pH-Responsive Core−Shell Particles and Hollow Spheres Attained by

Langmuir 2005, 21, 1531-1538

1531

pH-Responsive Core-Shell Particles and Hollow Spheres Attained by Macromolecular Self-Assembly Youwei Zhang,†,‡ Ming Jiang,*,† Jiongxin Zhao,‡ Zhouxi Wang,‡ Hongjing Dou,† and Daoyong Chen† Department of Macromolecular Science and The Key Laboratory of Molecular Engineering of Polymers, Ministry of Education of China, Fudan University, Shanghai 200433, China and College of Materials Science and Engineering, Donghua University, Shanghai 200051, China Received August 22, 2004. In Final Form: November 17, 2004 According to our “block-copolymer-free” strategy for self-assembly of polymers, noncovalently connected micelles (NCCM) with poly(-caprolactone) (PCL) as the core and poly(acrylic acid) (PAA) as the shell in aqueous solutions were attained due to specific interactions between the component polymers. The micellar structure was then locked in by the reaction of PAA with diamine. Afterward, hollow spheres based on PAA network were obtained by either core degradation with lipase or core dissolution with dimethylformamide of the cross-linked micelles. The cavitation process was monitored by dynamic light scattering, which indicated a mass decrease and size expansion. The hollow structure is confirmed by transmission electron microscopy observations. The resultant hollow spheres are pH- and salt-responsive: there is a substantial volume increase when pH changes from acid to base, and vice versa. The volume change takes place dramatically over the pH-range from 5.8 to 7.5. Furthermore, this volume-pH-dependence is found to be completely reversible provided the effect of ionic strength is excluded. The volume change can be adjusted by changing the shell thickness and the cross-linking degree of the hollow spheres. The salt effect on the hollow sphere size depends on pH: with increasing salt concentration the size shows an increase, a decrease, and a little change in acidic, basic, and neutral media, respectively.

Introduction Polymeric assembled materials with well-defined structure on sub-micrometer or nanometer scales have been a subject of intensive studies due to their broad potential applications. Among these materials, hollow spheres have attracted special attention because of their unusual large capacity of encapsulation.1 Self-assembly of block copolymers followed by shell cross-linking and subsequent core degradation is frequently used to prepare shape- and sizeinvariant hollow spheres.2 “Layer-by-layer (LBL)” technique3 using alternate depositions of oppositely charged species on various templates and subsequent sacrificing core can produce diverse hollow spheres. In addition, using vesicles or inorganic particles as templates for in situ polymerization to produce hollow spheres have been reported.4 Recently, some new approaches to the hollow spheres have been reported: assembly of amphiphilic graft copolymers at the oil-water interface,5a assembly of * To whom correspondence should be addressed. E-mail: [email protected]. † Fudan University. ‡ Donghua University. (1) (a) Meier, W. Chem. Soc. Rev. 2000, 5, 295. (b) Jenekhe, S. A.; Chen, X. L. Science 1998, 279, 1903. (c) Marinakes, S. M.; Anderson, M. F.; Ryan, J. A.; Martin, L. D.; Feldheim, D. L. J. Phys. Chem. B 2001, 105, 8872. (2) (a) Ding, J. F.; Liu, G. J. Chem. Mater. 1998, 10, 537. (b) Ding, J. F.; Liu, G. J. J. Phys. Chem. B 1998, 102, 6107. (c) Steward, S.; Liu, G. J. Chem. Mater. 1999, 11, 1048. (d) Huang, H. Y.; Remsen, E. E.; Kowalewski, T.; Wooley, K. L. J. Am. Chem. Soc. 1999, 121, 3805. (e) Zhang, Q.; Remesen, E. E.; Wooley, K. L. J. Am. Chem. Soc. 2000, 122, 3642. (f) Sanji, T.; Nakatsuka, Y.; Ohnishi, S.; Sakurai, H. Macromolecules 2000, 33, 8524. (g) Nardin, C.; Hirt, T.; Leukel, J.; Meier, W. Langmuir 2000, 16, 1035. (3) (a) Caruso, F.; Caruso, R. A.; Mo¨hwald, H. Science 1998, 282, 111. (b) Donath, E.; Sukhorukov, G. B.; Caruso, F.; Davis, S. Angew. Chem., Int. Ed. 1998, 37, 2202. (c) Caruso, F.; Lichtenfeld, H.; Donath, E.; Mo¨hwald, H. Macromolecules 1999, 32, 2317. (d) Caruso, F.; Mo¨hwald, H. J. Am. Chem. Soc. 1999, 121, 6039. (e) Gao, C.; Donath, E.; Mo¨hwald, H.; Shen, C. Angew. Chem., Int. Ed. 2002, 41, 3789.

