time (6). Several wave isotherms are a
This paper presents several studies essentially on the solid/liquid interface. The studied system is the liquid binary mixture water– 2,5dimethylpyridine (2,5DMP) in contact with an amorphous silica (Aerosil type). The study of wetting along the two-liquid bulk phases coexistence in contact with silica allows the determination of a wetting transition temperature, which is an important parameter for the adsorption behavior at coexistence. The adsorption isotherms in a diluted phase and in a concentrated phase present some very significant features. In diluted solutions the isotherms show a succession of plateaus; these transitions of layers persist in concentrated phases presenting a sequence of maxima and minima of relative adsorption (Gibbs adsorption) of 2,5DMP with respect to water. In concentrated solutions, they interfere with the prewetting that gives an original shape to the isotherms in concentrated solutions. These adsorption measurements were completed by activity measurements which allow one to obtain the chemical potential of mixtures and linked information. The whole study shows the behavior of colloidal suspensions close to bulk phase changes. © 1998 Academic Press Key Words: adsorption; layering; wetting; prewetting; mixtures; phase transitions; critical divergence.
INTRODUCTION
The behavior of a fluid adsorbed on solid substrates is of interest in many fields. Industry uses adsorption in catalysis and separation processes and in the characterization of porous solids. Physicists are interested in the properties of matter on an atomic scale and in two-dimension phase transitions that adsorbed phases undergo. In recent years, much effort has gone into studying structured mixtures and other similar systems. Cahn’s first paper (1) about the wetting transition and its connection with adsorption has been followed by numerous papers, but experiments in a solid/liquid binary mixture did not follow this abundance. Our study is the first experimental evidence of layering and prewetting transitions in such an interface. Figure 1 summarizes Cahn’s ideas. The wetting transition is characterized by a wetting temperature T W. For T C , T , T W, wetting is complete (one of the two liquid phases completely 1
To whom correspondence should be addressed.
0021-9797/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
46
47
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
FIG. 1. Classical Cahn’s diagram for wetting and prewetting. T W is the wetting transition temperature. T C is the critical temperature. The figure is drawn for liquid–liquid demixing of binary mixture with lower critical solution temperature. b is the bulk wetting phase. PW shows the prewetting line which ends at T SC, the surface critical point. Coexistence curve and wetting behavior correspond to the water (1)-2,5DMP(2) system, against silica.
generalization of this in the case of superposition with layering. Layering at solid–liquid interfaces is a new observation but can be considered as a generalization of observations made at solid– gas interfaces (7). The growth of the adsorption of the j-component, G j , in the surface in contact with a single phase close to undergoing a j-rich phase separation, is the classical nucleation idea. To observe the influence of a wetting transition on this is a new approach. Again, to observe surface phase changes other than layering, such as prewetting, is also a new result. The approach of the experimental results obtained on the water–2,5DMP/silica system, which are rather intricate, is simplified when one remembers these references about what was already observed and what is obviously new. Finally, throughout the paper, our attention is focused on the twocomponent aspect of the system. In some systems, and very often in dilute mixtures, the solute can be considered to behave as a gas, and the results or theories made for single gases can be applied. This is far rarer in, say, equimolar mixtures and still rarer when interactions of each component with the surface are different, and the interactions between components are intricate. This is the case for our system, which allows interesting considerations. EXPERIMENTAL
Chemical 2,5DMP is supplied by Aldrich and purified by distillation. Water was purified by Milli Q device. Silica is Aerosil 200 supplied by Degussa. Silica was kept in a dry place; it was not dried before use; several water-content measurements by the Karl Fischer method gave a 3% water content. Adsorption Measurements The adsorption measurements were made by the gravimetric method, by shaking about 20 g of solution with a sample of
about 0.4 g of a fine silica powder, at a fixed temperature until the adsorption equilibrium was reached. Preliminary experiments about the kinetics of adsorption, namely adsorption data versus time obtained from the same initial solution concentration, showed that about 2 h after the contact between solid and solution, data became constant (i.e., the equilibrium was reached when the final concentration was obtained). The contact time, which could have been shorter for diluted solutions, was standardized at 2 h. To avoid any loss of organic compound in vapor, the tubes containing the mixtures were completely filled. After centrifugation, the supernatant was analyzed by ultraviolet absorption spectroscopy or differential refractometry. Because the refractive index is affected by the silica solubility, in that case, samples were prepared with silica saturated water, which was used alone as a reference in the refractometer. The relative adsorption is calculated as G 21 5
n 0~ x 02 2 x 2! , ms~1 2 x 2!
[1]
where n 0 is the total initial mole number of liquid components; x 02 and x 2 are the initial and supernatant mole fractions, respectively; m the mass of silica, and s the specific area, determined by BET adsorption isotherm for nitrogen. G21 was defined by Gibbs (8) as
G 21 5 G 2 2
x2 G x1 1
with
Gi 5
E
`
r i~ z!dz,
[2]
0
where z is the distance from the wall in a direction perpendicular to the surface, and r i ( z) is the individual density profile of species i.
