7214
Langmuir 2005, 21, 7214-7219
Cylindrical Micelles of r-Fluorocarbon-ω-hydrocarbon End-Capped Poly(N-acylethylene Imine)s Stephan Kubowicz,† Andreas F. Thu¨nemann,*,‡ Ralf Weberskirch,§ and Helmuth Mo¨hwald† Max Planck Institute of Colloids and Interfaces, MPI KGF Golm 14424 D-Potsdam, Germany, Federal Institute for Materials Sciences and Testing, Richard Willsta¨ tter Strasse 11, 12489 Berlin, Germany, and Lehrstuhl fu¨ r Makromolekulare Stoffe, Technische Universita¨ t Mu¨ nchen, 85747 Garching, Germany Received April 13, 2005. In Final Form: June 3, 2005 Micelles of ABC block copolymers with varying degrees of polymerization of the B block (n) and constant lengths of the A and C blocks were investigated by small-angle X-ray scattering (SAXS), analytical ultracentrifugation (AUC), surface tension measurements, and isothermal titration calorimetry. The copolymers consisting of hydrophilic poly(N-acylethylene imine)s, end-capped with a hydrophobic fluorocarbon and a hydrocarbon block, are polymeric surfactants (γ ) 35 mN/m). They form cylindrical micelles with radii of 3.0 nm (n ) 35), 3.8 nm (n ) 57), and 4.0 nm (n ) 72). Their lengths are about 20 nm. The micelles can be doped with 1,4-diiodoperfluorobutane for the polymers with n ) 57 and 72 but not for n ) 35. We assume that the doped micelles form distinct fluorocarbon domains, which are able to incorporate selectively the fluorocarbon dopant. The work presented here is a contribution to the development of multicompartment micelles.
Introduction The concept of multicompartment micelles that are able to mimic basic properties of natural systems such as serum albumins, which are capable of transporting poorly watersoluble compounds in the blood by their selective uptake and release, has been first proposed by H. Ringsdorf1 and is an intriguing example for the use of bottom-up strategies in nanotechnology. Applications in medicine, pharmacy, biotechnology, etc. seem to be possible, but the preparation and control of stable multicompartment micellar systems is still at the very beginning. Even less is known about the properties of the existing examples. In the past years, various approaches have been presented by Sta¨hler et al.,2 Weberskirch et al.,33 and Laschewsky et al.44 to realize it experimentally. All approaches are based on the mutual incompatibility of fluorocarbon and hydrocarbon chains within a watersoluble polymer and were reviewed by Laschewsky.1 More recently, Lodge et. al. presented a cryo-TEM investigation of micelles formed by a mixed-arm star block terpolymer in water.5 They were the first who visualized clearly two separated compartments in a single micelle. Darker appearing regions of the TEM pictures were assigned to fluorocarbon-rich compartments and brighter ones to hydrocarbon-rich compartments. In the literature, there are several studies in which micelles of fluorocarbon/hydrocarbon hybrid surfactants * To whom correspondence should be addressed. E-mail:
[email protected]. † Max Planck Institute of Colloids and Interfaces. ‡ Federal Institute for Materials Sciences and Testing. § Technische Universita ¨ t Mu¨nchen. (1) Laschewsky, A. Curr. Opin. Colloid Interface Sci. 2003, 8, 274281. (2) Stahler, K.; Selb, J.; Candau, F. Langmuir 1999, 15, 7565-7576. (3) Weberskirch, R.; Preuschen, J.; Spiess, H.; Nuyken, O. Macromol. Chem. Phys. 2000, 201, 995-1007. (4) Kotzev, A.; Laschewsky, A.; Adriaensens, P.; Gelan, J. Macromolecules 2002, 35, 1091-1101. (5) Li, Z.; Kesselman, E.; Talmon, Y.; Hillmyer, M.; Lodge, T. Science 2004, 306, 98-101.
