Langmuir 1999, 15, 1059-1066
1059
Study of Structure Formation in Aqueous Solutions of Poly(ethylene oxide)-Poly(propylene oxide)-Poly(ethylene oxide) Block Copolymers by Measuring Rate Constants of the Thermal Cis-Trans Isomerization of an Azobenzene Dye and Self-Diffusion of Copolymer Molecules Kathrin Gille,† Helmut Knoll,*,† Frank Rittig,‡ Gerald Fleischer,‡ and Jo¨rg Ka¨rger‡ Wilhelm-Ostwald-Institut fu¨ r Physikalische und Theoretische Chemie, Fakulta¨ t fu¨ r Chemie und Mineralogie der Universita¨ t Leipzig, Linne´ strasse 2, D-04103 Leipzig, Germany, and Institut fu¨ r Experimentelle Physik I, Fakulta¨ t fu¨ r Physik und Geowissenschaften der Universita¨ t Leipzig, Linne´ strasse 5, D-04103 Leipzig, Germany Received August 4, 1998. In Final Form: October 29, 1998 In aqueous solutions of poly(ethylene oxide) (PEO)-poly(propylene oxide) (PPO)-PEO triblock copolymers (Pluronics P85, F88, and L64), structure formation (micellization) on increasing temperature was followed by determination of rate constants kiso of the thermal cis-trans isomerization of 4,4′-nitroanilinoazobenzene by means of flash photolysis in H2O and D2O. The kinetic solvent isotope effect kiso,H2O/kiso,D2O indicates that the azobenzene dye molecules are solubilized in a water-rich environment. From the nearly constant solvatochromic UV/vis absorption band maxima λmax of the dye, it is concluded that the S shape of the ln kiso vs 1/T curves is mainly due to microviscosity changes on micellization. Critical micelle temperature values derived are in satisfactory agreement with those from self-diffusion coefficients of the copolymer molecules dependent on temperature determined by means of pulsed field gradient nuclear magnetic resonance measurements. The self-diffusion experiments allow conclusions on the size of the diffusing particles in H2O and D2O and the influence of dye molecules on aggregation. The hydrodynamic radii of the diffusing species are larger in H2O than in D2O. The reason is seen in the stronger hydrogen bonds between EO units and D2O compared to those between EO units and H2O. On gelation of 25% (w/v) F88 in water at 31° C, the bulk viscosity increases sharply but the microviscosity around the dye molecules does not.
Introduction Poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) triblock copolymers behave like nonionic amphiphiles in aqueous solutions. Industrial products such as Pluronics (BASF) or Poloxamers (ICI) find widespread applications in their uses as wetting, thickening, emulsifying, coating, solubilizing, stabilizing, dispersing, lubricating, and foaming agents1 and in recent times also as drug delivery systems.2 There is a growing interest in fundamental research on aqueous solutions of such triblock copolymers with respect to their properties as amphiphiles and to comparison with the usual nonionic detergents such as CnEOm.3 Copolymers are dissolved in water as unimers at low temperature, where water is a good solvent for both poly(ethylene oxide) (PEO) and poly(propylene oxide) (PPO) blocks. With increasing temperatures the solvent properties of water become less favorable for PPO. Therefore, micelles are formed consisting of a less polar core of PPO
blocks which also contain some water and a water-swollen corona of PEO blocks, above a critical temperature (cmt) and/or a critical concentration (cmc). The PPO blocks are responsible for the entropy-driven micellization process which is accompanied by a desolvation of the PPO units. Whereas the length of the PPO block determines the size of the core and the temperature dependence of the micellization, the ratio between EO and PO monomeric units is responsible for the differently formed phases.4-7 Not only micellar but also more complicated e.g., lamellar and hexagonal, phases are observed in aqueous solutions of Pluronics.7-10 At high concentrations of micelles, on increasing temperature, a solid, glass-clear gel may be formed. On fluid-gel transition a dramatic change of bulk viscosity occurs.4,11,12 Different scattering techniques, various spectroscopic and other light-related methods, and thermodynamic and macroscopic as well as microscopic methods have been (4) Almgren, M.; Brown, W.; Hvidt, S. Colloid Polym. Sci. 1995, 273, 2.
* Corresponding author. † Fakulta ¨ t fu¨r Chemie und Mineralogie der Universita¨t Leipzig. ‡ Fakulta ¨ t fu¨r Physik und Geowissenschaften der Universita¨t Leipzig. (1) (a) Pluronic and Tetronic Surfactants; Technical Brochure; BASF Corp.: Parsippany, NJ, 1989. (b) Chu, B.; Zhou, Z. In Nonionic Surfactants, Polyoxyalkylene Block Copolymers; Surfactant Science Series Vol. 60; Nace, V. M., Ed.; Marcel Dekker: New York, 1996; p 67. (2) Kabanov, A. V.; Nazarova, I. R.; Astafieva, I. V.; Batrakova, E. V.; Alakhov, V. Yu.; Yaroslavov, A. A.; Kabanov V. A. Macromolecules 1995, 28, 2303. (3) Alexandridis, P.; Holzwarth, J. F.; Hatton, T. A. Macromolecules 1994, 27, 2414.
