Shell Micelles. Interaction

8394

J. Phys. Chem. B 2007, 111, 8394-8401

Interpolymer Complexes Based on the Core/Shell Micelles. Interaction of Polystyrene-block-poly(methacrylic acid) Micelles with Linear Poly(2-vinylpyridine) in 1,4-Dioxane Water Mixtures and in Aqueous Media† Pavel Mateˇ jı´cˇ ek,* Mariusz Uchman, Jana Lokajova´ , Miroslav Sˇ teˇ pa´ nek, and Karel Procha´ zka Department of Physical and Macromolecular Chemistry, Faculty of Science, Charles UniVersity, AlbertoV 2030, 128 40 Prague 2, Czech Republic

Milena Sˇ pı´rkova´ Institute of Macromolecular Chemistry, Academy of Sciences of the Czech Republic, HeyroVsky´ Sq. 2, 16206 Prague 6, Czech Republic ReceiVed: December 11, 2006; In Final Form: February 15, 2007

The size and structural changes of nanoparticles formed after the addition of poly(2-vinylpyridine), PVP, to block copolymer micelles of polystyrene-block-poly(methacrylic acid), PS-PMA, were studied by light scattering and atomic force microscopy. Due to the strong hydrogen bonding between PVP and PMA segments, complex structures based on the core/shell micelles form in mixed selective solvents. As proven by a combination of light scattering and atomic force microscopy, individual PS-PMA micelles are “glued” together by PVP chains. The dialysis against solvents with a high content of water results in transient increase in polydispersity and turbidity of originally clear solutions. However, the precipitated polymer material dissolves in basic buffers and stable soluble nanoparticles reform in aqueous media. The behavior of their solutions was studied in a broad pH range by light scattering, atomic force microscopy and capillary zone electrophoresis.

Introduction The self-assembly of amphiphilic diblock copolymers, such as polystyrene-block-poly(methacrylic acid), PS-PMA, or polystyrene-block-poly(2-vinylpyridine), PS-PVP, both in melts and solutions has been amply studied for many years.1,2 Block copolymers find many potential applications in biomedicine as components of systems for the targeted drug delivery and as stimuli-responsive functional materials for construction of nanodevices.3 The most important principles controlling the association processes and the behavior of formed nanoparticles have been revealed and generally understood. Thanks to recent advances in polymer synthesis, new structures have been synthesized (multiblock copolymers, star-like copolymers, etc.), and their self-assembly has been studied.2 At present, the interest of researchers focuses on micelle-based interpolymer complexes with the aim to prepare various stimuli-responsive structures.4-8 Many published studies describe the formation of complex micelles, in which specific interactions between individual components lead to the formation of mixed micellar cores.4,5 In some cases, the interpolymer complex is formed in the middle layer of onion micelles.6 Very interesting systems are “schizophrenic block copolymers”, which reverse the structure of associates upon changes of conditions (pH, temperature, etc.).4a,7 One of promising strategies for the preparation of complex nanostructures is the interaction (usually electrostatic) and the complex formation of already formed micelles with linear homopolymers.8 In the Webber group, the interaction of kinetically frozen polystyrene-block-poly(2-vinylpyridine), PS† Part of the special issue “International Symposium on Polyelectrolytes (2006)”. * Corresponding author. Tel: +420221951292. Fax: +420224919752. E-mail: [email protected].

