Langmuir 1994,10, 1110-1121
1110
Surface Forces between Silica Surfaces in Cationic Surfactant Solutions: Adsorption and Bilayer Formation at Normal and High pH Mark W.Rutland' and John L. Parker? Laboratory for Chemical Surface Science, Institute for Surface Chemistry, P.O.Box 5607, 114 86 Stockholm, Sweden, and Department of Chemhtry, Royal Institute of Technology, S-100 44 Stockholm, Sweden Received October 4,1993. I n Final Form: January 21, 2994" Forceshave been measuredbetween glass surfacesin cetyltrimethylammoniumbromide (CTAB)solutions at pH 10,and the adsorption, inferred from the fitted surface potentials and the measured adhesion, has been shown to be strongly dependent on the surface charge and the competing ions. At very low CTAB concentrations adsorption can only be detected after the surfaces are left in contact, and this results in a time-dependent adhesionconsistent with diffusion-controlledadsorption of CTAB in the annulus around the contact area. At slightly higher concentrations (between 3 X 1@ and 6 X 10-8 M),a hydrophobic monolayer forma and purely attractive hydrophobic interactions are measured,indicatingthat the surfaces are close to electroneutrality. This concentration is much lower than that required to achieve neutral surfaces at pH 5.6 for glass or for mica surfaces. It is argued that after charge reversal the forces can be fitted with DLVO theory, assuming a plane of charge away from the surface. Bilayer formation occurs at the critical micelle concentration (cmc) (1X 10-3 M)for low pH and below the cmc at elevated pH. The density of surfactant in the outer layer of the bilayer on silica at normal and elevated pH is much lower than that on mica at a comparable bulk CTAB concentration. As a result it is possible to disrupt the bilayers and push the surfaces into a hydrophobic contact, The thickness of the bilayer and the surface charge density obtained from fitting the force law as well as the nature of the force law during compression of the bilayer lead to the conclusion that CTAB adsorbs to glass as patches of bilayers or surface aggregates. Dynamic rearrangements within the surface layer have also been observed.
Introduction The adsorption of surfactants to surfaces of opposite charge plays a crucial role in many industrial applications. Although the broad principles governing the nature of the adsorption are well understood, there is, as yet, no possibility of accurately predicting the nature of the aggregate formed by surfactant molecules on a surface, in the same way that the solution structure of an aggregate can be predicted from the geometry of the molecule.' This is because the interaction of the surfactant molecule with the surface, the surface roughness, heterogeneity, and charging behavior all have to be considered as well as intermolecular forces and packing considerations between the molecules. Many studies of surfactant adsorption have been undertaken over the last 30 years, using isotherm^,^^^ contact angles,' fluorescence decay: neutron reflection? small angle neutron scattering: and calorimetry.* It is apparent from these studies that no simple picture of the adsorption can be drawn in terms of simple monolayers
* To whom correspondence should be addreesed at the Institute for Surface Chemistry. + On leavefrom the Departmentof Applied Mathematice, Research School of Physical Sciences and Engineering, Australian National University, GPO Box 4, ACT 2601, Australia. Abstract published in Advance ACS Abstracts, March 1,1994. (1) Israelachvili,J. N.; Mitchell,D. J.;Ninham, B. W.Biochim.Biophys.
Acta 1977,470,185-201. (2) Rupprecht, H. Kolloid-2. 1971, 249, 1127. (3) Gao, Y.;Du. J.: Gu, T. J. Chem. Soc., Faraday Trans. 1 1987,83,
2671. (4) Gonzalez, G.; Travalloni-Louvisee,A. M. Langmuir 1989,5, 26. (5) Levitz, P.; Van Damme, H.; Keravie, D. J. Phys. Chem. 1984,88, 2228. (6) Rennie, A. R.; Lee, E. M.; Simister,E. A.; Thomas, A. K. Langmuir 1990,6,1031. (7) Cummine. P. G.:. Stades. _ .E.: .Penfold. J. J. Phvs. Chem. 1990.94. . . 3740:
(8)Partyka, S.; Lindheimer, M.; Zaini, S.; Keh, E.; Brun, B. Langmuir
1988, 2, 101.
and bilayers. One of the few systems where adsorption is of this nature is in the adsorption of cationic surfactants to which has a molecularly smooth, homogeneous surface which becomes negatively charged in aqueous solution. Over the last decade or so the surface force technique has been used to probe the nature of the forces mica surfaces exert on each other in different solutions. A large proportion of this research has involved the effect of cationic surfactant adsorption on the surface forces, so the adsorption of cationic material to mica has been well characterized. Perhaps the most studied cationic surfactant is CTAB (cetyltrimethylammonium bromide) which adsorbs to mica and reduces the surface charge. Charge neutralization occurs at -3.3 X 10-6 M, and at higher concentrations additional surfactant adsorbs and the surface begins to recharge.15 At a concentration of about half the critical micelle concentration (cmc)bilayers form with a measured thickness of -3.2 nm.15 A t concentrations well above the cmc the Debye screening length agrees well with a model which excludes micelles from the double layer and includes monomers and counterions only.12 At very high concentrations structural forces due to the micelles are observed at large separations.*e The results for CTAB adsorption to glass follow roughly the same trend as the results for mica. However, there are a number of dramatic and important differences between the results obtained with the two different (9) Israelachvili,J. N.; Pashley, R. M.Nature (London)1982,300,341. (10) Pashley, R. M.; Israelachvili, J. N. Colloids Surf. 1981,2, 169. (11) Pashley, R. M.; McGuiggan, P. M.; Ninham, B.W.; Evans, D. F. Science 1985,229, 1088. (12) Pashley, R. M.; Ninham, B. W. J. Phys. Chem. 1987, 91, 2902. (13) Pashley, R. M.; McGuiggan, P. M.; Horn,R. G.; Ninham, B.W. J. Colloid Interface Sci. 1988,126, 569-578. (14) Herder, P. C. J. Colloid Interface Sci. 1990, 134, 336. (15) Kbkicheff,P.; Christenson, H. K.; Ninham, B.W. Colloids Surf. 1989,40, 31-41. (16) Richetti, P.; Kbkicheff, P. Phys. Reu. Lett. 1992,68,1951-1954.
0743-746319412410-1110$04.50/0 0 1994 American Chemical Society
Surface Forces between Silica Surfaces
Langmuir, Vol. 10, No. 4, 1994 1111
substrates. At concentrations very close to the cmc the evidence for bilayer formation on glass is inconclusive whereas on mica bilayers form at a concentration of less than halfthe cmc. At all the concentrations so far studied it is possible to push two glass surfaces into a strongly hydrophobic contact whereas this becomes impossible in the mica experiments after formation of a bilayer. Moreover, the point a t which electroneutrality occurs is greatly different, warranting an investigation of the effect of the surface charge and ion-exchange properties on adsorption, as detected by the change in the surface force. Results are presented of surface force measurements between silica glass surfaces a t pH 10 in a range of CTAB solutions up to the cmc, and an investigation of the forces between bilayers at pH 5.6 and 10 has been undertaken. Materials and Methods Surface forces were measured with a new type of surface force apparatus and a brief description follows. One surface is mounted at the end of a bimorph force sensor,1' and the other is mounted at the end of a piezoelectric tube. The bimorph is enclosed in a Teflon sheath, and this is mounted inside a small chamber (volume -10 mL) clamped to a translation stage which is used to control the coarse position of the piezoelectric tube and the upper surface. A complete description of the apparatus will be given in a forthcoming publication.18 The analysisof the raw experimental data follows the procedure given in ref 17. The surfaces are pushed together until they come into contact. When in contact the motion of the piezoelectrictube is transmitted directly to the sensor,and this straight line is used to calibrate the sensitivity of the bimorph. Using this procedure, it is not possible to define the separation with respect to the contact between the two surfaces in dry atmosphere. As a result it is not possible to determine the thickness of an adsorbed layer if it is not pushed away from the contact region during the experiment. Even when the adsorbed layer can be pushed away, it may be that the force required to do so is close to the maximum measurable force, in which case the surfaces may not attain the region where the motion of the piezo is transmitted directly to the bimorph. This situation arises if the layer is compressible, or if additional work is required to expel the layer after the initial rupture where the force required to achievemonolayer contact is larger than that which can be applied by the apparatus. The calibration of the bimorph sensor remains constant from one measurement to the next, and so this number can be obtained from previous force runs. Glees surfaces were prepared by cutting a 3-cm length of a 2-mmglass rod. The rod was then cleaned with ethanol and the end melted in a gas burner until a molten droplet of glass formed with a radius of 2 mm. Highly smooth glass surfaces can be prepared in this manner. Atomic force microscopy (AFM)images of these surfaces are featureless with a RMS roughness of 0.13 nm and a maximum variation in height of 0.19 nm across a 200nm scan range.19 Two such surfaces were mounted in the apparatus and aligned so that the centers of both spheres were as close to parallel with the axis of motion of the piezoelectric tube as possible. At the end of the experiment the radii of both spheres are measured with a micrometer, and the average radius is given by the geometric mean of the two. The macroscopic radius of the sphere is measured (R = 2 0.1 mm), and the fact that this is equal to the microscopic radius has been proven by hydrodynamic and pull-off force measurements. Water was purifiedwith Millipore RO and MilliQ water purification systems; CTAB was obtained from Fluka, NaOH was from BDH, and both were used without further purification. The experiments were begun with water in the measuring chamber. CTAB was added in steps to the measuring chamber to reach the final desired concentration. In order to determine whether equilibrium adsorption had been reached, the forces (17) Parker, J. L. Langmuir 1992,8,551. (18) Parker, J. L. Manuscript in preparation. (19) Parker, J. L.; Freberg, J. R. E. Manuscript in preparation.