(5) Armstrong, J. K.; Parsonage, J.; Chowrhy, B.; Leharne, S.; Mitchell, J.; Beezer, A.; Lo¨hner, K.; Laggner, P. J. Phys. Chem. 1993, 97, 3904. (6) Hvidt, S. Colloids Surf., A 1995, 112, 201. (7) Wanka, G.; Hoffmann, H.; Ulbricht, W. Macromolecules 1994, 27, 4145. (8) Alexandridis, P.; Olsson, U.; Lindman, B. Langmuir 1998, 14, 2627. (9) Zhang, K.; Khan, A. Macromolecules 1995, 28, 3807. (10) Schille´n, K.; Brown, W.; Johnsen, R. M. Macromolecules 1994, 27, 4825. (11) Brown, W.; Schille´n, K.; Hvidt, S. J. Phys. Chem. 1992, 96, 6038. (12) Brown, W.; Schille´n, K.; Almgren, M.; Hvidt, S.; Bahadur, P. J. Phys. Chem. 1991, 95, 1850.
10.1021/la980968x CCC: $18.00 © 1999 American Chemical Society Published on Web 01/22/1999
1060 Langmuir, Vol. 15, No. 4, 1999
used in order to study the influence of temperature, polymer concentration, molecular weight, and length of the propylene oxide and ethylene oxide blocks on structure formation in water solutions of triblock copolymers; see, e.g., refs 3-18. To the best of our knowledge, reactivity parameters of a unimolecular electronic ground-state reaction dependent on structure formation of triblock copolymer/water solutions have not yet been studied. With respect to the use of triblock copolymer/water solutions as media for chemical reactions, the combined influence of temperature and structure formation in triblock copolymer/water solutions on reaction rates is of interest. A unimolecular isomerization reaction seems to be an appropriate model reaction. Very recently we prepared the unstable cis isomers of various azo dyes from the stable trans isomers by means of microsecond flash photolysis and studied the rate constants of the immediately following thermal cis-trans backisomerization in pure solvents and micellar solutions;19 see also ref 20. Besides the influence of polarity and hydrogen bonding which can be monitored by the solvatochromic absorption band maximum λmax of the trans isomer, the rate constants of 4,4′-nitroanilinoazobenzene were found to be the most sensitive against bulk viscosity. A rate-increasing effect due to the dynamics of hydrogen bond forming and breaking cannot be excluded.19,20 In aqueous solutions of nonionic surfactants, we observed a decrease of the rate constants kiso with increasing surfactant concentrations. As the UV/vis absorption band maxima λmax of the trans isomers changed only marginally, it was concluded that the decrease of the isomerization rate was mainly due to an increase of microviscosity in the water/surfactant interface layer. In this study we extended our investigations on rate constants of the thermal cis-trans isomerization of azo dyes in nonionic surfactant solutions by kinetic measurements with 4,4′-nitroanilinoazobenzene in water solutions of some Pluronics. From such rate measurements, cmt values can be derived, as the aggregation behavior of the Pluronics is reflected by an S-shaped curve of ln kiso vs 1/T. By means of pulsed field gradient (PFG) NMR,21-24 selfdiffusion coefficients of triblock copolymers which decrease on micellization as azo dye isomerization rate constants kiso do, can be measured with a “probe molecule free method”, see, e.g., refs 17 and 25. The self-diffusion coefficients are related to the macroscopic bulk viscosity and provide hydrodynamic radii and therewith the size of the diffusing unimers and micelles, respectively. The aim of the present study was to follow the micellization process (structure formation of the triblock copolymer molecules in water solutions) by measuring the changes of kiso values on increasing temperature and to compare the results with self-diffusion coefficients and (13) Malka, K.; Schlick, S. Macromolecules 1997, 30, 456. (14) Nivaggioli, T.; Tsao, B.; Alexandridis, P.; Hatton, T. A. Langmuir 1995, 11, 119. (15) Mortensen, K.; Pedersen, J. S. Macromolecules 1993, 26, 805. (16) Wanka, G.; Hoffmann, H.; Ulbricht, W. Colloid Polym. Sci. 1990, 268, 101. (17) Fleischer, G. J. Phys. Chem. 1993, 97, 517. (18) Chu, B. Langmuir 1995, 11, 414. (19) Knoll, H.; Gille, K. J. Inf. Recording 1998, 24, 203. (20) Shin, D.-M.; Schanze, K. S.; Whitten, D. G. J. Am. Chem. Soc. 1989, 111, 8497. (21) Stilbs, P. Prog. NMR Spectrosc. 1987, 19, 1. (22) Callaghan, P. T. Principles of Nuclear Magnetic Resonance Microscopy; Clarendon Press: Oxford, U.K., 1992. (23) So¨dermann O.; Stilbs, P. Prog. NMR Spectrosc. 1994, 26, 445. (24) Price, W. S. Concepts Magn. Reson. 1997, 9, 299; 1998, 10, 197. (25) Bahadur, P.; Li, P.; Almgren, M.; Brown, W. Langmuir 1992, 8, 1903.
