Spontaneously Formed Nonequilibrium Vesicles of

Langmuir 2004, 20, 6611-6618

6611

Spontaneously Formed Nonequilibrium Vesicles of Cetyltrimethylammonium Bromide and Sodium Octyl Sulfate in Aqueous Dispersions Mats Almgren* and Stanislav Rangelov† Uppsala University, Department of Physical Chemistry, P.O. Box 579, SE-751 23 Uppsala, Sweden Received March 25, 2004. In Final Form: May 13, 2004 It is well-known that vesicles form in mixtures of cationic and anionic surfactants. We have investigated mixtures of cetyltrimethylammonium bromide (CTAB) and sodium octyl sulfate (SOS) with the latter in excess over a long time, about 500 days. We have followed the growth of the aggregates by light scattering and checked the morphologies by cryogenic transmission electron microscopy (cryoTEM). All samples showed a monotonic growth with decreasing rate (the change of size was about linear on a logarithmic time scale). In series of samples with weight ratio 30:70 of CTAB/SOS and total surfactant concentration between 0.5 and 3 wt %, the size increased with the surfactant concentration up to 2 wt % and decreased thereafter; cryoTEM examination revealed that the samples contained a majority of open bilayer structures at the highest concentrations. Part of the sample at 2 wt % was diluted to 0.5 wt % after 60 days. The size measured after dilution was slightly smaller than before but well above that found in the directly prepared 0.5 wt % sample, and the particle size in the three samples continued to grow in parallel. Structures other than unilamellar vesicles were observed also in samples at 2 wt % total surfactant concentration at CTAB/ SOS ratios close to the borders of the vesicle lobe in the (quasi) ternary phase diagram as published (Yatcilla, M. T.; Herrington, K. L.; Brasher, L. L.; Kaler, E. W.; Chiruvolu, S.; Zasadzinski, J. A. J. Phys. Chem. 1996, 100, 5874). The results clearly show that the spontaneous vesicle populations do not represent equilibrium populations. They also suggest that the vesicle lobes in the phase diagram mainly represent areas where a lamellar phase is easily dispersed in the form of vesicles in an aqueous solution.

Introduction Spontaneous formation of vesicles occurs on mixing solutions of anionic and cationic surfactants, with either one in excess. Many such systems have been investigated, with respect to the position and extension of the vesicle lobes in the (quasi) ternary phase diagram, the size distribution of the vesicles, and the kinetics of their formation.1-6 Recently a number of investigations have compared experimentally determined vesicle size distributions to theoretical ones, seemingly quite successfully.7 The theoretical distributions were based on equilibrium size distributions from a simple theory of balance between the curvature elasticity of the bilayers, allowing for asymmetric bilayers with a spontaneous curvature, and the entropy of mixing of the vesicles.8 As a result, estimates of the spherical bending modulus and the spontaneous curvature were obtained from the analysis. Vesicles in the system cetyltrimethylammonium bromide (CTAB)/ sodium octyl sulfate (SOS)/water with a CTAB/SOS ratio of 3:7 were found to have a low bending modulus and a * To whom correspondence should be addressed. E-mail: [email protected]. † Present address: Institute of Polymers, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria. (1) Yatcilla, M. T.; Herrington, K. L.; Brasher, L. L.; Kaler, E. W.; Chiruvolu, S.; Zasadzinski, J. A. J. Phys. Chem. 1996, 100, 5874. (2) Kaler, E. W.; Murthy, A. K.; Rodriguez, B. E.; Zasadzinski, J. A. N. Science 1989, 245, 1371. (3) Kaler, E. W.; Herrington, K. L.; Murthy, A. K.; Zasadzinski, J. A. J. Phys. Chem. 1992, 96, 6698. (4) Xia, Y.; Goldmints, I.; Johnson, P. W.; Hatton, T. A.; Bose, A. Langmuir 2002, 18, 3822-3828. (5) Shioi, A.; Hatton, T. A. Langmuir 2002, 18, 7341. (6) Marques, E. F. Langmuir 2000, 16, 4798. (7) Coldren, B.; van Zanten, R.; Mackel, M. J.; Zasadzinski, J. A.; Jung, H.-T. Langmuir 2003, 19, 5632-5639. (8) Jung, H. T.; Coldren, B.; Zasadzinski, J. A.; Iampietro, D. J.; Kaler, E. W. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1353-1357.