Figure 1. Surface forces between silica glass spheres in CTAB solutions at pH 10 as a function of distance and surfactant concentration (mol/L). The force is normalized by the mean radius of curvature. The circles indicate where the surfacesjump into contact due to instability of the spring. In each case the zero of separation is taken from the region of constant compliance, and thus corresponds to silica-silica contact for the lowest concentration, monolayer-monolayer contact at intermediate concentration, and bilayer-bilayer contact at the highest concentration. No information on the thickness of the monolayers is obtained. were measured until successive runs agreed with one another, and this required more than 3 h in most cases and at low concentrations more than 24 h equilibration after addition. Note that different results are obtained if a concentrated solution is added rather than using stepwise addition." Forces were fitted with simple DLVO theory using a numerical solution to the nonlinear Poisson-Boltzmann equation and a nonretarded van der Waals interaction.21 When the CTAB concentration was greater than or equal to the cmc, the measured Debye lengths were consistent with the monomers and counterions only model of Pashley12 and no fitting was applied as the simple symmetric model used is not valid. (The exclusion of micelles leads to an asymmetric distribution of monomer and counterions in the double layer.)
Results Forces below the cmc. Surface forces between glass surfaces below the cmc a t pH 5.6 have been reported elsewhere.22 Briefly, adsorption of CTAB at very dilute concentrations causes the force between glass surfaces to become attractive at small surface separations and the surfaces become adhesive in contact. The surface charge on glass is neutralized at a CTAB concentration of 5 X 10-4 M. Further addition of surfactant causes the surfaces to recharge due to continued adsorption. At concentrations just below the cmc the surfaces jump from a separation of 3.0 nm into an adhesive contact. The forces measured between glass surfaces at pH 10 for all the concentrations studied are shown in Figure 1. The measured force (F)is scaled by the mean radius of curvature (R) of the surfaces, and this is related to the interaction free energy between plane parallel plates of unit area by FIR = uE. The cross-cylinder geometry is normally used in surface force measurements, and the scaled forces (FIR = 2uE) differ by a factor of 2. At low (20) Chen, Y. L.; Chen, S.; Frank, C.; Israelachvili, J. N. J. Colloid Interface Sci. 1992, 153, 244. (21) Chan, D. Y.; Paahley, R. M.; White, L. R. J.ColloidZnterface Sci. 1980, 77, 283. (22) Parker, J. L.; Yaminsky, V.; Claesson, P. M. J.Phys. Chem. 1993, 97,7706.
Rutland and Parker
1112 Langmuir, Vol. 10, No. 4, 1994
1
0
0.1
0.2
0.4
0.3
I
n
E
g1
e
L
0.1
0
10
20
30
40
50
Separation (nm) Figure 2. Surface forces as a function of separation between two silica glass spheres immersed in 6 X lo-' M CTAB solution at pH 10. The arrows indicate the direction of movement of the surfaces. The two upper curves are indistinguishable and correspond to approach runs. The two lower curves were measured on separation after 9.3 (upper) and 18.2 s in contact. The surfacesjumped out from small adhesive minima at contact (linear part of outward runs), and the electrostatic component of the force is reduced due to diffusion-controlled CTAB adsorption during the time in contact. The inset shows the adhesion between silica glass surfaces immersed in 6 X lo-' M CTAB solution at pH 10 plotted as a function of t1/2 (from ref 29). For large t the curve is linear with t1/2, indicating that the process is diffusion controlled. concentrations of surfactant the force is entirely repulsive and can be described at large separations with PoissonBoltzmann theory for interacting double layers. (Note: The calculated interaction free energy was scaled by ?r to compare with FIR for two interacting spheres.) At small separations DLV023124 theory predicb an attraction due to van der Waals forces, but there is instead an additional repulsion which has been ascribed to hydration As the CTAB concentration is increased to about 2 X 10-6 M, the electrostatic component of the force disappears and the surfaces jump into zero separation under the influence of a hydrophobic attraction. Increasing the CTAB concentration further leads to the reappearance of an electrostatic repulsion, but at short separations the force continues to be attractive. Above 10-4 M a small repulsion is observed prior to the attractive jump in, which is not accounted for by DLVO theory. This indicates that a bilayer is forming on the surface and at higher concentrations the force required to overcome this repulsion and achieve adhesive contact increases. The force curves measured on both approach and M CTAB are shown in Figure 2. The separation in 6 X two upper curves were both measured as the surfaces approached and are indistinguishable. However, on separation the outward curves do not follow the inward run as they do in the absence of surfactant. When the (23) Derjaguin, B.; Landau, L. Acta Physiochem. 1941, 14, 633. (24) Verwey, E. G. W.; Overbeek, J. T. G. The theory of the stability
of lyophobic colloids; Elsevier: Amsterdam, 1948.
(25) Horn,R. G.; Smith,D. T.;Haller, W. Chem. Phys.Lett. 1989,162, 404-408. (26) Peschel, G.; Belouschek,P.; Muller, M. M.; Muller, M. R.; Konig, R. Colloid Polym. Sci. 1982, 260, 444-451. (27) Rabinovich, Y. I.; Derjaguin, B. V.; Churaev, N. Adu. Colloid Interface Sci. 1982, 16,63-78. (28) Ducker, W. A,; Senden, T. J.; Pashley, R. M. Nature (London) 1991, 353, 239.
0
5
10 15 20 25 30 35 40 45
Distance (nm) Figure 3. Surface forces measured on approach between silica glass surfaces in CTAB solution at pH 10. The upper curve (open symbols) was measured in 2 X 10-8M CTAB. The solid lines are a DLVO fit of the data assuming interaction at constant charge (upper) and constant potential (lower),giving a surface potential, \ko, of -88 mV and a Debye length, rl,of 18.5 nm. The lower curve (closed symbols)was measured 50 min after addition of CTAB to a concentration of 3 X 10-8M and can be fitted with = 18.5nm. DLVO theory (solidlines),giving \ko = -74 mV and r1 Aftsr a further 13-h equilibration, the electrostatic component was immeasurably small and an extra attractive contribution to the force was observed, corresponding to a hydrophobic interaction (see Figure 4). surfaces remained in contact for 9.35 s, the upper curve was measured, whereas if the surfaces remained in contact for 18.2 s, the lowest curve was obtained. In each case, on separation, there is an adhesion of -1 and -3 mN/m, respectively. Also, the electrostatic component of the force is progressively reduced with the length of time spent in contact which is consistent with a time-dependent adsorption of surfactant around the contactzone.29 The inset of Figure 2 shows how the adhesion between the surfaces in 6 X le7 M CTAB increases with time ( t )in contact. On the horizontal axis is plotted t-1/2, and it can be seen that for longer contact times the adhesion is linear with W2. The surface forces measured on approaching two silica glass surfaces in dilute CTAB solution at pH 10are shown in Figure 3. The upper curve (open symbols)was measured in 2 X 10-6 M CTAB. The solid lines are a DLVO fit of the data assuming interaction a t constant charge (upper) and constant potential (lower), giving a surface potential, \ko, of -88 mV and a Debye length, lcl, of 18.5 nm. The lower curve (closed symbols) was measured 50 min after addition of CTAB to a concentration of 3 X 10-6 M and can be fitted with DLVO theory (solid lines), giving *O = -74 mV and r 1= 18.5 nm. At the higher surfactant concentration the interaction follows DLVO theory and turns attractive at approximately 6 nm from contact. However, at about 2-nm separation another force contribution renders the interaction repulsive once again. Although this force was reproducible over a time scale of l/2 h (measurement time approximately 30 s), it was not the equilibrium interaction as, after a further 13-h equilibration, the electrostatic component was immeasurably small and a strongly attractive force was observed as shown in the next figure. (29) Parker, J. L.; Rutland, M. W. Langmuir 1993,9, 1965.