are utilized to solubilize oils and fluorocarbons simultaneously. It has been demonstrated that the size of these micelles is larger than the size of nondoped globular micelles.6 In this study, we concentrate on three different telechelic polymers based on poly(N-acylethylene imine) with varying degrees of polymerization (n) with n ) 35 (P1), 57 (P2), and 72 (P3). Each polymer is end-capped with a fluorocarbon segment [CF3(CF2)7CH2CH2-] (A blocks) and a hydrocarbon segment [CH3(CH2)15-] (C blocks) (cf. Figure 1). Whereas, in the first publication dealing with this polymer system, the main focus was to study phase separation of the fluorophilic and lipophilic domain during the self-association process in water by a combination of 19 F NMR and pyrene fluorescence studies, the structure of the micellar assemblies at a lower polymer concentration between 1 and 10 g/L remained unclear.3 Moreover, with the idea of using these polymeric aggregates as a transport vehicle in mind, we studied herein their ability to solubilize a fluorophilic dopant in aqueous media. Materials and Methods Polymers. A detailed description of the polymer synthesis and the molecular characterization is given elsewhere.5 In short, 2-methyl-2-oxazoline was polymerized by initiation with trifluoromethanesulfonic acid 1-ethylperfluorodecyl ester and terminated with N-hexadecylpiperazine. The number averaged molecular weights of P1, P2, and P3 were Mn ) 3700, 5600, and 6900 g/mol, respectively, as determined by 1H NMR endgroup analysis. The degrees of polymerization were n ) 35 (P1), 57 (P2), and 72 (P3) (Mw/Mn values were 1.20, 1.10, and 1.40, respectively, as determined by GPC in chloroform with polystyrene standards). The intrinsic viscosities of the polymers as determined from Huggins’ plots (not shown) were 5.3 mL g-1 (P1), 9.4 mL g-1 (P2), and 12.3 mL g-1 (P3). The corresponding Huggins’ constants were kH ) 7.1 (P1), 9.3 (P2), and 0.3 (P3). Doping of Micelles. The polymer P2 (68.1 mg, 1.21 × 10-5 mol) was dissolved in 10 mL of water. Then, 10 mL acetone was (6) Matsuoka, K.; Moroi, Y. Curr. Opin. Colloid Interface Sci. 2003, 8, 227-235.
10.1021/la050987o CCC: $30.25 © 2005 American Chemical Society Published on Web 07/07/2005
Cylindrical Micelles
Figure 1. Chemical structure of the R-fluorocarbon-ωhydrocarbon poly(N-acylethylene imine). The degrees of polymerization are n ) 35 (P1), 57 (P2), and 72 (P3). The lower part of the figure displays a simplified sketch of the aggregation of the polymer chains to cylindrical micelles with a hydrophilic shell (light brown) and a core with fluorocarbon-rich domains (red) and hydrocarbon-rich domains (blue).
Langmuir, Vol. 21, No. 16, 2005 7215 stirring rate of 310 rpm. The measurements were performed at a constant temperature of 20 °C. Small aliquots of the sample (typically 10 µL) were successively injected into the solution of the working cell. The first injection was usually set to a volume of 2 µL. Because of possible dilution during the equilibration time preceding the measurement, the first injection was ignored in the analysis of the data. Each injection produced a characteristic peak in the heat flow (in J/s) because of released or absorbed heat. In the analysis, a baseline was subsequently subtracted from the data, which corresponded to the signal between consecutive injections when no change in the heat flow was detected. Integrating each of the peaks provides the heat per injection. The data analysis was performed using the Origin software provided by MicroCal. The surface tension was determined using the pendant drop method.8 A profile analysis tensiometer PAT1 from SINTECH was operated at a temperature of 20 °C. The shape of the drop and its surface tension γ are correlated by the Gauss-Laplace equation
(
γ added, and the solution was treated with ultrasound for 5 min in an ultrasound bath. A stock solution of 1,4-diiodoperfluorobutane was prepared by dissolving 1 mL (1.13 × 10-4 mol) in 10 mL of acetone. An amount of 1 mL of the stock solution, containing 1.13 × 10-5 mol of 1,4-diiodoperfluorobutane, was added to the solution of P2. This mixture was stirred gently and heated to 50 °C for 12 h to evaporate the acetone to allow micelle formation and to load them simultaneously with 1,4-diiodoperfluorobutane. Evaporated water was replaced, and a clear micellar solution was obtained. The same procedure was carried out with P1 and P3. We observed also a clear micellar solution in the case of P3 (60.6 mg, 8.8 × 10-6 mol) after mixing with 8.82 × 10-6 mol of 1,4-diiodoperfluorobutane (dissolved in 0.663 mL of stock solution). In contrast, P1 does not form a clear solution after evaporating the acetone. The P1 and 1,4-perfluorobutane