nanoparticles of silica and gold cooperatively with block copolypeptides,5b and formation of core-shell particles based on polycondensation of organosilanes followed by removal of nonbonded chains from the core.5c To make the load and release of the resultant hollow spheres controllable, it is desirable to develop further socalled smart hollow spheres, which can switch their structure reversibly from a closed to an open state with the help of external stimuli. Recently, several such hollow spheres have been reported to be responsive to changes in temperature,4f pH,6 and salt,7 etc. In fabricating stimuli-responsive hollow spheres in water, poly(acrylic acid) (PAA) is a good candidate because its chain conformation evidently changes with the degree of the protonation of the carboxyl groups, which depends on the acidity and slat concentration of the medium. Taking advantage of these properties, PAA hydrogel exhibiting a dramatic volume change with the external pH and ionic strength has been extensively studied.8 In (4) (a) Hotz, J.; Meier, W. Langmuir 1998, 14, 1031. (b) Meier, W. Chimia 1999, 53, 214. (c) Okubo, M.; Konishi, Y.; Minami, H. Colloid Polym. Sci. 1998, 276, 638. (d) Emmerich, O.; Hugenberg, N.; Schmidt, M.; Sheiko, S. S. Adv. Mater. 1999, 11, 1299. (e) Jang, J.; Ha, H. Langmuir 2002, 18, 5613. (f) Zha, L. S.; Zhang, Y.; Yang, W. L.; Fu, S. K. Adv. Mater. 2002, 14, 1090. (g) Kamata, K.; Lu, Y.; Xia, Y. J. Am. Chem. Soc. 2003, 125, 2384. (h) Beil, J. B.; Zimmerman, S. C. Macromolecules 2004, 37, 778. (5) (a) Breiteukamp, K.; Emrick, T. J. Am. Chem. Soc. 2003, 125, 12070. (b) Wong, M. S.; Cha, J. N.; Choi, K. S.; Deming, T. J.; Stucky, G. D. Nanoletters 2002, 2, 583. (c) Jungmann, N.; Schmidt, M.; Ebenhoch, J.; Weis, J.; Maskos, M. Angew. Chem., Int. Ed. 2003, 42, 1714. (6) (a) Sauer, M.; Streich, D.; Meier, W. Adv. Mater. 2001, 13, 1649. (b) Ma, Q.; Remsen, E. E.; Kowalewski, T.; Schaefer, J.; Wooley, K. L. Nano Letter 2001, 1, 651. (c) Che´cot, F.; Lecommandoux, S.; Gnanou, Y.; Klok, H. A. Angew. Chem., Int. Ed. 2002, 41, 1339. (d) Dou, H. J.; Jiang, M.; Peng, H. S.; Chen, D. Y. Angew. Chem., Int. Ed. 2003, 42, 1516. (e) Hu, Y.; Jiang, X. Q.; Ding, Y.; Chen, Q.; Yang, C. Z. Adv. Mater. 2004, 16, 933. (7) (a) Chu, L. Y.; Yamaguchi, T.; Nakao, S. Adv. Mater. 2002, 14, 386. (b) Ibarz, G.; Dahne, L.; Donath, E.; Mo¨hwald, H. Adv. Mater. 2001, 13, 1324.

10.1021/la047912p CCC: $30.25 © 2005 American Chemical Society Published on Web 01/15/2005