Langmuir, Vol. 10, No. 4,1994 1113
Surface Forces between Silica Surfaces
I
&a n
E
2 E
W
"I-L 0 A
3.0 x 1 0 6 M
o 6.0 x 10-6M 0 9.2 x 10.6M
-1 -1.5
2.0 x 10-5M
0
5
10
15
20
25
Separation (nm) Figure 4. Surfaceforces measured over the concentrationrange 3 X 1O-e to 2 X 10-6M CTAB at pH 10. The solid limescorrespond to DLVO theory with the parameters (bottomto top) % = 0 mV (i.e.,onlythevanderWaalecontribution),90= +30mV (constant potential and constant charge), and *O = +49 mV (constant potential and constant charge). The arrows on the fits show where theory predicts the interaction to become attractive. The other arrow showwhere thejumpa occur in the measured forces. At the concentrations3 X 1O-eand 6 X 1o-BM CTAB, adsorption to the surfacea has neutralized them but the attractive force is much longer ranged than can be explained by dispersion forces, indicating that the surfaces are hydrophobic. When the concentration ia raised to 9.2 X 1O-e M, the surface charge is overcompensated by CTAB so the surfaces become slightly positivelycharged. Increasing the concentrationeven furtherto 2.0 X 1W M causes the surfaces to become even more highly charged, and they can be fitted assuming a potential of 49 mV and that the plane of charge is located 3 nm away from the hydrophobic, monolayermonolayer contact. The forces measured on approach in the concentration range 3 X 10-8 to 2 X 106 M CTAB are shown in Figure 4. The solid lines correspond to DLVO theory with the parameters (bottom to top) WO = 0 mV (Le., only the van der Waals contribution), 90= +30 mV (constant potential and constant charge), and QO= +49 mv (constant potential and constant charge). At the concentrations 3 X 10-8and 6 X 10-8M CTAB (circles and filled triangles) adsorption to the surfaces has neutralized them, but the attractive force is much longer ranged than can be explained by dispersion forces. Long-ranged attractive forces have routinely been observed in surfactant systems10fm2and have been explained as a "hydrophobic interaction" which is discussed later. When the concentration is raised to 9.2 X 10-8 M (open diamonds), the surface charge is overcompensated by CTAB so the surfaces become slightly positively charged. Increasing the concentration even further to 2.0 X 106 M (closed diamonds) causes the surfaces to become even more highly charged, and they can be fitted assuming a potential of 49 mV and that the plane of charge is located 3 nm away from the hydrophobic, monolayer-monolayer contact. The value of 3 nm refers to the separation between the surfaces with respect to monolayer-monolayer contact, and this corresponds to a 1.5-nm layer on each surface. The surfaces jumped in from a position much closer than that predicted by the constant-potential solution for DLVO theory, so it (30)Israelachvili, J. N.; Pashley, R. M. J. Colloid Interface Sci. 1984, --.--(31) C l a w n , P. M.; Blom, C. E.; Herder, P. C.; Ninham, B. W. J. 98.600-614. ----
Colloid Interface SCL 1986,114, 234. (32) Christeneon, H. K.; Clneseon, P. M. Science 1988, 239, 390.
0
5
10
15 20 25 30 35 40
Separation (nm) Figure 5. Force measured at pH 10 in 4 X l W M CTAB. The open symbolscorrespond to an inward run, and the closed symbols
were measured on separating the surfaces. The calibration for constant deflection was taken from the previous run since the surfaces did not achieve monolayer contact, and the distance was adjusted so the force walls overlapped at equivalent forces. The solid lines are DLVO fits for constant charge (upper) and constant potential (lower),giving a surface potential of 59 mV assumingthe plane of charge to be 3.0 nm from monolayer contact and plane of operation of the van der Waals interaction to be at 2.5 nm.
is not necessary to invoke a hydrophobic force to explain the attraction. At a CTAB concentration of 2 X 1VM (Figure 5, open diamonds) there is a small repulsive barrier in the force just prior to the point where the surfaces jump into contact. This is due to the formation of a second layer of surfactant molecules adsorbed to the hydrophobic monolayer to form a bilayer. The force can be fitted very well by the constantcharge solution for DLVO theory assuming that the plane of charge is 3.0 nm from zero separation and that the plane from which the van der Waals force operates is 2.5 nm. If the surfaces are brought in such that the repulsive barrier is not overcome (i.e., to anF/R value equal to about 2 mN) and are then separated, a force such as that depicted by the closed diamonds is observed. The surfaces jump out from the force wall located at about 3 nm to approximately 7 nm, indicating that there is a slight adhesion manifest a t the contact of the bilayers. After the jump out the outward and inward runs are superimposable, indicating that any effect on the surface charge of bringing the head groups near contact is immediately reversed. Since the surfaces do not reach the regime of constant deflection (Le., hard contact), in this case the bimorph calibration was taken from the previous force runs, and the zero of separation adjusted so that the non-DLVO repulsions superimposed. The adhesion and the surface potential as a function of the CTAB concentration are shown in Figure 6. Although a much more laborious technique than measuring or streaming potentials, the potential data (part A) obtained in this way avoid the problem of the slipping plane, and ita uncertain thickness, which plagues measurements using electrophoresis. At the same time it should be recognized that the values of the potential shown here are by no means absolute since the simple DLVO theory used for fitting the forces is only an a p p r o x i m a t i ~ nand ~ ~there ~ ~ ~is some uncertainty and subjectivity in the location of the plane (33yKjellander,R.; Marcelja, S. Chem. Phy.9. Lett. 1987, 142, 486.
1114 Langmuir, Vol. 10, No. 4, 1994
Rutland and Parker
:I
80 1
I
20
g
l 5 ~ I 100
-40 -60
2
-80 -100
-100
I
6
5
3
4
2
7
6
-log Concentration
,
g
10
Q
o
\ A '
I
6
5
5
4
3
2
-log Concentration
4
140 >
1
Y
. 3
2
-log Concentration
Figure 6. (A) Surface potential of silica at pH 10plotted against the logarithm of the CTAB concentration. (B)Pull-offforce as a function of CTAB concentration. The different symbols correspond to two different experimentswith different bimorph strips. of charge. The curve appears to have an inflection at zero potential, but this is probably an artifact due to incomplete equilibration. With the electrostatic driving force for adsorption removed, equilibration is expected to take a long time at such low concentrations. The pull-off force, or adhesion, changes dramatically as a function of the CTAB concentration (part B) and can be seen to correlate well with the surface potential data. The different symbols correspond to two different experiments with different bimorph sensors. Both seta of data give information on the amount of CTAB adsorbed, but the adhesion also gives information on the molecular orientation. At very low concentrations of surfactant, the adhesion is zero due to the presence of a hydration force and the potential is equivalent to that of silica at pH 10. As adsorption of CTAB increases the potential increases and plateaus with the formation of a second layer, whereas the adhesion increases, undergoes a maximum, and then decreases again with the formation of a bilayer. The adhesion corresponds to the adhesion after pushing throughlaway the bilayer except at the highest concentrations where a large force was required to push through the bilayer. (After initial buckling of the bilayer the adhesion is a function of the applied load, asymptoting to a value equivalent to monolayer-monolayer contact.) The data from Figure 6 (triangles) are shown along with data obtained at pH 5.6 on silicaZ2(circles) and a t pH 5.7 on micals (open diamonds) in Figure 7. The point at which the surfaces undergo charge reversal occurs at a much lower CTAB concentration for pH 10 than for pH 5.7 on silica or that at which CTAB neutralizes mica surfaces at pH 5.7. The curve has much the same shape as that for mica; however, the value of the potential at which the curve plateaus is a great deal lower, suggesting that the nature of the surface aggregates formed at high concentrations is different from that on mica. (34)Marcelja, S. In Interactions between interfaces in liquids; Charvolin, J., Joanny, J. F.; Zinn-Justin, J., Eds.; Elsevier: New York, 1990.