PVP micelles with poly(styrenesulfonate) was studied in acidic aqueous solutions.8a A similar type of structures was studied also by Pergushov et al.8b,c He investigated a system of micelles with non-frozen cores and coronas formed by a weak polyacid and used a strong polybase as a complex-forming homopolymer. The stabilization of superstructures by nonelectrostatic interaction, e.g., by hydrogen bonds has been a much less studied topic. To our knowledge, only one paper which is directly related to our work was published so far: Zhang et al. studied a hydrogen bond-mediated adsorption of poly(4-vinylpyridine) chains on the kinetically frozen polystyrene-block-poly(acrylic acid) micelles in ethanol-dimethylformamide mixtures.8d The strategy based on the electrostatic complex formation has a strong limitation because at least one or better both hydrophilic blocks should be strong polyelectrolytes to ensure the solubility of both components in a broad range of pH. Otherwise, it is difficult to elaborate a feasible protocol for the reproducible preparation of well-defined electrostatically stabilized complexes. In the case of weak polyelectrolytes, the stability and solubility of interpolyelectrolyte complexes is usually restricted to a narrow pH region only.9 In this paper, we would like to demonstrate that the solubility problems can be overcome by using dialysis in mixed solvents. We combine the strategies of the polyion-based and hydrogen bond-based complex formation with the goal to prepare new polymeric associates from simple systems, the behavior of which is wellknown and understood. We have chosen the PS-PMA micelles and PVP, because PVP forms a complex with PMA.10,11 The aims of the study may be outlined as follows: We want to (i) prepare hydrogen-bond stabilized interpolymer complexes based on well-defined PS-PMA micellar precursors, where the PMA shells are slowly and gradually saturated by PVP in 1,4-

10.1021/jp0685075 CCC: $37.00 © 2007 American Chemical Society Published on Web 04/07/2007

Complexes Based on the Core/Shell Micelles

J. Phys. Chem. B, Vol. 111, No. 29, 2007 8395

dioxane-rich mixtures with water, and (ii) elaborate a reproducible protocol of their transfer toward aqueous solutions based on dialysis. (iii) The main goal is the preparation of stable and monodisperse nanoparticles in aqueous media, and the study of their properties and stability in a broad pH region. The research was performed in several steps and hence also the presentation (the part Results and Discussion) is divided in two main parts: (i) experiments concerning a slow saturation of PMA shells by PVP and studies of the behavior of mixed systems in 1,4-dioxane-rich solutions and (ii) studies of the behavior of aqueous solutions after dialysis from 1,4-dioxane/ water mixtures. Experimental Section Materials. Polymer Samples. Two polystyrene-block-poly(methacrylic acid) diblock samples: PS-PMA1, with molar masses of PS and PMA blocks corresponding to 24.6 × 103 and 18.8 × 103 g/mol and PS-PMA2 with 40.6 × 103 and 28.3 × 103 g/mol, respectively, were synthesized at the University of Texas at Austin. Details on the synthesis and characterization are given in ref 12. Poly(2-vinylpyridine) (PVP) was purchased from Aldrich. Two polymer samples with different degree of polymerization were used. The molar masses are Mw ) 37.5 × 103 g/mol (PVP1) and 159.0 × 103 g/mol (PVP2) with polydispersity, Mw/ Mn, 1.07 and 1.05, respectively, as provided by the manufacturer. Preparation of Polymer Solutions. The PS-PMA copolymer was dissolved (c ) 1 g/L) in a mixture of 1,4-dioxane (80 vol %) with water, in which spherical multimolecular micelles with swollen PS cores and PMA shells form spontaneously upon dissolution. Then the PVP solution (1 g/L in an identical solvent mixture) was slowly added to the PS-PMA solution. The addition of a large amount of PVP provokes the precipitation, but quite stable soluble complexes may be prepared in a relatively broad region of PVP-to-PMA ratios. The PVP content in the PS-PMA/PVP mixtures ranged from 0% to 60% (w/w) and from 0% to 10% (w/w) for the PS-PMA1 and PS-PMA2 samples, respectively. All solutions with different PVP content were dialyzed stepwise against 60% 1,4-dioxane in water, 40% 1,4-dioxane in water, 20% 1,4-dioxane in 0.025M disodium tetraborate aqueous solution and finally against 0.05 M disodium tetraborate aqueous solution and pure water. The aqueous solutions of PS-PMA1/PVP were dialyzed in 0.05 M tetraborate solution with 0.1 M NaCl and in 0.03 M HCl solution with 0.1 M NaCl for LS measurements. The PS-PMA1/PVP and PSPMA2/PVP solutions were dialyzed into Britton-Robinson buffers with pH ranging from 2 to 12 and ionic strength of ca. 0.1 mol/L. Methods. Light Scattering. The light-scattering setup (ALV, Langen, Germany) consisted of a 633 nm He-Ne laser, an ALV CGS/8F goniometer, an ALV High QE APD detector and an ALV 5000/EPP multibit, multitau autocorrelator. The solutions of PS-PMA and PVP were filtered through 0.20 µm Acrodisc filters before their mixing. The measurements were carried out for different concentrations, PVP fractions, and angles. Static light-scattering (SLS) data were treated by the standard Zimm and Berry methods. The refractive index increments of copolymers, (dn/dc), were calculated as the weight-average of literature data.13,14 For pure PS-PMA, values of (dn/dc) reported earlier after dialysis which accounts for the preferential sorption of 1,4dioxane were used,14c i.e., 0.112 for PS-PMA and 0.189 for PVP. The refractive index increments for different hydrogen bond-stabilized complexes formed by PS-PMA and PVP were calculated as the pertinent weight-averages of PS-PMA and PVP values.