form clearly separate phases. Methods. The small-angle X-ray scattering (SAXS) measurements were performed with an Anton Paar HR-PHK pinhole camera, which was equipped with a xenon-filled two-dimensional detector (512 × 512 pixel, ∆θ ∼ 0.02°), at a temperature of 20 °C. The sample-detector distance was 610 mm. The scattering vector is defined in terms of the scattering angle θ and the wavelength λ of the radiation (Cu KR ) 0.154 nm); thus, q ) 4π/λ sin(θ/2). The density measurements were performed with a density meter (DMA 60/602, Paar, Austria). It is based on a vibrating mass whose resonance frequency is influenced by the density of the polymer solution (oscillating U-tube principle).7 The correctness as given for the density of water was ( 0.5 × 10-6 g/cm3, and the temperature was held constant at 20 ( 0.01 °C. Analytical ultracentrifugation was carried out using a Beckman-Coulter Optima XL-I ultracentrifuge (Beckman Coulter, Palo Alto, CA) at 20 ((0.2) °C and 50 000 min-1 (corresponding to an acceleration of 180000g) for sedimentation velocity experiments (sample cell was a 12 mm double sector cell of carbonfilled Epon). Detection of the micelles was carried out by applying UV-vis absorption detection and simultaneously using a Rayleigh interference optics. The measured radial positions of the sedimenting boundary were transformed into a sedimentation coefficient using s ) ln(ri/rm)/(ω2t), where ω is the angular velocity, r is the radial distance from the center of rotation with the indices i ) sedimenting boundary and m ) meniscus, and s is the sedimentation coefficient. The isothermal titration calorimetry (ITC) measurements were carried out with a VP-ITC microcalorimeter from MicroCal (Northhampton, MA). Two identical spherical cells, a reference cell and a sample cell, both with a volume of 1.442 mL, were enclosed in an adiabatic jacket. The working cell was filled with the sample solution, and the reference cell was filled with the solvent used to prepare the sample solution. The sample was injected stepwise into the working cell with a syringe of a total volume of 288 µL. The sample cell was constantly stirred at a (7) Kratky, O.; Leopold, H.; Stabinger, H. Z. Angew. Phys. 1969, 27.
)
1 1 + ) ∆P0 + ∆Fgz R1 R2
where R1 and R2 are the main radii of the drop, ∆P0 is the pressure difference at the apex, ∆F is the difference of the density between the drop and its surroundings, g is 9.81 m s-2, and z is the coordinate in the vertical direction. Viscosity measurements were performed using an Ubbelohde viscometer with automatic dilution (Viscoboy 2; Lauda) at 20 °C. The dynamic light scattering measurements were carried out using a Malvern Instruments particle sizer (HPPS-ET 5002) (Malvern Instruments, U.K.) equipped with a He-Ne laser (λ ) 632.8 nm). The scattering data were recorded at 25 °C in backscattering modus at a scattering angle of 2θ ) 173°. The aqueous sample solutions were placed into a squared 10 × 10 mm disposable polystyrene cuvette. Prior to measurement, the sample was filtered with a 0.45 µm Millipore syringe filter to clear it from dust particles.
Results and Discussion Micellar Structure. We know that P1, P2, and P3 are micelle-forming polymers, as expected from their amphiphilic ABC structure with a water-soluble B block surrounded by water-insoluble A and C blocks.9 However, the structures of their micelles are unknown. Therefore, we measured the SAXS of their aqueous solutions first when the micelles were empty and second after doping them with 1,4-diiodoperfluorobutane as a model compound for fluorophilic dopants. Typical SAXS curves are shown in Figure 2 (O and 4). The SAXS curves were analyzed by fitting them with scattering functions of cylindrical and cylindrical core-shell micelles applying scattering functions described by Fo¨rster et al., which consider the polydispersity of micelles.10-12 Briefly, the scattering intensity, I(q), of an isotropic solution with a low micelle density is equal to NpK2P(q), where Np is the number of particles, P(q) is the form factor, and K is the contrast factor. The average form factor P(q) of micelles with different radii, R, displaying a distribution of radii f(R) is
P(q) )
∫0∞ F(q,R)2f(R) dR
(8) Andreas, J. M.; Hauser, E. A.; Tucker, W. B. J. Phys. Chem. 1938, 42, 1001-1019. (9) Nuyken, O.; Weberskirch, R.; Bortenschlager, M.; Schonfelder, D. Macromol. Symp. 2004, 215, 215-229. (10) Forster, S.; Timmann, A.; Konrad, M.; Schellbach, C.; Meyer, A.; Funari, S.; Mulvaney, P.; Knott, R. J. Phys. Chem. B 2005, 109, 1347-1360. (11) Forster, S.; Hermsdorf, N.; Bottcher, C.; Lindner, P. Macromolecules 2002, 35, 4096-4105. (12) Forster, S.; Burger, C. Macromolecules 1998, 31, 879-891.