I
6
5
4
3
2
-log Concentration
Figure 7. Surface potential and adhesion data as a function of CTAB concentration for silica at pH 10 (fiied triangles, data from Figure 6))silica at pH 5.6 (filled circles),n and mica at pH 5.7 (open diamonds).ls In the latter two cases there was no background electrolyte. For the silica surfaces the potentialwas fitted assuming the plane of charge was 1.6 nm from monolayermonolayer contact. The observation that the adhesion at monolayer contact undergoes a maximum was also made for the adhesion between mica surfaces in CTAB solutions1s and must be due to a combination of effects involving surfactant molecules in reversed orientation within the monolayer, molecules trapped between the surfaces in contact, and the difference in the surface energy after the surfaces are separated. The values for the adhesion between two spheres should be doubled for comparison with crossed cylinders, and if this is done the maximum value is considerably smaller (a factor of 2) than that reported between hydrophobized mica surfaces; see for example ref 11. Forces between Bilayers. Surface forces in CTAB solutions measured at pH 5.6 at concentrations ranging from 8 X 10" to 1.2 X 1 0 - 2 M CTAB are shown in Figure 8 (filled diamonds, 8 X 10-4 M;open diamonds, 1 X 10-9 M; filled squares, 2 X le3M;open squares, 6 X 10-9M). The arrows indicate the position at which the surfaces jump into an adhesive contact. At 8 X 10-4 M the force is well fitted by DLVO theory at large separations. The surfaces jump into an adhesive contact from a separation of -3 nm after a very small extra repulsion. This separation is larger than expected from DLVO theory if the force is calculated with the plane of charge set to the position where the surfaces finally come into contact (Le., the contact between the hydrophobic monolayers). The pull-off force of 65 mN m-* is consistent with the surfaces achieving a hydrophobic contact; see for example ref 11. The fact that the surfaces are charged indicates that there is a substantial amount of surfactant adsorbed with the head groups exposed to the aqueous phase. This must be due either to molecules present in the monolayer with a reversed orientation or to material adsorbed weakly to the monolayer; in these experiments it is not possible to determine between these two alternatives. After the cmc is reached the evidence for bilayer formation is far more conclusive. The measured surface forces at the cmc and
Langmuir, Vol. 10, No. 4, 1994 1115
Surface Forces between Silica Surfaces
0
5
10
15
20
Separation (nm) Figure 8. Measured force scaled by the radius of curvature between two glass surfaces immersed in aqueous solutions (pH 5.6) of CTAB (filled diamonds, 8 X lo-' M; open diamonds, 1 X 10-8 M; fiied squares, 2 X 10-8 M open squares, 6 X 10-8 M). The arrows indicate the position at which the surfaces jump into an adhesive contact.
at concentrations above the cmc are very similar. The magnitude of the double-layer force is constant, and the decay length of the interaction varies according to the counterions and monomers only model.12 In all cases the forces turn steeply upward at small separations, indicating a strong steric interaction with the surfactant molecules. When the force between the surfaces exceeds -9 mN/m, the surfaces move inward. If they are then separated, an adhesion is measured which is a complicated function of the force applied. These results are entirely consistent with the formation of bilayers on the surfaces. When the surfaces are pushed to bilayer contact, the force required to deplete the contact zone of surfactant depends on the packing of the surfactant molecules in the outer layer of the bilayer. If the molecules are loosely packed, then it is relatively easy to push the molecules away from the contact zone. Increasing the surfactant concentration in the bulk liquid surrounding the surfaces leads to an increase in the adsorption and hence an increase in the packing density of molecules in the outer layer of the bilayer. This is indeed what is observed in the results shown in Figure 8. The height of the repulsive barrier increases very rapidly as the bulk surfactant concentration approaches the cmc. This is a simple consequence of the bilayer becoming increasingly tightly packed by adsorption. The force curve is sloped after the step, and this indicates that the bilayer compression occurs in two stages: an initial rapid step followed by a slow compression of the remaining material between the surfaces. At the maximum measurable force which can be applied by the apparatus the surfaces can be compressed 2.0 nm. On the basis of the raw data at a single concentration, it is impossible to determine the thickness of the adsorbed layer. However, if we consider the data from several force runs, we can confidently assign a minimum thickness for the second adsorbed layer. At a CTAB concentration of 6 X 10-4M (not shown) the surfaces jump into an adhesive contact from a separation of 3.0nm. On increasing the surfactant concentration only slightly (to 8 X 10-4M), the surfaces jumped from precisely the same separation. Increasing the surfactant concentration still further (1 X 10-8M)resulted in the appearance of a hard wall, and the position of this hard wall relative to the hydrophobic contact can be determined. The plane of origin of the surface charge must either
0
5
10
15
20
Separation (nm) Figure 9. Data from Figure 8 at CTAB concentrations of 8 X lo-' and 1 X 10-9 M plotted on a log scale. The solid lines are = 8.5 nm. The fits to DLVO theory with \ko = 80 mV and r1 plane of origin of the charge is held fixed at 3.2 nm with respect to contact between hydrophobicsurfaces, and the three different fits are with the plane of origin of the van der Waals interaction set to 2.4 (solid line which continues to smallest separation) 2.6 and 2.8 nm (lowest solid line).
remain at the same position, if surfactant adsorbs to make the outer layer of the bilayer more tightly packed, or shift outward. Hence,we adjusted the D = 0 so that the position of the steric barriers overlap, Le., the marked step in the force curve at 103 M (triangles) is superimposed on the small steric barrier at 8 X 10-4M shown in Figure 9. This procedure then gives an estimate of the minimum thickness (of 1.5 nm) of the outer adsorbed surfactant layer. Also shown in Figure 9 are three separate theoretical fits to the data at 8 X 10-4 M. The surface potential was held fixed, and the origin of the van der Waals interaction was set to 2.8,2.6, and 2.4 nm with reference to the contact between the hydrophobic surfaces. The best fit is obtained with the last separation chosen. At the slightly higher surfactant concentration the jump in disappeared as the height of the force barrier became larger; however, there was no appreciable change in the electrostatic repulsion. The forces measured at a CTAB concentration of 8 X 10-4 M (pH 10)are shown in Figure 10. The force runs were recorded in eight consecutive measurements, and the data from all eight are shown. The force can be fitted from DLVO theory above about 4 nm, giving a surface potential of 75 mV and a Debye length of 9.4 nm. The formation of a bilayer can be clearly observed by the large repulsive deviation from DLVO theory starting about 3 nm from contact and reaching a steep wall at approximately 2.8 nm. The two-step nature of the bilayer removal is very pronounced and consists of a step at constant force (at about 10 mN/m) into asecond repulsive regime which leads monotonically to adhesive monolayer contact. The adhesion between the surfaces before the initial force barrier is overcome is minimal, but as the surfaces enter the second, repulsive regime the adhesion increases as a function of the force applied, as depicted in Figure 11. This indicates that intersurface hydrophobic interactions increase during this part of the interaction. Figure 12 shows how the forces depend on the time the surfaces are permitted to equilibrate between force runs. Eight force runs are shown in a 1 X 10-8 M solution of CTAB at pH 5.6 with a period between the measurement of 20 s. If the period is 30 s between runs,then no difference is observed between the forces, indicating that an "equilibrium" force is being measured and that the bilayer has sufficient time to re-form. In the case of the 20-5period,
1116 Langmuir, Vol. 10, No. 4, 1994
Rutland and Parker
10
h
E
$ 1
0.1 0
5
15
10
20
25
30
Distance (nm)
Figure 10. Measured force scaled by the radius of curvature between two glass surfaces immersed in aqueous solutions of CTAB, 8 X lo-' M, pH 10. The solid lines are DLVO fits to the data. 45
I
/I
Figure 12. Surface forces measured on repeated approach of two surfaces immersed in CTAB, 1 x 1W M. The forces were measured by continuouslyramping the surfaces in and out with a saw-toothwave form, and the period betweenthemeasurementa was 20 s. If a period of 30 s was left between each of the force runs,then an equilibrium force curve was measured, see Figures 8 and 9.