1532

Langmuir, Vol. 21, No. 4, 2005

comparison with the hydrogels in bulk or the microspheres, hollow spheres or nanocages2d of PAA are expected to have larger specific surface area and, consequently, larger and quicker responses to the environmental stimuli. In recent years, our group developed a “block-copolymerfree” strategy to prepare so-called noncovalently connected micelles (NCCM).9,10 Homopolymer or random copolymer can be used as the building blocks, and therefore, in the resultant micelles, the core and shell are connected by hydrogen bonding. Recently, Pispas reported that such NCCM can be produced on the basis of the hydrogen bonding interactions of poly(2-vinylpyridine) with sulfonic ended polystyrene and polyisoprene, respectively.11 Since there are no chemical bonds connecting the core and shell in the NCCM, it is easy to obtain hollow spheres from the NCCM simply by subsequent shell cross-linking and core dissolution. For example, the hollow spheres based on poly(4-vinylpyridine) (PVPy) and poly(vinyl alcohol) (PVA) were prepared from the NCCMs of PVPy-PS(OH) [PS(OH) ) poly{styrene-co-[p-(1,1,1,3,3,3-hexafluoro-2-hydroxylpropyl)-R-methylstyrene]}] and PVA-CPB (CPB ) carboxyl-ended polybutadiene) pairs, respectively.10b,d In this paper, we report our efforts along our blockcopolymer-free strategy to obtain PAA-based hollow spheres. Biocompatible, biodegradable, and nontoxic polyester-poly(-caprolactone) (PCL) was chosen as the counterpart of PAA. Self-assembly of PCL and PAA was realized, leading to the NCCM with PCL as the core and PAA as the shell in dimetylformamide (DMF)/water (v/v, 1/10). After cross-linking the shell, hollow spheres or nanocages based on PAA were obtained by either simple biodegradation or dissolution of the PCL core. The resultant hollow spheres displayed perfect pH- and saltsensitivity. Experimental Section Samples. Polycaprolactone with Mw ) 33 000 is a product of Scientific Polymer Product Inc., carrying a hydroxyl end and a carboxyl end. And, PCLs with Mw of 7000, 10 000, and 30 000 were synthesized by bulk polymerization with 1, 4-butanediol and butyl titanate as starting agent and catalyst, respectively, provided by the University of Science and Technology of China. PCL with Mw ) 33 000 is used except for the case specified. PAA was synthesized by AIBN-initiated free radical polymerization, with Mn ) 1.51 × 105 and Mw/Mn ) 2.99 estimated by size exclusion chromatography in tetrahydrofuran (THF). Micelles Preparation. A stock solution of PAA (5 mg/mL) in water was further diluted to the desired concentration. Micelle solution was typically prepared by adding 1 mL of PCL/DMF solution dropwise very slowly (ca. 0.2 mL/min) into 10 mL of stirred PAA/water solution at 25 °C. The mixture was kept stirring overnight and then kept for 7 days at room temperature before measurements. Shell Cross-Linking of Micelles. A typical shell crosslinking process was as follows: to the stirred micelle solution (8) (a) Cleary, J. L.; Bromberg, E.; Magner, E. Langmuir 2003, 19, 9162. (b) Lee, W. F.; Lin, Y. H. J. Appl. Polym. Sci. 2001, 81, 1360. (c) Yoo, M. K.; Sung, Y. K.; Lee, Y. M.; Cho, C. S. Polymer 2000, 41, 5713. (d) Ende, M. T. A.; Peppas, N. A. J. Controlled Release 1997, 48, 47. (9) NCCM, similar with the micelles of block copolymers, has a coreshell structure. However, there is no equilibrium between the singlecomponent molecules and the core-shell aggregates as the NCCM structure is mainly kinetically controlled in its preparation process. (10) (a) Wang, M.; Zhang, G. Z.; Jiang, M.; Cheng, D. Y. Macromolecules 2001, 34, 7172. (b) Wang, M.; Jiang, M.; Ning, F. L.; Cheng, D. Y. Macromolecules 2002, 35, 5980. (c) Liu, X. Y.; Jiang, M.; Yang, S. L.; Chen, M. Q. Angew. Chem., Int. Ed. 2002, 41, 2950. (d) Zhang, Y. W.; Jiang, M.; Zhao, J. X.; Zhou, J. Macromolecules 2004, 37, 1537. (e) Duan, D. H.; Cheng, D. Y.; Jiang, M.; Gan, W. J. J. Am. Chem. Soc. 2001, 123, 12097. (f) Kuang, M.; Duan, H. W.; Wang, J.; Chen, D. Y.; Jiang, M. Chem. Commun. 2003, 496. (g) Duan, H. W.; Kuang, M.; Wang, J.; Chen, D. Y.; Jiang, M. J. Phys. Chem. B 2004, 108, 550. (11) Orfanou, K.; Topouza, D.; Sakellariou, G.; Pispas, S. J. Polym. Sci., Part A: Polym. Chem. 2003, 41, 2454.

Zhang et al. M1 (50 mL, 0.069 mmol of acrylic acid unit) was added dropwise an aqueous solution of the catalyst, 1-(3-(dimethylamino)propyl)ethylcarbodiimde methiodide (20 mg/mL, 250 µL, 0.017 mmol). The mixture was stirred for about 30 min before aqueous solution of the cross-linker, 2,2-(ethylenedioxy)bis(ethylamine) (10 mg/ mL, 125 µL, 0.0087 mmol), was added dropwise. Then the reaction mixture was kept stirring for 12 h at room temperature followed by dialyzing against water for 3 days to remove impurities including catalyst, unreacted cross-linker, and DMF. Core Degradation of Shell-Cross-Linked Micelles. pH of the cross-linked micelle solution was first adjusted to a weak basic range (pH 8-11) by adding 0.1 M NaOH aqueous solution, and then the mixture was kept under mild stirring for about 1 h before adding an appropriate amount of dust-free lipolase aqueous solution (Novozymes Co.). And the solution was allowed to react for about 5 days to ensure the full degradation of the PCL core; after that, the solution was dialyzed against water for 3 days and kept still for about 10 days; finally, the insoluble degradation product was removed by discarding the lower precipitant, and the upper hollow sphere solution was obtained. Measurements FTIR spectra were obtained on a Nicolet Magna 550 spectrometer as KBr pellets and film on aluminum foil for polymer and micelle solution, respectively. Malvern Autosizer 4700 laser light scattering (LLS) spectrometers were used. In dynamic light scattering (DLS), the line-width distribution, G(Γ), can be calculated from the Laplace inversion of the intensity-intensity time correlation function, G(2)(q,t).12 The inversion was carried out by the CONTIN program. G(Γ) can be converted into a translational diffusion coefficient distribution G(D) or a hydrodynamic diameter distribution f(Dh) via the Stokes-Einstein equation. Dh ) (kBT/3πη)D-1, where kB, T, and η are the Boltzman constant, the absolute temperature, and the solvent viscosity, respectively. Transmission electron microscopy (TEM) observations were performed on a Philips CM 120 electron microscope at an accelerating voltage of 80Kv. The samples were prepared by placing 5µl micelle solutions on copper grids coated with a thin carbon film and allowing them to freeze-dry for 24 h.

