pubs.acs.org/Langmuir © 2010 American Chemical Society
Mixed, Multicompartment, or Janus Micelles? A Systematic Study of Thermoresponsive Bis-Hydrophilic Block Terpolymers Andreas Walther,*,†,‡ Christopher Barner-Kowollik,§ and Axel H. E. M€uller*,† †
Makromolekulare Chemie II and Bayreuther Zentrum f€ ur Kolloide und Grenzfl€ achen, Universit€ at Bayreuth, D-95440 Bayreuth, Germany, ‡Molecular Materials, Department of Applied Physics, School of Science and Technology, Aalto University, 02150-Aalto, Finland, and §Preparative Macromolecular Chemistry, Institut f€ ur Technische Chemie und Polymerchemie, Karlsruhe Institute of Technology (KIT), Engesserstr. 18, 76128 Karlsruhe, Germany Received March 24, 2010. Revised Manuscript Received May 3, 2010 We present a systematic investigation of the extent of compartmentalization in micelles formed by a series of bishydrophilic block terpolymers with two outer water-soluble segments. The corona blocks are constructed from poly(ethylene oxide) (PEO) and the thermoresponsive poly(N-isopropyl-acrylamide) (PNiPAAm). The fraction of PNiPAAm is varied to establish its influence on the supramicellar aggregation and corona phase behavior. We demonstrate that—when the collapse of PNiPAAm is triggered—a clustering of micelles into superstructures only occurs when the contour length of the thermoresponsive block is longer than that of the PEO chains. The volume fractions play a minor role. The extent of superstructure formation increases with the amount of heating cycles, pointing to a rearrangement of micelles with a mixed corona into a phase-segregated corona. The collapse of PNiPAAm is exploited to artificially raise the incompatibility and drive phase segregation. A uniform population of biphasic Janus micelles cannot be obtained. After repeated heating cycles, the mixture consists of a range of multicompartment architectures, whose patch distribution can be derived from aggregate structures found in cryo-TEM obtained at high temperature. In the last section, we relate our results to previously studied systems and attempt to derive some generalities. First, we try to answer the question of how likely it is in terms of thermodynamics to obtain truly selfassembled Janus micelles. Furthermore, our results can provide an estimation for the volume ratio or/and block lengths required in micelles composed out of two corona blocks to induce supramicellar aggregation when a hydrophilic-tohydrophobic phase transition is triggered in one of the blocks.
Introduction Soft nanotechnology—represented in the form of small polymer particles—provides some of the best means to tackle present problems in fields ranging from materials science to biomedicine.1-5 Micelles are currently investigated for their capability to selectively transport drug payloads within the body or as catalyst carriers or for the formation of ordered patterns. Their possible application areas and increasing capabilities are, however, further reaching when the architecture is changed from simple micelles composed of one hydrophobic and one hydrophilic block to multicompartment micelles constituted of at least three different blocks with the option of having smart, environmentally sensitive To whom correspondence should be addressed Axel.Mueller@ uni-bayreuth.de;
[email protected]. (1) Sawant, R. M.; Hurley, J. P.; Salmaso, S.; Kale, A.; Tolcheva, E.; Levchenko, T. S.; Torchilin, V. P. Bioconjugate Chem. 2006, 17, 943–949. (2) Rosler, A.; Vandermeulen, G. W. M.; Klok, H.-A. Adv. Drug Delivery Rev. 2001, 53, 95–108. (3) Kataoka, K.; Harada, A.; Nagasaki, Y. Adv. Drug Delivery Rev. 2001, 47, 113–131. (4) Hoffman, A. S.; Stayton, P. S. Prog. Polym. Sci. 2007, 32, 922–932. (5) Rapoport, N. Prog. Polym. Sci. 2007, 32, 962–990. (6) Rao, J.; Luo, Z.; Ge, Z.; Liu, H.; Liu, S. Biomacromolecules 2007, 8, 3871– 3878. (7) Wang, D.; Wu, T.; Wan, X.; Wang, X.; Liu, S. Langmuir 2007, 23, 11866– 11874. (8) Wang, D.; Yin, J.; Zhu, Z.; Ge, Z.; Liu, H.; Armes, S. P.; Liu, S. Macromolecules 2006, 39, 7378–7385. (9) Cai, Y.; Armes, S. P. Macromolecules 2005, 38, 271–279. (10) Liu, S.; Armes, S. P. Langmuir 2003, 19, 4432–4438. (11) Weaver, J. V. M.; Armes, S. P.; B€ut€un, V. Chem. Commun. 2002, 2122– 2123. (12) Liu, S.; Armes, S. P. Angew. Chem., Int. Ed. 2002, 41, 1413–1416.
Langmuir 2010, 26(14), 12237–12246
segments.6-12 “Smart” nanoparticles undergo structural changes while sensing environmental stimuli, such as changes in temperature, pH, or ionic strength.1,13-26 Those particles provide the ability to further increase capabilities in controlled release applications, multicomponent storage and delivery, adhesion control at interfaces, and the controlled formation of superstructures.27-36 In particular, the introduction of non-centrosymmetric Janus (13) Zhang, L.; Bernard, J.; Davis, T. P.; Barner-Kowollik, C.; Stenzel, M. H. Macromol. Rapid Commun. 2008, 29, 123–129. (14) Hsu, Y.-H.; Chiang, W.-H.; Chen, C.-H.; Chern, C.-S.; Chiu, H.-C. Macromolecules 2005, 38, 9757–9765. (15) Topp, M. D. C.; Dijkstra, P. J.; Talsma, H.; Feijen, J. Macromolecules 1997, 30, 8518–8520. (16) Gu, J.; Cheng, W.-P.; Liu, J.; Lo, S.-Y.; Smith, D.; Qu, X.; Yang, Z. Biomacromolecules 2008, 9, 255–262. (17) Zhang, L.; Nguyen, T. L. U.; Bernard, J.; Davis, T. P.; Barner-Kowollik, C.; Stenzel, M. H. Biomacromolecules 2007, 8, 2890–2901. (18) Giacomelli, C.; Le Men, L.; Borsali, R.; Lai-Kee-Him, J.; Brisson, A.; Armes, S. P.; Lewis, A. L. Biomacromolecules 2006, 7, 817–828. (19) Zhang, W.; Shi, L.; Ma, R.; An, Y.; Xu, Y.; Wu, K. Macromolecules 2005, 38, 8850–8852. (20) Rodriguez-Hernandez, J.; Lecommandoux, S. J. Am. Chem. Soc. 2005, 127, 2026–2027. (21) Sfika, V.; Tsitsilianis, C.; Kiriy, A.; Gorodyska, G.; Stamm, M. Macromolecules 2004, 37, 9551–9560. (22) Gil, E. S.; Hudson, S. M. Prog. Polym. Sci. 2004, 29, 1173–1222. (23) Sumerlin, B. S.; Lowe, A. B.; Thomas, D. B.; McCormick, C. L. Macromolecules 2003, 36, 5982–5987. (24) Yusa, S.; Shimada, Y.; Mitsukami, Y.; Yamamoto, T.; Morishima, Y. Macromolecules 2003, 36, 4208–4215. (25) Liu, S.; Weaver, J. V. M.; Save, M.; Armes, S. P. Langmuir 2002, 18, 8350– 8357. (26) Plamper, F. A.; McKee, J. R.; Laukkanen, A.; Nykaenen, A.; Walther, A.; Ruokolainen, J.; Aseyev, V.; Tenhu, H. Soft Matter 2009, 5, 1812–1821. (27) Walther, A.; Andre, X.; Drechsler, M.; Abetz, V.; M€uller, A. H. E. J. Am. Chem. Soc. 2007, 129, 6187–6198.
