1326
Langmuir 1988, 4 , 1326-1331
the optical absorption is used to determine nNa3. The solution to these three quantities provides values for N , n, and a. Some words of caution are in order here. For a given size of agglomerate and primary particle size, it may or may not be possible to make both small q and R,q > 1 measurements with visible light. Also, polydispersity of the sample will complicate the analysis in some situations. Finally, the procedure requires absolute intensities to obtain n W a 6 , so additional experimental effort may be involved. Generally, the results reported in the literature for measurements in flames are based on treating the soot particles as spheres. For example, Santoro et al.3 utilize Rayleigh scattering theory for a sphere to infer the particle volume fraction, volume equivalent diameter D, and number concentration n on the basis of light extinction measurements and on V-V scattering measurements at 90". A qualitative indication of the effect of the scattering model on the inferred results can be obtained from the curves in Figure 3. We assume lo9 clusters/cm3 with a = 25 nm and h = 500 nm for each of the three cluster sizes, 12,32, and 119 primary particles. We treat the calculated scattering curves as data utilizing the method of Santoro
Table I agglomerate model
N 12 32 119
effective sphere model D, nm n, da 9.8 x 10-8 55 1.1 x 109 2.6 x 10-7 73 1.3 x 109 9.7 x 10-1 91 2.5 x 109
D, nm 57 79 123
n, cm-3 1.0 x 109 1.0 x 109 1.0 x 109
Both models predict the same value for the volume fraction.
et al. to infer the average particle size and number concentration. The particle volume fraction 4 is the same for the effective and agglomerate models since the light extinction measurement for strongly absorbing small particles is independent of the particle structure. As indicated in Table I, the effect of structure becomes pronounced for clusters of size 30 or more and affects the inferred number concentration to the greatest extent. The effective sphere model overestimates the number concentration for cluster size 119 by a factor of 2.5. Perhaps a more severe limitation of the effective sphere model is that it provides no information regarding the size of the primary spheres, and this quantity is of greatest interest in regard to surface growth and oxidation of soot.
Phase Transitions at the Silica/Water-2,6-Lutidine Mixture Interface L. Ghaicha, M. Privat," L. Tenebre, R. Bennes, E. Tronel-Peyroz, and J. M. Douillard Laboratoire de Physico-Chimie des Systdmes Polyphas&, U A 330, CNRS-Route de Mende, B.P. 5051, 34033 Montpellier, Cedex, France Received April 5, 1988. I n Final Form: July 14, 1988 The interface between silica and water-2,6-lutidine mixtures has been.studied at several concentrations and temperatures near the conditions of the wetting transition. Ellipticity measurements on a plane silica surface, as well as adsorption measurements on silica microspheres, show a transition for several concentrations when the temperature is raised. Results are correlated with others from the literature and qualitatively explained with the Cahn-Rowlinson-Widom theory.
I. Introduction Phase transitions at the surfaces are a phenomenon already known in very particular cases. The adsorption of gases in the surface of solids is known to give a transition between a gaseous state and a liquid state of the surface phase.' Spread layers of fatty molecules onto water may be considered as gaseous phases or condensed phases, with a phase transition from one to another2 when the surface pressure is varied. Fairly recently, another case was considered: the case of the wetting transitions. If two liquid phases, a and /3, at mechanical and thermodynamical equilibrium, are imperfectly wetting a third (solid) phase, y, there exists a contact angle B between the three phases. By varying the thermodynamical conditions (phase composition and/or temperature) one may observe the intrusion of, say, the /3 phase between the a and the y phases. Finally, the (1) Larher, Y.; Haranger, D. Surf. Sci. 1973, 39, 100 (characteristic
work amongst numerous others). (2) Defay, R.; Prigogine, I.; Bellemans, A.; Everett, D. H. Surface Tension and Adsorption; Longmans: London, 1966.