Gille et al.
hydrodynamic radii determined by PFG NMR. D2O is the inherent solvent for most PFG NMR methods applied on triblock copolymer solutions, but the differences introduced by the solvent D2O in comparison to the natural solvent H2O were hardly considered. Therefore, the results of measurements under the same conditions in both H2O and D2O, by both kinetic and PFG NMR methods should be compared and the influence of dye probe molecules on the micellization process tested by PFG NMR measurements. Experimental Section Materials. 4,4′-Nitroanilinoazobenzene (azo; Aldrich) was dissolved in acetone and the solution decanted from the salt residue. 4,4′-(imidocarbonyl)bis(N,N-dimethylaniline) hydrochloride (auramine; Aldrich) was recrystallized from ethanol. Poly(propylene oxides) PPO 425 and PPO 2000 (M ≈ 425 and 2000 g mol-1, respectively; Aldrich) were used as received. The following Pluronic triblock copolymers PEO-PPO-PEO were used without further purification: P85 (lot WPAN-628B, EO26PO40EO26), F88 (lot WPDN616B, EO96PO39EO96) from BASF, Parsippany, NY, and L64 (PE 6400, lot 55-1400, EO13PO30EO13) from BASF AG, Ludwigshafen, Germany. These Pluronics very probably contain traces of diblock copolymers and other impurities. The copolymers were dissolved in Milli-Q water, making a transparent, homogeneous solution. D2O (99.7%; Laborchemie Berlin) was freshly distilled over KMnO4. All concentrations are given in percent (w/v, 25 °C) in this paper. Apparatus and Procedures. UV/Vis Spectroscopy. Spectra were recorded with a Lambda2 spectrometer (Perkin-Elmer; the experimental error of λmax was estimated to (2 nm). Stationary fluorescence intensities of auramine were determined with a Fluoromax-2 fluorometer (Spex), using an excitation wavelength of 410 nm and an emission wavelength of 500 nm. The samples were placed in Teflon-stoppered quartz cuvettes. For cis-trans isomerization, a homemade microsecond flash-photolysis apparatus (fwhm 25 µs, up to 180 J/flash) equipped with a digital storage oscilloscope PM3533 (Fluke) and 10 cm cuvettes (water jacket, bath thermostat (0.2 K) was used. The irradiation of the dye solutions (about 2 × 10-6 mol dm-3) by the flash resulted in a bleaching of the ππ* chargetransfer (CT) band because of the transformation of part of the stable trans form into the cis isomer. The reaction systems were completely reversible on repeated flashes. First-order rate constants of the isomerization kiso were calculated from the time-dependent absorbance At and the absorbance A0 at t ) 0, and A∞ for infinity time, by a least-squares fit according to eq 1 including absorbance data from at least 3 half-lives.
(At - A∞)/(A0 - A∞) ) exp(-kisot)
(1)
In those cases where the differences between fitted data according to eq 1 and experimental data showed systematic deviations, the experimental At curves were fitted to the biexponential eq 2, with the four fit parameters a1, a2,
(At - A∞)/(A0 - A∞) ) a1 exp(-kiso,1t) + a2 exp(-kiso,2t) (2) kiso,1, and kiso,2 and component 2 being slower. Absorbance data from 4 half-lives were included, and 0.9 > a1 > 0.6, 0.4 > a2 > 0.1 and a1 + a2 g 0.96 was obtained in these
Structure Formation in PEO-PPO-PEO Block Copolymers
cases. Minor contributions of a2 < 0.1 were neglected by this procedure. For repeated experiments with the same solution, the standard deviation of kiso was found to be less than (5%, and for experiments with independently prepared solutions, it was about (15%. PFG NMR. With PFG NMR, the thermal Brownian motion of the proton-bearing entities in the sample is monitored in a space region on the order of micrometers and a time interval between a few milliseconds up to about 1 s. The experimentally measured quantity, the spinecho attenuation due to the application of two field gradients separated by the diffusion time t, is given by
Sinc(q,t) ) exp(-q2tD)
(3)
in the narrow pulse approximation (δ/3 , t, which was always fulfilled at our experimental conditions). q ) γδg is a generalized scattering vector, with g denoting the magnitude of the field gradient pulse and δ its width. γ denotes the gyromagnetic ratio of the proton, the nucleus under study. PFG NMR is a generalized scattering experiment, and the spin-echo attenuation is equivalent to the intermediate incoherent scattering function Sinc(q,t).26,27 In eq 3, the existence of one kind of species in the sample is assumed, with the self-diffusion coefficent D characterizing its Brownian motion. PFG NMR, also used as Fourier transform PFG NMR, has been often applied to study the self-diffusivity in surfactant systems; for a review, see ref 23. Details of PFG NMR as used by us may be found in refs 17 and 26-28. We have used a home-built spectrometer FEGRIS 400 operating at a resonance frequency of 400 MHz. The stimulated echo pulse sequence π/2-τ-π/2-t′-π/2-τecho was used (π/2: radio-frequency-pulses), with the field gradient pulses applied after the first and third radiofrequency-pulses, respectively. In one experiment, δ was fixed (maximum value 1.85 ms) and g was incremented, with the maximum g being 25 T/m. The diffusion time t ) t′ + τ was typically 13 ms (τ ) 3 ms). The parameters δ and t′ were chosen such that at the maximum field gradient the echo was attenuated down to a few percent. The temperature was controlled within (0.5 K. The NMR samples were prepared by weighing the respective amount of polymer and water into NMR sample tubes of 7.2 mm o.d. and subsequently flame-sealing the tubes. H2O and D2O were used as solvents. In the case of H2O, the subtraction of the NMR signal of the water protons was easily possible since the water self-diffusion is about 1 order of magnitude faster than the polymer self-diffusion. The slight polydispersity of the Pluronics polymers leads to a small nonexponentiality of the echo attenuation in the unimer as well as the micellar state (a slight distribution of self-diffusivities). The measured echo attenuations were fitted with a stretched exponential:
Sinc(q,t) ) exp(-(q2tD)β)
(4)
The parameter β e 1 is a measure of the nonexponentiality. The smaller β is, the broader is the distribution of self-diffusivities. The β values in our experiments have always been larger than 0.9, indicating only a slight distribution of the unimer and micelle sizes. In the Results and Discussion section, the fitted D values are given. (26) Fleischer, G.; Fujara, F. NMR 1994, 30, 159. (27) Ka¨rger, J.; Pfeifer, H.; Heink, W. Adv. Magn. Reson. 1988, 12, 1. (28) Ka¨rger, J.; Fleischer, G. TRAC 1994, 13, 145.