Published on Web 05/13/2010
DOI: 10.1021/la101173b
12237
Article
Walther et al. Scheme 1. Schematic Drawing of the Self-Assembly of Block Terpolymers
A block terpolymer (a) with two outer hydrophilic blocks can self-assemble into mixed micelles (b), partly demixed or multicompartment micelles (c) or completely demixed, biphasic Janus micelles (d). The amphiphilicity upon triggering the hydrophilic-to-hydrophobic transition may result in superstructure formation.
colloids with a biphasic corona has led to rapid developments of a multitude of applications among multicompartment structures.37 The generation of nanometer-sized Janus micelles remains a challenging task. Major advances have only been reported for relatively large micrometer-sized Janus particles.38-45 Nevertheless, for instance, in materials science, the need for extremely small colloids is evident as the structuring of materials is desired to occur on the mesoscale. Thus, micrometer-sized colloids are not very useful as building blocks, amplifying the need for extremely small particles. So far, the generation of soft matter based Janus particles often involves complex routes with multistep reactions and dissatisfying yields.46,47 The most appealing way to create such colloids would be to use the self-assembly of a single block terpolymer with two outer hydrophilic blocks and an inner hydrophobic block, serving as micellar core. Such an approach can also overcome problems associated with the co-micellization of two diblock copolymers, which usually need attractive forces to allow their coassembly into one micelle. In the case of rather incompatible corona blocks (28) Walther, A.; Hoffmann, M.; M€uller, A. H. E. Angew. Chem., Int. Ed. 2008, 47, 711–714. (29) Walther, A.; Matussek, K.; Mueller, A. H. E. ACS Nano 2008, 2, 1167– 1178. (30) Li, Z.; Hillmyer, M. A.; Lodge, T. P. Macromolecules 2006, 39, 765–771. (31) Lodge, T. P.; Rasdal, A.; Li, Z.; Hillmyer, M. A. J. Am. Chem. Soc. 2005, 127, 17608–17609. (32) Kubowicz, S.; Baussard, J.-F.; Lutz, J.-F.; Thuenemann, A. F.; von Berlepsch, H.; Laschewsky, A. Angew. Chem., Int. Ed. 2005, 44, 5262–5265. (33) Lutz, J.-F.; Laschewsky, A. Macromol. Chem. Phys. 2005, 206, 813–817. (34) Li, Z.; Kesselman, E.; Talmon, Y.; Hillmyer, M. A.; Lodge, T. P. Science 2004, 306, 98–101. (35) Walther, A.; Drechsler, M.; Rosenfeldt, S.; Harnau, L.; Ballauff, M.; Abetz, V.; M€uller, A. H. E. J. Am. Chem. Soc. 2009, 131, 4720–4728. (36) Walther, A.; Drechsler, M.; M€uller, A. H. E. Soft Matter 2009, 5, 385–390. (37) Walther, A.; M€uller, A. H. E. Soft Matter 2008, 4, 663–668. (38) Hong, L.; Jiang, S.; Granick, S. Langmuir 2006, 22, 9495–9499. (39) Roh, K.-H.; Martin, D. C.; Lahann, J. Nat. Mater. 2005, 4, 759–763. (40) Nisisako, T.; Torii, T. Adv. Mater. 2007, 19, 1489–1493. (41) Nisisako, T.; Torii, T.; Takahashi, T.; Takizawa, Y. Adv. Mater. 2006, 18, 1152–1156. (42) Shepherd, R. F.; Conrad, J. C.; Rhodes, S. K.; Link, D. R.; Marquez, M.; Weitz, D. A.; Lewis, J. A. Langmuir 2006, 22, 8618–8622. (43) Dendukuri, D.; Pregibon, D. C.; Collins, J.; Hatton, T. A.; Doyle, P. S. Nat. Mater. 2006, 5, 365–369. (44) Nie, Z.; Li, W.; Seo, M.; Xu, S.; Kumacheva, E. J. Am. Chem. Soc. 2006, 128, 9408–9412. (45) Seo, M.; Nie, Z.; Xu, S.; Mok, M.; Lewis, P. C.; Graham, R.; Kumacheva, E. Langmuir 2005, 21, 11614–11622. (46) Nie, L.; Liu, S.; Shen, W.; Chen, D.; Jiang, M. Angew. Chem., Int. Ed. 2007, 46, 6321–6324. (47) Cheng, L.; Zhang, G.; Zhu, L.; Chen, D.; Jiang, M. Angew. Chem., Int. Ed. 2008, 47, 10171–10174.