0743-7463/88/2404-1326$01.50/0
surface layer on the y solid phase may be suddenly changed; i.e., a surface phase transition may O C C U ~ . ~ These three kinds of surface transitions have been very frequently observed. There is another one more awkward to observe. This is a phase transition or surface transition in the layer at the contact between a solid phase and liquid mixture. The question arises when one observes the adsorption isotherm. Frumkin himself asked the q ~ e s t i o nand , ~ two of us also asked the question in the particular case of the adsorption isotherm a t the liquid vapor interface.6 Independently, in a theory of the van der Waals type, Cahn3 suggested an explanation of the wetting transition and also a prediction, in this case, of a surface transition in the monophase preceding the wetting. He called this the "prewetting" transition. (3) Cahn, J. W. J. Chem. Phys. 1977,66, 3667. (4) Moldover, M. R.; Cahn, J. W. Science (Washington, D.C.)1980, 207, 1073. (5) Frumkin, A. N. Zh.Fiz. Khim. 1938, 12, 337. (6) Privat, M.; Bennes, R. J. Colloid Interface Sci. 1982, 90, 454.
0 1988 American Chemical Society
~ ~
Phase Transitions at SilicalWater-2,6-Lutidine Interface
Langmuir, Vol. 4, No. 6, 1988 1321
The first version of the theory has been reworked and extended by Rowlinson and Widom.' We give below a summary of this new version, because it can be extended to the study of a binary liquid system and because it does not need, for a simple preamble, all the apparatus of statistical thermodynamics. But one must keep in mind that this theory was made without taking into account the role of an eventual wall, say a solid support in contact with fluid phases, and that this role is very important." Also, the study is made starting from a density of free energy excess, and it is not quite rigorous for obtaining the surface tension. More advanced treatments were attempted by several authors in the field of statistical thermodynamics and particularly by Sullivan. He gives a very satisfactory summation of this kind of theory in a paper with M. Telo Da Gama8 where the weaknesses outlined above are overcome, and a very complete literature review is given. Our work consists of relating the surface transition to the wetting transition in the water-2,6 lutidine system. The wetting transition for this biphasic system has been observed by Pohl and G ~ l d b u r g . ~The behavior of the lutidine layer on the surface of silica microspheres for the monophasic system was studied by turbidimetry by Beysens and co-workers.'wb They found a sudden thickening of the diameter of the spheres, when the temperature is raised, near the consolute curve and near the critical point. They thought this thickening to be due to the thickening of the lutidine layer and therefore to have shown the prewetting transition. Surprisingly, in the two-phase domain the phase wetting is the "water-rich" phase and not the "lutidine-rich" phaseag Here, we present results on the adsorption and the ellipsometry on the surface between silica (plane or microsphere shaped) and monophasic solutions of 2,6-lutidine with water, varying temperature and composition.
\k = 'ko(5)
11. Cahn-Widom-Rowlinson Theory of Wetting and Prewetting Transitions 1. van der Waals Description of an Heterogeneous System. The chemical system is composed of two phases, a and p, separated by a planar interfacial zone, where a surface tension exists. There are c components, and each component has the density pias@,with C U , ~= CY if the point considered is in the a phase and a,p = p if the point considered is in the p phase. pi is the local value of the density of the component i; it varies with the locus. The surface phase is the heterogeneous part of the system; the CY and p phases are homogeneous. The heterogeneous behavior of the system is thus oriented along the z axis perpendicular to the plane of the interface: pi = pi(z); 5 = 5(z) (1) 3 is the vector of the densities. One defines a local free energy density \k, which is supposed to vary with 5 as in an homogeneous phase \k = \k(z,T)= \k(ij(z),T) (2) If it is supposed that the composition gradient is small compared with the inverse of the intermolecular distance, \k may be developed as (7)Rowlinson, J. S.;Widom, B. Molecular Theory of Capillarity; Clarendon: Oxford, 1984. (8)Sullivan, D. E.; Telo da Gama, M. M. Fluid Interfacial Phenomena; Croxton, C. A,, Ed.; Wiley: New York, 1986. (9)Pohl, D. M.;Goldburg, W. I. Phys. Rev. Lett. 1982,48,1111. (10)(a) Beysens, D.; Esteve, D. Phys. Reu. Lett. 1985,54,2123. (b) Beysens, D.; Houessou, C.; Perrot, F. On Growth and Form. Fractal and non fractal at pattern in physics; Stanley, H. F., Ostrowsky, N., Eds.; Martinius Nijhoff Boston, 1986;pp 211-217. (11)Durian, D. J.; Franck, C. Phys. Reu. Lett. 1987,59, 555. (12)Sullivan, D. E.Symp. Trans. Faraday SOC.1981,16, 191.