Langmuir, Vol. 15, No. 4, 1999 1061
Viscosimetry. Viscosities were measured with a LS100 rheometer (Paar-Physica) and an Ubbelohde viscosimeter. Results and Discussion UV/Vis Spectroscopy. The azo dye molecules are insoluble in pure water, so that partition between water and micelles has not to be considered. Small amounts of the dye (≈2 × 10-6 mol dm3) can be solubilized in most aqueous Pluronic solutions below the cmc/cmt. Triblock copolymer molecules are assumed to be dissolved only as single molecules (unimers) under these conditions. Probably the dye molecules are loosely wrapped in the hydrated PPO and PEO blocks of unimers, or solubilization occurs in smaller aggregates formed from or induced by impurities. Aggregation induced by the azo dye probe molecules themselves will be discussed below. The solvatochromic ππ* CT absorption band maximum λmax of the trans form of the azo dye spans a range between 438 nm in heptane (kiso ) 1.4 × 10-3 s-1) and 484 nm in dimethylformamide (kiso ) 4.9 s-1) and between 461 nm in pure acetone (kiso ) 0.2 s-1) and 485 nm in acetone/ water (1:4, v/v; kiso ) 180 s-1), respectively.19 The λmax values of 483 ( 2 nm found in all Pluronic solutions with >1% in the temperature range of micellization are near the value in an acetone/water solution (1:4, v/v) and indicate a rather polar microenvironment. However, in micelles of Pluronics the dye molecules exist in a less polar microenvironment or are less oriented along the polarity gradient than in micelles of simple surfactants with PEO headgroups such as Triton X-100 or C12EO8, where λmax ≈ 493 nm was observed.19 For this extremely polar environment, a radial alignment of the dye can be assumed, where the nitro group is located in the PEO/ water interface. As the longest axis of the azo dye molecule (trans) is about 1.9 nm,29 comparable to the PEO corona thickness of 1.8 nm in P85 micelles,30 for radial alignment in the PEO corona of P85 micelles of λmax > 490 nm would be expected. From the λmax values it is therefore concluded that the dye molecules probably either are parallel to the PEO/water interface within the PEO corona or are radially aligned with the anilino substituent sticking into the PPO/ PEO interface. From a further experiment it follows that solubilization of the entire dye molecule in the PPO core is also unlikely. The dye shows λmax ) 476 nm in pure PPO 2000 which consists of 35 PO units on average, similar to the Pluronics used in this study. We attempted to match λmax ) 483 nm by addition of water to PPO 2000. At a water content larger than 3% with λmax ) 479 nm, the solution became turbid on further addition of water, in contrast to the transparent water solutions of Pluronics in this study. Isomerization Rate Measurements. S-Shaped curves were obtained in the Arrhenius plots of the rate constants kiso. The data for 2.3% F88 in H2O and D2O, respectively, are shown as an example in Figure 1. At lower temperatures, nearly constant or with temperatures weakly increasing, rate constants were measured. Then a temperature range was reached where the rate constants decreased sharply because of micellization on increasing temperature up to a second inflection point. Above this temperature Arrhenius-like behavior for isomerization in micelles again was observed. The monoexponential kinetic curves show that the dye molecules are essentially solubilized by one type of aggregate of F88 at any (29) Gas-phase AM1 calculations give 1.7 nm between the centers of the most distinct atoms of the dye (trans) and 1.5 nm for the cis isomer, respectively. Dietz, F., personal communication. (30) Mortensen, K.; Brown, W. Macromolecules 1993, 26, 4128.
1062 Langmuir, Vol. 15, No. 4, 1999
Figure 1. Arrhenius plots of isomerization rate constants kiso for 2.3% F88 in H2O and D2O. Dotted lines were drawn by hand to estimate the first inflection point as a measure for the cmt.