Figure 1. AFM scan on a mica surface of PS-PMA1 micelles (ξ ) 0) deposited from 80 vol % 1,4-dioxane in water.

DLS data analysis was performed by fitting the measured normalized intensity autocorrelation function g2(t) ) 1 + β|g1(t)|,2 where g1(t) is the electric field correlation function, t is the lag-time, and β is a factor accounting for deviation from the ideal correlation. An inverse Laplace transform of g1(t) with the aid of a constrained regularization algorithm (CONTIN) provides the distribution of relaxation times, τA(τ). Effective angle- and concentration-dependent hydrodynamic radii, RH(q,c), were obtained from the mean values of relaxation times, τm(q,c), of individual diffusive modes using the Stokes-Einstein equation. To obtain true hydrodynamic radii, the data have to be extrapolated to a zero scattering angle. Atomic Force Microscopy (AFM). All measurements were performed in the tapping mode under ambient conditions using a commercial scanning probe microscope (Digital Instruments NanoScope dimensions 3), equipped with a Nanosensors silicon cantilever. Polymeric particles were deposited on a freshly cleaved mica surface by a fast dip coating in a dilute polymeric solution. The samples were dried in a vacuum oven at ambient temperature for ca. 3 h. Capillary Zone Electrophoresis (CZE). The electrophoretic measurements were carried out using Hewlett-Packard 3D CE apparatus (Hewlett-Packard, Waldbronn, Germany), equipped with an in-line variable-wavelength detector. Two capillaries were used: (i) an untreated fused silica capillary for measurements in the basic buffer, and (ii) a silica capillary coated by Polybrene for measurements in the acidic buffer. The background electrolytes contained (i) 5 mM borate buffer (I ) 10 mM and pH 9.2), or (ii) 30 mM chloracetic acid and 6 mM lithium hydroxide (I ) 9.55 mM and pH 2.5). Thiourea was used as an EOF marker of the electroosmotic flow. Results and Discussion Polymer Systems in a Mild Selective Solvent Mixture. First, the PS-PMA micelles and PVP chains in a 80% 1,4dioxane/water mixture were characterized separately before they were allowed to interact with each other. The PS-PMA copolymer forms monodisperse compact spherical micelles. They were characterized by SLS and DLS. The standard data treatment (the Zimm plot) gives for PS-PMA1: Mw ) 13.0 × 106 g/mol, Rg ) 18.3 nm and RH ) 29.3 nm. The micelles can be deposited on the mica surface from the mixed solution producing well separated round particles which can be clearly imaged using the soft tapping mode AFM technique (Figure 1). The nonassociated PS-PMA chains (unimers) were not detected by AFM, despite the fact that they coexist in the equilibrium with micelles in selective solvents.14 The absence