48
HAMRAOUI AND PRIVAT
RESULTS AND DISCUSSION
Adsorption Isotherms Figure 2a shows the adsorption isotherms at 25, 38, 48, and 56°C in diluted solutions between the pure water and coexistence mole fraction range. The adsorption grows layer by layer and increases greatly near bulk coexistence at constant temperature. Figure 2b shows an enlargement of the two first plateaus of the isotherms at 25°C. The height of both plateaus is about 2.9 3 10210 mol z cm22, which is quite a noticeable value, namely the value for a monolayer of 2,5DMP. Error bars, which have been calculated following the usual rules from formula [1] and uncertainties on the experimental parameters are certainly overestimated, as is the case in such estimations. However, it is impossible to ignore the existence of two plateaus in Fig. 2b even when taking into account the overestimated error bars. The same observation could be made for each plateau in Fig. 2a. The enlargement also shows the existence of a metastable second plateau (the highest on Fig. 2b). On the isotherms of Fig. 2, vertical parts lead to each new layer and metastability lines extend almost all plateaus: these are typical characters of surface phase separations. In addition, when T 3 T C, plateaus disappear, and the adsorption diverges (9) with the correlation length which governs the adsorption film’s thickness. These features allow the identification of the stairlike isotherms to layering. From a theoretical point of view, these layering transitions are obtained on solids without any structure, and this is caused by the packing effect (10) in the liquid bulk phase, and the liquid–solid demixing that the adsorbed phase undergoes. This packing effect has been used, theoretically, by Ball and Evans (10) to obtain a layering from a one-component gas system on one structured solid, near a sublimation curve, in contact with a solid surface. It is the layered adsorption which gives the wetting film after the wetting transition. Our case is somewhat different. In our case, we must distinguish between the adsorption modes of the two kinds of molecules. The organic ones obviously adsorb or desorb by a layering mechanism (see empirical model in Ref. 11). In this case, layering results from the orientational and agregative properties of this single molecule because of its p electronic system. The water molecules which are essentially responsible for the wetting and consequently for the prewetting seem to only accompany this behavior of the organic and are able to separate in surface in a different, less structured physic state—liquid instead of solid. It seems that these surface properties are caused by long-range hydrogen bonds. In Fig. 3, we represent isotherms in concentrated solutions. Error bars cannot be shown on the graph scale. Isotherms
FIG. 2. (a) Adsorption isotherm from dilute solutions (wetting bulk phase) at 25, 38, 48, and 56°C, as relative excess of 2,5DMP (2) with respect to water (1), G21, versus the mole fraction of 2 in bulk, at equilibrium. (b) Enlargement of the beginning of the curve at 25°C.
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
49
FIG. 3. (a) Adsorption isotherm as relative excess of 2,5DMP (2) with respect to water (1), G 21 5 f( x 2 ), in concentrated solutions at 40, 44, and 48°C. (b) Magnified adsorption isotherm as relative excess of 2,5DMP (2) with respect to water (1), in concentrated solutions, close to the bulk demixing at 40, 42, 44, 46, and 48°C. The line is just an eye guideline, amongst results showing a typical metastability.
50
HAMRAOUI AND PRIVAT
In order to verify the existence in the system of the molecular packing which could be responsible for layering, we calculate (5) the experimental Kirkwood integrals (12) whose definitions are
G ij 5
E
`
4 p r 2@ g ij~r! 2 1#dr,
i 5 1, 2,
0
FIG. 4. Adsorption at coexistence: (a) at the critical concentration, below T C; (b) at coexistence in the wetting bulk phase showing the critical divergence (continuous) caused by the correlation length divergence at T C; (c) at coexistence in the nonwetting bulk phase, the divergence (discontinuous) is at T W.
exhibit a succession of maxima and minima and sorts of “waves” around a zero G21 value. As for dilute solutions, isotherms exhibit strong metastability phenomena, which makes it difficult to obtain beautifully shaped waves systematically. A typical example is Fig. 3b, 44°C at x 2 5 0.3 where the minimum is indicated by only two points and the curve’s shape has been imagined by analogy with other isotherms. At concentrations where no phase transition occurs as on Fig. 3a, 44°C around x 2 5 0.48, waves are perfectly drawn. As a first step, the study of 2,5DMP adsorption variation along the coexistence curve versus temperature shows a discontinuous divergence of adsorption close to 46°C (9). Figure 4c is the adsorption at coexistence in the nonwetting bulk phase, g, whereas Fig. 4b is in the wetting bulk phase, and Fig. 4a at the critical concentration, below T C. Figures 4a and 4b show the critical divergence at T C caused by the critical density correlation divergence. The divergence in Fig. 4c is at T W; it is linked to the wetting film nucleation whose mechanism is discussed later. The discontinuity observed for this divergence is a very significant result and indicates that the wetting transition is first order and certainly preceded by a prewetting transition. This justifies our undertaking of the prewetting transition study. In a second step, the isotherm’s shape can be analyzed. In concentrated solutions, the negative slopes can be shown to correspond to the achieved surface enrichment by at least one layer (6, 11), and these parts of waves have the same behavior as the plateaus (i.e., shrinking when x 2 tends toward its coexistence value and disappearance close to a bulk critical end point); consequently, the layering is an intrinsic behavior of the system and this is why it persists in concentrated solutions. The positive slopes are caused by the enrichment of the surface with water when one is going to bulk demixing, because minimal G21 are negative. The prewetting transition must be sought on such parts of the isotherm because the adsorbed phase is water-rich near coexistence compositions.