7216
Langmuir, Vol. 21, No. 16, 2005
Kubowicz et al.
Figure 2. SAXS curves of empty micelles P2 (O) and doped with 1,4-diiodofluorobutane (4) and best-fitted curves (s). The scattering curve of the empty micelles is in agreement with cylindrical micelles (R ) 3.8 nm, σR ) 0.8 nm, and L g 20 nm). The scattering of the doped P2 is fitted by a cylindrical coreshell micelle (Rm ) 5.0 nm, Rc ) 4.5 nm, σR ) 1.8 nm, and L g 20 nm).
The polydispersity of the micellar radii was calculated in terms of the Schulz-Zimm distribution
f(R) )
(
)
Z+1 R h
(
)
(Z + 1)R RZ exp R h Γ(Z + 1)
Z+1
where R h is the mean of the distribution and Z determines its width. Γ(x) is the Γ function. The root-mean-square deviation from the mean radius is given by σR ) R h (Z + 1)-1/2.13 The shape of the best-fitted curves for the scattering of the nondoped micelles is in agreement with the form factor of homogeneous cylinders. An example of P2 is shown in Figure 2. For simplicity, the form factor of the cylinders with lengths L was factorized
P(L,R,q) ) F2(L,R,q) ) F 2| (L,q)〈F 2⊥(R,q)〉 where F|(L,q) is the orientation averaged longitudinal contribution of the scattering amplitude and F⊥(R,q) is the cross-sectional part of the scattering amplitude.10 The square of the longitudinal scattering amplitude is given by
F 2| (L,q) )
(
)
sin(t) sin(qL/2) dt t qL/2
∫0qL
2 qL
2
The square of the cross-sectional scattering amplitude of a homogeneous cylinder, averaged for a Schulz-Zimm distribution of R, is given by
3Z+1Z+2 , ,2,3; - 4q2Rz2 〈F 2⊥(R,q)〉 ) 3F2 , 2 2 2
[
]
where 3F2 is a hypergeometric function, Z determines the width of the Schulz-Zimm distribution, and Rz ) R h /(z + 1). It must be mentioned that the polydispersity was considered only for the radii of the micelles and not for their length. This is justified because F 2⊥(R,q) dominates the shape of the scattering curve at higher q values, where we can measure the scattering intensity precisely. In contrast, F 2| (L,q) influences the shape at very small q, where we were limited in measuring the intensity because of the beam stop and parasitic scattering. When this model was applied, we found that in the absence of the dopant the scattering curve of P1 is in (13) Kotlarchyk, M.; Chen, S. J. Chem. Phys. 1983, 79, 2461-2469.
agreement with cylinders of R ) 3.0 nm and σR ) 0.5 nm. The radii of the micelles of P2 and P3 were determined to be 3.8 and 4.0 nm, respectively, i.e., they are significantly thicker. Their size distributions with σR ) 0.8 nm (P2) and σR ) 1.0 nm (P3) are also broader. The lengths of all micelles could be estimated to be at least about 20 nm. Because of the sampling theorem, this is about the maximum distance that can be determined at the smallest q values of the curves (0.15 nm-1). The next question was whether doping of the micelles is possible and whether it influences their size and shape. It was found that the micelles of P1 cannot be filled with 1,4-diiodoperfluorobutane, i.e., micellar solution and dopant form macroscopically separated phases. In contrast, the micelles of P2 and P3 can be doped as indicated by forming clear homogeneous micellar solutions. The scattering of the doped micelles was investigated with an equimolar polymer-dopant ratio. We used 1,4-diiodoperfluorbutane as dopant material for two reasons. First, it has a liner fluorinated alkyl chain with a length shorter than that of the fluorinated block of the polymer. This makes the incorporation of the dopant in fluorocarbon-rich domains probable, while its solubilization in hydrocarbon-rich domains is unlikely. Second, the iodine atoms of the dopant significantly enhance the contrast for SAXS experiments. This improves the quality of the scattering curves. The SAXS curves of the doped micelles are best-fitted with cylindrical core-shell micelles.10 An example is shown in Figure 2 (upper curve) and a sketch of the micelle in Figure 1. The doped micelles of P2 have a radius of Rm ) 5.0 nm and a core radius of Rc ) 4.5 nm. The values of P3 are Rm ) 6.0 nm and Rc ) 5.0 nm. The radii distributions are broad for both (σR ) 1.8 nm for P2 and σR ) 2.1 nm for P3). It was found that the cores display a higher electron density than the shells as expected from filling the micelles with the electron-rich dopant. The SAXS data do not allow determining