12238 DOI: 10.1021/la101173b
A and C, a mixture of AB and BC block copolymers would lead to two individual populations of micelles.48,49 Herein, we present results concerning the occurring phasesegregation of micelles based on bis-hydrophilic block terpolymers with a hydrophobic middle block and two outer hydrophilic blocks. We demonstrate that reorganization in the corona takes place. One corona block is thermoresponsive, which allows the micelles—if phase-segregated—to develop a strong amphiphilicity or to further assemble into superstructures upon switching with temperature. We additionally tuned the block ratio of the two outer blocks to derive a relationship, for when the fraction and block length of the thermoresponsive block, PNiPAAm, is sufficient to induce clustering of individual micelles into superstructures when its hydrophilic-to-hydrophobic transition is triggered. Reasonably, after reaching a critical threshold value, the collapsing PNiPAAm chains cannot be efficiently shielded anymore by the PEO chains against aggregation. The thermoresponsiveness with the collapse of one corona block can be considered as a means to artificially increase the Flory-Huggins interaction parameter, χ, between the segments due to the exclusion of the solvent. This situation is sketched in Scheme 1. The block terpolymer (a) can either self-assemble into mixed (b), patchy (c), or completely biphasic Janus micelles (d). A mixture of those structures may exist and external stimuli may allow alteration of their distribution as indicated by the arrows. An amphiphilic character can be introduced by triggering a hydrophilic-to-hydrophobic transition, i.e., via the thermally induced collapse of PNiPAAm chains. The chains (here, pink) collapse into domains. Depending on the patch size and volume fraction, the multicompartment micelles or Janus micelles can self-assemble into superstructures. Finally, we will relate our model system in which we can tune the corona composition, to previous studies that dealt with micelles possessing two different corona blocks50-59 and attempt to derive some generalities. Aside, our results can provide an (48) Palyulin, V. V.; Potemkin, I. I. Macromolecules 2008, 41, 4459–4463. (49) Halperin, A. J. Phys. (Paris) 1988, 49, 131. (50) Hoppenbrouwser, E.; Li, Z.; Liu, G. J. Macromolecules 2003, 36, 876–881. (51) Gohy, J. F.; Khousakoun, E.; Willet, N.; Varshney, S. K.; Jerome, R. Macromol. Rapid Commun. 2004, 25, 1536–1539. (52) Zhang, W.; Shi, L.; Ma, R.; An, Y.; Xu, Y.; Wu, K. Macromolecules 2005, 38, 8850–8852. (53) Chen, X.; An, Y.; Zhao, D.; He, Z.; Zhang, Y.; Cheng, J.; Shi, L. Langmuir 2008, 24, 8198–8204. (54) Zhang, R.; Liu, G.; Yan, X. J. Am. Chem. Soc. 2005, 127, 15358–15359.
Langmuir 2010, 26(14), 12237–12246
Walther et al.
Article
estimation of the volume ratio or/and block lengths required in micelles, composed out of two corona blocks, to induce supramicellar assemblies when a hydrophilic-to-hydrophobic phase transition is triggered in one of the blocks.
Experimental Section Polymers. The block terpolymers used in this study were prepared by radical addition-fragmentation chain transfer (RAFT) polymerization starting from a PEO macro-RAFT agent according to a procedure published by us earlier.60 Formation of Aggregates. Polymer solutions in dioxane (p.a. grade) (5 g 3 L-1) were exchanged to Milli-Q water in a dialysis cell using dialysis membranes with a MWCO of 5-10 kDa. Typically, 10 mL of polymer solution was exchanged two times with 10 L of Milli-Q water for 24 h each. The polymer concentration at the end of dialysis was determined by quantitatively removing the micellar solution, weighing its quantity, and then calculating the concentration based on volume of the solution and mass of polymer. The final dioxane concentration can be calculated to below 1 10-6 mL dioxane in 1 mL water. Further dilution with water to the final concentrations reduces this value further, typically to below 0.5 10-6 mL dioxane in 1 mL water. For the cryogenic transmission electron microscopy (cryoTEM) studies, a drop of the sample dissolved in water was placed on a lacey transmission electron microscopy (TEM) grid, where most of the liquid was removed with blotting paper, leaving a thin film stretched over the lace. For samples frozen from temperatures above ambient temperature, a custom-built environmental chamber with controlled heating and 100% humidity was used. The specimens were instantly vitrified by rapid immersion into liquid ethane and cooled to approximately 90 K by liquid nitrogen in a temperature-controlled freezing unit (Zeiss Cryobox, Zeiss NTS GmbH, Oberkochen, Germany). The temperature was monitored and kept constant in the chamber during all of the sample preparation steps. After freezing the specimens, the specimen was inserted into a cryo-transfer holder (CT3500, Gatan, M€ unchen, Germany) and transferred to a Zeiss EM922 EF-TEM instrument. Examinations were carried out at temperatures around 90 K. The transmission electron microscope was operated at an acceleration voltage of 200 kV. Zero-loss filtered images (ΔE=0 eV) were taken under reduced dose conditions. All images were registered digitally by a bottom-mounted CCD camera system (Ultrascan 1000, Gatan) combined and processed with a digital imaging processing system (Gatan Digital Micrograph 3.9 for GMS 1.4). Dynamic light scattering (DLS) was performed on an ALV DLS/ SLS-SP 5022F compact goniometer system with an ALV 5000/E cross-correlator and a He-Ne laser (λ0 =632.8 nm). An automatic thermostatting control system was used for the temperature sweeps. The samples were filtered with a 450 nm Nylon filter prior measurements. The heating steps were chosen to either 1 or 2 K. Prior to each measurement, the sample was allowed to equilibrate for 10 min. The data evaluation of the dynamic light scattering measurements was performed with the CONTIN algorithm.
Results and Discussion The system analyzed in the present study consists of a series of bis-hydrophilic block terpolymers, with one inner hydrophobic (55) Ma, R.; Wang, B.; Xu, Y.; An, Y.; Zhang, W.; Li, G.; Shi, L. Macromol. Rapid Commun. 2007, 28, 1062–1069. (56) Hu, J.; Liu, G. Macromolecules 2005, 38, 8058–8065. (57) Schmalz, H.; Schmelz, J.; Drechsler, M.; Yuan, J.; Walther, A.; Schweimer, K.; Mihut, A. M. Macromolecules 2008, 41, 3235–3242. (58) Guiying, L.; L., S.; Rujiang, M.; Yingli, A.; Nan, H. Angew. Chem., Int. Ed. 2006, 45, 4959–4962. (59) Fang, B.; Walther, A.; Wolf, A.; Xu, Y.; Yuan, J.; M€uller, A.; H., E. Angew. Chem., Int. Ed. 2009, 48, 2877–2880. (60) Walther, A.; Millard, P.-E.; Goldmann, A. S.; Lovestead, T. M.; Schacher, F.; Barner-Kowollik, C.; M€uller, A. H. E. Macromolecules 2008, 41, 8608–8619.