+K
(3) where \ko(z) is the local free energy density in an homogeneous solution with densities ij but not in a thermodynamical equilibrium (5 # 5, and ED). K is a functional of ij(z). The surface tension is calculated from the excess function: A\k = \ko(ij)+ K - @(e) (4) where *(e) = Cipipi(e)is a function of the equilibrium chemical potentials pi(e). Denoting (5) -WG) = q0(ij)- We) one has
AF is the free energy excess. Its minimum value versus ij is the surface tension u.
Therefore, we will use Was a local variable: it may be demonstrated to be independent of the surface of separation. It is zero for = 5" and ij = and may be calculated by the statistical thermodynamical methods and expressed with critical coefficients when working close to a critical point. 2. van der Waals' Equation. In van der Waals' hypothesis, K in eq 6 is
z
(7)
with m a constant and comes
AF
=
5 = ij(z).
s+-[
Thus the equation be-
-W[ij(z)]+ i m (
-m
gr]
dz
(8)
This is valid for a gradient region in p(z) separating two homogeneous fluid phases, the separation surface being a plane perpendicular to the z axis. It is evident that one of the limit conditions is different in the case of a "wall", but the discussion about the solutions is the same, and we present only this one. 3. Solving van der Waals' Equation. The surface tension (r is the minimum value of AF. It can be found by variational calculus, i.e., the Euler Lagrange equation: d2ij d(W(ij)) m- =-(9) dz2 dP in the case on one component and by numerical methods in the case of several components. Equation 9 has the form of the Newton equation. p acts as a coordinate, p' = dp/az acts as the velocity, z is the time, m is the mass, W(5)is the potential energy, K is the kinetic energy, - W + K is the Lagrangian, and (r is the action. Taking into account the boundary values Zltu = ijl'@ = 0 and W(ij")= W(;@) = 0 for reasons of homogeneity (ref 7 , p 55), mathematics show that the sum W + K (the Hamiltonian of the system) is a constant equal to zero. In a system with two densities, the function W(pl(z),p2(z)) is three dimensional (Figure 1). The values of W(;"), W(ij@), and W(ijY)are represented by the tops of peaks tangent to the zero plane. Solving eq 9, we obtain the functions of pl(z) and p&) and their derivatives p'' and pt2,which minimize AF and give u (we find the "trajectory" and the "velocity" of the particle moving in the field W). The set of the values of
1328 Langmuir, Vol. 4, No. 6, 1988 w(p) 0
t
Ghaicha et al.
Pa
Figure 1. Representation of the variation of W , the local free energy excess, versus densities p1 and pP Solid lines give sets of points where W has the same value. The plane plop2 gives the zero value for W , and the three peaks are tangent to this plane. Other values of W are negative. Dashed lines with arrows are the “trajectories” to go from a bulk phase to another; the bulk phases cy, p, and y correspond with the zero values of W ,according to Rowlinson and Widom.’