temperature. These are unimers below the temperature of the first inflection point, which aggregate to micelles within the temperature range between the first and second inflection points. Solubilization in unimers existing besides micelles seems unlikely in the case of F88. From the first inflection points at the lower temperature, the cmt values were derived for this and other samples with an uncertainty of (1 °C; cf. Table 1. In all cases kiso values were lower in D2O than in H2O. Some of the curves determined in D2O were slightly shifted to lower temperatures with lower cmt values than those in H2O. Therefore, a maximum of the kinetic solvent isotope effect on increasing temperature was observed, because of the beginning of micelle formation at lower temperature in D2O; see Table 2. In the case of 0.2% and 1.0% solutions of P85 and a 4.6% solution of L64, the biexponential eq 2 gave a better fit for the lower temperatures compared to eq 1. a2 of the slower component became smaller with increasing temperature. This result indicates that in favorable cases two different coexisting hosts with different microenvironments for the dye molecules can be detected by means of different rate constants, as shown in Figure 2 for a L64 solution. It was observed both in H2O and in D2O. This noticeable distribution of dye molecules between two different hosts vanishes upon increasing the temperature in favor of the faster component. Before the second inflection point was reached upon increasing the temperature, essentially all dye molecules were solubilized in one aggregate type, with a microenvironment giving rise to a rate between the former faster and the latter slower, respectively. The distribution of dye molecules between two different types of hosts comes probably from additional aggregates formed by impurities or by the most hydrophobic constituents of the polydisperse L64 sample. These additional aggregates were incorporated into the micelles when these were grown to some size. The biexponential behavior is in accord with very recent studies on L64, and the step in both the slower rate constants on increasing the temperature might be due to an independent micelle formation of an impurity.31 In estimating the cmt values of Table 1, we considered the faster component to be assigned to the unimers. In Figure 3 Arrhenius plots of kiso for three concentrations of P85 in H2O are shown. The first inflection point (31) (a) Marinov, G.; Michels, B.; Zana, R. Langmuir 1998, 14, 2639. (b) According to a private communication by J. F. Holzwarth, the impurity is probably a triblock copolymer comparable to L61. (c) Purifying the L64 sample by three extractions with hexane did not remove the biexponential character of the kinetic curves in the lower temperature range.
Gille et al.
of the curves occurs with rising concentration at lower temperatures, in accordance with cmt ) f(T) measurements by the solubilization method.3 Our estimates of the cmt values together with literature values are given in Table 2. Contrary to 2.3% and 4.6% F88 (not shown), with P85 there are significantly lower rate constants for 4.6% compared to 0.2% and 1.0% solutions in the premicellar temperature range where generally the existence of unimers at all concentrations is assumed. Probably aggregation promotion by dye molecules is effective in these more concentrated solutions32 (see below), or a sufficient concentration of oligomers or aggregates from impurities are effective as hosts for the low concentration of dye molecules. Moreover, the kiso values above the second inflection point in Figure 3 decreased with increasing concentrations of P85. They also reflect that isomerization occurs in micelles of different microenvironments for the dye. Taking into account that the polarity-dependent λmax values of 483 ( 2 nm do not change significantly on micellization and in micellar solutions of different concentrations, the changes and differences of kiso should mainly be due to changes and differences of microviscosity. We use the term microviscosity as other authors do33 to distinguish the viscosity of the environment of probe molecules which is effective in hindering molecular diffusion or rotation (here rotation around the azo bond during isomerization) from other domains of microstructured solutions or from the bulk viscosity of the solution. Different rate constants should therefore be due to hosts from various polymer molecules (unimers) and/or different aggregation numbers in the premicellar range. Because of the low concentration of dye molecules, these hosts must not be the majority of unimers but probably better solubilizing oligomers or aggregates of impurities in the premicellar range. Between the first and second inflection points (Figures 1-3), the decreasing kiso values may be explained with increasing microviscosity because of increasing aggregation numbers as found in SANS studies.15,30,34 At temperatures above the second inflection point, the difference of kiso between 1 and 4.6% P85 should also be due to larger aggregation numbers at the higher concentration, in accord with the hydrodynamic radii determined in PFG NMR experiments; see below. By means of SANS studies at 40 °C, also different aggregation numbers were obtained in 1% and 5% solution, respectively,34 namely, 37 and 55. Cis-trans isomerization rate constants were empirically correlated with bulk viscosity according to kiso ∼ 1/ηR.35 We have calculated the average of rate constants from two temperatures next to below the first (kiso,uni) and above the second inflection point (kiso,mic) for 4.6% solutions of the Pluronics. From (kiso,uni/kiso,mic)1/R ) ηmic/ηuni ) f, the ratio f of the microviscosities on change from unimers into micelles may be roughly estimated, neglecting the temperature difference between the unimer and micellar states. In a recent investigation we had determined R ) 0.52 (25 °C) in pure solvents including liquid polymers.36 When R ) 1 is taken as an upper limit, lower limits of f follow and are given in Table 3. An increase of microviscosities of about 1 order of magnitude was estimated. Assuming a water-like viscosity of 1 mPa s around the (32) Gadelle, F.; Koros, W. J.; Schechter, R. S. Macromolecules 1995, 28, 4883. (33) Nakashima, K.; Anzai, T.; Fujimoto, Y. Langmuir 1994, 10, 658. (34) Goldmints, I.; von Gottberg, F. K.; Smith, K. A.; Hatton, T. A. Langmuir 1997, 13, 3659. (35) Benniston, A. C.; Harriman, A. J. Chem. Soc., Faraday Trans. 1994, 90, 2627. (36) Gille, K.; Knoll, H.; Quitzsch, K. Int. J. Chem. Kinetics, in press.
Structure Formation in PEO-PPO-PEO Block Copolymers
Langmuir, Vol. 15, No. 4, 1999 1063
Table 1. cmt Values of Triblock Copolymer Solutions in H2O and D2O Determined from Kinetic and PFG NMR Measurements, Respectively, and Literature Data H2O/D2O 0.2% from ln kiso ) f(1/T)
L64 P85 F88 P85 L64 P85 F88
from PFG NMR log D ) f(1/T) from solubilization3 in H2O
a
1.0%
34.5/34.5
30.0/29.0
34
30/31/26a
33.7
29.5
2.3%
4.6%
32.5/32
26.0/24.5 24.5/23.5 29.0/28
5%
27/27 26.5 25.5 30.5
34b
[4,4′-nitroanilinoazobenzene] ) 4 × 10-4 mol dm-3. b 2.5%.