Langmuir 2010, 26(14), 12237–12246
Table 1. Overview of the PEO-b-PnBuA-b-PNiPAAm (EBN) Block Terpolymers Employed in the Present Study60 sample codea
weight fractionsb
103 Mn (PDI)c
E114B105N56 E20B54N26 25.1 (1.05) E18B49N33 28.1 (1.07) E114B105N82 E16B43N41 31.9 (1.07) E114B105N116 E14B36N50 37.5 (1.07) E114B105N165 E13B35N52 38.9 (1.05) E114B105N178 E12B33N55 40.8 (1.04) E114B105N195 E9B23N68 58.4 (1.09) E114B105N350 a Subscripts denote the number-average degree of polymerization of the respective blocks, DPn. b Subscripts denote the weight fractions of each block. The weight fractions are not used as sample codes in the following. c Number-average molecular weights calculated based on the true molecular weight of the PEO macro-CTA (MALDI-ToF) and the weight fractions of the block terpolymers determined by 1H NMR. The polydispersity indices were obtained by GPC in N-methylpyrrolidone using a polystyrene calibration curve.
Chart 1. Structure of Block Terpolymer Used
and two outer hydrophilic blocks. We selected poly(n-butyl acrylate) (PnBuA) as hydrophobic middle block as it ensures sufficient dynamics due to the low glass transition temperature (Tg ≈ -46 C). Poly(ethylene oxide) (PEO) and poly(N-isopropylacrylamide) (PNiPAAm) were selected as outer blocks. Even though there is in principle the possibility of hydrogen bonding between both blocks, they are not known to strongly and specifically interact in solution. PNiPAAm and also PEO may, however, form hydrogen-bonded complexes with polymers containing significantly more acidic moieties.61 In general, polyelectrolytes and strongly interacting blocks would exhibit a strong tendency for a mixed corona. PNiPAAm is a thermoresponsive polymer with lower critical solution temperature (LCST) behavior, exhibiting cloud points just below the human body temperature. The synthesis of these PEO-PnBuA-PNiPAAm block terpolymers can be accomplished via reversible addition-fragmentation chain transfer (RAFT) polymerization, as reported in our preceding publication.60 We chose the RAFT technique for its superior performance in the polymerization of a wide variety of polar and apolar monomers and excellent blocking efficiencies during the polymerization to complex block terpolymers.62 We prepared a series of block terpolymers simply by taking samples from the same polymerization mixture when polymerizing the third block. The living/controlled polymerization method enables predicting the evolution of the block length during the polymerization. With increasing reaction time, the third block grows, and taking samples during this process allows for a continuous tuning of the volume fraction of one of the two outer blocks while keeping the inner hydrophobic block constant. The herewith synthesized polymers are shown in Table 1, and Chart 1 shows the structure. All polymers are characterized by a narrow (61) Koussathana, M.; Lianos, P.; Staikos, G. Macromolecules 1997, 30, 7798– 7802. (62) Barner-Kowollik, C. Handbook of RAFT polymerization; Wiley-VCH: Weinheim, 2008.
DOI: 10.1021/la101173b
12239
Article
Walther et al.
Figure 1. DLS temperature ramps of the first group (c = 5 g/L). Plots c,d summarize the important data points of plots a,b. Plots a,b show the evolution of hydrodynamic radii as a function of temperature for 4 cycles for the following polymers: E114B105N56 (a,c), E114B105N82 (b,d). The DPn of the end blocks are shown on top of the figure (a,b): The temperature ramps range from 25 to 45 C and back to 25 C. Filled and open symbols describe the heating and cooling trace, respectively. Symbols for ramps 1-4 are shown as inset in plot a. The lower graphs (c,d) show the evolution of the hydrodynamic radii as a function of the heating cycle. The symbols 9, b, 2 correspond to the value determined before the ramp (25 C), at 45 C, and at the end of each run (25 C), respectively. The trend lines in plots c,d serve to guide the eye.
molecular weight distribution with polydispersity indices below 1.09. The PNiPAAm block length was varied between 56 and 350 units, whereas the PEO block is constant at 114 units. Thus, the block length and the volume fraction of the PNiPAAm block range from shorter or comparable to longer and much larger than the PEO block, respectively. For a detailed description and characterization of the synthesized blocks employed in the present study, the reader is referred to our previous report.60 Consequently, the terpolymer system described above allows a systematic study of the effect of block length and volume fraction on the corona phase behavior of the resulting micelles. The effect of repeated temperature cycles on the corona phase behavior can be deduced, and the onset of supramicellar aggregation as a function of PNiPAAm block can be investigated. All polymers do not readily dissolve in water despite the low glass transition temperature of PnBuA. Several weeks are necessary to obtain stable colloidal solutions. Although such an observation shows that the micelles principally exhibit dynamic character, the process cannot be considered time-efficient. Therefore, the micelles were prepared by dialysis from the nonselective solvent dioxane into water. All polymers form spherical micelles due to their large hydrophilic-to-hydrophobic ratios,63,64 which, according to the weight fractions in bulk, is between 1 and 3.4 for the smallest and (63) Zhulina, E. B.; Borisov, O. V. Macromolecules 2002, 35, 9191–9203. (64) Riess, G. Prog. Polym. Sci. 2003, 28, 1107–1170.
12240 DOI: 10.1021/la101173b
largest block terpolymer, respectively. The z-average hydrodynamic radii range from 20 to 35 nm. Considering the molecular size of the block terpolymers, the structure can be assigned to spherical micelles. The spherical structure is confirmed by cryogenic transmission electron microscopy (cryo-TEM, see below). To study the temperature-dependent phase behavior, we decided to primarily employ dynamic light scattering (DLS) measurements in a computer-controlled thermostatted cell. We chose to restrict the measurements to 90 and use the CONTIN algorithm to evaluate the autocorrelation functions. The resulting intensityweighted distribution of apparent hydrodynamic radii is used as a precise and reliable tool to monitor evolutions of micelle populations, aggregate formation, and growth of micelles. On account of the expected polydispersity both in size and in aggregate type, we refrained from recording angular dependent DLS data. A moderate polydispersity can already make a reliable fitting of angulardependent DLS data difficult, hence complicating unambiguous conclusions. In contrast, we decided to support the conclusions by cryo-TEM measurements obtained at high and ambient temperatures. This quasi in situ imaging method allows us to check for the supramicellar aggregate structures in solution and provides complementary information to the evolution of sizes of the micelles and their aggregate. Temperature-Dependent Dynamic Light Scattering. The heating steps in the DLS temperature ramps were typically set to 2 K and the equilibration time after a specific temperature had Langmuir 2010, 26(14), 12237–12246
Walther et al.