whether the lengths of the micelles increase because of filling with the dopant or not. It can only be said that their length is not shorter than 20 nm. The fact that only the micelles of P2 and P3 can be doped may be interpreted as resulting from their structures. Their interior is probably bulky enough to form separated hydrocarbon and fluorocarbon domains as indicated by NMR investigations.3 The fluorocarbon domain of the micelle solute the fluorocarbon dopant resulting in a homogeneous micellar solution. In contrast, the micelles of P1 are very likely too thin (3.0 nm) to allow the formation of fluorocarbon domains that are large enough to incorporate the fluorocarbon dopant. Mixed solutions of P1 and the dopant form consequently two macroscopically separated phases. An important question is the loading capacity of the micelles of P2 and P3. The preparation of the doped micelles was carried out by solvent exchange from a water/acetone mixture to pure water. When using this procedure, it is problematic to determine the maximum loading capacity because of the formation of nonequilibrium structures. An overloading of the micelles with dopant is easily possible, forming micelles in a metastable state. A typical titration experiment, for example, to load the micelles with the dopant is impossible. Therefore, we do not give values for the limit of the loading capacity of the micelles. We performed dynamic light scattering measurements in addition to SAXS to determine the length of the micelles. The field correlation functions, g1(t), are shown in Figure 3, where it can be seen that the decay of g1(t) decreases in the line of P1-P3 as expected from increasing micellar sizes. The correlation functions were analyzed by using
Cylindrical Micelles
Langmuir, Vol. 21, No. 16, 2005 7217
Figure 3. Field-correlation functions of micellar solutions of P1-P3 determined by dynamic light scattering (4, O, and 0). The solid lines represent the cumulant fits of the data.
Figure 4. Field-correlation functions of micellar solutions of P2 before and after doping with 1,4-diiodoperfluorobutane. The solid lines represent the cumulant fits of the data.
the cumulant expansion ln(g1(t)) ) -Γt + (k/2)Γ2t2 - (k′/ 6)Γ3t3 + ..., with Γ ) Dapp(q)q2. We denote the apparent translation diffusion coefficient Dapp, as we measure at a fixed scattering vector of q2 ) 6.96 × 10-4 nm-2. The resultant Dapp values were (4.50 ( 0.04) × 10-11 m2 s-1 (P1), (4.25 ( 0.06) × 10-11 m2 s-1 (P2), and (2.29 ( 0.06) × 10-11 m2 s-1 (P3). The diffusion coefficient and frictional coefficient, C, are correlated by
D)
kBT C
where kB and T are the Boltzmann constant and the absolute temperature, respectively. For rodlike particles with length L and radius R, the frictional coefficient is given by14
C)
3πηL ln(L/(2R)) + γ
where η is the viscosity and γ is a function of L and R, which can be approximated for 1 < L/R < 10 by14
R R γ(L,R) ) 0.312 + 1.122 + 0.4 L L
2
()
We calculated the length of the micelles using the micellar radii already determined by SAXS, approximate D ≈ Deff and taking the viscosity of water η ) 0.887 mPa s at a temperature of 298 K. This yields L ) 18 nm (P1), 20 nm (P2), and 36 nm (P3). The decay of the correlation function of the doped micelles is slower as observed for the empty micelles (cf. Figure 4). The apparent diffusion coefficients are (1.2 ( 0.1) × 10-11 m2 s-1 (P2) and (1.3 ( 0.1) × 10-11 m2 s-1 (P3). In comparison, the diffusion coefficients of the loaded micelles are significantly lower than for the empty micelles. This was expected because the size of the micelles increases because of the incorporation of the dopant. The calculated lengths are about 40 nm (P2) and 60 nm (P3). Density of the Micelles. The next question arising was whether the micelles display not only different shapes but also different densities. Density measurements of the micellar solutions were carried out to answer this question. The densities of the micellar solutions in dependence of the polymer concentrations were determined using a densitometer based on the oscillating U-tube principle. Linear lines were fitted to the data points as shown in (14) Garcia de la Torre, J. Analytical Ultracentrifugation in Biochemistry and Polymer Science; The Roal Society of Chemistry: Cambride, U.K., 1992.