broad size distribution around a spontaneous radius of curvature of 37 nm, whereas in a system containing a fluorinated component, CTAB/FC7 (sodium perfluorooctanoate)/water at CTAB/FC7 ) 2:8 by wt, much higher values of the bending modulus were deduced, resulting in a narrow size distribution centered around a spontaneous curvature of 23 nm. There are some unsolved problems, however. One is the equilibrium state: are the vesicles not only formed spontaneously but also represent a true equilibrium state, as required by the theoretically derived size distributions? This question has been addressed earlier, for example, by Marques6 who showed that vesicles in mixtures of didodecylammonium bromide (C12TAB) and sodium dodecyl sulfate, formed along several different pathways, had different shape and size distributions. These differences remained even after about 3 years, implying that the size distributions were not equilibrated. A second problem is the nature of the so-called “vesicle phases”, or the lobes within which vesicles are found to form spontaneously in the phase diagram. Is there any clear difference between these vesicle solutions and the solutions containing vesicles that are obtained by adding a cationic surfactant solution to a lamellar phase of lecithin or other zwitterionic lipids?9,10 In the latter case, the vesicle solution is commonly regarded as a dispersion of a lamellar phase in an aqueous solution, even though it forms very easily, spontaneously by usual standards and also thermodynamically, because the vesicle dispersion represents a state of lower free-energy than the starting solutions, even if it not necessarily is the true equilibrium state. Mixtures of a cationic and an anionic surfactant in water contain varying concentrations of salt from the surfactant (9) Gustafsson, J.; Ora¨dd, G.; Almgren, M. Langmuir 1997, 13, 69566963. (10) Almgren, M. Biochim. Biophys. Acta 2000, 1508, 146-163.

10.1021/la049211y CCC: $27.50 © 2004 American Chemical Society Published on Web 06/26/2004

6612

Langmuir, Vol. 20, No. 16, 2004

counterions and, in addition, some free surfactant ions. Because of this, mixtures of the three components are only quasi-ternary systems. By neutralizing the acid of the anionic component with the OH- form of the cationic surfactant, a true catanionic surfactant is obtained. Binary phase diagrams of such compounds in water and true ternary systems obtained by also adding an ionic surfactant were studied early on in Lund.11 The phase behavior of the ternary system is similar in many respects (allowing for the comparatively high solubility of the surfactants) to that of zwitterionic lipids with added ionic surfactant. In both there is a prominent lamellar phase that swells extensively as it becomes charged by an excess of the surfactant. Vesicle formation was not looked for in the early study of the catanionic system but is known to occur spontaneously in lipid/surfactant systems, as mentioned previously. Further addition of surfactant leads to the formation of mixed micelles, sometimes threadlike, and an area where micelles coexist with the lamellar phase (possibly dispersed as vesicles). In excess water (or brine), the swollen lamellar phase obtained after addition of a charged amphiphile to zwitterionic or nonionic lipids is usually very easily dispersed into vesicles. In the system C12TAB/egg phosphatidylcholine (EPC)/100 mM NaCl, for example, at a total concentration of 0.75 wt % and molar ratio C12TAB/ EPC of 2:1, unilamellar vesicles with a broad size distribution (radii from 7 to 200 nm) were observed by cryogenic transmission electron microscopy (cryoTEM).9 The vesicles were found to be well separated at an average distance of about 25 nm. This is a larger separation than what would be expected from only electrostatic repulsive forces in 0.1 M salt. It may be a consequence of the confinement in a thin film that occurs in cryoTEM; the largest vesicles actually appear to be so large that they distend the film. In any case, repulsive forces separate the vesicles and keep them unilamellar. On an increase of concentration, the repulsive forces between the vesicles become more important. In the vicinity of the close-packing concentration, which is lower the larger the vesicles, the interactions might affect the size and shape of the vesicles. It is possible that the extensions of the vesicle lobes in the phase diagram of cationic/anionic surfactant mixtures are partly determined by this close-packing limitation,1 and if so the extension of the lobes would tend to be smaller for larger vesicles. A change of the morphology of the aggregates in the vicinity of the lobe borders can be expected and is of particular interest to investigate. In CTAB/EPC/brine and SOS/CTAB, as well as some other systems, the originally formed vesicles are often found to grow slowly, seemingly approaching a limiting size depending on the composition and other conditions. Sometimes this limiting size is tacitly assumed to represent equilibrium. A simple test of the assumption would be to change the conditions so that a decrease of vesicle size would be required to restore equilibrium. Such tests have seldom been performed. In other words, growth of vesicles to produce larger entities is often observed, but processes that decrease the size are much less frequent. The vesicles must then be regarded as kinetically trapped. An objective of the present investigation is to test if the catanionic systems are different from the zwitterionic lipid/ ionic surfactant systems in this respect. An exception is the interesting results of Zemb and Dubois that indicate that vesicles in true catanionic systems with a slight excess of one charged component but no salt could be made to reversibly change their size, up and down, by changing (11) Jokkela, P.; Jo¨nsson, B.; Khan, A. J. Phys. Chem. 1987, 91, 3291.