Article
Figure 2. DLS temperature ramps (c = 5 g/L) for the second series. Plots c,d summarize the important data points of plots a,b. Plots a,b show the evolution of hydrodynamic radii as a function of temperature for at least 4 cycles for the following polymers: E114B105N116 (a,c), E114B105N165 (b,d). The DPn of the end blocks are shown on top of the figure (a,b): The temperature ramps range from 25 to 45 C and back to 25 C. Filled and open symbols describe the heating and cooling traces, respectively. Symbols for ramps 1-7 and 10 are shown as inset. The lower graphs (c,d) show the evolution of the hydrodynamic radii as a function of the heating cycle. The symbols 9, b, 2 correspond to the value determined before the ramp (25 C), at 45 C, and at the end of each run (25 C), respectively. The dashed lines highlight the hysteresis phenomenon. The trend lines in plots c,d serve to guide the eye and clearly show the increase in size.
been reached was 10 min. Preliminary experiments had shown that the evolution of hydrodynamic radii does not depend on the heating rate. Traces of heating steps of 2 K superimpose with heating steps of 1 K. Therefore, we used 2 K steps, as this rate still gives sufficient data points but is twice as time-efficient.60 The temperature-responsive behavior of the micelles, when subjected to repeated temperature ramps, is depicted in Figures 1-3. Their behavior can be clearly divided into three groups. Before turning to the main scope of the present study, it should be noted that the first heating cycle involves a relaxation phenomenon. This peculiarity is most pronounced for the longest PNiPAAm chains. The populations of the micelles, as obtained after dialysis, contain some bimodality that is most pronounced for the longest chains of the third block. During the first heating cycle, the smaller micelles present in the solution mixture are completely transferred into slightly larger ones, resulting in a unimodal distribution.60 The bimodality or broadened distribution after dialysis may originate from a kinetic trapping of some smaller micelle species during the dialysis process. After this first temperature cycle, all polymers exhibit unimodal micelle distributions of similar size, and thus, the samples are even more comparable. The overall temperature-responsive behavior of the various micelles obtained from the different block terpolymers can be Langmuir 2010, 26(14), 12237–12246
divided into three different groups. These three groups exhibit distinctly different response in dependence of the block length of the third block, PNiPAAm. Figures 1 and 2 display the temperature ramps and the evolution of hydrodynamic radii as a function of the number of heating cycles for the first two groups, respectively. The first set of terpolymers possesses significantly shorter segments of the thermoresponsive PNiPAAm. They show nearly constant hydrodynamic radii for repeated temperature cycles (Figure 1). Temperature-induced aggregation is clearly absent, as this would result in a drastic increase of hydrodynamic radii. During the phase transition, the small PNiPAAm chains collapse, but an intermicellar interaction still does not take place due to the smaller length of the collapsed PNiPAAm chains as compared to PEO. The PEO chains are still sufficiently long to shield the collapsed PNiPAAm patches against aggregation. While E114B105N56 exhibits no changes in size, a slight increase of the structures can be monitored for E114B105N82. Consequently, there is a slight growth of the micelles with each temperature cycle. This growth is, however, less significant in comparison to the second set of terpolymers (Figure 2). Whereas the increase in micelle size is only a few percent (ca. 15%) for four temperature cycles of E114B105N82, the micelles of the second set double their size for the same amount of repetitions. DOI: 10.1021/la101173b
12241
Article
Walther et al.
Figure 3. DLS temperature ramps (c = 5 g/L). Plots 3d-f summarize the important data points of plots 3a-c. Plots a-c show the evolution of hydrodynamic radii as a function of temperature for 4 cycles for the following polymers: E114B105N178 (a,d), E114B105N195 (b,e), E114B105N350 (c,f). (a-c) The temperature ramps range from 25 to 45 C and back to 25 C. Filled and open symbols describe the heating and cooling trace as shown as inset in (a). The lower graphs (d-f) show the evolution of the hydrodynamic radii as a function of the heating cycle. The symbols 9, b, 2 correspond to the value determined before the ramp (25 C), at 45 C, and at the end of each run (25 C), respectively. The encircled areas in plots a-c correspond to the hydrodynamic sizes at room temperature before each start of the temperature ramp. The trend lines in plots d-f highlight the widening gap between the hydrodynamic radii at room and elevated temperature.