W ( p l ( z ) and p 2 ( z ) ) corresponding to the minimum in AF is a line going through the three-dimensional picture, passing a or @ or y if they are stable phases, with p’ being zero and p’ obtaining its higher values in the bottom of the valleys separating the peaks. (The particle moves in the field from say a to P with a maximum velocity where the field W is minimum and a zero velocity at the limit points a and 4.) It is simply because, as W + K is zero, a minimum for W corresponds with a maximum in K , i.e., a quadratic combination of P ’ ~and pf2 which is maximum. A zero value of W (for a stable homogeneous phase) corresponds with a zero value of K , i.e., for prl and pf2 (independently, the stable phases are homogeneous). Therefore, we may have several cases: (i) The a,8, and y phases do exist, and to go from y to a there is a “direct path”; i.e., the p1 and p2 functions which minimize AF are not compatible with p2@. It is a simple case, with just an adsorption microscopic gradient between y and a. (ii) The a,P, and y phases exist, but to go from y to a the p1 and p2 functions which minimize A F do include p2@, and therefore pll and pf2 are zero for these values. This is the case where a macroscopic P phase exists between y and a: it is the case of the perfect wetting of a or y by
P.
(iii) The P phase does not exist as a bulk stable phase, but the thermodynamic conditions are close to the conditions where @ does exist. Then, there may exist a “path” from y to a that is situated in the neighborhood of P: in the sum W + K , W is near W@, which is zero, and so p f l and also close to zero (going from y to a to have an ”action” minimum, the “particle” goes near P, not on P, with a “velocity” infinitely small but not zero). In this case, there exists a sort of microscopic phase, a layer of thick adsorption. It is called the “prewetting” state. (iv) Two “paths” may be also equally possible, for instance, cases i and iii. Two difference states may then coexist at the interface between y and a: this is a surface separation of phases. 4. Surface Transitions. Case 1. If the thermodynamic conditions are changed it may be possible to go from state 1 to state 2; this is a wetting transition with two actions equal in a single point. Case 2. If the system goes from state 1to state 3, it is only a prewetting transition or Cahn transition (ref 7, p 227).
Case 3. From state 1to state 4 there is a surface-phase transition (ref 7, p 227). 111. Experimental Section 1. Choice of the System. We have studied the water-2,6lutidinebinary system because of its phase diagram. This diagram shows a coexistence curve for two liquid phases13 with a lower critical point (LCST) easy to study: T, = 34 “C and xca = 0.064 (T,is the critical temperature, x,” is the critical mole ratio of the bulk phase).14J5 Several striking phenomena appear near this critical point. They have been studied by light scattering, viscosity measurement~,’~!’~ ultrasonic velocity, and adsorption.” Pohl and Goldburg have shown a wetting transition, from a perfect wetting to a partial wetting of the silica by a water-2,6-lutidine mixture, by raising the temperature. Sigl and F e d studied its modification by addition of KCl.23 Beysens and co-workerslOJ1have studied the turbidimetry of microspheres of silica floating in such a mixture. They observed on raising the temperature a sudden enhancement of the radius of the microspheres due to the thickening of the adsorption layer, which is probably rich in lutidine,enhancement which causes the microspheres aggregation. This phenomena is attributed to a low adsorption-high adsorption transition. 2. Chemicals. The 2,6-lutidine,or 2,6-dimethylpyridine,was produced by Merck, 98%,for synthesis. It was redistilled,yielding a 99.5% product; the last 0.5% consists of isomers of lutidine, as revealed by gas chromatography analysis. The water was first purified on an ion-exchange column and then distilled over potassium permanganate in sulfuric acid; the purity test consisted of checking the value for the surface tension of the water at 25 “C: y = 72.0 mN m-l. All glass containers were washed with a freshly prepared siilfochromic mixture and then amply rinsed with distilled water. Water-2,6-lutidine mixtures were prepared immediately before use and kept in closed containers. The synthetic silica microspheres used to study the adsorption of the mixture on a solid surface were supplied by Rhbne-Poulenc (Spherosil XOA 400). The silica particles are spheres with a diameter of about 40 Fm. The specific area is 540 m2g-l, measured by the BET method. 3. Adsorption Measurements at the Solid-Liquid Interface. Every sample contains the same silica quantity (1g) and
the same volume of solution. The composition of the mixture varied between 1%and 40% by weight. Samples were shaken during 3 h in an incubator kept at the desired temperature (*0.1 “C). The adsorbed quantity was calculated by difference after measuring by UV adsorption the proportions of the components in the supernatant obtained by centrifugation. The relative excesses is then given by the formula (10)
where no is the total number of moles in the mixture, xZo is the initial mole ratio of the 2,6-lutidine,x2 is the last mole ratio after adsorption, m is the quantity of silica in grams, and s is the specific area of the silica (cm2g-l). (13)Gulari, E.;Collings, A. F.; Schmidt, R. L.; Pings, C. J. J. Chem. Phys. 1972,56,6169.