Results and Discussion Surfactant-Free Particles of PCL in Water. It was reported that PCL nanoparticles could be prepared by adding dropwise a dilute PCL solution in acetone into a large amount of aqueous solution containing a surfactant.13 Here, the PCL nanoparticles are stabilized by the surfactant molecules absorbed on the particle surfaces. However, we found that adding dropwise a dilute solution of PCL/DMF into a large excess of milliQ water without the surfactant under stirring led to faint blue opalescence, indicating the formation of nano- or microsized particles as well. PCL solutions in THF or dioxane led to similar results. It was found that the larger the dielectric constant of the solvent is, the smaller the size of PCL particles is: when using DMF, THF, and dioxane as the solvent, the hydrodynamic diameters of the PCL particles are 89, 154, and 170 nm, respectively (Table 1, Supporting Information). It can be rationalized that the particles are stabilized by the hydrophilic carboxyl and hydroxyl end groups of PCL, and the degree of ionization or polarization of the groups increases with the polarity of the solvent, leading to a finer dispersion of the PCL particles. In addition, the size of the PCL particles in DMF/water (v/v, 1/10) displayed little change in the range from pH 3.9 to 9.0.However, when the pH of the solution decreased to 3.5, where the ionization of carboxyl group is completely suppressed, two peaks began to appear in its hydrodynamic diameter distribution curve, indicating aggregation of the particles (data not shown). (12) Chu, B. Laser Light Scattering: Basic Principles and Practicle, 2nd ed.; Academic Press: New York, 1991. (13) Gan, Z. H.; Fung, J. T.; Jing, X. B.; Wu, C.; Kuliche, W. K. Polymer 1999, 40, 1961.

Core-Shell Particles and Hollow Spheres

Langmuir, Vol. 21, No. 4, 2005 1533

Table 1. Preparation Conditions and Compositions of (PCL)-PAA Micelles, Cross-Linked Micelles, and Hollow Spheresa concn of PAA/ concn of PCL/ PCL/PAA water (mg/mL) DMF (mg/mL) in feed (w/w) M1 M2 M3 C1 C1a C2 C2a C3

0.1 0.5 1.0 0.1 0.1 0.5 0.5 1.0

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

1/1 1/5 1/10 1/1 1/1 1/5 1/5 1/10

targeted shell cross-linking degree (%)

25 50 25 50 25

a Hollow spheres H1, H1a, H2, and H3 were obtained from C1, C1a, C2, and C3 by core degradation, respectively; hollow sphere H1′ was obtained from C1 by core dissolution.

Self-Assembly of PCL and PAA in Water. To construct the micelles composed of PCL and PAA, typically, 1 mL of PCL/DMF solution (1.0 mg/mL) was added dropwise into 10 mL of PAA/water solution (0.1 mg/mL) under stirring, leading to a bluish tint. The DLS measurements demonstrated that a peak associated with the particles with sizes around 50-200 nm appeared and the peak corresponding to the free PAA chains (10-40 nm) disappeared (Figure 1, Supporting Information). In addition, the average hydrodynamic diameter 〈Dh〉 (106 nm) was larger than the particles composed of PCL alone formed under similar conditions (89 nm). The results indicate the formation of NCCM with PCL as the core and PAA as the shell (denoted as (PCL)-PAA). This result is similar with the NCCM composed of CPB and PVA, formed in water driven by the hydrogen bonding interaction between the components.10d In the present case, as soon as the PCL solution was dropped into PAA aqueous solution, the medium for PCL chains switched from a good solvent to a very poor one; the hydrophobic PCL chains aggregated rapidly. Meanwhile, driven by the interaction between the carboxyl of PAA and the end groups of carboxyl, hydroxyl, and the ester groups lying on the periphery of the PCL particles, PAA chains gathered and surrounded the particles. Some PAA chains may play a role as a stabilizer by physically absorbing to the PCL particles. Thus, the PCL particles would be stabilized by the soluble PAA chains, leading to the formation of NCCM. A series of NCCM solutions with different compositions were prepared by adding 1 mL of PCL/DMF solution into 10 mL of PAA/water with different concentrations. The micelles and the subsequent shell-cross-linked micelles and hollow spheres are denoted as M, C, and H, respectively (Table 1). Furthermore, all the micelles were prepared with PCL having a hydroxyl end and a carboxyl end, and Mw ) 33 000 except for the study of the dependence of the molecular weight of PCL, where hydroxyl-terminated PCL was used. We studied the influence of the molecular weight of PCL and the solvent in the initial solution on the size of the resultant (PCL)-PAA micelles. The results showed that the solvent affects the size of (PCL)-PAA micelles in a trend similar to that of the pure PCL particles; that is, for DMF, THF, and dioxane, the average hydrodynamic diameters of the micelles are 106, 160, and 191 nm, respectively. Furthermore, the size of the micelles goes up with increasing molecular weight of PCL (Figure 2, Supporting Information). As the molecular weight increases, self-association of the polymer chains strengthens, and the relative content of the hydrophilic end groups in PCL chains comes down, both the facts favor larger aggregates.