The second group is composed of micelles with a majority weight fraction of PNiPAAm within the corona. The weight fraction of PNiPAAm exceeds that of PEO already by a factor of 2.5 to 3.5, thus occupying major parts of the corona. Note however, that the actual contour length of PEO is 1/3 longer than for PNiPAAm, as its backbone is composed of three atoms. Thus, the PEO and PNiPAAm chains are of almost equal contour length for E114B105N165. We can now observe a different behavior for these micelles of the second group. The micelles start growing linearly when subjected to repeated temperature ramps. This can nicely be seen in the trendlines in Figure 2c,d as compared to Figure 1c. The growth itself is linear for 10 repetitions as can be seen for E114B105N116 (Figure 2c). The overall growth of the micelles depends on the chain length of the PNiPAAm block as deduced by observing all trendlines in Figure 1d and Figure 2c,d. The growth is more pronounced for longer PNiPAAm chains. Whereas a block length, DPn = 165 for PNiPAAM, leads to an increase to ÆRhæz = 50 within 4 cycles (Figure 2d), a shorter block of DPn =116 requires almost twice as many cycles (7) to reach the same size (Figure 2c). The origin of the growth of the micelles is triggered by the dramatic change of the hydrophilic-to-hydrophobic balance when the PNiPAAm chains collapse. A decrease in this ratio favors the formation of aggregates of lower curvature, i.e., larger micelles. Thus, the collapse of PNiPAAm has a more significant effect for longer chains of the temperature-responsive block and leads to the more pronounced growth observed. A growth of the micelles stabilizes the shape to the new energetic requirements induced by the collapse of PNiPAAm. We suggest that mainly the exchange of unimers is responsible for the growth of the micelles. There are several considerations that support such a notion: (a) We are concerned with a system of dynamic micelles, implying that the glass transition temperature 12242 DOI: 10.1021/la101173b
of the hydrophobic block is sufficiently lower than ambient temperature (Tg =-46 C). In fact, the polymers undergo direct dissolution in water. Detailed investigation by Colombani et al.65 point to the fact that poly(n-butyl acrylate)-block-poly(acrylic acid) micelles are dynamic in terms of unimer exchange but may be slow on the experimental time scale in adapting to external stimuli at room temperature. (b) Herein, the temperature of the system is raised significantly above room temperature, which increases the dynamics of the system and possibly the solubility of the block terpolymer. (c) The dynamics and solubility of unimers is expected to be greater for bis-hydrophilic block terpolymers with two end blocks than for an AB diblock copolymer with a hydrophobic terminus of similar composition. (d) Importantly, the small and steady increase for the polymers presented in Figures 1d and 2c,d cannot be caused by fusion. The highest probability for fusion via overcoming the brush-like repulsive corona of the micelles would be at its weakest state, meaning when PNIPAAm is collapsed and PEO has to shield a larger hydrophobic structure. However, there is no indication for clustering above the cloud point. Even annealing the structures at 45 C for a longer time does not allow for growth, pointing to an absence of fusion events. The small size increase at high temperature (typically 5-10 nm) can more likely be attributed to a stretching of the PEO chains at the core-corona boundary, whose available interface with the PnBuA core is reduced by the collapsed PNiPAAm. (e) For the following third set of micelles, we will also show that reversible aggregation and not fusion takes place upon triggering the collapse of PNIPAAm. (65) Colombani, O.; Ruppel, M.; Burkhardt, M.; Drechsler, M.; Schumacher, M.; Gradzielski, M.; Schweins, R.; M€uller, A. H. E. Macromolecules 2007, 40, 4351–4362.
Langmuir 2010, 26(14), 12237–12246
Walther et al.
The temperature ramps contain a hysteresis phenomenon that can be seen by comparing the values obtained at the end of each run (25 C, 2) with those obtained just before starting the new cycle (25 C, 9) (dashed lines in Figure 2c,d). The micelles shrink some nanometers during this process. Typically, several hours (1-3 h) of waiting time were allowed between two temperature ramps, and the ongoing relaxation in size therein can be related to some more time-consuming repacking of kinetically trapped chain conformations in both corona and core. Prolonged aging, for instance, overnight, does not further change the size. From all these observations, we conclude that these two sets of micelles do not cluster into superstructures. The overall structure is sufficiently shielded by the PEO chains, regardless of the amount of temperature cycles. On the other hand, however, they start growing according to the change of the hydrophilic-to-hydrophobic ratio. The growth is continuous, steady and in small intervals, pointing to some control in these steps. Continuing this growth process results in larger micelles, as seen by DLS, and may eventually lead to a shape change toward elongated micelles. Although, it is generally interesting to study this well-developed growth process in more detail, here we want to set the scope on the corona phase behavior and the appearance of higher level aggregation of multicompartment micelles. We now turn to the third set of micelles. This group again behaves very differently and provides new insights into the behavior of responsive coronas composed out of two polymers. All polymers possess PNiPAAm end blocks exceeding the PEO chain in terms of maximum contour length and weight fraction. The DLS data are shown in Figure 3. The striking difference compared to the other two groups of micelles is the drastic increase in hydrodynamic size upon passing the cloud point of PNiPAAm. The z-average hydrodynamic radii above the cloud point already double during the first temperature cycle and the gap between the sizes at 25 and 45 C further widens (observe the trend lines in Figure 3d-f ). For instance, the fourth heating cycle for E114B105N350 leads to an approximate 5-fold increase in aggregate size from ÆRhæz ≈ 30 to 150 nm upon passing the cloud point. The micelle size at room temperature stays nearly constant, and a pronounced growth as for the micelles of the second group does not occur. The encircled areas in Figure 3a-c point to the hydrodynamic sizes at room temperature before the start of every cycle. Thus, the behavior is significantly different compared to sets one and two that only show an increase of 3-10 nm when the collapse of PNiPAAm is triggered. This large increase in size for the third group can by no means be explained by a slight chain stretching of PEO. Clearly, a higher level aggregation of the micelles into superstructures takes place. This will be further confirmed by cryo-TEM. The behavior is well-developed, follows a trend, and is thus not a chaotic phenomenon (Figure 3d-f ). The hydrodynamic sizes at high temperature increase almost linearly for the cycles investigated. Although DLS only yields apparent hydrodynamic radii, a clear increase in size can be unambiguously concluded. At a given cycle, the sizes at elevated temperatures are larger for the polymers with the longer PNiPAAm blocks. A clustering of the micelles takes place as promoted by the increasing length and volume fraction of the PNiPAAm block. The PEO chains are no longer sufficient to shield the large hydrophobic quantities of PNiPAAm. Considering that the micelles return to similar size at room temperature, independent of the temperature cycle, we can rule out fusion of various hydrophobic cores upon heating. As the most important consequence of the widening gap and increased aggregation, there must be an effect of the heating cycles Langmuir 2010, 26(14), 12237–12246