Almgren and Rangelov

the temperature.12 The strong and unscreened electrostatic forces seem to be decisive for the size of the vesicles in this case. To get some insight into the nature of these systems, in particular in what respects they differ from polar lipid/ surfactant systems10 and if the vesicle size distributions really are equilibrated, we have investigated the SOS/ CTAB/water system at compositions belonging to the vesicle lobe with an excess of SOS. Two series of samples were prepared, one at CTAB/SOS ) 3:7 by weight at different total concentrations from 0.5 to 3 wt % and one series at a total concentration of 2 wt % and varying ratio. The change of size was followed over more than 1 year by dynamic light scattering (DLS) measurements, and the morphology of the samples was assessed with cryoTEM after about 40 days. Experimental Section Materials. CTAB and SOS of high purity were obtained from Serva and BDH, respectively, and used without further purification. The critical micelle concentrations of pure SOS and CTAB determined by the surface tension method were approximately 3 and 0.03 wt %, respectively. These values agree well with the literature data (3.0 and 0.029 wt %, respectively)31 and are in conformity with the high purity of the materials. Sample Preparation. Samples were prepared by first making stock solutions of each surfactant in purified water (Millipore Super-Q-System). Prior to preparing samples, the stock solutions were filtered through a 0.2-µm filter and then mixed at the desired ratio followed by diluting the resulting dispersions to the desired concentration. After brief vortexing, the dispersions were not subjected to any type of mechanical agitation. They were kept at 25 °C. Light Scattering. As described previously,13 the light scattering setup consisted of a 488-nm Ar ion laser and detector optics with an ITT FW 130 photomultiplier and an ALV-PM-PD amplifier-discriminator connected to an ALV-5000 autocorrelator built into a computer. The cylindrical scattering cells were sealed and then immersed into a large-diameter thermostated bath containing the index matching fluid Decalin. Measurements were made at different angles in the range 50-130° at 25 °C. In depolarized light scattering experiments the scattered light passed through a Glan-Thompson polarizer in the detection line, thus, allowing separate evaluation of the polarized (VV) and depolarized (VH) components of the scattered light. The dynamic data were analyzed by fitting the experimentally measured g2(t), the normalized intensity autocorrelation function, which is related to the electrical field correlation function g1(t) by the Siegert relationship:14

g2(t) - 1 ) β|g1(t)|2

(1)

where β is a factor accounting for deviation from ideal correlation. For polydisperse samples, g1(t) can be written as the inverse Laplace transform (ILT) of the relaxation time distribution, τA(τ):

g1(t) )

∫τA(τ) exp(-t/τ) d ln τ

(2)

where t is the lag time. The relaxation time distribution, τA(τ), is obtained by performing ILT using the constrained regularization algorithm REPES,15 which minimizes the sum of the squared differences between the experimental and calculated g2(t). A mean diffusion coefficient D90 is calculated from the second moment of each peak as D90 ) Γ90/q902, where q90 is the magnitude of the scattering vector q90 ) (4πn/λ) sin(θ/2) and Γ90 ) 1/τ90 is the relaxation rate of each mode at an angle 90°. Here, θ ()90°) is (12) Zemb, Th.; Dubois, M. Aust. J. Chem. 2003, 56, 971-979. (13) Rangelov, S.; Brown, W. Polymer 2000, 41, 4825. (14) Chu, B. Laser Light Scattering, 2; Academic Press: New York, 1991. (15) Jakes, J. Czech J. Phys. B 1988, 38, 1305.

Spontaneously Formed Nonequilibrium Vesicles

Langmuir, Vol. 20, No. 16, 2004 6613

the scattering angle, n is the refractive index of the medium, and λ is the wavelength of the light in a vacuum. The hydrodynamic radius at θ ) 90°, R90, was calculated using the Stokes-Einstein equation (eq 3):

R90 ) kT/(6πηD90)

(3)

kT is the thermal energy factor, and η is the temperaturedependent viscosity of water. CryoTEM. Transmission electron microscopy observations were conducted on a Zeiss EM 902 A instrument operating at 80 kV. The procedure for the sample preparation has been discussed.16,17 In short, a drop of the dispersion is deposited on an electron microscopy copper grid coated by a perforated polymer film. Excess liquid is blotted by a filter paper, leaving a thin film of the dispersion on the grid. The film on the grid is vitrified by plunging the grid into liquid ethane. The vitrified sample is then transferred in a cold stage to the microscope for observation at low temperature.