and which measure long range packing of molecules between themselves (G 11 , G 22 ) and around each other (G 12 ). Calculations have been made from experimental data and for water– 2,5DMP mixtures in bulk on both sides of the phase diagram using the chemical potential values (13) through their derivatives versus x 2 (5). In dilute phases, the tendency for 2,5DMP molecules to associate increases close to the coexistence, but this tendency persists in concentrated solutions in bulk phase change proximity. Therefore, as a surface effect, this packing tendency must be the layering cause in the concentrated zone, particularly the formation cause of the second, third, and further layers (10). Starting from this experimental example, it seems that layering is a more general phenomenon than one thought, but its presence in the adsorption of a binary system can take place in two ways according to the interactions specificity of every molecule with the solid. It happens that in our system, the kind of molecules undergoing layering is not the same as the one responsible for prewetting. In cyclohexane–methanol system in contact with vapor, both were identical, so that the theoretical description founded for a single component could be adapted (10, 14). To continue analyzing the isotherms shape, we now consider the impact of several factors. A current isotherm form is given in Fig. 5a where a classical “single wave” s shaped G21 isotherm is shown [i.e., with a single maximum and a single minimum and the possible G2 and 2G1 curves which lead to G21 through the equation G 21 5 G 2 2 ( x 2 /x 1 ) G 1 . In such a representation, the effect of x 2 on the combination of G2 and 2G1 is clearly seen: at low x 2 , G 21 , and G2 are confused; when x 2 is close to 0.5, strong differences appear and G21 may become negative (we chose to show such an example). In addition, the effect of bulk demixing is shown. According to the phase nucleation process in surfaces, G2 grows close to the left-hand side of the miscibility gap (a 2-rich new phase forming), G1 grows close to the right-hand side (a 1-rich new phase is forming). In Fig. 5b, G2 is supposed to form by layering in the whole concentration scale. G1 is supposed to be slightly affected by this, so the result is a succession of maxima and minima on G21: one obtains an isotherm identical to Fig. 5a (without miscibility gap) for each G2’s step, except for the very small x 2 values where G21 exactly reproduces G2 variations. In Fig. 5c, a surface phase transition is shown both on G1 and G21 (normally if it is a liquid–liquid demixing, it also appears
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
51
FIG. 5. (a) “Single wave”-shaped isotherm (i.e., single maximum and a single minimum) and the possible G2 and 2G1 curves which lead to G21. (b) Isotherm “wavelike” form with a series of maxima and minima on G21, which result from G2 supposed to form by layering and G1 supposed to be slightly affected by this so that the result is a series of maxima and minima on G21. (c) A surface phase transition is shown both on G1 and G21 (normally if it is a liquid–liquid demixing, it appears also on G2) for curves a and a9; only on G1 for curves b and c on the right-hand side of the miscibility gap, close to coexistence. Such a behavior (G1 only affected) is rather expected in 1-rich mixtures.
on G2). Such a behavior (G1 only affected) is rather expected in 1-rich mixtures. Superposition between phenomena of Figs. 4b and 4c is also to be expected and corresponds to a physical reality in some conditions in our system. According to this analysis, if we only had layering in concentrated isotherms, our isotherm’s shape would take the form plotted in Fig. 5b. In Fig. 5c, we have the scheme when only the prewetting behavior appears; as T approaches T W from below, the prewetting step in the isotherm occurs. It becomes closer to x 5 x co (coexistence molar fraction) as T becomes closer to T W whereas the associated jump in adsorption increases in value and possibly diverges at T 5 T W. In binary liquid mixtures, if the molecules in majority in the wetting film
were responsible for layering, we could have a prewetting transition occurring by layering. The wetting film formation would start by a pure monolayer in the bulk single phase and grow layer by layer until macroscopic thickness in the twophase bulk. This is probably what was observed by Bonn et al. (14) for the cyclohexane–methanol mixture at the liquid–vapor interface. If the molecules in majority in the wetting film do not undergo layering but the other, the detailed wave shape of the isotherm should be affected in the single bulk phase diagram zone where prewetting should occur. Some diagrams must be used to prove the modification in wavelike isotherm, and we used some with our system as shown next. The theory predicts that the prewetting transition occurs between the T W wetting temperature transition and the T SC
52
HAMRAOUI AND PRIVAT
CONCLUSION
FIG. 6. Layering steps width for G2 at several temperatures. When T 3 T C, the step width tends to zero. Between 40 and 46°C, the points cross the ordinary curve (11).
surface critical temperature, and their corresponding compositions. To experimentally localize the prewetting transition, we use a property of layering, which is the underlying disappearance of adsorption plateaus, when T 3 T C, and its interference with prewetting. To do this, we plotted the first plateau’s width [extracted from the form of waves and the model (5, 11)] versus temperature in Fig. 6. If we had pure layering, the curve in Fig. 6 should tend toward zero continuously; but in reality this tendency is upset between 40 and 46°C, and the width becomes zero at 46°C. This shows that at this temperature the first minimum (starting from the coexistence curve) is on the coexistence curve, corresponding to strong adsorption of water on the surface, which confirms the wetting phase germination notion, on the surface, in the bulk single phase and in the demixing route. Now we experimentally localize the still supposed prewetting transition between 40 and 46°C. The examination of isotherms at 40, 42, 44, and 46°C shows that vertical parts exist in this isotherm, that they disappear for T . 46°C and T , 40°C or at least change in a significant way. These vertical segments reveal surface phase transitions whose interpretation is given by our model. Starting from bulk coexistence, the first is caused by layering (i.e., surface liquid–solid demixing), the second by prewetting (i.e., surface liquid–liquid demixing), and so on. All this happens as though the surface system oscillated on a surface phase diagram showing a eutectic curve interrupted by a liquid–liquid demixing loop (11). In Figs. 7 and 8, we have plotted the experimental surface phase diagram of the identified surface liquid–liquid (i.e., prewetting transition). The surface critical temperature is 40°C where both adsorbed water-rich and 2,5DMP phases become indistinguishable, and 46°C is the wetting temperature transition (15). The diagram form is very similar to the liquid–liquid bulk phase diagram: a lower consolute point and high dissymetry of the curve, with a steep aisle on the 2,5DMP-poor concentration side. This may be a confirmation of the prewetting hypothesis as a liquid–liquid surface demixing.