8396 J. Phys. Chem. B, Vol. 111, No. 29, 2007

Figure 2. Weight-averaged molar mass, Mw, of PS-PMA1/PVP in 80 vol % 1,4-dioxane in water as a function of ξ for PVP1 (curve 1, filled points and straight line) and PVP2 (curve 2, hollow points and dashed line). Inset: Radius of gyration, Rg, of PS-PMA1/PVP in 80 vol % 1,4-dioxane in water as a function of ξ for PVP1 (curve 1′, filled points and straight lines) and PVP2 (curve 2′, hollow points and dashed lines).

of the unimer chains at the surface can be explained by the fact that their concentration is low, and their size is very small, preventing their detection by the AFM tip. In the second step, the light scattering study of poly(2vinylpyridine), PVP, dissolved in 80% 1,4-dioxane/water mixture was carried out. Neutral PVP is molecularly soluble in the form of random coils in the solvent mixture, the corresponding hydrodynamic radii, RH, are well measurable by DLS: 7.1 and 9.1 nm for the shorter PVP1 the longer PVP2 chains, respectively. During the step-by-step addition of PVP to PS-PMA1 micelles, the structure of polymeric associates changes considerably as evidenced by LS measurements. The PVP segments interact with the shell-forming PMA segments and form hydrogen bonds in 1,4-dioxane-rich media. The interaction is hindered by steric constraints; nevertheless, the complex forms quite fast (on the time scale of minutes), presumably first in the peripheral part. We assumed that PVP slowly penetrates in the middle and inner part of the shell forming a compact interpolymer complex. In Figure 2, the dependence of the apparent weight-average molar mass, Mw, on the ratio of 2-vinylpyridine units to methacrylic acid units, ξ, is shown (the apparent Mw values were obtained from the Berry plots which were reasonably linear in most cases). The weight-average aggregation number of all PS-PMA/PVP associates in the solution increases in the presence of PVP. It rises gradually with ξ and exceeds the value that is 2 orders of magnitude higher than that of PS-PMA micelles. The radius of gyration (Inset of Figure 2) increases from 18.3 nm for PS-PMA1 micelles to about 300 or 400 nm for PS-PMA1/PVP associates at ξ ca. 0.25. At higher ξ, the system becomes turbid and polydisperse. The behavior of turbid solutions in organic media was not studied. The static light scattering measurements show that the addition of PVP and the formation of hydrogen bonds between PMA and PVP destabilize the PS-PMA micelles in mixed selective solvents and leads to the formation of large clusters. There is almost no difference between the systems with PVP1 or PVP2 which is in agreement with results of Talingting et al.8a The Mw and Rg values in Figure 2 are only slightly lower for PVP2, although PVP1 contains ca. 360 segments and PVP2 ca. 1500 ones (for comparison, PS-PMA1 copolymer contains ca. 220 PMA segments, and the PS-PMA1 micelles consist of ca. 310 chains in a 1,4-dioxane(80%) water mixture). Additional information was obtained by dynamic light scattering, DLS. A careful analysis of autocorrelation functions reveals three individual modes corresponding to the diffusive motion (see Figure 3, where a typical distribution of relaxation times is shown). The hydrodynamic radii, corresponding to individual modes extrapolated to a zero scattering angle are

Mateˇjı´cˇek et al.

Figure 3. A typical distribution of correlation times of PS-PMA1/ PVP in 80 vol % 1,4-dioxane in water for ξ ) 0.04 and PVP1 sample at the scattering angle of 90°. Three individual modes are assigned as No 1, No 2 and No 3.