Results General Remarks and Visual Observations. In the present work, we investigated the aggregate morphology and the effect of aging on the size and stability of vesicles formed from oppositely charged surfactants, CTAB and SOS. The compositions and concentrations of the samples were well within the vesicle lobe according to the ternary phase diagram for CTAB/SOS/water published by Yatcilla et al.1 The samples were prepared at seven different total surfactant concentrations, 0.5, 1.0, 1.3, 1.6, 2.0, 2.5, and 3.0 wt %, at a fixed CTAB-to-SOS weight ratio of 30:70 and in four different ratios, 37:63, 33:67, 20:80, and 15: 85, at a fixed total surfactant concentration of 2.0 wt %. Stock solutions were mixed at desired ratios and quickly diluted to the desired concentration. After brief vortex mixing the samples were left undisturbed for long time. The initially prepared samples looked similar. At all concentrations and mixing ratios they appeared as moderately to highly opalescent dispersions. All samples contained easily dispersible turbid wisps. As the samples aged, distinct changes visible to the naked eye were observed only in the most dilute dispersions. At the mixing ratio 30:70 and total concentrations 0.5 and 1.0 wt %, a crystalline precipitate appeared after about 360 and 445 days, respectively. The precipitate was not analyzed but probably consists of cetyltrimethylammonium octylsulfate. Yatcilla et al.1 observed precipitate after a few days in samples of several compositions with SOS in excess, at very low total surfactant concentrations. Apparently, the bilayer vesicles are less stable thermodynamically than the salt, but in our samples the product of the activities of the free surfactant ions in the solution outside the vesicles is only slightly larger than the solubility product so that nucleation and precipitation require a very long time to occur. It should be noted, however, that even vesicles that are metastable with respect to the solid phase could reach a size distribution that represents a (quasi)equilibrium if the precipitation of the solid is slow enough. The question of chemical stability is relevant to alkyl sulfate surfactants in solution because they are subject to hydrolysis to the corresponding long-chain alcohols.18 The kinetics of the hydrolysis processes have been carefully studied by Kurz.19 From the results it can be concluded that the hydrolysis is extremely slow at room temperature and low acidity. The chemical stability of the samples (16) Almgren, M. Aust. J. Chem. 2003, 56, 959-970. (17) Almgren, M.; Edwards, K.; Gustafsson, J. Curr. Opin. Colloid Interface Sci. 1996, 1, 270-278. (18) Muramatsu, M.; Inoue, M. J. Colloid Interface Sci. 1976, 55, 80. (19) Kurz, J. L. J. Phys. Chem. 1962, 66, 2239-2246.

Figure 1. Relaxation time distributions measured at an angle of 90° for dispersions with a CTAB/SOS mixing ratio 30:70 at (a) 1.0 wt % and 31 days incubation time and (b) 0.5 wt % and 99 days incubation time.

was ensured by keeping them at room temperature. It has already been noted by others that the surfactant degradation is not the cause of the slow changes in the aggregate morphology and dimensions observable upon long-term aging.1,6 The time evolution of the particle size was monitored with DLS, whereas the aggregate morphology was explored with cryoTEM. Slow Mode, Fast Mode, and Multiple Scattering. Typical relaxation time distribution spectra are presented in Figure 1a,b. As is seen, the dominant modes are accompanied by fast and slow modes of low amplitude. In contrast to the fast modes which are invariably present, the slow modes appear unsystematically. They are diffusive and correspond to particles with radii from about 100 nm, reaching even more than 1 µm. As a coarse estimate, the ratio Cslow/Cdominant is of the magnitude 10-3. We attribute the slow modes to the presence of easily dispersible turbid wisps. The latter may be fine precipitates or dilute flocks of multilamellar or giant vesicles or just concentration fluctuations of the aggregates. The positions of the fast modes varied slightly within the samples. It was revealed from the angular dependence (not shown) that the fast modes were not diffusive. Originally we considered them to reflect internal motions of the particles responsible for the slow mode. However, as clearly seen from Figure 1, the fast modes were invariably present even when the slow modes were absent.

6614



Figure 3. DLS relaxation rate as a function of sin2(θ/2) in the depolarized (VH) and polarized (VV) geometries.

Figure 2. Variations of the scattered light intensity measured at an angle 90° of the polarized (a) and depolarized (b) components with the concentration for dispersions with a CTAB/ SOS mixing ratio 30:70. The incubation times are (a) 99 days (squares) and 178 days (circles) and (b) 17 days (squares) and 24 days (circles).