These experimental studies of physical adsorption from a binary mixture on a colloidal suspension of silica shed a new light on the factors that can have influence. The use of Aerosil 200, whose surface is very homogeneous and whose porosity is very weak, allowed the observation of two kinds of surface phase changes, layering and prewetting. This, known at the solid– gas interfaces, was first observed at solid–liquid interfaces. Bulk demixing has been shown to have a strong influence on adsorption. Already observed, the critical adsorption (16) has been shown in a binary mixture, with the dissymetry induced by the existence of a wetting transition: only the adsorption in the phase constantly wetting silica diverges at T C. The other diverges at T W. Close to the coexistence, and along a given isotherm, the bulk demixing approach leads to remarkable adsorption enhancements, stronger as they get closer to T C. The wetting transition in the bulk two-phase system in contact with silica not only gives nonsymmetrical behavior to the adsorption at coexistence but also allows a prewetting transition. It manifests itself at a very low temperature and concentration range but has drastic effects on the isotherm’s shape and the adsorption values. This study confirms several theoretical views on the influential relation between wetting, prewetting, and phase changes on adsorption. The results oblige the scientist to take into account the double nature of molecules constituting the system: a simple transposition of a qualitative behavior of a single system cannot be made. Among the particle correlations to take
FIG. 7. Experimental surface phase diagram constituted from x s2 5 f( x 2 ) adsorption isotherms (5, 11), corresponding to a liquid–liquid surface demixing. x s2 is calculated according to a layering model for G2 (5); values have been selected on the second vertical segment appearing on isotherms at 40, 42, 44, and 46°C (Fig. 3b), which are attributed to liquid–liquid surface demixing. x s2 is drawn versus RT ln( x 2 /x co) 5 m 2 2 m co 2 5 D m . A liquid–liquid coexistence curve appears, with a critical point close to 40°C.
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
53
FIG. 8. (a) Experimental Cahn’s diagram, the solid line is the bulk phase diagram, and triangles represent the prewetting line. This line separates weak water adsorption zone on the left-hand side, from the strong adsorption zone. T W is the wetting transition temperature. a is silica, b is a bulk water-rich phase, and g is a bulk 2,5DMP-rich phase. (b) The same figure is magnified close to prewetting, with supplementary transition lines. Line 1 represents the first liquid–liquid surface transition [i.e., prewetting (second vertical line on the isotherms)]. Consequently, line 1 separates adsorption zones with five and four 2,5DMP layers; line 2 separates thick water-rich and water-less-rich zone and so on.
into account, 1–1, 2–2, and 1–2 correlations must be separately analyzed to explain results. The water–2,5DMP system shows in addition the importance of analyzing interactions in chemical terms: nothing can be understood in the surface packing tendency of 2,5DMP without taking into consideration the p electron properties of the molecule; wetting must be thought of in terms of hydrogen bonds, both between water molecules, water and 2,5DMP molecules, and water and silica. A lot of studies will be able to be carried out, either with other methods on the same system or on other systems undergoing demixing. The use of such surprising properties in applied fields is, as a result, worth undertaking. REFERENCES 1. Cahn, J. W., J. Chem. Phys. 66, 3667 (1977). 2. Dash, J. G., Phys. Rev. B 15, 3136 (1997). 3. Sullivan, D. E., Phys. Rev. B 20, 3991 (1979); J. Chem. Phys. 74, 2604 (1981); Faraday Symp. 16, 191 (1982).
4. Telo da Gama, M. M., and Evans, R., Mol. Phys. 48, 687 (1983). 5. Hamraoui, A., Privat, M., and Sellami, H., J. Chem. Phys. 106, 222 (1997). 6. Lane, J. E., in “Adsorption from Solution at the Solid/Liquid Interface” (G. D. Parfitt and G. H. Rochester, Eds.). Academic Press, London, 1983. 7. Duval, X., and Thomy, A., J. Chim. Phys. 67, 1101 (1970). 8. Defay, R., and Prigogine, I., with A. Bellemans, and translated by D. H. Everett, “Surface Tension and Adsorption.” Longmans, London, 1966. 9. Privat, M., Amara, M., Hamraoui, A., Sellami, H., and Mear, A. M., Ber. Bunsenges. Phys. Chem. 98, 620 (1994). 10. Ball, P. C., and Evans, R., J. Chem. Phys. 98, 4412 (1988). 11. Hamraoui, A., and Privat, M., J. Chem. Phys. 107, 6936 (1997). 12. Kirkwood, J. G., and Buff, F. P., J. Chem. Phys. 19, 774 (1951). 13. Bassiloua, V., Ghaicha, L., Privat, M., R. Bennes, and Tronel-Peyroz, E., J. Solut. Chem. 24, 935 (1995). 14. Bonn, D., Wegdam, G. H., Kellay, H., and Nieuwenhuizen, Th. M., Europhys. Lett. 20, 235 (1992). 15. Amara, M., Privat, M., Bennes, R., and Tronel-Peyroz, E., J. Chem. Phys. 98, 5028 (1993). 16. Privat, M., Amara, M., Bassiloua, V., Bennes, R., and Tronel-Peyroz, E., Langmuir 10, 3770 (1994).