(14)Cox, J. P.; Herington, E. F. G. Trans. Faraday SOC.1956,52,928. Rice, 0. K. Trans. Faraday SOC.1963, 59,2723. (15)Loven, A. W.; (16)Stein, A.;Davidson, S. J.; Allegra, J. C.; Allen, G. F. J . Chem. Phys. 1971,12,6164. (17) Gutschick, V. P.; Pings, C. J. J. Chem. Phys. 1971, 55, 3845. (18)Tenebre, L. Ph.D. Thesis, Montpellier, 1971. (19)Gladden, G. P.; Breuer, M. M. J. Colloid Interface Sci. 1975,53, 249. (20)Privat, M.; Tenebre, L.; Bennes, R.; Tronel-Peyroz, E.; Douillard, J. M.; Ghaicha, L., submitted for publication. (21)Tronel-Peyroz, E.; Bennes, R.; Douillard, J. M.; Privat, M. C. R. Acad. Sci., SBr 2 1984, 299, 1313. (22)Zarzycki, J. Le Verre et 1’Etat V i t r e w ; Masson: Paris, 1982. (23)Sigl, L.; Fenzl, W. Phys. Rev. Lett. 1986, 57, 2191.
Langmuir, Vol. 4, No. 6, 1988 1329
Phase Transitions at SilicalWater-2,6-Lutidine Interface
0
0
0
2
4
6
0
0
qo2
Figure 2. Variation of the surface excess rZlof the 2,6-lutidine on silica spheres versus the 2,Glutidinemole ratio in the solution, at several temperatures. 4. Ellipsometry. The measurements of the coefficients of
ellipticity ( p ) were carried out by using the apparatus constructed by Tenebre.18 The approach allows measurements of ellipticity coefficients, at Brewster's angle, on a horizontal surface, with a precision of fl X 10". The light source was a He-Ne laser (A = 6328 A). To study the silica-solution interface, the light was reflected on a plane silica piece lying at the bottom of a thermostated cell containing the solution. The piece of silica was washed with the sulfochromic mixture, carefully rinsed with distilled water, and kept in water for 24 h. Measurements were made for a given composition of the water-2,6-lutidine mixture, driving the temperature up and down and waiting for equilibrium for each value of the temperature. Ellipticity values were very well reproducible. Given values are relative to the ellipticity of the silica-water interface at the given temperature.