Article
on the corona structure of the micelles, as they serve as the essential building blocks for the aggregates. Initially, the micelles seem to only form a small number of loose aggregates as seen by the moderate increase in size upon passing the cloud point. On the contrary, larger aggregates are formed later on. Such a behavior can only be reasonably explained by an increased stickiness and patchiness of the micelles. The term patchiness is used to describe the presence of larger compartments due to phase separation of the two blocks in the corona of the micelle (see Scheme 1b-d). The tendency of micelles to interact with each other and cluster into supramicellar assemblies increases with the domain size of collapsed PNiPAAm on the surface of the PnBuA core. This is simply due to a less effective shielding of the PEO arms around increasingly large PNiPAAm domains. Consequently, the micelles undergo a transition from a presumably rather mixed micelle—or a micelle of very low patchiness—to a micelle with increasing size of patches and increased stickiness. The increase in size of the PNiPAAm domains can be understood considering simple surface tension effects. Upon collapse of the PNiPAAm chains, domains of it are formed on the PnBuA core. Since the then-hydrophobic PNiPAAm favors a minimum interface with water and the PnBuA core, there is a strong tendency to form increasingly large domains with a better, lower surface-to-volume ratio. A rearrangement of PNiPAAm chains on the PnBuA core is possible (yet slow) due to the low glass transition temperature of the PnBuA. The chains within the core must rearrange significantly to accommodate for the changed localizations of the attached PEO and PNiPAAm chains. This accounts for a high cooperativity in this phase transition. Consequently, the important question arises whether an initially mixed micelle can be transformed into a completely biphasic Janus micelle. We tried to assess the mixing of the coronal chains by 2D 1H-1H NOESY-NMR, which can be used to probe mixed and phase-segregated coronas via their intermolecular polarization transfer. Unfortunately, despite using 800 MHz NMR instruments,57 even the micelles directly after dialysis, which were most likely to show a mixed corona or very small patches, did not show any cross peaks. We attribute this to unfavorably high chain flexibility and overly fast dynamics within the corona. Thus, even a mixed corona can remain undetectable due to fluctuations during the characteristic time scales of the measurement. A mixed corona was found earlier by Voets et al.66 for complex core coacervate micelles with PEO and PNiPAAm corona composed of two diblock copolymers. In any case, it would not be possible to distinguish between a patchy micelle and a Janus micelle, as the interfaces are small in both cases. Cryo-TEM. To directly image the aggregates formed and the phase segregation of the corona, we performed cryo-TEM measurements for various samples. We placed the emphasis on the important question of how the supermicellar aggregates are structured and to confirm the presence of spherical micelles at some stages of the process. Figure 4 displays the images obtained for two polymers of the first and third groups at room temperature and images of all groups at high temperature. The micelles are all similar at room temperature at the initial conditions. The images corroborate and extend the conclusions from the DLS measurements. First, DLS indicated spherical micelles for all block copolymers. Those can be confirmed by cryo-TEM measurements of two polymers (E114B105N56 and E114B105N116). Considering that DLS yields a z-average hydrodynamic radius, (66) Voets, I. K.; Moll, P. M.; Aqil, A.; Jerome, C.; Detrembleur, C.; Waard, P. d.; Keizer, A. d.; Cohen Stuart, M. A. J. Phys. Chem. B 2008, 112, 10833–10840.
DOI: 10.1021/la101173b
12243
Article
Walther et al.
Figure 4. Representative cryo-TEM images of E114B105N56 (a) and E114B105N116 (d) vitrified from room temperature after the first cycle. Micrographs obtained for samples of E114B105N56 (b,c), E114B105N116 (e,f), and E114B105N178 (g-i) vitrified from 45 C at the fourth heating cycle.
the sizes of the micelles correspond to the values obtained from DLS. The inset in Figure 4a visualizes a corona of PNiPAAm and PEO surrounding the central PnBuA core. At higher temperatures, above the cloud point of PNiPAAm, the situation is different. The sample of the first (Figure 4b,c) and second (Figure 4d,e) group obtained from 45 C exhibit wellseparated micelles; fusion of particles does clearly not take place upon triggering the hydrophilic-to-hydrophobic transition of the PNiPAAm block. The absence of intermicellar aggregation could already be concluded from the DLS data, which only showed a moderate increase of 3-10 nm upon passing the cloud point of PNiPAAm. The equal spacing between the micelles in Figure 4b,c is a sign for efficient repulsion between the colloids. A corona can no longer be detected for the first and second groups (Figure 4b,c, e,f), which relates to a lower density of materials in the corona as the PNiPAAm is collapsed. Note that a PEO corona is usually very difficult to efficiently visualize in cryo-TEM. The cores of the micelles are enlarged and they are asymmetric. Several micelles can be seen with bumps or elongated character. The asymmetry is likely to originate from PNiPAAm patches collapsed onto the micelle. PnBuA and collapsed PNiPAAm cannot be distinguished 12244 DOI: 10.1021/la101173b
due to their very similar densities. The micelles of the second group are slightly larger than the micelles of the first group after four heating cycles, coinciding with their observed growth process (Figure 2c). Strikingly and as expected from the DLS data, the images obtained for a polymer of the third group are remarkably different (Figure 4g-i). Herein, the micelles are connected into superstructured branched chain-like assemblies. The branching points mostly connect three linear chains via two motifs. Either the corona (dotted circle) or the core (dashed circle) can be the focal point of a branching point in Figure 4g-i. Besides, there are also some square-like arrangements and various end-caps to the linear chains. Clearly, the connectivity of each structure originates from its patch distribution. We had recently shown this for a system of micelles of a single block terpolymer, having a fluorinated core and outer blocks of poly(tert-butoxystyrene) and poly(tert-butyl methacrylate), in organic media.59 Therein, precise staining of the compartments allowed discernment of the individual patches within the coronas of the micelles and the superstructures. The amount and the size of the patches give rise to its location as building block in the supramicellar assemblies. Langmuir 2010, 26(14), 12237–12246
Walther et al. Scheme 2. Patch Distribution of the Micelles and Their Localization within the Self-Assembled Chain-Like Structuresa
a Three types of patchy multicompartment micelles (a-c) and Janus micelle (d).