207, 46 –53 (1998)
CS985542
Impact of Bulk Phase Transitions and Wetting Transitions on Adsorption Isotherms in Binary Systems in Contact with Solid A. Hamraoui and M. Privat1 UMR CNRS 6521, De´partement de Chimie UBO, 6 avenue Le Gorgeu, B.P. 809, 29285 Brest Cedex, France Received October 3, 1997; accepted March 26, 1998
wets the wall), for T . T W wetting is incomplete (there is a nonzero contact angle between the two liquid phases against the wall). The wetting transition occurs between class I (complete wetting) and IIA (partial wetting), according to Telo da Gama and Evans’ classification. The notion of wetting class was introduced by Dash (2) and developed by Sullivan (3) for one component system and extended to binary mixture by Telo da Gama and Evans (4). As a consequence of the one density character of their model, Telo da Gama and Evans found in such a case that the bulk wetting phase is the preferentially adsorbed species-rich phase; we have shown in our experiments that the 2,5DMP is preferentially adsorbed from a dilute phase (5) (from a water-rich phase), although this water-rich phase is the bulk wetting phase, which shows the incontestable two-density character of our system. When the wetting transition is first order, it is preceded by a transition called prewetting, whose existence in the phase diagram is represented by the prewetting line (PW) in Fig. 1. This line separates weak and strong adsorption zones. When passing the prewetting line, the adsorbed phase undergoes a phase change in surface. The surface phase transition is similar to the bulk phase transition (gas–liquid, liquid–liquid, solid– liquid), and it leads to the prewetting transition which is equivalent to a germination, in surface, of the wetting bulk phase. In our case, prewetting must be a liquid–liquid transition because the wetting transition occurs along a liquid–liquid coexistence. Understanding the mechanism of the prewetting transition, when it exists, for systems such as silica/water– 2,5DMP helps to make the connection between interfacial and macroscopic properties. The surface can undergo other phase transitions: a liquid– solid transition may appear beside a liquid–liquid one. Both can lead to layering, even if liquid–solid separations are preferentially observed in the case of layering. To distinguish between them may be the task of an experimentalist. He must also try to connect with the interactional properties of the system. In the results, some of them should be considered as classical, or as the generalization of classical results. Single wave isotherms (i.e. with a maximum and a minimum) have been known for a long time (6). Several wave isotherms are a
This paper presents several studies essentially on the solid/liquid interface. The studied system is the liquid binary mixture water– 2,5dimethylpyridine (2,5DMP) in contact with an amorphous silica (Aerosil type). The study of wetting along the two-liquid bulk phases coexistence in contact with silica allows the determination of a wetting transition temperature, which is an important parameter for the adsorption behavior at coexistence. The adsorption isotherms in a diluted phase and in a concentrated phase present some very significant features. In diluted solutions the isotherms show a succession of plateaus; these transitions of layers persist in concentrated phases presenting a sequence of maxima and minima of relative adsorption (Gibbs adsorption) of 2,5DMP with respect to water. In concentrated solutions, they interfere with the prewetting that gives an original shape to the isotherms in concentrated solutions. These adsorption measurements were completed by activity measurements which allow one to obtain the chemical potential of mixtures and linked information. The whole study shows the behavior of colloidal suspensions close to bulk phase changes. © 1998 Academic Press Key Words: adsorption; layering; wetting; prewetting; mixtures; phase transitions; critical divergence.
INTRODUCTION
The behavior of a fluid adsorbed on solid substrates is of interest in many fields. Industry uses adsorption in catalysis and separation processes and in the characterization of porous solids. Physicists are interested in the properties of matter on an atomic scale and in two-dimension phase transitions that adsorbed phases undergo. In recent years, much effort has gone into studying structured mixtures and other similar systems. Cahn’s first paper (1) about the wetting transition and its connection with adsorption has been followed by numerous papers, but experiments in a solid/liquid binary mixture did not follow this abundance. Our study is the first experimental evidence of layering and prewetting transitions in such an interface. Figure 1 summarizes Cahn’s ideas. The wetting transition is characterized by a wetting temperature T W. For T C , T , T W, wetting is complete (one of the two liquid phases completely 1
To whom correspondence should be addressed.
0021-9797/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
46
47
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
FIG. 1. Classical Cahn’s diagram for wetting and prewetting. T W is the wetting transition temperature. T C is the critical temperature. The figure is drawn for liquid–liquid demixing of binary mixture with lower critical solution temperature. b is the bulk wetting phase. PW shows the prewetting line which ends at T SC, the surface critical point. Coexistence curve and wetting behavior correspond to the water (1)-2,5DMP(2) system, against silica.
generalization of this in the case of superposition with layering. Layering at solid–liquid interfaces is a new observation but can be considered as a generalization of observations made at solid– gas interfaces (7). The growth of the adsorption of the j-component, G j , in the surface in contact with a single phase close to undergoing a j-rich phase separation, is the classical nucleation idea. To observe the influence of a wetting transition on this is a new approach. Again, to observe surface phase changes other than layering, such as prewetting, is also a new result. The approach of the experimental results obtained on the water–2,5DMP/silica system, which are rather intricate, is simplified when one remembers these references about what was already observed and what is obviously new. Finally, throughout the paper, our attention is focused on the twocomponent aspect of the system. In some systems, and very often in dilute mixtures, the solute can be considered to behave as a gas, and the results or theories made for single gases can be applied. This is far rarer in, say, equimolar mixtures and still rarer when interactions of each component with the surface are different, and the interactions between components are intricate. This is the case for our system, which allows interesting considerations. EXPERIMENTAL
Chemical 2,5DMP is supplied by Aldrich and purified by distillation. Water was purified by Milli Q device. Silica is Aerosil 200 supplied by Degussa. Silica was kept in a dry place; it was not dried before use; several water-content measurements by the Karl Fischer method gave a 3% water content. Adsorption Measurements The adsorption measurements were made by the gravimetric method, by shaking about 20 g of solution with a sample of
about 0.4 g of a fine silica powder, at a fixed temperature until the adsorption equilibrium was reached. Preliminary experiments about the kinetics of adsorption, namely adsorption data versus time obtained from the same initial solution concentration, showed that about 2 h after the contact between solid and solution, data became constant (i.e., the equilibrium was reached when the final concentration was obtained). The contact time, which could have been shorter for diluted solutions, was standardized at 2 h. To avoid any loss of organic compound in vapor, the tubes containing the mixtures were completely filled. After centrifugation, the supernatant was analyzed by ultraviolet absorption spectroscopy or differential refractometry. Because the refractive index is affected by the silica solubility, in that case, samples were prepared with silica saturated water, which was used alone as a reference in the refractometer. The relative adsorption is calculated as G 21 5
n 0~ x 02 2 x 2! , ms~1 2 x 2!