Figure 4. Hydrodynamic radius, RH, of PS-PMA1/PVP in 80 vol % 1,4-dioxane in water as a function of ξ for PVP1 (curves 1, 3, and 5; filled points and straight lines) and PVP2 (curve 2, 4, and 6; hollow points and dashed lines) corresponding to mode No 1 (curves 1 and 2), No 2 (curves 3 and 4) and No 3 (curves 5 and 6).

Figure 5. Light-scattering-intensity-weighted fractions, ai, of individual modes in distributions of relaxation times of PS-PMA1/PVP1 in 80 vol % 1,4-dioxane in water as a function of ξ measured at 90°; a1 corresponding to the mode No 1 (curve 1, filled point), a2 corresponding to the mode No 2 (curve 2, hollow point), and a3 corresponding to the mode No 3 (curve 3, half filled point).

shown in Figure 4 for PVP1 (curves 1, 3, and 5) and for PVP2 (curves 2, 4, and 6) as a function of ξ. The scattered-intensityweighted fractions of individual modes obtained at 90°, a1-a3, are depicted in Figure 5 for PVP1 (for PVP2 almost identical, not shown). The changes in the hydrodynamic radius, RH, corresponding to mode No 1 (as marked in Figure 3) with ξ are interesting (see Figure 4, curves 1 and 2). At low fractions ξ, the measured RH drops suddenly from 29.3 nm (corresponding to pure PSPMA1 micelles) to 19.7 and 17.0 nm for PVP1 and PVP2, respectively. Then it increases slowly, and for high ξ it levels off at 33.0 ( 2.5 nm and 32.9 ( 4.7 nm for PVP1 and PVP2, respectively. The fraction a1 diminishes dramatically with increasing PVP content, but this mode persists until the macroscopic precipitation (a1 < 0.1 for ξ > 0.05; see Figure 5, curve 1). It is necessary to bear in mind that the fraction of the small particles in the complex system is underestimated by SLS. As the hydrodynamic radius of particles corresponding to this mode is similar to that of PS-PMA micelles, we assign the first

Complexes Based on the Core/Shell Micelles mode to individual core/shell micelles with shells containing the PMA/PVP complex. The shell collapse at low ξ is due to the incorporation of a low fraction of PVP stabilized by hydrogen bonds.8d Hence, the formed polymeric particles resemble the micelles with relatively dense chemically crosslinked shells. If we assume further that the original association number of PS-PMA1 micelles does not change and that the density of cores is around 0.9 g/mL (i.e., 10% swelling by the solvent), we get an approximate PS core radius only ca. 14.5 nm and shell thickness ca. 4 nm for very low ξ. Mode No 2 (Figure 3) does not change significantly with PVP/PMA fraction (see Figure 4, curves 3 and 4). The hydrodynamic radii are 140 ( 31 nm and 135 ( 34 nm, for PVP1 and PVP2 samples, respectively. It means that there is almost no influence of the PVP chain length and fraction on RH of the second type of particles. The intensity-weighted fraction, a2, increases steeply at low ξ, it levels off quite soon at ξ ca. 0.04 and does not change any more (Figure 5, curve 2). It is ca. 0.5 for PVP1 and ca. 0.4 for PVP2. The data suggest that some fraction of micelles form compact and relatively monodisperse aggregates interconnected by PVP, the size of which does not almost depend on ξ. The clusters survive in the solution until the macroscopic precipitation. Individual micelles are bound together by PVP chains which interact with PMA shells of different micelles and “glue” individual micelles to each other. Finally, we would like to discuss the origin of mode No. 3, which appears at ξ higher than ca. 0.02, and the fraction of which a3, increases slowly with ξ (see Figure 5, curve 3). We assume that this mode corresponds to large and polydisperse micellar clusters that are not stable enough under given conditions. Below ξ ) 0.25, they partially precipitate on the time scale of weeks. At higher PVP-to-PMA ratios, they precipitate fast on the time scale of minutes. The auxiliary study of the structure and size distribution of the micellar clusters show that the distributions and mainly the fractions of the large aggregates depend on the way of preparation. Not only the inverse mixing (i.e., addition of PS-PMA micelles to PVP solutions, which, however, does not fit to our preparation strategy of a slow saturation of PMA shells of micellar precursors), but also different mixing rates affect significantly the fraction of individual types of aggregates. The preparation of mixed nanoparticles at low concentrations generally suppresses the fraction of large aggregates. It proves an important role of interparticle interaction, but still it depends on the preparation procedure. Nevertheless, it is worth-mentioning that the already formed aggregates do not split upon dilution by a 1,4-dioxane-rich solvent mixture. In any case, the fraction of large particles increases slowly with time, but it seems to levels off. The gradual addition of PVP promotes the penetration of PVP within the PMA shells. The compact PMA/PVP layer forms slowly in the inner part of the shell which becomes denser and more rigid. Finally, the nanoparticles become kinetically frozen. The freezing of denser parts of the shell does not affect the flexibility of PMA and PVP chains at the shell periphery. They can still interact and mediate the interparticle interaction and aggregation of nanoparticles upon worsening of thermodynamic conditions. When discussing the role of large aggregates, it should be kept in mind that we present the intensity-weighted distribution of relaxation times (see Figures 3 and 5). It means that the contribution of large particles is strongly overestimated (the scattering is proportional to RH6). If we recalculate it to the mass-weighted distribution, the mass fraction of the huge clusters is only few