h
E
g
40
Lo
25
35 30
5 0 0
10
20
30
40
50
F(max)/R (mN/m)
Figure 11. Measured pull-off force scaled by the radius of curvature between two glass surfaces immersed in aqueous solutions of CTAB at 8 X lo-' M, pH 10,as a function of the load applied between the surfaces. The adhesion increases monotonically with the applied load after the force wall of approximately 10 mN/m (see Figure 10) is overcome. however, the electrostatic repulsion and steric bilayer repulsion are progressively reduced with approachnumber until the force becomes uniformly attractive with a longer range than predicted from dispersion forces alone. Discussion Surface force measurementsprovide three specific pieces of informtion about the structure and nature of an adsorbed surfactant layer. Firstly the surface charge density obtained from fitting the force curves and the known ionization constant for the surfactant can be used to estimate the area per surfactant molecule. Secondly the measured layer thickness can be used to provide a picture of the orientation of the molecules in the bilayer. Thirdly the adhesion force measurementsand the presence or absence of a hydrophobic interaction provide clues about the orientation and packing of the molecules adsorbed to the surface. Forces at pH 10 below the cmc. It is at first sight surprising that at low surfactant concentrations there is
an extra repulsive force operating between the surfaces at pH 10 since that is usually ascribed to the hydration of silanol groups on the silica surface.*28 However, in the case of mica, a hydration force is observed due to the adsorption of hydrated metal ions and it is probable that, at the pH considered here, the hydration force a r k largely from the binding of sodium ions (since sodium hydroxide was used to adjust the pH). A binding constant for sodium ions to various silica surfaces has been determined by Grabbe and Horn,% and using a simple model of ion binding incorporating the ion size, as proposed by Pashley,36it is possible to gain an idea of the expected sodium coverage. The input parameters for such a calculation are the two dissociation constants, the area occupied by an adsorbed sodium ion, and the negative site area. The dissociation constants for hydrogen ions were found by Grabbe and Horn to be uniformly higher than found in earlier studies. However, the values obtained by {potential measurements using a similar3' and more sophisticated model of the interface% are in perfect agreement (pK, = 5.8). Using this value for the proton dissociation constant, the sodium ion dissociation of Grabbe and Horn (essentially independent of the surface treatmentas),the ion size determined by Pashley, and a negative site area of 0.25 nm2 (close to an "average" value of Grabbe and Horn, taken from Hair39) showed that, by pH 7, a sodium ion concentration of between 0.1 and 1 mM leads to 90% of the maximum possible sodium coverage. Of course the same is expected at pH 10; however, as shown by Scales et aL38 such simple models lose their validity above pH 8. This is because the silanol groups cannot be considered 88 independent sites with identical dissociation constants (35)Grabbe, A.; Horn, R. G. J . Colloid Interface Sci. 1993,167,376383. (36)Pashley, R. M. J. Colloid Interface Sci. 1981,83,631-646. (37)Rutland, M. W.; Pashley, R. M. J. Colloid Interface Sci. 1989, 130,448-456. (38)Scales, P. J.; Grieser, F.; Healey, T. W. Langmuir 1992,8,966-
_.
W A I.
(39)Hair, M.L.J. Non-Cryst. Soli& 1978,19,299,
Langmuir, Vol. 10, No. 4, 1994 1117
Surface Forces between Silica Surfaces
since the pK, is a function of the degree of charging.40The ability of a site to dissociate is dependent on the charging of neighboring sites, particularly if the sites are anchored to the same silicon atom, but the different ionization constants of vicinal, geminal, and isolated silanol groups have not been measured." However, for silicic acid the second ionization constant has been determiend to be 2 orders of magnitude smaller than the first.42 Energies of OH bonds have been obtained from infrared spectra, and the relative numbers of different silanol species have been estimated; see for example refs 39,43,and 44. Thus, the surface sites can only be treated as having a single binding constant if the dissociated sites are distant from one another. In a more recent study the surface potential of fused silica as a function of pH has been determined by second harmonic generation (SH).& The ratio of charged and uncharged densities at the interface was determined as a function of pH, and it was possible to estimate the pK, of the silanolsites. Specificallytwo different sites were found, one with a pK, of 4.5 which occupies 197% of the sites and the other with a pK, of 8.5 which occupies 81% of the sites. At pH 10the surfacepotential obtained in this study was -85 mV in 0.5 M background electrolyte. NMR measurements have shown that the main driving force for sodium interaction with silica is pure electrostatic interaction rather than specific site binding.a The proton is the potential-determining ion, and so the potential obtained from force measurements at pH 10 (Le., 90 mV) is similar to that at higher salt concentrations obtained by SH. The observation of a hydration force at pH 10 which can be disrupted by the addition of calcium ions has been observed between silica surfaces4' using the colloid probe technique developed by Ducker,28148and using the same techniques used in the present study,49 which strongly suggests that the hydration force is due to bound metal ions, which have a lower affinity for the surface than protons, since calcium ions have no effect on the hydration force at low pH. Significant dissolution of silica is not expected to take place until after about pH 10.5,50and so no gel layer is present on the silica surface. Considering the method of preparation of the silica surfaces, however, this is not surprising, since a gel layer is usually only invoked to explain charging behavior on precipitated, amorphous silica.sl Furthermore, there appear to be no measurable effects due to the dissolution of the surface that one might expect at this pH. The time-dependent adhesion displayed in Figure 2 has been discussed in a recent publicationm and arises from the adsorption of surfactant in the annulus around the contact position, which is favored due to the possibility of hydrophobic interactions across the gap as well as laterally,as depicted in Figure 13. The time ( t )dependence (40) Strazhesko, D.; Strelko, V. B.; Belyakov, V. N.; Rubank, S. C. J. Chromatogr. 1974,102, 191. (41) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979; p 661.
(42) Scherban, J. D. Dokl. Akad. Nauk SSSR 1967,177,1200. (43) Hair, M. L.; Hertl, W. J. Phys. Chem. 1969, 73,4269. (44) Burneau, A.; BarrBs, 0.;Gallas, J. P.; Lavalley, J. C. Langmuir 1990,6,1364-1372. (45) Ong, S.; Wao, X.; Eieenthal, K. B. Chem. Phys. Lett. 1992,191, 327. (46) Jang, H.M.; Fuerstanau, D. W. Langmuir 1987,3, 1114. (47) Meagher, L. J. Colloid Interface Sci. 1992,152, 293-295. (48) Ducker, W. A.; Senden, T. J.; Paahley, R. M. Langmuir 1992,8, 1831. (49) Rutland, M. W. Manuscript in preparation. (50) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979; p 47 and references therein. (51) Iler, R. K. The Chemistry of Silica; Wiley, New York, 1979; p 357 and references therein.
F l a t t e d Contact
-
* Figure 13. A schematic illustration of the contact between surfaces and the presence of surfactant in the gap around the contact region. The negative midplane potential causes surfactant to accumulate in the gap. The aggregation of surfactant in the gap is favored when the gap is less than twice the length of the hydrocarbon tail as the hydrocarbon tails can overlap. The time requiredto attain the equilibrium surfactant concentration is limited by diffusion of surfactant into the gap, and this gives rise to the time dependence of the adhesion as shown in Figure -1-
2.
of the adhesion is consistent with diffusion limitation as, for large t, it is linear with t-1/2F21s3 The time required to achieve equilibrium is large in comparison to previous diffusion studies,S2tM but this is due to the low concentration of surfactarkand the unfavorable geometry of the slit into which material must diffuse. The sodium ion concentration far outweighsthe surfactant concentration, so at large surfaceseparationsthe predominantly adsorbed ions are sodium, and thus, on separation, the CTA+ ions desorb. However, this process is not instantaneous, so when the surfacesjump apart, the surface charge inferred from the magnitude of the surface force is reduced compared to that measured on approach. Since desorption and ion exchange take place while the surfaces are being separated, the forces cannot be fitted with a simple DLVO interaction; however, we speculate that a careful analysis of the forces measured on separation should provide information on the desorption kinetics. We hope to examine this possibility in a future publication. At pH 10 the adsorption of CTAB becomes significant enough to be observed on approach runs at 2 X 1o-S M (Figure 3). It is difficult to draw any firm conclusions about the mechanism of the attraction below 8 nm which must be due either to the adsorption of surfactant, causing disruption of the hydration force and thus reducing the range of the repulsion, and allowing the van der Waals attraction to be observed, or to a small hydrophobic attraction being superimposed on the (reduced)hydration force, or a combination of the two. However, when the surfaces are allowedto equilibrate for a further 13h (Figure 4),the surface charge was neutralized (the adsorbed species neutralizing the surface charge at this concentration are sodium ions, CTA+ions, and protons). The monotonically attractive force profile is longer ranged than predicted by DLVO theory, indicating that with this degree of adsorption the surfaces are sufficiently hydrophobic for a hydrophobic force to be observed. The mechanism of the hydrophobic force is much debatedams The adhesion data in Figure 7 and Table 1 suggest that the amount of CTAB adsorbed is still much less than a full monolayer; however, the coverage is sufficient to give rise to a hydrophobic interaction. At such low adsorbed amounts it is highly unlikely that there is any ordering of ~~
(52) Davies, J. T.; Collins Smith, J. A.; Humphreys,D. G.Proceedings of the 2nd International Conferenceon Surface Activity; Butterworthe: London, 1957; p 281. (53) van Hunsel, J.; Bleys, G.; Jooe, P. J. Colloid Interface Sci. 1986, 114,432-441. (54) Eriksson, J. C.; Ljunggren, S.; Claesson, P. M. J. Chem. SOC., Faraday Tram. 2 1989,85,163. (55) Attard, P. J. Phys. Chem. 1989,93,6441. (56)Podgomik, R. J. Chem. Phys. 1989,91, 5840. (57) Yushchenko, E. A.; Yaminsky, V. V.; Shchukin, E. D. J. Colloid Interface Sci. 1981, 96, 307. (58) Attard, P. J. Phys. Chem., submitted for publication. Evans, D. F.; Wennerstr6mlm, H. Langmuir 1993,9, (59) Tsao, Y.-H.; 779-785.