It was suggested, therefore, that they might derive from multiple scattering. The latter is a feature that is often present in light scattering experiments and frequently complicates the data analysis and interpretation. Both experimental techniques and data analysis approaches have been developed to avoid this problem.20-23 Several effects of multiple scattering were observed in the present systems. Figure 2a shows the variations of the scattered light intensity at θ ) 90° (I90vv) from solutions with a mixing ratio CTAB to SOS of 30:70 in the concentration range 0.5-3.0 wt %. Although the I90vv readings are influenced by the presence/absence of a slow mode, the curve pattern shown in Figure 2a is typical for multiple scattering.23,24 Furthermore, by placing a polarizer in the detection line (20) Schatzel, K. J. Mod. Opt. 1991, 38, 1849. (21) Phillies, G. J. D. J. Chem. Phys. 1981, 74, 260. (22) Phillies, G. J. D. Phys. Rev. A 1981, 24, 1939. (23) Stepanek, P. J. Chem. Phys. 1993, 99, 6384.

a depolarized scattering mode was detected.23 The relaxation rate of the latter is independent of the wave vector as is also characteristic of multiple scattering.25 The angular dependence of the relaxation rates for the depolarized (VH) and polarized (VV) experiments are shown in Figure 3. Interestingly, the position of the depolarized scattering mode does not coincide with the position of the fast mode in the polarized experiments. However, as noted elsewhere,23 because of the low amplitude of the latter its position as determined by ILT may be wrong. The next feature in conformity with the assumption for the presence of multiple scattering is the variation of I90vh, the intensity at θ ) 90° in the depolarized experiments, with the concentration of CTAB/SOS 30:70 dispersions (Figure 2b). The curve pattern is similar to that observed for I90vv (Figure 2a); however, the I90vh values are considerably lower. In fact, the depolarization ratio, I90vh/I90vv, for the whole aging period and at all concentrations and compositions, is in the range 0.008-0.038, which is below or close to the ratio where the multiple scattering becomes negligible (I90vh/I90vv >0.03).23 Thus, it can be concluded that the results from the present systems are not influenced by multiple scattering. It is important to note that the multiple scattering components are well separated in time from the dominant mode and do not interfere in the evaluation of the relaxation rates of the latter. Unfortunately, the slow modes are not always well separated from the dominant ones, which, in turn, may cause some scattering of the results. Intermediate Mode. After a period of almost 180 days, the relaxation time distributions changed in that intermediate modes with a position between the multiple scattering and the dominant modes started to appear much more frequently. Typical distributions are presented in Figure 4a. Such intermediate modes were only occasionally observed earlier, that is, before the 180th day. For comparison, up to the 180th day the frequency at which the intermediate modes appeared was less than 12% compared to 60% afterward. The intermediate modes correspond to hydrodynamic radii typically between 10 and 50 nm. The positions of the intermediate modes and, hence, the dimensions of the particles associated with them varied with time in an almost periodic manner. Examples of the variation of the relaxation rate, Γ, of the intermediate modes with time are presented in Figure 4b. As is seen, Γ decreases rapidly (corresponding to an increase of particle size), then the (24) Rangelov, S.; Almgren, M.; Tsvetanov, Ch.; Edwards, K. Macromolecules 2002, 35, 4770. (25) Melnichenko, Y.; Brown, W.; Rangelov, S.; Wignall, G.; Stamm, M. Phys. Lett. A 2000, 268, 186.


Langmuir, Vol. 20, No. 16, 2004 6615

Figure 5. Evolution on a logarithmic time scale of the apparent particle radius measured at an angle of 90° for selected samples with 30:70 CTAB to SOS, 3.0 wt % (open circles); 2.0 wt % (filled triangles); 0.5 wt % (filled circles); and 2.0 wt % diluted to 0.5 wt % (filled squares).

Figure 4. Examples of relaxation time distributions (a) and variations of the relaxation rate of the intermediate modes (b), measured at an angle of 90° for dispersions of CTAB/SOS. (a) Mixing ratio 30:70, 2.5 wt %, after 445 days; (b) 30:70 CTAB to SOS, 2.0 wt %, day 513. The lines through the data points were drawn to guide the eye.

intermediate modes disappear, apparently merging into the dominant modes, and for a short period of time the typical trimodal (i.e., multiple scattering, dominant, and slow modes) relaxation time distribution is manifested. Then a new intermediate mode with a relaxation rate slightly lower than that of the multiple scattering mode appears, tends to slow with time, and finally merges into the dominant mode. Interestingly, for all of the samples the observed periodicity was found to be in the range of 120-200 min and is considerably shorter than the “fast” growth of the initially formed particles (see the following). Dominant Mode. Large particles, vesicles, were found on the first observations after preparation of the samples. The contribution from the vesicles to the overall scattered light intensity dominates under the whole aging period. The apparent particle radii, determined from DLS, changed upon aging of the samples. A considerable increase of the dimensions was observed during the first 30 days. After that, the particles seemed to grow much slower. The impression of two distinct stages disappears in a semilogarithmic plot, however, as shown for some samples in Figure 5. A monotonic increase in size with

Figure 6. Variation of the vesicle radius with the total surfactant concentration at a 30:70 CTAB-to-SOS mixing ratio after incubation for 49 (filled triangles) and 304 (open triangles) days.