Table 2. Kinetic Solvent Isotope Effect kiso,H2O/kiso,D2O of the Thermal Cis-Trans Isomerization Dependent on Temperaturea kiso,H2O/kiso,D2O T/°C 20 25 30 35 40 45 50
L64 4.6% 2.9 4.7 4.5 3.1 2.8 2.3
P85
F88
1%
4.6%
5 7 4.2 3.0 2.8
1.8 2.1 1.7 1.4 1.3 1.3
2.3% 2.5 3.2 3.8 3.5 2.3 2.2
4.6% 2.3 2.3 3.0 4.2 2.2 1.7
a k iso,H2O/kiso,D2O of the fast rate components are given for 1.0% P85 and 4.6% L64.
Figure 3. Arrhenius plots of isomerization rate constants kiso (only fast component) for different concentrations of P85 in H2O and in PPO 425 + 3% H2O. Table 3. Ratio f ) ηmic/ηuni of Estimated Microviscosity Changes for 4,4′-Nitroanilinoazobenzene Solubilized by Unimers of Triblock Copolymers Which Aggregate to Micelles (Assumed To Be Lower Limits with r ) 1; See Text)a
f(H2O) f(D2O)
Figure 2. Arrhenius plots of isomerization rate constants kiso for 4.6% L64 in H2O (open symbols) and D2O (filled symbols). Lines are guides for the eyes: solid lines connect rate constants kiso,1, and dotted lines connect rate constants kiso,2, according to eq 2.
dye solubilized by unimers, viscosities between 5 and 24 mPa s follow for the microenvironment of the dye in micelles, the hydrated PEO corona of the micelles. These values are less than those in the micellar PPO core determined by pyrene fluorescence,14 as expected. The kinetic solvent isotope effect kiso,H2O/kiso,D2O indicates a water-rich microenvironment of the dye molecules and can be explained with stronger hydrogen bonds in D2O compared to H2O, leading to a more compact corona and higher microviscosity. This idea is in accord with the lower hydrodynamic radii found in D2O solutions in comparison to H2O solutions in the PFG NMR experiments. The result of a further kinetic experiment supports the idea that the solubilization site of the azo dye is essentially the PEO corona and not the PPO core: the rate constants kiso in a homogeneous solution of PPO 425 with 3% H2O (λmax ) 485 nm) were determined as lower by a factor of 6 than in micelles of a 4.6% P85 solution, which are the smallest kiso values determined in Pluronics solutions; see Figure 3.
P85
F88
L64 4.6%
1%
4.6%
2.3%
4.6%
23 14
10 7
6 5
9 8
10 10
a Rate constants k iso,uni (lower temperatures, 20-30 °C) and kiso,mic (higher temperatures, 30-35 °C) were compared, neglecting the temperature differences.
Next we compare the microenvironment of the isomerizing azo dyes in unimers (and other premicellar hosts in the case of L64) and micelles of the Pluronics according to the kiso values. kiso,premic and kiso,mic were plotted vs the PEO block length (Figure 4). From Figure 4 it is obvious that P85 offers a more rigid microenvironment for the isomerizing dye compared to F88 with longer PEO blocks, respectively, both in the premicellar and in the micellar temperature range. The rate constants from L64 solutions, however, do not fit to a uniform trend, probably in part due to the shorter PPO block which forms micelles less readily.4 Moreover, the hydrodynamic radii of L64 unimers and micelles are smaller than those of P85 as derived from self-diffusivities.37 The kinetic solvent isotope effect is stronger than that for P85 and F88 (see Table 3), and a better access of water to the dye can be assumed. This could influence the microenvironment of the isomerizing dye in the direction of a lower microviscosity and therefore lead to faster isomerization rates in both premicellar and micellar solutions of L64. (37) Besides larger self-diffusivities, echo attenuations are more nonexponential and therefore the data are not given in the results.
1064 Langmuir, Vol. 15, No. 4, 1999
Gille et al.
Figure 4. ln(kiso) vs PEO block length in premicellar (premic) and micellar (mic) aggregates, in H2O and D2O, respectively. For L64 besides kuni, kiso,2 values are given as data for the premicellar range.
Figure 6. Self-diffusion coefficients D of unimers and micelles of P85 measured by PFG NMR vs 1/T for different concentrations in H2O and D2O. Lines are guides for the eyes only.
Figure 5. ln(kiso) and relative fluorescence intensities I/I0 of auramine vs 1/T in 25% F88 in H2O and D2O. The vertical line at T ) 304 K indicates the temperature of the fluid/gel transition.
Gelation. At high concentrations of the triblock copolymer F88 (25%), where all probe molecules are solubilized in micelles at temperatures above 20 °C, a clear solid gel is formed on increasing temperature. The bulk viscosity increased with a sharp breakpoint in the viscosity vs temperature curve at 31 °C.38,39 Gel formation occurs at close packing of micelles, and an interpenetration of the PEO chains starts. Therefore, the outer part of the PEO corona of the micelles as a possible environment for the dye molecules might experience some change and accordingly also the isomerization rate. Moreover, Wanka et al.16 reported a peculiar behavior of pluronics in the gel phase as compared to a solution deduced from viscoelastic data. Therefore, we determined kiso of the dye in F88 (25%) dependent on temperature to probe the PEO corona around the dye probe molecules on fluid/gel transition. λmax ) 483 nm was constant below and above the gelation temperature. The Arrhenius plot of ln kiso vs 1/T shown in Figure 5 is linear across the gelation temperature with a constant kinetic solvent isotope effect, indicating that there is no coupling between rate-determining microviscosity and bulk viscosity. This result is in accord with the conclusion from UV/vis spectroscopy that the solubilization site of the azo dye probe molecules is preferentially the inner part of the PEO corona. Microviscosity in the micelles (38) Scheller, H. Ph.D. Thesis, Universita¨t Leipzig, Leipzig, Germany, 1997. (39) Scheller, H.; Fleischer, G.; Ka¨rger, J. Colloid Polym. Sci. 1997, 275, 730.