1118 Langmuir, Vol. 10, No. 4,1994
Rutland and Parker
Table 1. Surface Potentials and Charges Obtained from Fitting
debye surface adhesion lengtho charge FdR (nm) (mCm-2) (mN/m) (MI Normal pH 9.2 X 10-8 -22 97 (99) 0.15 7 0 44 (44) 0 9-11 4.6 X 106 1.4 X lo-' 18 26 (26.5) 0.47 12-13 50 17 (17) 2.32 14-16 3.1 X lo-' 85 14 (14) 6.38 41 4.8 X lo-' 6.5 X lo-' 90 12 (12) 8.24 52-60 8.0 X lo-' 90 10.6 (10.6) 9.24 61 1.00 x 10-8 80 8.5 (9.5) 8.38 36 pH 10 b -90 18.5 (18.6) -0.25 6.0 X 10-7 11 18.5 (18.6) -0.42 2x104 -88 18.5 (18.6) -0.40-0 32 3X10-8 -74-0 16 (18) 0.51 67-70 2x106 45-49 16 (16.7) 1.19 65-67 6XlV 55 15 (15.7) 1.42 71-74 lXlo-' 51 13 (14) 2.02 67 2x104 53 11.5 (11.7) 3.31 4X10-4 59 55-61 8 X W 75 9.4 (9.3) 6.73 0.4-3.7 The decay lengths given in parentheses are calculated from the CTAB concentration assuming full dissociation and for pH 10 from the CTAB concentration and 2.6 X lo-' M NaOH added to adjust the pH. The data for normal pH are taken from ref 22 for concentrationsbelow 1 X 10-8M.b The adhesion at low concentrations is time dependent; see text and Figure 2. CTAB concentration
surface potential (mV)
chains on the surface, implying that the hydrophobic interaction seen here is not due to dipolar effects arising from chain packing consideration^.^^ If the surfactant were adsorbed in small patches of extended molecules-a hemimicellar arrangement-one would expect the strength of the interaction to increase with increasing coverage, and also that a step would be observed in the force curve as the chains were forced to tilt under a high compressive load. However, no step was observed, and increasing the surfactant concentration to 6 X 10-6 M had no effect whatsoever on the range of the hydrophobic attraction, so it is unlikely that aggregates form. The surface potential also remained constant so that an inflection was observed in the potential vs concentration curve (Figure 6), whereas the adhesion did increase. Thus, increased adsorption of CTAB took place, but there was an approximately 1:l replacement of other adsorbed cations so the surface potential did not measurably change. At higher CTAB concentrations the behavior is as for mica, with further adsorption taking place by ion exchange with sodium ions and due to lateral hydrophobic interactions between the adsorbed molecules so the potential increases to a 2ositive value. The adhesion data show that until about 1X lV M the adsorption takes place in a monolayer orientation. It can clearly be seen in Figure 4 that after the surfaces undergo charge reversal the attractive force which pulls them into contact is the hydrophobic attraction; however, as the surface potential increases with increasing adsorption it becomes harder to interpret the mechanism of the jump in. If the plane of charge is set at monolayer-monolayer contact, then the long-range part of the curve can be well fitted, with the surfacesjumping in from a larger separation than predicted by DLVO theory. This interpretation has routinely been used in the past,l4J5vmand the extra attraction has been ascribed to the hydrophobic interaction. Then, when a bilayer forms, the curves are fitted by changing the plane of charge to bilayer contact. However, such a sharp transition is not physically probable, and the maximum (60) Rutland, M. W.; Waltermo, A.; 8, 176-183.
Claesson, P. M. Langmuir
1992,
in the adhesion suggests that material is trapped between the surfaces in contact. It is likely that the hydrophobic surface is "screened" by weakly adsorbed surfactant molecules, or submicellar aggregates which are pushed away as the surfaces approach. Thus, an equally good fit can be attained by assuming that the plane of charge in fact lies two molecular lengths away from monolayer contact (Le.,3.2 nm). In this case, the surfaces no longer jump at distances longer than predicted by DLVO theory. What is obtained is really then an "effective potential" as it is not possible to distinguish between a charge on the monolayer and a diffuse charge weakly attached at some distance from the monolayer. The closer one is to bilayer formation the better is the approximation that the plane of charge is at a distance from the surface. The first evidence for the formation of a second layer occurs at 4 X 1V M (Figure 5). The force can be fitted extremely well with simple DLVO theory assuming a threelayer system for the van der Waals contribution with a J. The small adhesion Hamaker constant of 1 X associated with the bilayer may be due to dispersion forces, or to the fact that the bilayer is patchy so that there is a small hydrophobic attraction simultaneously with the steric repulsion. The nature of the bilayer formed on silica glass is discussed in a later section. Comparison with pH 5.6. The adsorption of CTAB is strongly dependent on the surface charge, and this is illustrated in the potential as a function of concentration shown in Figure 6 and values given in Table 1. The potential-concentration plot is related to the adsorption isotherm, and with care, the adsorption on different surfaces can be compared. The measurements for mica and silica at pH 5.7 were conducted in the absence of background electrolyte which makes the determination of a surface potential very difficult at low concentrations due to the long Debye length of the solution. However, it is clear that the point at which charge neutralization occurs is very dependent on the charging properties of the surface. The surface charge of silica is generally thought to arise due to the dissociation of protons from surface silanol groups to which they are chemically bound. However, at high pH it has been suggested that other factors than the dissociation of protons determine the surface charge, for example, the formation of vicinal silanol pairs with negative charge resulting from a change in coordination number from 4 to 5 of a surface silicon atom?' There is also more complex ion exchange with silicate ions. The charge on mica arises from isomorphic substitution of aluminum for silicon atoms in the silicate lattice, which in the crystal is neutralized almost exclusively by potassium ions. In neutral aqueous solution, potassium ions at the exposed surfaces are completely ion-exchangedby protons. Some differences between the potential behavior of mica and silica in CTAB at pH 5.7 seen in figure are discussed in a previous publication.22 At pH 10the glass surfaces are quite negatively charged, with sodium ions expectedto be the most densely adsorbed cations. It has been shown that for both mica and silica the proton binds to the surface much more strongly than alkali-metal ions.-*62 It follows, therefore, that CTAB displaces bound sodium ions more readily than protons; thus, adsorption and charge neutralization occur at lower concentrations at high pH. This is also demonstrated by the fact that the adhesion rises much more sharply with concentration at pH 10than at pH 5.6.22 At both pHs the maximum adhesion is the same (approximately 70 mN/ (61) Kirichenko,L. F.; Vysotaki,Z. Z. Dokl. Akad. Nauk SSSR 1967, 175, 635. (62) Scales, P. J.; Grieser, F.; Healey, T. W. Langmuir 1990,6,582589.