time is clearly indicated. Partly, the scatter in the results is due to the presence of the additional intermediate modes which, as mentioned previously, might influence the position of the dominant mode. Vesicle close packing has been suggested to set the limit of the vesicle phase at high surfactant concentrations.1 A possible consequence might be that a growing vesicle population reaches a final size where further growth is hindered by the repulsive interactions between the vesicles. If this were the case, one would expect inverse proportionality between the vesicle size and the surfactant concentration. This is not what we observed. Figure 6 shows the variations of the vesicle size (after 49 and 304 days) with the total surfactant concentration at a CTABto-SOS ratio of 30:70. As is seen, the dependence is far from inversely proportional; instead. the observed size increased up to a maximum at 2 wt % after both time periods. Evidently, the size of the vesicles is not set by the close-packing condition. The concentration dependence of the vesicle size displayed in Figure 6 offers a possibility to test the hypothesis that the vesicles have reached an equilibrium size distribution. After dilution of an aged concentrated sample, the vesicles would evolve to the distribution characteristic of the equilibrium state of the diluted sample. Such an experiment was performed with a part of the sample 30:

6616



Figure 7. Slope (filled squares) and radius after 1 h (open circles) according to eq 3 from the semilogarithmic growth curves as in Figure 5. (a) Varying compositions at a total surfactant concentration of 2.0 wt %. (b) Varying total surfactant concentrations at a CTAB/SOS ratio of 30:70. The straight lines are linear correlations; the broken line is a guide to the eyes.

70, 2.0 wt %, which after 49 days was diluted to 0.5 wt %. The results are presented in Figure 5. Not only did the radius not decrease after dilution but also, on the contrary, it was found to slowly increase with time. Thus, samples of the same composition and concentration develop distinctively different particle dimensions as an effect of the formation path, and we can conclude that the size distributions are not in equilibrium. From the plots of Rh of the dominant mode on a logarithmic time scale for all samples, as exemplified in Figure 5, the slope k and Rh0 values according to eq 4 were determined.

Rh ) Rh0 + k log(t/1h)

(4)

The values from the series with CTAB/SOS ) 30:70 are presented in Figure 7b and those from a total concentration of 2.0 wt % at different compositions in Figure 7a. The growth rate increases with the total surfactant concentration and with decreasing fraction of CTAB in the mixtures. CryoTEM. The particle morphology was probed with cryoTEM. The micrographs were taken about 60 days after the preparation of the samples, that is, relatively stable dimensions had been attained. Representative micrographs are shown in Figure 8. In general, the cryoTEM results corroborate the DLS investigation and reveal some additional aspects such as the presence in some areas of

open bilayer structures as well as more complex morphologies. Figure 8a-c shows images from samples with a CTAB/SOS ratio of 30:70. At low concentrations, from 0.5 to 2.0 wt %, the images are dominated by spherical unilamellar vesicles with dimensions that correspond well to the dimensions calculated from the dominant modes in DLS. It should be emphasized that micelles, which are usually observable as small dots in the background, were not found. This is in good agreement with the conclusion based on the analysis of the DLS data that the fast mode is not diffusive and most probably derives from the multiple scattering. Occasionally, deformed vesicles were observed, as in Figure 8b from 1.3 wt %, otherwise samples with the same composition but different concentrations mainly differed in the number of vesicles found. At the highest concentrations, however, at 2.5 and 3.0 wt %, only few vesicles were found in the micrographs, which instead showed the presence of open bilayer fragments of varying size and shape (Figure 8c). Figure 8d-f presents micrographs from a series of samples with a constant surfactant concentration of 2.0 wt % and compositions varying from 2:8 to 37:63 CTAB/ SOS. At the highest SOS excess, 15:85, the only large objects identified in the micrographs were a few rodlike micelles (not shown). At 20:80 and 30:70 CTAB/SOS, mostly well-shaped vesicles were found, similar to those in Figure 8a,d, and also some vesicles with openings. At the highest SOS concentration, also a few bilayer fragments were present (Figure 8d). At CTAB/SOS ) 33:67 and 37:63, the vesicles were polydisperse in size: large vesicles with mean dimensions that are in agreement with the DLS results are abundant and coexist with a population of small (500 nm) vesicles (Figure 8f). Very large vesicles were only occasionally observed, but because they tend to escape entrapment in the thin films used in cryoTEM, they may be much more frequently present in the bulk solution. In these samples, more complex structures were also found, such as those shown in Figure 8e. Another micrograph (not shown) presented vesicles with sparse perforations; some of which seemed to contain some sort of irregular bilayer structure. The slow modes that are almost invariably present in the relaxation time distribution spectra apparently stem from very large vesicles. It is difficult to capture such large particles in the cryoTEM investigations, as a result of both their low number density and the size limitations of the method. We cannot expect to find vesicles larger than those shown in Figure 8f. Discussion The morphology and stability of particles formed by mixing oppositely charged surfactants with much different chain lengths were explored by DLS and cryoTEM. As shown by others,1,2 mixing of surfactants with asymmetric tails favors an enlargement of the vesicle region. In the present study, the investigated concentration and composition ranges are located well within the vesicle lobe of the published phase diagram. Our observations, both visual and experimental, agree qualitatively with observations on similar systems published earlier.1,3,6 Shortly after the brief mixing of the surfactant stock solutions, large particles are formed, shown by cryoTEM to be mainly vesicles. Initially, the vesicles increased rapidly in size and seemed to reach stable dimensions after about 30 days. There was still a slow growth, however. Observations over long times indicated that the radius increased linearly with the logarithm of time, with a slope of 10-25 nm,