IV. Results and Discussion 1. Adsorption on the Silica Spheres. In Figure 2 we show the results of the measurements of surface excesses at the silica solution interface at 22,25, and 32 "C for mole ratios between 1.5 X and 9.25 X and at 40,43, and 46 "C for mole ratios less than 1.5 X lop2. Comparison with the literature can be made only for 25 "C (Figure 2).19 Our values are slightly higher than Gladden's values: it can be explained by the difference between the two kinds of silica used. The shape of the curve is the same, and it is seen that the adsorption maximum occurs for a solution composition very different from the critical composition. It is a different case from the adsorption a the solution-vapor interface.m However, adsorption values for the solid-liquid interface are in doubt for mole ratios higher than the value that gives the maximum: experimental errors are strongly increasing from this point. The principal error term that appears when calculating the relative error of rZlin eq 11is 6xz/(xzo - xz). axz, the error of the experimental value of xz, is due to the dilution the solution must suffer before spectrometric titration, during the titration itself, and the combination of these errors through calculus to have xz from these data. As xzo and x 2 are increasing, for a Ax2 = x2" - x 2 of the same order,
30
TTC)
50
P
Figure 3. Variation of the surface excess r21of the 2,6-lutidine on silica spheres versus the temperature for several mole ratios of 2,6-lutidine. For three of them, rzlvalues become very high and a break appears in the curve. to have a comparable accuracy on this term it would be necessary to get much better absolute accuracy for xz and xzo,which is impossible experimentally. As an example, starting from a solution with xzo = 1.78 X one gets a Ax2 = 2.35 X which is very small and difficult to measure with an error of about 20%. Starting for a solution with xZo = 6.77 X 1O-I) one gets Axz = 6.77 X 30 times greater, which would apparently be easier to measure, but the error is then around 45%! Fortunately, in intermediate cases the precision is far better. Thus, for this discussion we will use only the values obtained for xZa values less than 3 X The values of xzu limiting the experiments at 40,43, and 46 "C are given by the phase separation observed for these temperatures. These values have been drawn versus temperatures in Figure 3. For dilute solutions (xza loT2)excesses are fairly constant. For the mole ratio xza = 1.30 X linearly increases with temperature. For mole ratios 1.35 X 1.40 X 1.40, and 1.60 X rZllinearly increases until a temperature where it suddenly diverges. Breaks on the curves then appear. For the mole ratio 2 X this phenomenon is no longer seen, but the curve is limited by the bulk-phase separation. 1.40 X The breaks seen for xZa = 1.35 X and 1.60 X mean a change for the fashion of adsorption, which suddenly takes very high values. Such a mode for the adsorption has not been observed at the liquid-vapor interface; although surface excesses were even stronger, their variation was continuous.z0 2. Ellipticities at the Solid-Liquid Interface. Parts a and b of Figure 4 show the curves p = f(T)for several mole ratios. For both mole ratios xZu = 0.0152 and 0.0169 (8.45 and 9.21 w t %), i.e., between 0.013, the value of xZa where the wetting transition line intercepts the coexistence curve, and
-
1330 Langmuir, Vol. 4, No. 6,1988
Ghaicha et al.
PW*,
I o4
,i
997%by weight
0 -
-5
-
a
25
30
R'C)
35
i '11
I
2
I
30
40
l("C)
Figure 4. Ellipticity of the surface between the water-2,6-lutidine mixture and a silica plane versus temperature (a, top) for solution 9.97% by weight and (b, bottom) for solutions 9.21%, 8.42%, 7.66%,6.87%,and 5.79% by weight. A break appears for solutions 9.21% and 8.62%.