Figure 5. Density of aqueous solutions of P1, P2, and P3 as a function of their concentrations.
Figure 5. The slope is 1 - F0/F, were F0 and F are the densities of the solvent and the micelles, respectively. We determined the densities to be 1.263 ( 0.005 g/cm3 (P1), 1.261 ( 0.002 g/cm3 (P2), and 1.250 ( 0.006 g/cm3 (P3). Obviously, there is no significant difference in the densities of the polymer micelles, and hence, their differences in size do not have an influence on the densities, which seems to be not obvious. Tentatively, we can assume that B blocks form a hydrophilic outer shell of the micelle, which has the same density for all micelles. The hydrocarbon blocks should decrease the density, whereas the fluorocarbon blocks increase it. However, the increasing and decreasing effects cancel each other because of the fact that the number of hydrocarbon and fluorocarbon blocks is the same. The density of the doped micelles of P2 and P3 is slightly lower (1.230 ( 0.005 g/cm3) than that of the empty micelles. This was a priori not to be expected. We assume that the volume increase of the micelles because of the dopant is slightly larger than the corresponding weight increase and therefore decreases the overall density of the micelle. Aggregation Number. A reliable experimental method for the characterization of micelles is the analytical ultracentrifugation. Therefore, we performed sedimentation velocity measurements of the micellar solutions at different concentrations (1-10 g L-1). It was found that the micelles sediment well at a rotation rate of 50 000 min-1. An example is given in Figure 6. It shows the sedimentation curves of micelles of P3 at a concentration of 6 g L-1 at different times. The sedimentation constants were determined from plots of the logarithm of the radial positions of the sedimentation front as a function of ω2t (see inset of Figure 6). It was found that the sedimentation velocity of the micelles decreases slightly with an increasing concentration (not shown), which can be ascribed to
7218
Langmuir, Vol. 21, No. 16, 2005
Figure 6. Ultracentrifugation sedimentation curves of P3 at a concentration of 6 g L-1. The inset displays the inflection points of the sedimentation curves (O) and their linear fit (s) resulting in an averaged sedimentation coefficient of s ) 3.67 × 10-13 s.
an increasing viscosity of the solution. Therefore, we extrapolated the sedimentation coefficient to the limit of zero concentration (s0) using s ) s0/(1 - ksc), where ks is a constant and c is the polymer concentration.15 The calculated s0 values are 3.3 × 10-13 s (P1), 3.8 × 10-13 s (P2), and 4.0 × 10-13 s (P3). These values were used to estimate the apparent molecular weight, Mapp, of the micelles by15
Mapp )
s0NAkBT D(1 - vF0)
where ν is the specific volume of the micelles (known from the density measurement, ν ) 1/F) and F0 is the density of water. The calculated Mapp values of the micelles are 0.8 × 105 g/mol (P1), 1.1 × 105 g/mol (P2), and 2.1 × 105 g/mol (P3). When the molecular weight of the polymers (Mn) is taken into account, we estimated the aggregation numbers of the micelles to be about 23 (P1), 19 (P2), and 31 (P3). That means that the aggregation number of the micelles is relatively low and similar for the three polymers. This is probably resulting from the identical A and C blocks. It was observed further that the aggregation number increases in the direction of higher polymer concentrations. We determined s0 values of 4.0 × 10-13 s (P2) and 4.3 × 10-13 s (P3) for the doped micelles, which lead to apparent molecular weights of 4.3 × 105 g/mol (P2) and 4.4 × 105 g/mol (P3). The corresponding aggregation numbers were 104 and 73 for P2 and P3, when taking into account an equimolar amount of dopant. The aggregation numbers of the doped micelles are much larger than those for the empty ones, which is induced by solubilization of the dopant. Obviously, the dopant leads to a reduction of free energy resulting in an increase of the aggregation numbers and swelling of the micelles. Unfortunately, we were not able to determine the geometry of the micelles directly from the ultracentrifugation measurements by interpreting, for example, the Wales/van Holde ratios.16 This is, in principle, possible. However, it is based on a number of assumptions and valid only for selected polymers as proven first by Creeth and Knight.17 A more detailed ultracentrifugation mea(15) Schachmann, H. K. Ultracentrifugation in Biochemistry; Academic Press: New York, 1959. (16) Wales, M.; van Holde, K. E. J. Polym. Sci. 1954, 14, 81. (17) Creeth, J. M.; Knight, C. G. Biochim. Biophys. Acta 1965, 102, 549-558.