Langmuir, Vol. 20, No. 16, 2004 6617

Figure 8. CryoTEM micrographs of samples aged for about 60 days of different concentrations and compositions. (a) CTAB/SOS 30:70, 1.6 wt %; (b) 30:70, 1.3 wt %; (c) 30:70, 3.0 wt %; (d) 20:80, 2.0 wt %; (e) 33:67, 2.0 wt %; (f) 37:63, 2.0 wt %. Bar length 200 nm.

depending on concentration and composition (Figure 5). The growth occurred faster at a higher surfactant concentration and with a larger fraction of SOS in the surfactant mixture (Figure 7). It should be noted that the light scattering results also agree with the indication from earlier studies that micelles are not found within the vesicle lobes. The observations of a continued growth already show that the vesicles have not attained a true equilibrium state. As remarked above, this conclusion was further strengthened by the results shown in Figure 5, where different sizes were obtained for two solutions with the same composition and concentration prepared in different ways. The general impression from the cryoTEM investigations is that the dispersed CTAB/SOS systems, with the broad variation of particle size, shape, and morphology, are similar to systems of polar lipids and ionic surfactants. The vesicles and the open structures form easilys spontaneouslysas a result of the interfacial charge but are not at equilibrium because the size distribution depends on the history of the samples. Marques6 made a thorough study of the time evolution of spontaneous catanionic vesicles. His results are often claimed to show that these vesicle are equilibrium structures, but the fact

is that his study proves that they are not, as the following citation clearly shows: “...for thermodynamically stable aggregates, an identical size distribution should be obtained irrespective of sample history. However, it is likely that kinetic factors induce the observed differences in this particular catanionic mixture.” Consequently, apart from the very facile formation of the vesicles, there is nothing extraordinary with the spontaneous vesicles, and there is little reason to regard the so-called “vesicle phase” as anything but a region in the phase diagram where vesiculation is facile and the dispersion stable. The results from the cryoTEM study also show that open bilayer structures were more prevalent than vesicles in some of the most concentrated samples. The fact that very few vesicles were found at the highest total surfactant concentrations of the 30:70 composition is particularly significant (Figure 8c). The presence of bilayer disk structures explains the decrease of the hydrodynamic radius indicated by the DLS measurements at concentrations above 2 wt %. There is no reason to consider them as artifacts from the sample preparation. Moreover, similar structures have been observed in many other systems.32,33 At lower surfactant concentrations, adsorption may change the composition substantially. Close to the limit

6618


of the vesicle region at low surfactant concentrations, it is possible that the composition in the thin vitrified film has changed from that in the bulk solution, because of adsorption of the surfactants on the polymer support or at the liquid interface. One could suspect that the cationic surfactant is adsorbed to a larger extent than the anionic surfactant because it is more hydrophobic and because interfaces usually acquire a negative charge. The fact that no vesicles were observed by cryoTEM in the sample with the largest SOS excess, contrary to the DLS results, may simply be due to that the composition in the vitrified film has changed so that it corresponds to a point in the phase diagram outside the vesicle lobe. For the most concentrated of the 30:70 samples, however, such an explanation is not possible. It seems more likely that close packing sets a limit to the size of closed vesicles at these concentrations and that open structures are favored. Equilibration of the vesicle size distribution can occur along two main routes: either by an exchange of monomers through the aqueous solution or by fission/fusion processes. The first mechanism would leave the number of vesicles unchanged; unlike micelles, vesicles cannot be assumed to form or disintegrate by only stepwise association of monomers: a large number of monomers are required in the smallest conceivable vesicle. The exchange process is related to Ostwald ripening and has been considered by Olsson and Wennerstro¨m.26 They pointed out that for onecomponent bilayer vesicles, that is, without a spontaneous curvature, the curvature energy is independent of vesicle size within the usual harmonic approximation. There is no driving force for a change of size, therefore, unless curvature effects of higher order or thermal undulations of the bilayer are important. In any case, for true equilibrium vesicles a mechanism such as fusion/fission that permits a change of the number of vesicles must be operative. This is the situation considered by Jung et al.8 Because they were concerned with anionic/cationic surfactant vesicles, the compositions of the inner and outer monolayers of the bilayer could be different and give rise to a spontaneous curvature of the bilayer. One would expect the component in excess to be mainly present at the outside because the excess charge will give a larger effective area per headgroup that can be accommodated on the outside better than on the inside. In this case, the harmonic expansion of the curvature energy truncated in the usual way at the harmonic term gives rise to a driving force for a change toward the size corresponding to the spontaneous curvature. In fact, even with a spontaneous curvature of zero there will be an optimum mean size because a growth of the vesicles also implies a decrease in the number of vesicles and, thus, a decrease in the total curvature energy. The curvature energy, thus, favors large vesicles, which is counteracted by the entropy of mixing of the vesicles. For asymmetric bilayers with a spontaneous curvature, the vesicles will be distributed around the spontaneous radius of curvature, with the width of the distribution determined by the spherical bending modulus: a modulus close to the thermal energy gives a broad distribution, whereas a bilayer with a large modulus forms vesicles with radii more narrowly distributed around the spontaneous radius. According to the theoretical work mentioned, there should be some preference for a certain vesicle size, but the free-energy minimum would be rather shallow in most cases and the thermodynamic driving force for changes (26) Olsson, U.; Wennerstro¨m, H. J. Phys. Chem. B 2002, 106, 51355138.