[1]
where n 0 is the total initial mole number of liquid components; x 02 and x 2 are the initial and supernatant mole fractions, respectively; m the mass of silica, and s the specific area, determined by BET adsorption isotherm for nitrogen. G21 was defined by Gibbs (8) as
G 21 5 G 2 2
x2 G x1 1
with
Gi 5
E
`
r i~ z!dz,
[2]
0
where z is the distance from the wall in a direction perpendicular to the surface, and r i ( z) is the individual density profile of species i.
48
HAMRAOUI AND PRIVAT
RESULTS AND DISCUSSION
Adsorption Isotherms Figure 2a shows the adsorption isotherms at 25, 38, 48, and 56°C in diluted solutions between the pure water and coexistence mole fraction range. The adsorption grows layer by layer and increases greatly near bulk coexistence at constant temperature. Figure 2b shows an enlargement of the two first plateaus of the isotherms at 25°C. The height of both plateaus is about 2.9 3 10210 mol z cm22, which is quite a noticeable value, namely the value for a monolayer of 2,5DMP. Error bars, which have been calculated following the usual rules from formula [1] and uncertainties on the experimental parameters are certainly overestimated, as is the case in such estimations. However, it is impossible to ignore the existence of two plateaus in Fig. 2b even when taking into account the overestimated error bars. The same observation could be made for each plateau in Fig. 2a. The enlargement also shows the existence of a metastable second plateau (the highest on Fig. 2b). On the isotherms of Fig. 2, vertical parts lead to each new layer and metastability lines extend almost all plateaus: these are typical characters of surface phase separations. In addition, when T 3 T C, plateaus disappear, and the adsorption diverges (9) with the correlation length which governs the adsorption film’s thickness. These features allow the identification of the stairlike isotherms to layering. From a theoretical point of view, these layering transitions are obtained on solids without any structure, and this is caused by the packing effect (10) in the liquid bulk phase, and the liquid–solid demixing that the adsorbed phase undergoes. This packing effect has been used, theoretically, by Ball and Evans (10) to obtain a layering from a one-component gas system on one structured solid, near a sublimation curve, in contact with a solid surface. It is the layered adsorption which gives the wetting film after the wetting transition. Our case is somewhat different. In our case, we must distinguish between the adsorption modes of the two kinds of molecules. The organic ones obviously adsorb or desorb by a layering mechanism (see empirical model in Ref. 11). In this case, layering results from the orientational and agregative properties of this single molecule because of its p electronic system. The water molecules which are essentially responsible for the wetting and consequently for the prewetting seem to only accompany this behavior of the organic and are able to separate in surface in a different, less structured physic state—liquid instead of solid. It seems that these surface properties are caused by long-range hydrogen bonds. In Fig. 3, we represent isotherms in concentrated solutions. Error bars cannot be shown on the graph scale. Isotherms
FIG. 2. (a) Adsorption isotherm from dilute solutions (wetting bulk phase) at 25, 38, 48, and 56°C, as relative excess of 2,5DMP (2) with respect to water (1), G21, versus the mole fraction of 2 in bulk, at equilibrium. (b) Enlargement of the beginning of the curve at 25°C.
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
49
FIG. 3. (a) Adsorption isotherm as relative excess of 2,5DMP (2) with respect to water (1), G 21 5 f( x 2 ), in concentrated solutions at 40, 44, and 48°C. (b) Magnified adsorption isotherm as relative excess of 2,5DMP (2) with respect to water (1), in concentrated solutions, close to the bulk demixing at 40, 42, 44, 46, and 48°C. The line is just an eye guideline, amongst results showing a typical metastability.
50
HAMRAOUI AND PRIVAT
In order to verify the existence in the system of the molecular packing which could be responsible for layering, we calculate (5) the experimental Kirkwood integrals (12) whose definitions are
G ij 5
E
`
4 p r 2@ g ij~r! 2 1#dr,
i 5 1, 2,
0
FIG. 4. Adsorption at coexistence: (a) at the critical concentration, below T C; (b) at coexistence in the wetting bulk phase showing the critical divergence (continuous) caused by the correlation length divergence at T C; (c) at coexistence in the nonwetting bulk phase, the divergence (discontinuous) is at T W.
exhibit a succession of maxima and minima and sorts of “waves” around a zero G21 value. As for dilute solutions, isotherms exhibit strong metastability phenomena, which makes it difficult to obtain beautifully shaped waves systematically. A typical example is Fig. 3b, 44°C at x 2 5 0.3 where the minimum is indicated by only two points and the curve’s shape has been imagined by analogy with other isotherms. At concentrations where no phase transition occurs as on Fig. 3a, 44°C around x 2 5 0.48, waves are perfectly drawn. As a first step, the study of 2,5DMP adsorption variation along the coexistence curve versus temperature shows a discontinuous divergence of adsorption close to 46°C (9). Figure 4c is the adsorption at coexistence in the nonwetting bulk phase, g, whereas Fig. 4b is in the wetting bulk phase, and Fig. 4a at the critical concentration, below T C. Figures 4a and 4b show the critical divergence at T C caused by the critical density correlation divergence. The divergence in Fig. 4c is at T W; it is linked to the wetting film nucleation whose mechanism is discussed later. The discontinuity observed for this divergence is a very significant result and indicates that the wetting transition is first order and certainly preceded by a prewetting transition. This justifies our undertaking of the prewetting transition study. In a second step, the isotherm’s shape can be analyzed. In concentrated solutions, the negative slopes can be shown to correspond to the achieved surface enrichment by at least one layer (6, 11), and these parts of waves have the same behavior as the plateaus (i.e., shrinking when x 2 tends toward its coexistence value and disappearance close to a bulk critical end point); consequently, the layering is an intrinsic behavior of the system and this is why it persists in concentrated solutions. The positive slopes are caused by the enrichment of the surface with water when one is going to bulk demixing, because minimal G21 are negative. The prewetting transition must be sought on such parts of the isotherm because the adsorbed phase is water-rich near coexistence compositions.