J. Phys. Chem. B, Vol. 111, No. 29, 2007 8397

Figure 6. (a, b) AFM scans on a mica surface of PS-PMA1/PVP deposited from 80 vol % 1,4-dioxane in water for (a) ξ ) 0.101 and PVP1, and (b) ξ ) 0.153 and PVP2.

percents. Because the behavior of the system is not fully reversible and depends on the history, it is likely that, e.g., concentration fluctuations during mixing, may lead to the freezing of a small fraction of less stable structures and to the formation of large aggregates. To get more insight in the associative behavior, we deposited micelles on the fresh mica surface and investigated the “dry” micellar aggregates by AFM. Even though the thermodynamic conditions and forces acting on the particles at the surface differ considerably from those in the solution and the aggregates can change, we have shown in our earlier studies that AFM yields valuable information on sizes, size distributions and processes of the secondary aggregation of micelles.15 In Figures 6, two AFM scans of PS-PMA1/PVP particles deposited from a 80% vol. 1,4-dioxane in water solution are shown for different ξ and PVP chain lengths. Although the AFM images show only a limited number of deposited clusters, it is quite obvious that the size distribution of deposited particles varies significantly with ξ. It changes from that of original PS-PMA micelles (see Figure 1) to a broad distribution of polydisperse micellar aggregates. The determination of the fraction of free PVP chains which are not bound to PMA shell-forming blocks represents quite serious problem. We cannot a priori preclude a certain part of free PVP coexisting with the PMA/PVP complex. Unfortunately, the LS methods do not detect small PVP coils in a system containing large PS-PMA/PVP aggregates. No structures similar to single macromolecules were detected on mica surface by AFM. It suggests that the fraction of free PVP chains is very low. The results of LS and AFM measurements allow us to propose a model of PS-PMA1/PVP1 clusters with very low PVP-to-

8398 J. Phys. Chem. B, Vol. 111, No. 29, 2007

Mateˇjı´cˇek et al.

SCHEME 1: Structure of PS-PMA/PVP Clusters for Short (Left) and Long (Right) PVP Chains in 1,4-dioxane-rich Selective Solvent (ξ ca. 0.02) Corresponding to the Mode No 2 in Figure 3a

a PS, PMA, and PVP blocks are depicted by blue, black, and red, respectively.