was independently probed by means of fluorescence intensities I of auramine in F88 (25%) solution relative to I0 determined in water; see Figure 5. It was shown that auramine fluorescence changes with bulk viscosity in glycerol/water mixtures,40 and this was recently used to characterize water surfactant solutions.41 I/I0 shown in Figure 5 reflects a small linear decrease of the microviscosity around the auramine molecules with increasing temperature without discontinuity in accord with the rate measurements. PFG NMR. When probe molecules are used to characterize their microenvironment, the question always arises as to what extent the probe molecule itself influences its microenvironment. In our case of probing the microenvironment by means of kiso, only those species were covered in which a dye molecule was solubilized. According to the low dye concentration and an aggregation number of about 50 in micelles as a rough estimate, less than 30% of the final micelles at 0.2% P85, and even much less in the other samples, were occupied by a probe molecule. To compare the results from kinetic measurements with a probe molecule free method, or where all of the microstructure-forming copolymer molecules are probe molecules, the PFG NMR method was applied on selected solutions of the P85 copolymer. Because the signal of the solvent water is invisible in the NMR experiments when using D2O or easy to subtract from the total NMR signal in the case of H2O, the transition unimer-micelle is readily observed in the experiment. The dependence of the self-diffusion coefficient D on temperature is shown as an Arrhenius plot in Figure 6. The unimer as well as micelle diffusion at the low- and hightemperature side scale within experimental accuracy with T/η0, where η0 denotes the viscosity of pure water. In the concentration range investigated, the temperature dependence of the viscosity of the polymer solution is almost (40) Oster, G.; Nishijima, Y. J. Am Chem. Soc. 1956, 78, 1581. (41) Miyagishi, S.; Kurimoto, H.; Ishihara, Y.; Asakawa, T. Bull. Chem. Soc. Jpn. 1995, 68, 135.
Structure Formation in PEO-PPO-PEO Block Copolymers
entirely determined by the temperature dependence of the viscosity of pure water. Therefore, the observed scaling of D with T/η0 is a result of a nearly temperatureindependent hydrodynamic radius of the unimers and micelles. Coming from low temperatures, the first inflection point at which D starts to decrease with increasing temperature is the cmt. At the temperature of the second inflection point, at which D again starts to increase with increasing temperature, the micellization is complete. In our echo attenuations, neither an indication of biexponetial behavior nor a broadening of the distribution of selfdiffusivities in this temperature region, measured by a reduced β value of the fit with eq 4, is observed. The echo attenuations deviate slightly from monoexponentiality, but a biexponential behavior could not be revealed. The fit according to eq 4 worked very well, which is often the case in samples with a slight distribution of self-diffusivities and traces of low molar mass substances in the samples.42 Such a behavior was already found in our previous investigation17 and in the literature.43,44 The triblock chains have a residence time in one micelle shorter than our shortest diffusion time of a few milliseconds; they change between the unimer and the micellar states several times in the experiment. Therefore, we always measure an averaged self-diffusion coefficient D h
D h ) puniDuni + pmicDmic
(5)
where puni and pmic and Duni and Dmic are the fractions of chains and the self-diffusivities in the unimer and micellar states, respectively. This case of “fast exchange” is confirmed by dynamic bulk modulus measurements by Hvidt6 at a similar Pluronic (P94) and relaxation measurements by Holzwarth45 and Zana et al.46 who have found time constants on the order of microseconds. In accordance with SANS data,34 the transition temperature range observed in self-diffusion extends further to higher temperatures than that of kiso measurements. kiso is decreased to the micellar value already if the first micelles of final size are formed, and a preferred solubilization of the dye in larger aggregates can be assumed. Because of the considerable contribution of puniDuni to the measured D h , the temperature range between the two inflection points of the self-diffusion experiments is larger than that in the case of the rate constants kiso. This rather broad temperature region of micellization has its origin in the thermodynamic equilibrium between unimers and micelles of increasing aggregation number34 with increasing temperature, and possibly composition and chain length heterogeneity of the triblocks, and also in the hydrophobic impurities.4 In Figure 7, the hydrodynamic radii in the different solutions dependent on temperature are shown. They were calculated with the Stokes-Einstein equation
RH ) kT/6ηD
(6)
where η is the viscosity of the solution. The validity of the Stokes-Einstein equation in this form for colloidlike particles was recently demonstrated in the literature.47,48 (42) Fleischer, G.; Rittig, F.; Konak, C. J. Polym. Sci., Polym. Phys. Ed., in press. (43) Bjo¨rling, M.; Herlo¨f-Bjo¨rling, A.; Stilbs, P. Macromolecules 1995, 28, 6970. (44) Walderhang, H.; Nystro¨m, B. J. Phys. Chem. B 1997, 101, 1524. (45) Goldmints, I.; Holzwarth, J. F.; Smith, K. A.; Hatton, T. A. Langmuir 1997, 13, 6130. (46) Michels, B.; Waton, G.; Zana, R. Langmuir 1997, 13, 3111. (47) Richtering, W.; Mu¨ller, H. Langmuir 1995, 11, 3699.