and which measure long range packing of molecules between themselves (G 11 , G 22 ) and around each other (G 12 ). Calculations have been made from experimental data and for water– 2,5DMP mixtures in bulk on both sides of the phase diagram using the chemical potential values (13) through their derivatives versus x 2 (5). In dilute phases, the tendency for 2,5DMP molecules to associate increases close to the coexistence, but this tendency persists in concentrated solutions in bulk phase change proximity. Therefore, as a surface effect, this packing tendency must be the layering cause in the concentrated zone, particularly the formation cause of the second, third, and further layers (10). Starting from this experimental example, it seems that layering is a more general phenomenon than one thought, but its presence in the adsorption of a binary system can take place in two ways according to the interactions specificity of every molecule with the solid. It happens that in our system, the kind of molecules undergoing layering is not the same as the one responsible for prewetting. In cyclohexane–methanol system in contact with vapor, both were identical, so that the theoretical description founded for a single component could be adapted (10, 14). To continue analyzing the isotherms shape, we now consider the impact of several factors. A current isotherm form is given in Fig. 5a where a classical “single wave” s shaped G21 isotherm is shown [i.e., with a single maximum and a single minimum and the possible G2 and 2G1 curves which lead to G21 through the equation G 21 5 G 2 2 ( x 2 /x 1 ) G 1 . In such a representation, the effect of x 2 on the combination of G2 and 2G1 is clearly seen: at low x 2 , G 21 , and G2 are confused; when x 2 is close to 0.5, strong differences appear and G21 may become negative (we chose to show such an example). In addition, the effect of bulk demixing is shown. According to the phase nucleation process in surfaces, G2 grows close to the left-hand side of the miscibility gap (a 2-rich new phase forming), G1 grows close to the right-hand side (a 1-rich new phase is forming). In Fig. 5b, G2 is supposed to form by layering in the whole concentration scale. G1 is supposed to be slightly affected by this, so the result is a succession of maxima and minima on G21: one obtains an isotherm identical to Fig. 5a (without miscibility gap) for each G2’s step, except for the very small x 2 values where G21 exactly reproduces G2 variations. In Fig. 5c, a surface phase transition is shown both on G1 and G21 (normally if it is a liquid–liquid demixing, it also appears
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
51
FIG. 5. (a) “Single wave”-shaped isotherm (i.e., single maximum and a single minimum) and the possible G2 and 2G1 curves which lead to G21. (b) Isotherm “wavelike” form with a series of maxima and minima on G21, which result from G2 supposed to form by layering and G1 supposed to be slightly affected by this so that the result is a series of maxima and minima on G21. (c) A surface phase transition is shown both on G1 and G21 (normally if it is a liquid–liquid demixing, it appears also on G2) for curves a and a9; only on G1 for curves b and c on the right-hand side of the miscibility gap, close to coexistence. Such a behavior (G1 only affected) is rather expected in 1-rich mixtures.
on G2). Such a behavior (G1 only affected) is rather expected in 1-rich mixtures. Superposition between phenomena of Figs. 4b and 4c is also to be expected and corresponds to a physical reality in some conditions in our system. According to this analysis, if we only had layering in concentrated isotherms, our isotherm’s shape would take the form plotted in Fig. 5b. In Fig. 5c, we have the scheme when only the prewetting behavior appears; as T approaches T W from below, the prewetting step in the isotherm occurs. It becomes closer to x 5 x co (coexistence molar fraction) as T becomes closer to T W whereas the associated jump in adsorption increases in value and possibly diverges at T 5 T W. In binary liquid mixtures, if the molecules in majority in the wetting film
were responsible for layering, we could have a prewetting transition occurring by layering. The wetting film formation would start by a pure monolayer in the bulk single phase and grow layer by layer until macroscopic thickness in the twophase bulk. This is probably what was observed by Bonn et al. (14) for the cyclohexane–methanol mixture at the liquid–vapor interface. If the molecules in majority in the wetting film do not undergo layering but the other, the detailed wave shape of the isotherm should be affected in the single bulk phase diagram zone where prewetting should occur. Some diagrams must be used to prove the modification in wavelike isotherm, and we used some with our system as shown next. The theory predicts that the prewetting transition occurs between the T W wetting temperature transition and the T SC
52
HAMRAOUI AND PRIVAT
CONCLUSION
FIG. 6. Layering steps width for G2 at several temperatures. When T 3 T C, the step width tends to zero. Between 40 and 46°C, the points cross the ordinary curve (11).
surface critical temperature, and their corresponding compositions. To experimentally localize the prewetting transition, we use a property of layering, which is the underlying disappearance of adsorption plateaus, when T 3 T C, and its interference with prewetting. To do this, we plotted the first plateau’s width [extracted from the form of waves and the model (5, 11)] versus temperature in Fig. 6. If we had pure layering, the curve in Fig. 6 should tend toward zero continuously; but in reality this tendency is upset between 40 and 46°C, and the width becomes zero at 46°C. This shows that at this temperature the first minimum (starting from the coexistence curve) is on the coexistence curve, corresponding to strong adsorption of water on the surface, which confirms the wetting phase germination notion, on the surface, in the bulk single phase and in the demixing route. Now we experimentally localize the still supposed prewetting transition between 40 and 46°C. The examination of isotherms at 40, 42, 44, and 46°C shows that vertical parts exist in this isotherm, that they disappear for T . 46°C and T , 40°C or at least change in a significant way. These vertical segments reveal surface phase transitions whose interpretation is given by our model. Starting from bulk coexistence, the first is caused by layering (i.e., surface liquid–solid demixing), the second by prewetting (i.e., surface liquid–liquid demixing), and so on. All this happens as though the surface system oscillated on a surface phase diagram showing a eutectic curve interrupted by a liquid–liquid demixing loop (11). In Figs. 7 and 8, we have plotted the experimental surface phase diagram of the identified surface liquid–liquid (i.e., prewetting transition). The surface critical temperature is 40°C where both adsorbed water-rich and 2,5DMP phases become indistinguishable, and 46°C is the wetting temperature transition (15). The diagram form is very similar to the liquid–liquid bulk phase diagram: a lower consolute point and high dissymetry of the curve, with a steep aisle on the 2,5DMP-poor concentration side. This may be a confirmation of the prewetting hypothesis as a liquid–liquid surface demixing.