Kubowicz et al.
Figure 7. Surface tension of the equilibrium states of the polymer surfactant P2 [interpolation of limtf∞(γ(t))] in dependence of the concentrations of the polymers (0). The straight lines indicate the interpolations used to determine the cmc, which is indicated by arrow. The inset shows the time dependency of the surface tension of P2 (polymer concentration was 5 × 10-6 mol/L). A constant surface tension was observed after about 104 s.
surement to reveal the shape of the micelles is therefore beyond the scope of the current work. Critical Micelle Concentrations (cmc’s). Micelles are typically formed when the polymer concentration is higher than its cmc, and they dissolve below the cmc as long as they are not frozen micelles.18,19 Here, we measured the surface tension of the polymers as a function of the polymer concentration for determining their cmc’s. The pendant drop method8 was used because one can study the kinetic of the surface formation, whereas the surface itself stays untouched after the drop has formed. This is an important experimental detail for surface-active polymers; hence, they often need long times to reach constant surface-tension values. It can be seen in the inset of Figure 7, which shows the time dependence of the surface tension of P2 (polymer concentrations was 5 × 10-6 mol/L), that a constant surface tension was observed after about 104 s. The same time for reaching constant surface-tension values was observed for a different concentration and also for the two other polymers. These long times can only be understood if the kinetics is not limited by diffusion but by the time to form a micelle. We conclude that the time needed for forming equilibrium structures of constant surface tension at the water/air interface is in the order of about 3 h. This time is the same for P1-P3 within experimental accuracy. A series of surface-tension measurements was carried out for each polymer, and the surface tension of the equilibrium states was determined by interpolation of limtf∞(γ(t)). The surface tension of the equilibrium states of P2 in dependence of the concentrations of the polymers (0) is shown as an example in Figure 7. The straight lines indicate the interpolations used to determine the cmc, which is indicated by an arrow. It was expected that the cmc increases with the increasing length of the water-soluble B block. This expectation was confirmed by the cmc values, which are 1 × 10-5 mol/L (P1), 6 × 10-5 mol/L (P2), and 10 × 10-5 mol/L (P3). The surface tension of the polymer solutions above their cmc’s is about 35 mN/m. There seems to be a trend that the minimal surface tension is slightly lower when the length of the B block decreases (which is understandable because the area that the molecule occupies at the surface is (18) Forster, S.; Abetz, V.; Muller, A. Polyelectrolytes Defined Mol. Archit. II 2004, 166, 173-210. (19) Koutalas, G.; Pispas, S.; Hadjichristidis, N. Eur. Phys. J. E 2004, 15, 457-464.
Cylindrical Micelles
Langmuir, Vol. 21, No. 16, 2005 7219
Table 1. Molecular Characteristics of the Polymers and Dimension of the Cylindrical Micelles as Determined by SAXS and DLS before (Empty) and after Filling with 1,4-Diiodoperfluorobutane (Doped) dimension of cylindrical micelles doped
polymer
Mn (g mol-1)
number of N-acyl ethylene units
cmc (mol L-1)
cmc (g L-1)
empty
P1 P2 P3
3700 5600 6900
35 57 72
1 × 10-5 6 × 10-5 10 × 10-5
0.04 0.34 0.62
R ) 3.0 nm, L ≈ 18 nm R ) 3.8 nm, L ≈ 20 nm R ) 4.0 nm, L ≈ 36 nm
Figure 8. Experimental titration curve (inset) observed from injecting P3 (10 mL aliquots) of 0.29 mmol/L into water at 25 °C (baseline corrected). The integrated heat per injection of P1-P3 (normalized with respect to the injected number of moles of macromolecules) as a function of the total concentration of polymer in the sample cell. The approximated enthalpies, ∆H, for dissolution of the micelles are indicated by arrows.