toward that size very weak, whatever the mechanism. It was observed in a system where CTAB was added to small EPC vesicles in 100 mM NaCl that the vesicles grew strongly (but slow enough for cryoTEM investigations) after addition of the surfactant, with open bilayers as intermediates. A mechanism similar to the proposals of Lasic27 and Fromhertz28 was suggested. Openings are created in the original small vesicles, with the rims of the holes stabilized by the surfactant. The holes may widen, transforming the vesicles into bilayer disks. These open bilayer structures may grow by merging and close again as they become large enough, that is, when the gain in free energy from removing the hydrophobic rim by closure exceeds the increase of the curvature energy of the vesicle compared to the open bilayer.29 A similar mechanism was suggested by Hatton et al.5,30 to be operative in the early stages of the formation of vesicles in CTAB/SOS after mixing micellar solutions of the component surfactants. It is tempting to assume that the slow growth follows the same type of process as the fast initial processes that were studied and discussed by Hatton et al.5,30 There is an abundance of open bilayers or bilayer flakes in many of the micrographs. It is conceivable, therefore, that growth of the vesicles can occur via merging of bilayer flakes. The rate of the growth process would be slowed strongly as the size of the vesicles becomes larger compared to the available free space; open bilayer structures would then be favored, as observed. Over the time period investigated the increase of the hydrodynamic radius appeared to be close to linear on a logarithmic time scale; we have no kinetic model that explains this behavior, and it is quite possible that the trend would not continue indefinitely. The rate of the growth, as given from the slopes of the semilogarithmic plots in Figure 5, was found to increase both with the concentration of surfactant (Figure 7) and, at a constant total surfactant concentration, with the fraction of SOS. The latter was not expected because an increase of SOS would increase the net charge of the bilayers and make them more repulsive. A very puzzling observation was the periodic emergence and decline of intermediate modes that started to appear frequently after half a year (Figure 6). True periodicity of the dynamics cannot be expected in the breakdown and build up of vesicles. It would require both amplification and synchronization and seems very unlikely in these systems. The roots of the periodicity must, therefore, be sought elsewhere. The observed turbid wisps may be important in this respect; the slow movement of these veils of scattering material in the sample may well lead to a periodicity on the order of hours. The wisps are elusive and very easily dispersed. We have not been able to study them in further detail. Acknowledgment. This work was supported by a grant from the Swedish Research Council. LA049211Y (27) Lasic, D. D. Biochim. Biophys. Acta 1982, 692, 149; J. Theor. Biol. 1987, 124, 35. (28) Fromhertz, P. Chem. Phys. Lett. 1983, 94, 259. (29) Edwards, K.; Almgren, M.; Gustafsson, J.; Karlsson, G. J. Colloid Interface Sci. 1993, 161, 299. (30) Xia, Y.; Goldmints, I.; Johnson, P. W.; Hatton, A. T.; Bose, A. Langmuir 2002, 18, 3822. (31) Mukerjee, P.; Mysels, K. J. Critical Micelle Concentrations of Aqueous Surfactant Systems. Natl. Stand. Ref. Data Ser. (U.S., Natl. Bur. Stand.) 1971, 36. (32) Almgren, M.; Edwards, K.; Karlsson, G. Colloids Surf., A 2000, 174, 3. (33) Edwards, K.; Johnsson, M.; Karlsson, G.; Silvander, M. Biophys. J. 1997, 73, 258.

Spontaneously Formed Nonequilibrium Vesicles of

Recommend Documents