These experimental studies of physical adsorption from a binary mixture on a colloidal suspension of silica shed a new light on the factors that can have influence. The use of Aerosil 200, whose surface is very homogeneous and whose porosity is very weak, allowed the observation of two kinds of surface phase changes, layering and prewetting. This, known at the solid– gas interfaces, was first observed at solid–liquid interfaces. Bulk demixing has been shown to have a strong influence on adsorption. Already observed, the critical adsorption (16) has been shown in a binary mixture, with the dissymetry induced by the existence of a wetting transition: only the adsorption in the phase constantly wetting silica diverges at T C. The other diverges at T W. Close to the coexistence, and along a given isotherm, the bulk demixing approach leads to remarkable adsorption enhancements, stronger as they get closer to T C. The wetting transition in the bulk two-phase system in contact with silica not only gives nonsymmetrical behavior to the adsorption at coexistence but also allows a prewetting transition. It manifests itself at a very low temperature and concentration range but has drastic effects on the isotherm’s shape and the adsorption values. This study confirms several theoretical views on the influential relation between wetting, prewetting, and phase changes on adsorption. The results oblige the scientist to take into account the double nature of molecules constituting the system: a simple transposition of a qualitative behavior of a single system cannot be made. Among the particle correlations to take
FIG. 7. Experimental surface phase diagram constituted from x s2 5 f( x 2 ) adsorption isotherms (5, 11), corresponding to a liquid–liquid surface demixing. x s2 is calculated according to a layering model for G2 (5); values have been selected on the second vertical segment appearing on isotherms at 40, 42, 44, and 46°C (Fig. 3b), which are attributed to liquid–liquid surface demixing. x s2 is drawn versus RT ln( x 2 /x co) 5 m 2 2 m co 2 5 D m . A liquid–liquid coexistence curve appears, with a critical point close to 40°C.
ADSORPTION ISOTHERMS IN BINARY SYSTEMS
53
FIG. 8. (a) Experimental Cahn’s diagram, the solid line is the bulk phase diagram, and triangles represent the prewetting line. This line separates weak water adsorption zone on the left-hand side, from the strong adsorption zone. T W is the wetting transition temperature. a is silica, b is a bulk water-rich phase, and g is a bulk 2,5DMP-rich phase. (b) The same figure is magnified close to prewetting, with supplementary transition lines. Line 1 represents the first liquid–liquid surface transition [i.e., prewetting (second vertical line on the isotherms)]. Consequently, line 1 separates adsorption zones with five and four 2,5DMP layers; line 2 separates thick water-rich and water-less-rich zone and so on.
into account, 1–1, 2–2, and 1–2 correlations must be separately analyzed to explain results. The water–2,5DMP system shows in addition the importance of analyzing interactions in chemical terms: nothing can be understood in the surface packing tendency of 2,5DMP without taking into consideration the p electron properties of the molecule; wetting must be thought of in terms of hydrogen bonds, both between water molecules, water and 2,5DMP molecules, and water and silica. A lot of studies will be able to be carried out, either with other methods on the same system or on other systems undergoing demixing. The use of such surprising properties in applied fields is, as a result, worth undertaking. REFERENCES 1. Cahn, J. W., J. Chem. Phys. 66, 3667 (1977). 2. Dash, J. G., Phys. Rev. B 15, 3136 (1997). 3. Sullivan, D. E., Phys. Rev. B 20, 3991 (1979); J. Chem. Phys. 74, 2604 (1981); Faraday Symp. 16, 191 (1982).
4. Telo da Gama, M. M., and Evans, R., Mol. Phys. 48, 687 (1983). 5. Hamraoui, A., Privat, M., and Sellami, H., J. Chem. Phys. 106, 222 (1997). 6. Lane, J. E., in “Adsorption from Solution at the Solid/Liquid Interface” (G. D. Parfitt and G. H. Rochester, Eds.). Academic Press, London, 1983. 7. Duval, X., and Thomy, A., J. Chim. Phys. 67, 1101 (1970). 8. Defay, R., and Prigogine, I., with A. Bellemans, and translated by D. H. Everett, “Surface Tension and Adsorption.” Longmans, London, 1966. 9. Privat, M., Amara, M., Hamraoui, A., Sellami, H., and Mear, A. M., Ber. Bunsenges. Phys. Chem. 98, 620 (1994). 10. Ball, P. C., and Evans, R., J. Chem. Phys. 98, 4412 (1988). 11. Hamraoui, A., and Privat, M., J. Chem. Phys. 107, 6936 (1997). 12. Kirkwood, J. G., and Buff, F. P., J. Chem. Phys. 19, 774 (1951). 13. Bassiloua, V., Ghaicha, L., Privat, M., R. Bennes, and Tronel-Peyroz, E., J. Solut. Chem. 24, 935 (1995). 14. Bonn, D., Wegdam, G. H., Kellay, H., and Nieuwenhuizen, Th. M., Europhys. Lett. 20, 235 (1992). 15. Amara, M., Privat, M., Bennes, R., and Tronel-Peyroz, E., J. Chem. Phys. 98, 5028 (1993). 16. Privat, M., Amara, M., Bassiloua, V., Bennes, R., and Tronel-Peyroz, E., Langmuir 10, 3770 (1994).