Figure 7. Weight-average molar mass, Mw, of PS-PMA1/PVP as a function of ξ in basic solution (pH 9.2) for PVP1 (curve 1, filled squares) and PVP2 (curve 2, filled points), and in acidic solution (pH 2) for PVP1 (curve 1, hollow squares) and PVP2 (curve 2, hollow points).

PMA fraction, ξ ca. 0.02 (left-hand side of Scheme I) and PSPMA/PVP2 (right-hand side) in 1,4-dioxane-rich aqueous solutions. In order to distinguish the short and long PVP chains better, the chain ends are marked by red circles. Behavior of Complexes in Aqueous Solutions. The stability of the PS-PMA/PVP systems decreases with increasing water content in the mixed solvent. Nevertheless, the dialysis of turbid suspensions of polymeric particles in basic aqueous buffers leads to stable solutions. The mixtures with 0 e ξ e 0.25 yield alkaline solutions of fairly monodisperse particles without large clusters. As no observable precipitation of PVP occurs at pH >4.8, we assume that all PVP chains are incorporated in PMA shells due to strong bonding between the non-dissociated PMA segments and deprotonated PVP segments. In systems with ξ > 0.25, the micellar clusters form also in borate buffers. First we studied the stability of aqueous solutions in alkaline media. We studied the behavior of two samples (PS-PMA1/ PVP and PS-PMA2/PVP) differing in molar mass. The absolute values of individual characteristics are different, but the trends presented in the paper are similar to each other. For the sake of brevity we included only the data for PS-PMA1. The data for PS-PMA2 are available upon request. The Mw versus ξ curves for two systems (PS-PMA1/PVP1 and PS-PMA1/PVP2) measured by LS are shown in Figure 7 (curves 1 and 2, respectively). The curves for PVP1 and PVP2 are almost identical which means that the length of the PVP block does not play important role as concerns the structure of final particles. For the sake of clarity, we will discuss the parts of the curve for ξ < 0.25 and for ξ > 0.25 separately. If we compare the first part with the corresponding curve in a 1,4dioxane-rich aqueous mixture (cf. Figure 2), we can see that molar masses of particles transferred in alkaline buffers are in

Figure 8. Hydrodynamic radius, RH, of PS-PMA1/PVP as a function of ξ in basic solution (pH 9.2, I ca. 0.1) for PVP1 (curve 1, filled squares) and PVP2 (curve 2, filled points), and in acidic solution (pH 1.5, I ca. 0.1) for PVP1 (curve 1, hollow squares) and PVP2 (curve 2, hollow points).

Figure 9. Relative molar mass, Mw/Mwbas, of PS-PMA1/PVP as a function of pH for PVP1 (filled triangles) and PVP2 (hollow triangles) with ξ ) 0.165. Inset: Relative molar masses, Mwac/Mwbas of PS-PMA1/ PVP as a function of ξ for PVP1 (curve 1′, filled squares), and for PVP2 (curve 2′, hollow points).

general considerably lower than those for the same ξ in the mixed solvent. It means that the micellar clusters already detected in mixed solvents split in the alkaline buffer. We assume that the limited amount of the originally soluble PVP chains that “glue” the micelles in mixed solvents shrink in water at pH >4.8 burying themselves in the PMA shells of individual micelles. Therefore, the large micellar clusters disintegrate. For ξ > 0.25, the Mw versus ξ curve rises steeply and levels off for ξ > 0.4. It is obvious that the redispersion of the already precipitated polymer yields very large and polydisperse micellar clusters. Nevertheless, all precipitated material dissolves in alkaline buffers without significant traces of the left-over polymer. It means that all PVP has been dispersed and incorporated into polymeric nanoparticles. Further, we investigated the stability of aqueous dispersions in a broad pH range. First we transferred the dispersions directly in a strongly acidic medium (HCl solution, pH ca. 1.5). It is worth-mentioning that no precipitation occurs after the pH drop in a wide range of ξ. However, the molar masses of stable nanoparticles in acid and alkaline media differ considerably. The corresponding Mw versus ξ curves for PS-PMA1/PVP1 and PS-PMA1/PVP2 are shown in Figure 7 (curves 3 and 4, respectively). Similar to the behavior in alkaline media, there is almost no observable influence of the PVP chain length. DLS data shown in Figure 8 support the conclusions drawn from SLS measurements. The ratio of molar masses of corresponding nanostructures in acidic and alkaline media, Mwac/ Mwbas (for the same ξ), is depicted in inset of Figure 9 as a function of ξ. For ξ < 0.25, the ratio is higher than 1, and the curve increases with ξ. It means that individual micelles, which are soluble at high pH, are destabilized and form micellar aggregates at low pH. The stability decreases with increasing content of PVP. For ξ > 0.25,