Surface Forces between Silica Surfaces
Langmuir, Vol. 10, No. 4,1994 1119
m), but at pH 10 the maximum value is achieved at approximately 1 X 106 M, whereas at pH 5.7 it does not achieve this value until 1 X 1 0 3 M, approximately the cmc. Thus, at pH 5.6 bilayers are not observed until the cmc as the first layer is insufficiently hydrophobic and densely packed. In this respect, the silica surface at high pH behaves more like the mica surface since a packed hydrophobic layer permits the formation of a bilayer structure at approximately half the cmc. The major difference between the mica and silica results lies in the difference between the hydrophobicities of the monolayer and the surface potential of the bilayers (see below). The explanation for both these phenomena is the same. As mentioned above the mica surface site is a negative lattice site which is always neutralized by a purely ionic bond with a cationic species, and thus it displays classic ion exchange charging behavior. Thus, as the concentration of CTAB dominates, all the sites become occupied by CTA+ ions. This is possible due to the fact that the maximum packing size of the CTA+ ion is smaller than the site area on mica.20 The result is a uniform, hydrophobic monolayer of CTAB which displays a large hydrophobic adhesion and on which a bilayer may form at higher concentrations. As mentioned earlier the adsorption sites on silica are very different from those on mica and are dissociating silanol groups. If there is a significant amount of dissociated silanol groups, the physisorption of CTA+ions to the negative sites is possible, and thus monolayer formation is observed. However, the proximity of the silanol sites to one another has two implications. Firstly, the CTA+ ion occupies a larger area than the silanol site density, so even if all the sites were to dissociate, they could not all be neutralized by CTA+ions;thus, a 1:l bilayer from simple adsorption kinetics is impossible (this applies to almost all cations). Secondly, the facts that a chemical bond is formed between the proton and the oxygen of the silanol group, that there is a significant hydrogen bonding between neighboring undissociated silanol groups, and that a much higher pK, is expected for the second ionization of vicinal silanol pairs mean that simple ion dissociation behavior is not observed and much less than a full coverage of adsorbed cations is permitted. It may well be that only isolated silanol groups on the surface are able to undergo ionic bonding with CTA+ ions. Thus, the monolayer is not as well developed as on mica, giving a somewhat lower hydrophobicadhesion. As a result no well-defined second, or outer, layer is formed at higher surfactant concentrations. The other possible explanation is that the site density on the silica is much lower than we assume. However, when the surfaces are treated with a monofunctional fluorocarbon chlorosilane, the adhesion between these hydrophobic surfaces in water (130-140 N/m,g3which gives a surface energy of 33 mJ/m with FIR = 4rE9 is close to the values obtained for the adhesion between surfactantcoated mica surfaces (400 N/m, which gives a surface energy of 38 mJ/m with FIR = 3.1rE69. Due to the different effective elasticities of the glass surfaces and mica glue compositesystem, slightly different formulas are required to relate the adhesive forces to the corresponding free energies. For a complete discussion on this point, see ref 66. This implies that the density of hydrophobic groups is similar to the density of hydrophobic groups in a tightly
packed monolayer on mica. Thus, we can be certain that the density of silanol groups is high. Formation of Bilayers. The measured degree of dissociation, Q/N, is 0.2 for CTAB in micelles (where Q is the charge on the micelle and N the number of molecule^);^^*^ hence, approximately 20% of the CTAB molecules at a micelle interface are charged. The surface charge obtained for mica surfaces coated with a bilayer of CTAB at 1 X 10-3 MI3 is 0.0069 C m-2, giving an area per charge of 2.5 nm2. Using the measured (single) layer thickness of 1.5 nm and the volume of the CTA+ ion of 0.76 nm3 (obtained from estimates by Tanford@),a degree of ionization of 0.22is obtained. It is not at all unreasonable to expect that the dissociation may depend on the curvature of the charged interface. However, the fact that the Q/N value obtained for micelles (where the curvature is high, R = 40 nm) agrees well with the data obtained from force measurements with mica (R = 2 cm) strongly suggests that the charging properties are largely independent of the radius of curvature of the interface. An equivalent analysis for CTAB adsorbed to glass immediatelyencounters problems. The surface potential obtained from DLVO fits to the surface force measurements for CTAB bilayers adsorbed to glass at 1 X 10-3 M is 80-90 mV. This is in good agreement with the surface potential of 90 mV obtained with the colloid probe technique for measurements of bilayers on silica.70 The fact that these values are nearly half the value obtained for mica at the same concentration suggests a very different structure in the adsorbed layers on the two different surfaces. The surface potential gives an area per charge of 13.4 nm2, and if the degree of ionization is 0.22, then the area per head group is 2.95 nm2. Using the previous volume for the CTA+ ion, the thickness of the layer obtained is 0.25 nm. Such a small value would suggest that the CTAB is lying down on the hydrophobic surface with the hydrocarbon tails oriented parallel to the silica interface (in which case the assumption of Q/N = 0.22 would be completely invalid). However,the totalmeasured thickness of the outer two layers of the bilayers is -3.0 nm. This value is much larger than expected considering the thickness obtained on the basis of surface charge, but is very similar to the values obtained for mica, implying the molecules are adsorbed in an extended configuration. Thus, the measurements of the bilayer thickness suggest a tightly packed bilayer on glass whereas surface potentials suggest rather that a sparsely packed layer is formed. These apparently contradictory statements can be resolved if the surfactant adsorbed to the surface forms patches of bilayers. In each patch the orientation of the surfactant molecules is close to perpendicular to the interface, giving the appropriate layer thickness, whereas the reduced number of head groups in the outer layer would give a very much reduced surface potential. In such a case the fitting procedure used here gives at best only an effective potential since the surface is no longer smooth with a uniform charge distribution. The fact that the jump shown in Figure 9 is better fitted by assuming that the plane of origin of the van der Waals force is closer to monolayer contact than is the plane of charge also supports the idea of bilayer patches since the implication is that there is a considerable amount of water within the layer defined by the plane of charge. Rennie et al.6 measured the thickness of CTAB bilayers at 4 X
(63) Parker, J. L.;Claesson, P. M. Langmuir, in press. (64)Muller, V. M.; Yushchenko, V. S.; Derjaguin, B. V. J . Colloid Interface Sei. 1983, 92,92. (65) Johnson,K.L.;Kendall, K.; Roberts, A. D. Proc. R. SOC. London 1971, A324, 301. (66) Attard, P.; Parker, J. L. Phys. Rev. A 1992,46,795*7971.
(67) h a , R. J. J. Colloid Interface Sci. 1980, 78, 330. (68) Dorshow,R.;Briggs, J.; Bunton,C. A.; Nicoll,D. F.J.Phys. Chem. 1982,86,2388. (69) Tanford, C. The Hydrophobic Effect; J. Wiley and Sons: New York, 1980. (70) Senden, T.; Drummond, C. Langmuir, submitted for publication.
1120 Langmuir, VoZ.10, No. 4, 1994
10-4 M by neutron scattering. They concluded that the hydrocarbon region of the bilayer adsorbed to quartz was 2.8 nm thick and that the head groups were about 0.6 nm thick. They also concluded that the area per head group was 0.3 nm2 for CTAB adsorbed to quartz, that the fractional coverage of CTAB adsorbed (from 6 X lo4 M bulk concentration) to quartz was 80%, and that the surfactant formed aggregates on the surface with an upper limit to the size of the aggregates of 1pm. It has not been established whether the aggregates are flattened micelles or defective bilayers. However, the results from neutron scattering certainly confirm the surface patch model proposed to explain the force measurements. Adsorption isotherms have been measured for a number of cationic surfactants adsorbed to s i l i ~ a . ~ lFor - ~ ~tetradecyltrimethylammoniumbromide (C14TAB)adsorbed M) the on silica73 slightly above the cmc (3.5 X adsorbed amount is -3.5 pmol/m2 at pH 4.8. The adsorbed amount for a closely packed bilayer with an area per head group of 0.5 mm2 is 6.64 pmol/m2. At least for C14TAB the adsorbed amount on silica is about half of what is expected for a tightly packed bilayer. C14TAB differs from CTAB by two fewer CH2 groups in the aliphatic chain; thus, the adsorption due to hydrophobic interactions should be stronger for CTAB than C14TAB, giving a higher adsorption density for CTAB. However, the fact that the adsorption plateaus before a tightly packed bilayer is formed for C14TAB means that we can expect the same behavior for CTAB albeit at a significantly different concentration. Adsorption of CTAB to silica has also been studied at elevated pH, and even at pH 10 the adsorbed amount plateaus at concentrations above 1X 103Mat an adsorbed amount of 4.5 p m ~ l / m ~ .This ~ l value is well below that (6.64 pmol/m2)expected for a completely packed bilayer. Even lower values for the maximum adsorbed amount at normal pH’s are obtained by Zhu et al.72 The amount adsorbed cannot produce a fully packed bilayer, and this further supports the idea that CTAB adsorbs to silica as surface aggregates of some description. There is a considerable amount of indirect evidence for the formation of surface aggregates. The so-called hemimicelle structure for adsorbed surfactant layers has been invoked since the 1950s.74,75Several theoretical models also suggest that surfactant is aggregated on the surfaces; for instance, Bohmer et al.76 used an extended selfconsistent field lattice theory to account for neutron reflectivity measurements of glycol monododecyl ethers adsorbed to silica. The conclusion from this work was that nonionic surfactants form surface aggregates. Rupprecht and G u have ~ ~also argued that surface aggregates occur when ionic surfactants adsorb to surfaces. On the basis of the bilayer thickness and the surface potential alone, it is impossible to conclude definitively that surface aggregates are present in CTAB layers adsorbed to silica; however, since there seems no reason to doubt the accuracy of the mica results and since there is no peculiar change in the ionization of the layer, then surface aggregates appear to be the most probable explanation for the observations. (71) Bijsterbosch, B. H. J. Colloid Interface Sci. 1974,47, 186. (72) Zhu, B. Y.; Gu, T.; X., Z. J. Chem. SOC.,Faraday Trans. 1 1989, 85, 3819. (73) Wiingnerud, P.; Olofsson, G. J. Colloid Interface Sci. 1992,153, 392. (74) Gaudin, A. M.; Fuerstenau, D. W. Trans. AIME 1955,202,958. (75) Somasundaran,P.;Healy,T. W.; Fuerstenau,D. W. J. Phys. Chem. 1964,68,3562. (76) Biihmer, M. R.; Koopal, L. K.; Jansen, R.; Lee, E. M.; Thomas, R. K.; Rennie, A. R. Langmuir 1992,8,2228-2239. (77) Rupprecht, H.; Gu, T. Colloid Polym. Sci. 1991,269, 506.