smaller; thus, the packing density at the surface is higher, and therefore, the surface tension is lower), but the experimental accuracy is not sufficient to verify this. Nevertheless, we expected that the surface tensions of the polymers above their cms’s are about the same because the polymer possesses identical hydrophobic surfacetension-reducing blocks. The surface-tension measurements confirm that the SAXS and ultracentrifugation measurements were performed well above the cmc’s (c > 1 g/L). Thermodynamics of Micelle Formation. We were interested in the thermodynamics of the micelle formation and followed the heat flow during dilution of the micelles by ITC. A typical experimental titration curve obtained from dilution of a micellar solution of P3 is shown in the inset of Figure 8. The enthalpy curve displays the heat flow by consecutive injections of 10 mL aliquots of the micellar solution into water. For the first injections, the final concentration in the sample cell is below the cmc and we observe a sharp increase in the reaction enthalpy in the curves. Integration of the heat-flow peaks yields the heat of each injection shown in Figure 8 for P1-P3. Obviously, the demicellization is exothermic, and vice versa, the micelle formation is an endothermic process (∆H > 0). It can be seen that the micelle formation enthalpy at a temperature of 25 °C increases with an increasing length of the B block (∆HP1 < ∆HP2 < ∆HP3). It can be seen further that the enthalpograms are far from idealized forms (constant values of the heat at low and high concentrations).20 Because of the fact that no constant heat values at low concentrations are present, only lower limiting values of ∆HP1 ≈ 9 kJ/mol, ∆HP2 ≈ 11 kJ/mol, and ∆HP3 ≈ 19 kJ/mol can be determined. These values are close to the enthalpy of micellization of low molecular weight surfactants such as reported for alkyl sulfates,21
no doping possible (phase separation) Rm ) 5.0 nm, Rc ) 4.5 nm, L ≈ 40 nm Rm ) 6.0 nm, Rc ) 5.0 nm, L ≈ 60 nm
indicating that the alkyl chains dissolve first and dominate the process. The enthalpograms display constant heat values already at concentrations in the range of 1 × 10-5 mol/L (P1 and P2) to 3 × 10-5 mol/L (P2), which are much smaller than the cmc values determined by surface tension measurements (except P1, see Table 1). Tentatively, we interpret this as resulting from a noncomplete dissolution of the micelles. The times to reach constant surface tension values are much longer (ca. 104 s) than the time between two injections in the ITC experiments (2 × 102 s). Therefore, it seems to be plausible to assume that we do not measure the heat of the complete dissolution of the micelles, which is probably a three- or multistep process. For example, it is possible that the hydrocarbon domains dissolve first in a fast process, which produces the observed heat. This may be followed by a second process, the dissolution of the fluorocarbon domains, which is much slower, and the heat flow is also much lower; i.e., it cannot be detected by ITC. Such multistep dissolution processes can be accompanied by a shape transition, e.g., from cylindrical to spherical micelles as it has been reported for heptaethylenglycoltetradecyl ether (C14EO7) by Heerklotz et al.22 In particular, ionic surfactants have been found to show a three-step behavior forming spherical micelles at the cmc, which then grow cooperatively to rodlike micelles at the “second cmc” at higher concentrations.23 The enthalpy is also negative for the doped micelles and even larger than for the empty micelles, roughly by a factor of 2-3. This was expected because the presence of the dopant gives rise to an increase of the Gibbs free energy (more negative) of micellization of P1 and P2. This results in the decrease of their cmc’s. Moreover, the micellar structures are more organized forms. Conclusions We have shown that the shape of micelles formed by an amphiphilic ABC block copolymer is cylindrical and that the micelles can be filled with a fluorocarbon dopant when the hydrophilic middle block is long enough. This was interpreted as resulting from the necessity of the presence of a fluorocarbon compartment in the micelle. Constant surface tension values with minimum values of about 35 mN/m are observed after about 3 h. This long time was attributed to slow kinetics in forming an equilibrium polymer layer at the air/water interface. The micelle formation of the ABC polymers was found to be an entropically driven process. These novel structured nanosized objects may serve as models to simulate properties of biological structures such as transport proteins (albumins) and might be a base for advanced drug-delivery systems. Acknowledgment. The financial support of the Max Planck Society, the Fraunhofer Society, and the Federal Institute for Materials Science and Testing is gratefully acknowledged. LA050987O
(20) Blandamer, M. J.; Cullis, P. M.; Engberts, J. B. F. N. J. Chem. Soc., Faraday Trans. 1998, 94, 2261-2267. (21) Ropers, M. H.; Czichocki, G.; Brezesinski, G. J. Phys. Chem. B 2003, 107, 5281-5288.
(22) Heerklotz, H.; Tsamaloukas, A.; Kita-Tokarczyk, K.; Strunz, P.; Gutberlet, T. J. Am. Chem. Soc. 2004, 126, 16544-16552. (23) May, S.; Ben-Shaul, A. J. Phys. Chem. B 2001, 105, 630-640.