Complexes Based on the Core/Shell Micelles

Figure 10. (a, b) AFM scans on a mica surface of PS-PMA1/PVP deposited from pure water for (a) ξ ) 0.153 and PVP2, and (b) ξ ) 0.297 and PVP2.

the curve drops fast and then it reaches the value approximately 0.07, which suggests that the large micellar aggregates existing at high pH split into smaller particles. It can be explained by the protonation of PVP at low pH. An important fraction of PVP chains close to the shell periphery, which “glued” micelles in large clusters at high pH, became protonated and soluble in acidic water. The pH-dependent behavior was studied in detail for PSPMA1/PVP systems in the PVP-to-PMA range 0 < ξ e 0.25. Typical dependences are shown in Figure 9 for ξ ) 0.165. It is evident that mixed micelles are stable in a broad pH region. A significant destabilization accompanied by micellar aggregation was observed only in strongly acidic solutions (for pH 0.3 deposited from dilute HCl solutions visualized by AFM (Figure 11b) is interesting. The micelles are spherical and monodisperse. They are significantly larger than single micelles deposited from pure water for low ξ (cf. Figure 10a). It seems that each particle contains only one PS core decorated with a thick layer of PVP chains. The most striking is the fact that no free PVP chains, which are soluble at low pH and should be partly stretched as a result of their protonation, were detected by AFM, although the deposition of positively charged polymers on the negative mica surface is easy. As we have proven by independent auxiliary experiments (not shown), linear PVP chains deposit from very dilute polymeric solution in 0.1 and 0.01 M HCl on mica as asymmetrical patches on the mica surface after the evaporation of the solvent. Since no such structures were detected in PSPMA/PVP samples, we assume that the free PVP chains are not present in acidic aqueous solutions of PS-PMA/PVP with ξ < 1.20. The most probable structures of PS-PMA/PVP associates with various PVP content and in buffers with various pH proposed on the basis of LS and AFM measurements are depicted in Scheme 2 a-d. As the PS-PMA/PVP system exhibits a complex behavior in aqueous buffers, the ultimate proof that the micellar shell is

8400 J. Phys. Chem. B, Vol. 111, No. 29, 2007 SCHEME 2: Proposed Structure of PS-PMA/PVP Micelles with Low and High Fraction of PVP in Acidic and Basic Solutions: (a) ξ < 0.25, pH >5, (b) ξ < 0.25, pH 0.35, pH >5, and (d) ξ > 0.35, pH 4.5 and positively charged at pH 4.5. Their structure resembles that of pure PS-PMA micelles, except that the PS-PMA/PVP micelles contain a layer of the PMA/PVP complex in the inner part of micellar shells. The incorporation of PVP chains leads to an increase of the micellar molar mass. The micelles start to aggregate below pH ca. 4 due to decreased solubility of nondissociated PMA shell-forming chains and to unsufficiently small amount of stabilizing PVPH+ groups at the shell periphery (PVP forms mainly the complex with PMA). The systems with high PVP contents are unstable in basic and neutral buffers and precipitate as the solutions age in a few weeks. However, fairly monodisperse particles form, when the dispersions are transferred in acidic buffers with pH