Rutland and Parker Fully Packed Bilayers
Patchy Bilayers
Figure 14. Schematicillustration of the hemifusion and proposed state of the adsorbedlayers of CTAB on glass surfaces. (A) shows the initial contact, (B) the rupture of the layers, and (C) the squeezing process. The primed symbols refer to the case on glass spheres as used here and the others to the case of mica as proposed by Horn78and Israela~hvili.~~
Close examination of the force curves obtained as the bilayer is being compressedbetween the surfaces provides further clues as to the structure of the bilayer. The fusion of bilayers has been studied e x t e n ~ i v e l y .Helm ~ ~ ~ et ~ ~al. studied the fusion of partly depleted bilayers of CTAB adsorbed to mica with the traditional surface force measurement techniques.79 Hemifusionoccurs when two bilayer-coated surfaces are pressed together until at a particular value of the applied force the outer layers of the bilayer rupture and the surfaces come into monolayermonolayer contact. The surfaces then zip together, and the size of the contact region between the hydrophobic monolayer grows. The rate of growth of the contact region depends on the solubility of the surfactants comprising the bilayer. This is schematically illustrated in Figure 14. When the surfaces are separated, a very strong adhesion is measured which is characteristic of two surfaces in hydrophobic contact. There are two important differences between the observation made for the glass substrate and those made when mica is used. Firstly, the bilayers on glass can be sequeezed into a hydrophobic contact at all the CTAB concentrations so far studied. Helm et al. did not observe hemifusion when the bilayers were fully developed, Le., when the surfactant concentration in the bulk was greater than 0.4 mM. Secondly,the fusion process proceeds by a two-step mechanism: a rapid rupture step followed by a much slower squeezing process. The fact that the outer bilayers can be squeezed away from the contact region for all the concentrations studied so far is not unexpected. The lower packing density of surfactant adsorbed in the bilayer means that a smaller force is required to remove the layer. The reason that hemifusion is only observed at particular CTAB concentrations is a result of the surface flattening which occurs in the mica experiments. When glued mica surfaces are pressed together under a high load, they deform and a flattened contact region forms. The precise shape of the surfaces in contact is dictated by the surface forces which exist between the With bilayers of CTAB adsorbed to the surface, a hard wall is encountered when the bilayers contact, and as a result a large amount of surface deformationis expected. The surface deformation causes an overestimation of the surface force when the Derjaguin approximation is used.81 Thus, it is possible to apply very large forces without pushing over the force ~
~~~~~
(78) Horn, R. G. Biochim. Biophys. Acta 1984, 778,224. (79) Helm, A. C.; Israelachvili,J. N.; McGuiggan, P. M. Science 1989, 246,919-922. (80)Parker, J. L.; Attard, P. J. Phys. Chem. 1992,96,10398. (81) Derjaguin, B. V. Kolloid-2. 1934, 69, 155.
Surface Forces between Silica Surfaces barrier and achieving a minimum which occurs at a much lower force value. Hemifusion in the mica case results simply from the fact that at a particular force the monolayer yields and there is an enormous force stored in the deformation of the surface which pushes the remaining material away. The mica-glue-silica composite has a much lower effective elastic constant than do the small glass spheres used in the present experiments, so it is expected that the observed forces will be different, and in fact the force law for the glass results shows two distinct regimes during compression. This is best illustrated in the results obtained at pH 10, Figure 10. The first step in the force curve is a rapid change in surface separation with almost no change in force. This most probably corresponds to compression of the layers from their extended conformationand rupture. The next part of the force law is repulsive and corresponds to the squeezing out of the bilayer patches or admicelles. This mechanism is supported by the measurements of adhesion in Figure 11. The adhesion between the surfaces increases dramatically as a function of applied force only after the initial force barrier of approximately 10 mN/m is overcome, implying that hydrophobic interactions across the layers start at this point. Other ways of rationalizing the results seem far less probable. For example, one could imagine that the first step in the fusion process may occur by rearrangements of the surface patches. If the surface patches move to positions where there fs no opposing patch on the opposite surface, then a three-layer contact may form. In this situation there is amonolayer on eachsurface with an intermediate surfactantlayer in between. Another possible mechanism is that interdigitation occurs between molecules in the layers rather than patches. However, both these mechanisms result in head groups being pushed into a nonpolar environment and are thus very unlikely. It is far more likely that the first step in the fusion of the bilayers is due to the compression of the outer layer into a thinner but more uniformly dense layer, which could be achieved if the outer layer of molecules were to tilt during the compression or if the bilayer patches were sparse enough that the chains could be compressed from their frozen, extended configuration. Regardless of the mechanisms of the first step in the hemifusion process, the result is that there is an intermediate layer of surfactant left between the surfaces. Since in this system there is no glue layer, a repulsive force is measured as material is pushed away from the contact region as depicted in Figure 14. The number of hydrophobic contacts across the layer increases as the layer is squashed down and outward and the surfactant between the hydrophobic layers becomes more difficult to expel. Thus, the adhesion between the surfaces is dependent on the applied load, and eventually it plateaus, approaching the value expected for a "pure" hydrophobic contact. At pH 10 the force required to remove the bilayer at 8 X 10-4 M CTAB is comparable to the force required to remove the bilayer at 1 X 103 M CTAB at normal pH, and no bilayer is observed at 8 X 10-4 M CTAB at normal pH. This is due to the fact that the monolayer formed at high pH is slightly more densely packed than a monolayer formed at low pH, and so the hydrophobic interaction which binds the second layer of the bilayer to the surface is stronger. This results in a more densely packed second
Langmuir, Vol. 10, No. 4,1994 1121
layer at the same concentration, and bilayer formation at a lower surfactant concentration. The surface potential 90= 75 mV at pH 10 ie even lower than the results obtained at a higher concentration at normal pH, and this indicates an even lower density of charged groups in the outer layer of the monolayer. The adsorption density measured from isotherms at this pH ~~;~ again ~ this value for CTA+ to silica is -4 ~ c m 0 l . m once is much lower than the value expected for a tightly packed bilayer of CTAB, indicating patchwise adsorption. When the period between successive force runs is reduced, then a situation can arise where there is insufficient time between the runs for the bilayer structure to completely rebuild before the next contact between the surfaces. This is shown dramatically in Figure 12. With each successive approach the force decreases until an extra attraction not accounted for by van der Waals interactions is measured. If a longer period is left between the force runs, then a single equilibrium interaction is obtained. This gives us a measure of the kinetics involved in the formation of the second layer. When the surfaces are pushed into monolayer contact, the outer layer is pushed out from the contact area; however, it is not obvious whether it is pushed completely off the surface or merely packed into a denser layer away from the contact zone. The measurement of the increasingly attractive force on subsequent approaches suggests that there is very little material trapped between the hydrophobic surfaces in contact, as that would tend to screen the hydrophobic attraction. This effect is crucially dependent on the concentration and is most pronounced at the concentration at which steric forces due to bilayer formation are first observed.
Conclusion The adsorption of CTAB to silica glass is strongly dependent on the surface charge and the nature of competing inorganic cations. At very low concentrations a time-dependent, diffusion-controlled adsorption takes place in the thin annulus around the contact zone due to the favorable interaction across the gap, as well as laterally, between hydrocarbon tails. No evidence for dissolution of silica or for the presence of a gel layer was observed at pH 10. Above the concentration at which electroneutrality occurs the forces can be equally well fitted assuming a plane of charge away from the monolayer contact; in this case a hydrophobiccomponent of the force is not observed. The surface charge obtained from fitting surface force curves and measurements of adsorption isotherms of CTAB on glass consistently indicate that bilayers adsorbed to silica are not densely packed. However, the measured layer thickness indicates that the molecules are oriented perpendicular to the surface and fully extended. This can only arise if CTAB forms surface aggregates on the silica surfaces. The nature of the aggregates, i.e.,whether they resemble flattened micelles or patches of bilayers, has yet to be established. A measure of the time taken for the bilayers to reform has also been obtained. Acknowledgment. The authors wish to thank Per Claesson and Vasili Yaminsky for valuable advice and discussions.