Figure 1. Average hydrodynamic diameters of (PCL)-PAA micelles in DMF/water (v/v, 1/10) as a function of the composition. The initial concentrations of PAA/water solution (0) and PCL/DMF solution (9) are 0.1 and 1.0 g/mL, respectively.

Figure 1 depicts two groups of data describing variations of the micelle size with the initial concentrations of PCL and PAA solutions, respectively. In the first data group (9), the micelles were prepared with PCL/DMF of 1.0 mg/ mL and PAA/water with different concentrations. All the Dh distribution curves of the micelle solutions show one peak with polydispersity index (〈µ2〉/Γ2) varying from 0.14 to 0.26 and the 〈Dh〉 increases monotonically with the initial concentration of PAA. This is understandable as the higher the PAA concentration, the more PAA chains gather around the PCL particles through hydrogen bonding. In the second data group (0), the micelle size shows a complex dependence on the PCL concentration when PAA concentration keeps constant; i.e., with an increase in the initial concentration of the PCL/DMF solution, the micelle size decreases first and then increases after reaching the minimum. Over the concentration range from 1.5 to 3.5 mg/mL PCL, the size increases with the concentration. This is understandable, as the high concentration favors aggregation of the PCL chains in water. However, we do not fully understand the results at the very low concentration range; i.e., the size is decreased with increasing concentration. One possible reason is that, at the very low PCL concentration range, the distance between PCL chains is fairly long; thus, the micellar core formed may possess very loose packing, i.e., a larger size but lower density. All the DLS curves for the above two micelle groups display unimodal distributions; i.e., no individual PAA molecules are detected, indicating that in the composition range we studied, most of PAA molecules are incorporated into the particles. Of course, as the PAA chains are much smaller than that of the particles, the presence of a small amount of free PAA chains in the solutions could not be excluded. Although the present micellization is a kinetically controlled process, the resultant (PCL)-PAA micelles in DMF/water (v/v, 1/10) are very stable: the average hydrodynamic diameter changed a little in a month: for example, for micelles M1 and M2, 〈Dh〉 changed from 106 to 104 nm and from 117 to 120 nm, respectively. It is well-known that PAA chains undergo magnificent conformation changes, from extended chains to hypercoils as the pH decreases.14 Therefore, it is interesting to study the micellar stability against the pH of the solution. The (14) Chen, J. Y.; Jiang, M.; Zhang, Y. X.; Zhou, H. Macromolecules 1999, 32, 4861.

1534

Langmuir, Vol. 21, No. 4, 2005

Figure 2. Change of the 〈Dh〉 of micelle M1 in water and crosslinked micelle C1 with the pH of the solution, which was adjusted by adding 0.1 N HCl or 0.1 N NaOH solution to the starting solution M1 and C1 with pH about 5.3 and 5.4, respectively.

size of micelles M1 in solution with different pH was characterized by DLS, in which M1 with pH 5.3 was used as the starting solution. The results showed that the micelles were stable over the pH-range from 4.0 to 11.0, and the diameter varied from 101 to 98 nm. When the pH of the solution decreased from 2.9 to 2.2, a dramatic size increase from 137 to 241 nm occurred (Figure 2), which showed further aggregation of the micelles in strong acidic media. Shell-Cross-Linking of (PCL)-PAA Micelles. The PAA shell chains of the micelles were cross-linked simply by reacting with a desired amount of a cross-linker, 2,2(ethylenedioxy)bis(ethylamine) in the presence of 1-(3(dimethyl amino)propyl) ethylcarbodiimde methiodide (ETC) as a catalyst at room temperature.2e The crosslinking process of M1 was monitored by infrared spectrometry (Figure 3, Supporting Information). The spectra display that a small band characteristic of the carbonyl of the reaction product amide around 1660 cm-1 appeared after the reaction proceeded for 8 h and the peak area ratio of this carbonyl to those in ester groups of PCL and carboxyl groups in PAA (around 1730 cm-1) increased as the reaction continued. After cross-linking reaction, dialysis against water was performed to remove all the impurities. The success of locking the micellar structure was proved by the fact that after further core dissolution with DMF, a very faint bluish tint was retained for the shell-cross-linked micelle solution C1 in DMF/water (5/1, v/v), while under similar conditions, the solution of M1 without being treated with the cross-linking agent became completely transparent, indicating the disintegration of the micelles. The main features of the dependence of 〈Dh〉 on the crosslinking time of M1 and M2 shown in Figure 3 are as follows. First, as the reaction proceeded, 〈Dh〉 increased rapidly first and then decreased gradually. Comparing the curves of M2/C2 and M2a/C2a, it can be seen that the larger the amount of the cross-link reagent is, the more pronounced the initial increase is. This rapid size increase is probably associated with the reaction of PAA and the catalyst ETC. It is reported by Wooley et al.15 that the reaction of ETC with the acrylic acid functionalities will produce acylisourea and anhydride active intermediates. Quite recently, Chen et al.16 found that such intermediates cause the (15) Ma, Q. G.; Remsen, E. E.; Clark, C. C.; Kowalewski, T.; Wooley, K. L. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 085. (16) Gu, C. F.; Chen, D. Y.; Jiang, M. Macromolecules 2004, 37, 1666.