Herein, staining is not possible in cryo-TEM, and TEM imaging is limited anyway as the low Tg of the PnBuA block causes an insufficient stability and film formation of the micelles during deposition on a TEM grid. Reasonably, the PNiPAAm patches are the motifs connecting the micelles once they become hydrophobic and sticky. Scheme 2 summarizes how various multicompartment micelles can interact with their patches as supramicellar units in the chainlike assemblies.59 Patch Distribution. According to the assemblies found, building blocks for linear chains somehow prevail. Such an observation implies that micelles with two compartments opposing each other (Scheme 2a) are energetically preferred. Janus micelles (Scheme 2d) can be found as end-caps to the various linear chains. The structures shown as Scheme 2a-c are potential building blocks for branching points. Several factors contribute to the extent of the patchiness of micelles with two corona blocks. Following thermodynamic considerations, there is a strong interplay between the enthalpic reason for two polymers to phasesegregate in a micellar corona and the entropic effect of keeping them in a mixed state. Furthermore, the influence of repacking and ordering the chains within the micelle core can result in a significant entropic penalty. Clearly a fully demixed corona is preferred if both chains do not have any favorable interaction and for nonionic corona blocks. Due to the inherent electrostatic repulsion, polyelectrolyte chains would have a very strong tendency to spread around the full corona. Here, in this system, using a thermoresponsive block, we can conclude that reorientations within the corona takes place during the heating cycles (especially concluding from set 3). Importantly, the collapse of the thermoresponsive block can be used to artificially and drastically increase the incompatibility between the blocks. The exclusion of solvent from one of the corona blocks increases the incompatibility. Most likely, it would be difficult to find two well-dissolved corona blocks showing a superior incompatibility than when one block can exclude the solvent and the other one stays dissolved. Hence, a thermoresponsive block and repeated temperature cycles are powerful tools when aiming at high incompatibility and multicompartment micelles. The increase of the hydrodynamic radii of the assemblies of the third group for each heating cycle shows that an increasing Langmuir 2010, 26(14), 12237–12246
Article
amount of micelles is converted from mainly mixed micelles into micelles with preferentially two opposing patches. These give rise to the linear assembly, and importantly, only linear assembly increases the hydrodynamic size drastically (Figure 3). A full transition into Janus micelles does not take place. Reasonably, the gain in enthalpy from reducing the interface of two patches into one single patch of each type is rather low, whereas the entropic loss in both corona and core may be rather significant. This may prevent a complete transformation into Janus micelles and a direct self-assembly into those. In order to draw a more general conclusion on the corona phase behavior of multicomponent micelles, we need to consider already published systems. Interestingly, we recently found a very similar behavior for the micelles of a single block terpolymer having a fluorinated core and end blocks of poly(tert-butoxystyrene) and poly(tert-butyl methacrylate), in organic media.59 Furthermore, only patchy multicompartment micelles of block terpolymers or mixtures of diblock copolymers were obtained for a variety of different systems50-58,67 Despite different chemistry among these studies, there seems some generality to the observations. We now have screened a system of thermoresponsive micelles with large incompatibility, as induced by the collapse of one block, in a wide composition range, and we were not able to create a uniform population of Janus micelles. Although there will certainly be developments toward the preparation of Janus micelles via direct self-assembly, we can still draw some conclusions by summarizing published systems and our systematic investigations. It can be said that simple and direct dissolution of mixtures of diblock copolymers or a block terpolymer toward a pure fraction of Janus micelles is a very challenging task. The question of whether self-assembled systems of two diblock copolymers or block terpolymers can easily form Janus micelles has a more negative answer at this point. The development of a mixture of different multicompartment micelles and only a fraction of Janus micelles is more likely to occur. The only systems, suspected to be a Janus structure, were obtained by Voets et al. by forced coassembly of two oppositely charged diblock copolymers into a complex-core coacervate micelle with PEO and poly(acrylamide) as end blocks.68,69 The conclusions therein were largely supported by 2D 1H-1H NOESYNMR and the unusual asymmetry of the resulting particle, yet not by direct microscopy of the patch distribution or small angle neutron scattering.70 Theoretical and modeling attempts to describe the phase segregation of mixed micelles into phase-segregated micelles mostly yield an unreasonably high Flory-Huggins parameter necessary for phase segregation.71 Furthermore, the question of the existence of multicompartment micelles with various patches and the consideration of packing constraints inside the core are so far often neglected. Thus, in light of these and already published results, there is some need for refining the theoretical and modeling attempts.
Conclusions We have systematically studied the influence of the length of a thermoresponsive block on the behavior and phase segregation of (67) Xie, D.; Xu, K.; Bai, R.; Zhang, G. J. Phys. Chem. B 2007, 111, 778–781. (68) Voets, I. K.; de Keizer, A.; De Waard, P.; Frederik, P. M.; Bomans, P. H. H.; Schmalz, H.; Walther, A.; King, S. M.; Leermakers, F. A. M.; Cohen Stuart, M. A. Angew. Chem., Int. Ed. 2006, 45, 6673–6676. (69) Voets, I. K.; Fokkink, R.; De Keizer, A.; May, R. P.; De Waard, P.; Cohen Stuart, M. A. Langmuir 2008, 24, 12221–12227. (70) Futterer, T.; Vliegenthart, G. A.; Lang, P. R. Macromolecules 2004, 37, 8407–8413. (71) Charlaganov, M.; Borisov, O. V.; Leermakers, F. A. M. Macromolecules 2008, 41, 3668–3677.
DOI: 10.1021/la101173b
12245
Article
micelles based on bis-hydrophilic block terpolymers with two different outer hydrophilic blocks. The LCST behavior of one block segment is used to artificially increase the incompatibility within the corona-forming blocks. Phase separation of the corona can be triggered by the collapse of the PNiPAAm blocks via temperature raise. The extent of phase separation can be increased by repeating the heating cycles. Regardless of the length of the thermoresponsive block, a full transition to Janus micelles could not be induced. We attribute this to the energetic penalties in the core and the very minor energetic differences between multicompartment and Janus micelles inside the corona, which cannot counterbalance the entropic penalty. The ability of micelles to organize into superstructures or to be potentially used as interfacial stabilizers depends on the ratio of the two corona-forming blocks. As long as the PEO chains are longer or of similar length as the thermoresponsive block, a sufficient shielding of the complete micelle occurs and aggregation is prevented. This is even the case when PEO is the minor weight fraction. Our systematic study, together with results on individual systems known in the literature, points to the fact that truly self-assembled
12246 DOI: 10.1021/la101173b
Walther et al.
Janus micelles based on block terpolymers or mixtures of diblock copolymers are extremely difficult to target. Most likely, only a certain fraction of an individual system can be converted to completely biphasic particles. Consequently, materials and bioscience may have to rely on some of the more complex ways to generate the attractive properties of Janus particles in a very defined way. These results should promote refined theoretical and modeling attempts to describe such systems in a better and more comprehensive way. Acknowledgment. The authors are grateful for financial support from the Australian Research Council (ARC) and the Deutsche Forschungsgemeinschaft (DFG) for financial support in the form of a an International linkage project. A. Walther acknowledges support from the Bavarian Elite Support Program and C.B.-K. acknowledges continued funding in the context of the Excellence Initiative for leading German Universities. We thank K. Schweimer (Department of Biopolymers, University of Bayreuth) for the attempts to record 2D 1H-1H NOESY NMR spectra as well as Oleg Borisov (Universite de Pau) for discussions.
Langmuir 2010, 26(14), 12237–12246