Langmuir, Vol. 15, No. 4, 1999 1065
Figure 7. Hydrodynamic radii RH calculated according to eq 6 from the self-diffusivities (Figure 6) and viscosities. Lines and symbols are as in Figure 6.
The hydrodynamic radii of the micelles do not appreciably depend on temperature as the result of a compensation effect. The increase of the core radius upon heating15 is compensated for by a decreasing corona thickness due to a decrease of the solvation of the PEO chains, as was already discussed by Zhou49 and Wanka et al.16 The size of the micelles is larger in H2O than in D2O. This difference was up to now, to the best of our knowledge, not noticed in the literature. The observed difference between RH of P85 measured in D2O by PFG NMR17 and in H2O by QELS12 was in our ref 17 attributed only to the different averaging behavior of the two methods. This must now be questioned. Hydration of PEO stabilizes the transgauche-trans conformation of the PEO chain.50 The stronger hydrogen bonds between the EO oxygens and D2O as compared to H2O lead to a stronger coiling of the PEO chain parts and, consequently, increased segment density and a more compact PEO corona in D2O solutions. In a separate experiment, the addition of the azo dye in a much higher concentration (4 × 10-4 mol dm-3) of 1 dye molecule/5 molecules P85 leads to an increase of the size of the dye-solubilizing hosts (oligomers, aggregates of impurities) below the cmt as well as of the micelles of 50%. The NMR signal is not influenced by the dye molecules because of their still low concentration. The increase of the hydrodynamic radii starts at 26 °C, at 5 K lower compared to samples without dye molecules, giving evidence for an aggregation promotion effect.32 From the difference of the curves between the necessarily low probe concentrations in our isomerization experiments and the high concentration in this PFG NMR experiment, it can be concluded that aggregation of two unimers with a probe molecule occurs at lower temperature than aggregation between unimers with and unimers without probe molecule. The latter situation corresponds to our (48) Segre, P. N.; Meeker, S. P.; Pusey, P. N.; Poon, W. C. K. Phys. Rev. Lett. 1995, 75, 958. (49) Zhou, Z.; Chu, B. J. Colloid Interface Sci. 1988, 126, 171. (50) Wartewig, S.; Alig, I.; Hergeth, W. D.; Lange, I.; Lochmann, R.; Scherzer, T. J. Mol. Struct. 1990, 219, 365.
1066 Langmuir, Vol. 15, No. 4, 1999
reactivity experiments. The satisfactory agreement of cmt values determined by both PFG NMR and kinetic methods and those given in the literature,3 however, shows that the low dye concentration in the kinetic experiments does not significantly influence the onset of micellization. Summary and Conclusions Rate constants of the thermal cis-trans isomerization of 4,4′-nitroanilinoazobenzene in aqueous solutions of triblock copolymers determined by means of flash photolysis show an S-shaped behavior due to unimer-micelle transition upon increasing the temperature. Self-diffusion coefficients D of P85 triblock copolymer molecules in water determined by means of PFG NMR measurements behave similarly. cmt values were estimated from both the first inflection points of Arrhenius plots of rate constants kiso and the self-diffusion coefficients D in satisfactory aggreement with each other as well as with literature data. Because λmax values of the trans isomers indicating micropolarity change only marginally at the unimermicelle transition, the decrease of rate constants is essentially due to an increase of the viscosity in the microenvironment of the azo dye probe molecules. The remarkable kinetic solvent isotope effects kiso,H2O/ kiso,D2O of the thermal cis-trans isomerization reflect that the dye is solubilized in a water-rich environment. The tendency to aggregation is stronger in D2O than in H2O, and aggregation begins at a lower temperature in D2O than in H2O. The hydrodynamic radii of unimers as well as micelles of P85 in 1% and 5% solutions are larger by
Gille et al.
a factor of about 1.25 in H2O than in D2O. Because of weaker hydrogen bonds between the EO units and H2O compared to D2O, the PEO chains are more stretched and less hydrated in H2O solutions. In a PFG NMR experiment with 1% P85 and a high dye concentration, a 5 K lower cmt than that without dye was determined. Obviously the dye has an aggregationpromoting effect, which is not effective at low dye concentrations of the kinetic experiments. Whereas space restrictions for diffusion of triblock copolymer molecules are effective in gelated solutions of 25% F88, isomerization rate constants are not significantly influenced by gel formation. This indicates that the viscosity of the environment of the dye molecules is not influenced by gelation. Therefore, the inner part of the PEO corona seems to be the preferential solubilization site of the dye molecules in accord with UV/vis spectroscopy of the trans isomer. Acknowledgment. The authors thank BASF for providing the samples of Pluronics. K.G. acknowledges a grant and further support by the “Graduiertenkolleg Physikalische Chemie der Grenzfla¨chen” at Universita¨t Leipzig (Deutsche Forschungsgemeinschaft and Sa¨chsisches Staatsministerium fu¨r Wissenschaft und Kunst). F.R., G.F., and J.K. are grateful for financial support from the Deutsche Forschungsgemeinschaft (SFB 294). H.K. thanks the Fonds der Chemischen Industrie for financial support. LA980968X