Zhang et al.

Figure 3. Average hydrodynamic diameters of micelles M1/ C1, M2/C2, and M2a/C2a at different cross-linking times. The target degrees of cross-linking for C1, C2, and C2a are 25, 25, and 50%, respectively.

formation of “short-life” nanoaggregates composed of PEOb-PAA and ETC. Usually, the amidation reaction occurs instantaneously between anhydride intermediates and diamine cross-linker. In the present case, as the crosslinker was introduced 30 min after adding catalyst ETC, and it took time for the cross-linker to diffuse into the PAA shell; consequently, at the initial stage the size increase dominates. In addition, some free PAA chains, if any, may be connected to the PAA shell of the micelles by reaction with ETC; this also contributes to the initial size increase. As the reaction proceeded, the cross-linking degree of the shell increased, leading to a size shrinkage of the micelles. In addition, all the distribution curves in Figure 3 show one peak only (figure not shown); it means that, under the reaction conditions we used, intermicelle cross-linking leading to micellar aggregation could be avoided. We studied the 〈Dh〉 variation of cross-linked micelle solution C1 (with a targeted cross-linking degree of 25%) with the pH of the solutions. The pH was adjusted by adding 0.1 N HCl or 0.1 N NaOH solution to the starting solution C1 with a pH of about 5.4. As shown in Figure 2, the change of 〈Dh〉 of C1 with pH is similar to that of the original micelle M1sthere is little change in the range from pH 4.0 to 11, but C1 showed even larger size increase than M1 did in the strong acidic media. Cavitation of Shell-Cross-Linked Micelles. The key step to reach our target of PAA-based hollow spheres is the core cavitation. Taking advantage of the biodegradability of PCL, we obtained hollow spheres simply by degradation of the PCL core with lipase. DLS was used to trace the core degradation process. From the results for C1 shown in Figure 4, it is surprising to find that, accompanying the continuous increase in the micelle size, the light intensity, which is related to the micellar mass, rapidly increases in the first 3 h and then goes down quickly followed by a slow and long decrease. After about 60-70 h, both the micelle size and the light intensity show little changes. We observed that after lipase was added, the bluish tint of the micelles solution deepened, which indicated the interactions between lipase and PAA shells. Our subsidiary experiment also proved it: when lipase was added to the PAA solution, a faint bluish tint also appeared, indicating that soluble complex formed. Therefore, the interaction between lipase and PAA is responsible to the initial increase of the light intensity in Figure 4. In fact only after the lipase enters into the core to catalyze the

Core-Shell Particles and Hollow Spheres

Figure 4. Changes in light intensity and average hydrodynamic diameter during degradation of the core of C1 with lipase with initial pHs of 7.9 and 10.2 (insert). A 40 µL aliquot of dust-free lipase solution was added to 2 mL of micelles solution.

Figure 5. 〈Dh〉 vs pH (pH sensitivity) of the solutions of hollow spheres H1, H1a, and H2. The pH-value of the solution was adjusted by adding 0.1 N HCl or 0.1 N NaOH solution to the starting solution H1, H1a, and H2 with pH about 5.8, 6.3, and 6.0, respectively.