-
0.018, the ellipticity coefficient varies linearly and increases with T until a temperature where it becomes a constant (Figure 4b). For mole ratios less than 0.013 and for the mole ratio more than 0.018 the shape of the curve is quite different, without any break (Figure 4a,b). The phenomenon of the break was explored for xZa = 0.0152 for 30 different temperatures, going up and down. It was seen also for x~~ 0.0169. Thus, it exists in a very narrow mole ratio zone: 0.013 < xZDl< 0.0178. It is very weak, and even disappears, if the silica surface has not been cleaned enough with the sulfochromic mixture or if one has not waited long enough before taking a new measurement, in order that the silica surface becomes hydrated. Several tests were made and have shown that this was necessary. Thus, ellipticity coefficients for the two interfaces (solid-liquid and liquid-vapor) behave quite differently. For the solid-liquid interface, when the temperature is increased they increase from weakly negative values to weakly positive values and have a break in a well-defined zone of concentrations and temperatures. For the liquidvapor interface, they tend first quickly to very negative values then become positive for strong values of xzu.20 Experiments indicate that the break phenomenon depends on the nature of the interface and on the presence of a solid and its state. 3. Surface Transition. The two breaks observed on the ellipticity diagrams occurring at the mole ratios xZDl= 0.015 and 0.0169 and the temperatures 35.7 and 41.0 O C , together with the breaks observed on the curves of the surface excesses versus the temperatures, have been drawn on a "phases coexistence diagram" T = f(xzDl). They constitute a line which could be the transition line called the "prewetting" line by Cahn (Figure 5). xZa
-
x.,or
3
Figure 5. Phase diagram for the water-2,6-lutidine system, including the wetting transition (dashed line) and the surface and by ellipsometry (e), transition seen on the adsorptions (0) according to Cox and Herringt~n,'~ Pohl and G~ldburg,~ and results from Figure 3 and 4b. Table I. Surface Layer Thickness and Composition versus Temperature"
T,"C
I'21, 1Olo mol.cm-2
20 29 32 35 37 40
3.5 4.70 5.00 5.80 6.25 7.4
104p -4.8 -1.1 0.1 1.3
2.0 2.0
d, 8, 20.2 20.3 19.7 21.0 21.6 26.2
X"2
6.7 X 9.00 x 10-2 9.9 x 10-2 1.1 x 10-1 1.16 X lo-' 1.13 X lo-'
Bulk composition is 9.21 % by weight.
First, it is remarkable that two sets of independent experiments, one obtained by optical technique the other by analytical technique and on different solid surfaces, although chemically identical, give results quite in agreement. These results show the existence of a surface transition. It is of interest to analyze what kind of transition. We tried to analyze the ellipsometry results, for xzr' = 0.0169 (9.21% by weight), by using the method previously established.2l This method is based on particular application of the Drude theory and consequently has the limitations of this theory, which may be severe in a zone of great fluctuations as the near critical zone. The results shown in Table I have been established with ti compacities being the compaction values of the pure components, uioDlthe mole volume in the bulk of the pure components, ti the dielectric constant for the pure componenta, z the mean value o f t in the bulk, and tg the t value for the glass: tl = tz = 0.82, c1 = 1.174, z = 1.82, vloa = 18 cm3 mol-', t2 = 2.241, v20a = 116.1 cm3 mol-l, and tg = 2.1258. The results show that the layer thickness goes on increasing after the "transition", although the ellipticity should be a constant. This constant would be then the result of the balance of several factors. However, all this is not really conclusive, the Drude theory alone being
Phase Transitions at SilicalWater-2,6-Lutidine Interface
Along the path shown by the arrow they are two possiMe cases : I . Cabn’s scenarip . .. . . . .. .. ...
2 . our scenario
thinlayer
3
2
1
r
phase pertecj separation wetting ,
4
partial wetting
“p likespots
Figure 6. Representation of different scenarios around the wetting transition. a and p are general names for phases coexisting for the same temperature, higher than the critical temperature. The wetting transition appears in the diphasic part of the diagram, going from part 4 (where there is a partial wetting of B by a and of the glass by a) to part 3 (where a complete wetting of B and the glass by a exists). Part 3 is monophasic for the bulk but
‘special” for the surface between the glass and the solution: “B like” or “biphasic”. The scenarios give the state of both the solution and the surface in parts 1, 2, 3, and 4.