core degradation, and then the degraded product diffuses out of the shell, the micellar weight begins to decrease. That is why the intensity decrease is observed only 3 h after adding lipase. Due to the complexity of the particles containing multicomponents and that the relative amounts of the components keep varying during degradation, we are not able to estimate the amount of the degraded PCL with time from the change in the scattering light intensity. During the core degradation, the restriction imposed by the insoluble PCL chains on the PAA shell is gradually released. This, of course, makes the shell network expand gradually. This shell expansion during the core degradation seems a common phenomenon as it was observed in our previous investigations for the cavitation of the shellcross-linked NCCMs.10b-d Comparing the results shown in Figure 4 for the cases of pH 7.9 and 10.2 (in the insert), it can be seen that, in the basic pH-range, pH has a little effect on the process. As mentioned above, micelle M1 and shell-cross-linked micelle C1 are stable as there is no significant change in the micelle size against pH provided that the solution is not less than pH 4.0.(Figure 2). However, the corresponding hollow spheres show a very different pH-dependence of the size. Figure 5 shows the variation of the hydrodynamic diameters of the hollow spheres of H1, H1a, and

Langmuir, Vol. 21, No. 4, 2005 1535

Figure 6. Salt-concentration dependence of 〈Dh〉 of the hollowsphere solution H1 at different pH.

H2 with pH. The starting materials are the as-dialyzed solutions with pH 5.8, 6.3, and 6.0 for H1, H1a, and H2, respectively. In the case of H1, little size change occurs in both low (4.5-5.5) and high (>8) pH-ranges, while a dramatic size jump happens when pH increases from about 6 to 8. In this pH-range, the diameter increases from 126 to 534 nm, corresponding to a 75 times volume expansion. Such remarkable volume change is associated with the hollow structure of the spheres. H1a and H2 show pronounced size dependence on pH as well. However, in comparison with H1, H1a displays much smaller size variation; i.e., over the whole pH-range, the volume increases by 6-8 times only. This apparent difference demonstrates the significant effect of cross-linking degree on the swelling of the nanocages in water, as the desired degrees of cross-linking of H1 and H1a are 25 and 50%, respectively (Table 1). In addition, H2, with the same degree of cross-linking as H1 but much thicker shell, shows a similar but slightly smaller size expansion as H1 does with pH increase. Salt-Sensitivity of Hollow Spheres.7,17 The size of the hollow spheres also varies with the salt concentration of the solutions. In fact, the pH-value of the medium strongly affects the dependence of the size on the salt and then the ionic strength. As shown in Figure 6, over the salt concentration range we studied, in the acidic media, 〈Dh〉 increases with salt concentration. However, in the basic media, the opposite dependence was observed. In addition, in the neutral medium, 〈Dh〉 changes very little with the salt. Our speculation is as follows. In acid medium, PAA shell chains of the spheres are protonated, the added salt makes the spheres aggregates, causing an increase in size. While in basic medium, PAA shell chains are fully dissociated; the added salt decreases the repulsion between PAA chains, leading to the PAA chains shrinking and then a size decrease. Reversibility of pH-Dependence of Hollow Spheres. Considering the possible applications of the hollow spheres, the reversibility of their response to the pH stimuli is very important. The hydrodynamic diameter variation of H1 in a whole pH-cycle has been measured by DLS (Figure 7). The dark points correspond to the first half pH-cycle, the same as that for H1 in Figure 5, where the size increases and decreases when NaOH and HCl is (17) (a) Dedunaite, A.; Ernstsson, M. J. J. Phys. Chem. 2003, 107, 8181. (b) Winkler, K. G. Macromol. Symp. 2004, 211, 55. (c) De´jugnat, C.; Sukhorukov, G. B. Langmuir 2004, 20, 7265. (d) Thuneman, A. F.; Muller, M.; Dautzenbug, H.; Joanny, J. F. D.; Lowne, H. Adv. Polym. Sci. 2004, 166, 117.

1536

Langmuir, Vol. 21, No. 4, 2005

Figure 7. 〈Dh〉 of hollow sphere solution H1 as a function of pH. Dark and red squares refer to the first half and second half cycles, respectively. 0 is the starting point.

Zhang et al.

added into the H1 solution starting with pH 5.8 (0), respectively. The second half cycle (red points) was performed starting from both ends, i.e., by adding HCl and NaOH into the solutions with the highest pH and the lowest pH, respectively. In the second half cycle, the size also shows a pronounced increase with increasing pH. However, there is a significant difference in the size values between the first and second half cycles at the same pHvalue. It is interesting to see that, in the basic medium, the size in the second half is always smaller than that in the first half and in the acidic medium it is opposite. This difference can be mainly attributed to the effect of ionic strength, as the salt ionic strength in the second half is always higher than that in the first half. This opinion is supported by the following facts. First, dialysis against water to remove ionic impurities for the two solutions at the ends of the second half with respective pH 5.8 and 6.0 causes 〈Dh〉 to decrease to about 153 nm, very close to that in the first half cycle. Second, as shown in Figure 6, at

Figure 8. TEM images of M1 (a), C1 (b), H1 (c), H2 (d), and H3 (e).