certainly inadequate to describe phenomena in such a perturbated system. Adsorption data give two pieces of information: 2,6lutidine is the component which adsorbs preferentially on silica spheres; this adsorption behaves with the temperature as if there would be a sort of transition. The supposition must be made that the phenomenon is the same on the silica plane surface as that on the spheres. In other words, 2,6-lutidine is the component which is preferentially adsorbed, and it is the same transition we see by ellipsometry on the plane surface and by adsorption on the spheres. Thus we have two experimental facts to explain with the Cahn-Rowlinson-Widom theory. First, the surface transition observed corresponds to a strong adsorption of 2,6-lutidine in the sense of the relative adsorption rZl(and probably a thickening of the adsorption layer); it remains true, even if rZland r2do not vary in exactly the same way. Second, next, the wetting experiments of Pohl and Goldburg show that, in the conditions of the diphasic system, it is the “water-rich phase” (or “a” phase) which wets glass. At this stage, it may be noted that, although different in shape, the three glass surfaces have the same chemical state “fused silica like”,22and consequently it is quite justifiable to compare results obtained under these different conditions. Cahn’s classic scenario of prewetting the glass wall by a thick layer of P-phase composition breaks down, the P phase being the phase rich in component 2 which appears when raising the temperature and wetting both glass and a phase (as Figure 6 shows). The scenario assumes that the ,8 phase actually wets the wall better than the a phase; i.e., it assumes an interaction of the 0phase with both the wall and the a phase, which makes a special profile only “ P like” possible in the in-
Langmuir, Vol. 4, No. 6, 1988 1331 terface (case number 2 in the theoretical section of the paper). A way to explain both an enhancement of rZland why, when P appears, the lutidine-rich phase does not wet the wall is to suppose that a phase separation between an a-like surface phase and a P-like surface phase already exists in the surface phase in contact with the monophasic system. It is the transition number 3 in section 11. This induces the following scenario: On the glass, rZl being already great, the surface phase undergoes a separation and nuclei of the P phase appear. When xZa or T increases, the bulk-phase separation occurs; a rich surface phase stays on the surface, in contact with a (in all monophasic phase, a has a strong adsorption), and ,8 separates from surface nuclei or bulk nuclei, not wetting the glass but wetted by a. According to Pohl and Goldburg, the wetting transition is first order. Our results show that the surface transition in the monophasic domain is rather second order. The first result is consistent with the van der Waals theory developed by Sullivan et al.: it occurs with systems where the interactions between the “wall”and the solution are strong. It is the case with silica and polar liquids. Experimentally, we saw that the transition disappears with a “dirty” surface when the attraction force of the surface is weaker. But this theory permits also a second-order transition, and it is easy to understand how a strong interaction may induce a early phase separation in the surface phase, the bulk phase being very near of such a separation. It is of interest to examine now, whether this scenario is consistent with the adsorption and ellipsometry results. First, a surface-phase separation allows the surface to become richer in the presence of the silica, and it explains the change of the slope in the adsorption-temperature graphs. The complex nature of the layer permits several explanations for the ellipticity steadiness with temperature: increasing thickness with inverse refractive index variation, great fluctuations escaping measurement, and existence of two layers.
V. Conclusion Experiments of adsorption of 2,6-lutidine on silica and of ellipticity of silica in contact with water-2,6-lutidine mixtures show clearly a surface transition near the wetting transition and near the conditions of bulk-phase separation. This transition is not exactly the “Cahn transition”, because it is second order, and cannot be described as a “prewetting” transition. The phenomena can be qualitatively explained by a surface-phase separation which is energetically possible, as shown by analyzing the vicinity of the wetting transition and taking into account the influence of the silica on the interactions in the liquid phase. Finally, the experiments confirm the validity of the analysis in the van der Waals mode, particularly when it takes into account both interactions of the molecules in the liquid phase and interactions between the wall and the molecules of the liquid. It would be of interest to study mixtures with different interactions between the molecules and also with different walls and to try to test more finely the diverse interactions in the models. Acknowledgment. We thank Dr. D. Beysens for helpful discussions concerning this work. Registry No. 2,6-Lutidme,108-48-5;vitreous silica,60676-86-0.