Core-Shell Particles and Hollow Spheres

Langmuir, Vol. 21, No. 4, 2005 1537

Scheme 1. Schematic Illustration for Preparing pH-Responsive Hollow Spheres

acid and basic media, the solution shows opposite dependence of 〈Dh〉 on salt concentration. Therefore, we are able to conclude that by excluding the effect of salt, our PAA nanocages show perfect reversible size variation with medium pH. In addition, this reversible pH-responsible volume change mainly takes place in the range from pH 6.0 to 7.5; the upper and low limits happen to be the values of tumor tissue and normal physiological conditions, respectively.18 Therefore, we are expecting bright application prospects of the hollow spheres in biomedicine fields. Although both the first and second half cycles show a size increase with pH in the acid range, their mechanisms seem different. In the first half cycle, all the Dh distribution curves in the acid medium are unimodal (Figure 4, Supporting Information), while in the second half cycle, all the curves turn to be bimodal as a new peak associated with aggregates in a larger size (Figure 5, Supporting Information). This indicates that, in the first half, the sphere size decrease with decreasing pH is mainly due to the protonation of PAA, leading to less swelling, while in the second half, due to the presence of salt, the hollow spheres are no longer stable and thus intersphere aggregation takes place. TEM is a powerful tool for observing the internal structure of polymeric assembled objects. Parts a-c of Figure 8 display the TEM images of micelles M1, crosslinked micelles C1, and hollow spheres H1, respectively. Although micelles M1 display fairly spherical shape, no clear core-shell structure is revealed. This is probably due to the poor contrast between shell chains and the background as the density of the solvated PAA chains is low.10d After shell-cross-linking, the spherical shape is basically retained (Figure 8b), but still no core-shell structure is visualized. The interparticle connections shown in Figure 8b are probably formed during the preparation of the TEM specimen as DLS of C1 solution shows a unimodal distribution (figure is not shown). The hollow nature for H1, which is characterized by a thin shell and a fairly large inner cavity, is clearly evidenced in Figure 8c, as there is a much higher electron transmission in the center of the spheres than that around the periphery.10d In addition, the central cavity is also confirmed by the combination of DLS and SLS measurements: the 〈Rg〉/〈Rh〉 value of the spheres H1 is 1.026, very close to the value of 1.0 reported for thin shell hollow spheres.19 Comparing Figure 8c-e, it is interesting to find that the shell thickness strongly affects the ability to see hollow structure by TEM. For H1, in its parent micelle M1, the weight ratio of the shell to core is 1/1; there is a clear contrast between the center and periphery. However, for H2 with much thicker shell due to the ratio reaching 5/1, only vague contrast can be realized in the TEM image at a high magnification (the insert, Figure 8d). Finally, almost no contrast could be observed for H3, in which the weight ratio increases to 10/1. Obviously, this failure of

revealing the central cavity in TEM is because the core is too small relative to the shell to provide a clear contrast. In fact, for H2 and H3, completely different pH-dependence of the size from its parent cross-linked micelles C2 and C3 was also observed (data not shown), which clearly proved the existence of a central cavity. Taking advantage of no covalent bonds connecting the core and shell, we are able to attain the hollow sphere by simple core dissolution of the cross-linked micelles C1. This was realized by switching the medium from water to DMF/water (v/v, 5/1), where DMF is a common solvent for PCL and PAA. After swelling of the cross-linked micelles C1 in DMF/water for 21 days, typical hollow sphere morphology was observed, indicating the success of preparing hollow spheres by core dissolution (Figure

Figure 9. TEM micrographs of hollow sphere H1′ obtained by swelling C1 in DMF/water (v/v, 5/1) for 21 days.

9). However, compared with the core biodegradation, the core dissolution process took much longer time. Conclusion Using biodegradable PCL and pH-responsive PAA, the micelles with PCL as the core and PAA as the shell in water were fabricated due to the specific interactions between the components. After locking the micelle structure by shell cross-linking, and subsequent core degradation with lipase, we obtained PAA hollow spheres. These PAA-based nanocages shrink or expand when decreasing or increasing the pH of the solution, respectively (Scheme 1). The pH-dependence of the sphere size was found to be completely reversible provided that the effect of ionic strength is excluded. Both the shell cross-linked micelles and hollow spheres are expected to be useful in capsulation in various areas, particularly, biomedical fields. (18) Haag, R. Angew. Chem., Int. Ed. 2004, 43, 278. (19) (a) Wu, C.; Zhou, S. Q. Phys. Rev. Lett. 1996, 77, 3053. (b) Burchard, W. Adv. Polym. Sci. 1983, 48, 1.

1538

Langmuir, Vol. 21, No. 4, 2005

Acknowledgment. This work was supported by the National Natural Science Foundation of China (NNSFC Grant Nos. 50173006, 5033010). Supporting Information Available: Characterization data of PCL particles, hydrodynamic diameter distribution

Zhang et al. curves of micelle M1, (1.0 PCL)-0.1PAA micelles with different molecular weights of PCL and hollow spheres H1 in acid media, and FTIR spectra of M1 in cross-linking. This material is available free of charge via the Internet at http:// pubs.acs.org.

LA047912P