Influence of Shell Structure on Stability, Integrity, and Mesh Size of Polyelectrolyte Capsules: Mechanism and Strategy for Improved Preparation

Transcription

1 Chem. Mater. 2005, 17, Influence of Shell Structure on Stability, Integrity, and Mesh Size of Polyelectrolyte Capsules: Mechanism and Strategy for Improved Preparation Wen-Fei Dong, James K. Ferri,, Thorsteinn Adalsteinsson, Monika Schönhoff, Gleb B. Sukhorukov, and Helmuth Möhwald*, Max Planck Institute of Colloids and Interfaces, Golm/Potsdam, D Germany, Department of Chemical Engineering, Lafayette College, Easton, PennsylVania 18042, and Institute of Physical Chemistry, UniVersity of Münster, Corrensstrasse 30, D Münster, Germany ReceiVed January 17, 2005 Novel polyelectrolyte microcapsules are developed using alternating layer-by-layer (LBL) adsorption of oppositely charged polyelectrolytes onto charged templates followed by core removal. These capsules have a continuous wall with a hollow interior and exhibit tunable permeability, which can be utilized for both sustained release from the capsule and precipitation reactions from the bulk to the capsule interior. The purpose of this work is to understand core removal mechanism and elucidate the structure-property relationships governing capsule stability and yield, integrity, and mesh size. To this end, melamine formaldehyde (MF) templates with polyelectrolyte multilayers (PEMs) of poly(styrene sulfonate) (PSS) and poly(allylamine hydrochloride) (PAH) were investigated by confocal Raman microscopy, surface force microscopy, and confocal laser scanning microscopy. The capsule structure is dominated by two critical factors: the wall thickness and the core decomposition condition. After optimization of the core removal conditions, capsule permeability is shown to be strongly dependent on the thickness, which is controlled by the variation of solvent quality (i.e. salt concentration) or the number of bilayers. The critical multilayer thickness for capsule preparation is about 10 nm. Above this value, capsules of three different structures are obtained as a function of wall thickness. The mesh size of intact capsules is also shown to decrease from meso- to microporous as thickness increases. To improve the capsule integrity, a new mode of core decomposition is proposed and demonstrated by control of the acid excess during core release. Introduction Hollow polymer microcapsules (or hollow microspheres) present a unique architecture having a continuous wall with a hollow interior, which exhibit unique encapsulation and delivery properties for sensitive contents or active agents. Thus, they have potential application in the field of drug delivery, the food industry, biology and biomedicine, etc. 1-8,10-18 Ideal microcapsules 1,2 should be mechanically tough and elastic, be stable in a wide range of conditions, have high encapsulation efficiency, have controlled or switched permeability, and have an easy and facile preparation. Besides lipid vesicles 3 or inorganic hollow spheres, 4 several approaches have been used to achieve hollow polymer capsules, including the self-assembly of block * To whom correspondence should be addressed. helmuth.moehwald@ mpikg-golm.mpg.de. Tel: Fax: Max Planck Institute of Colloids and Interfaces. Lafayette College. University of Münster. (1) Möhwald, H.; Donath, E.; Sukhorukov, G. B. In Multilayer Thin Films: Sequential Assembly of Nanocomposite Materials; Decher, G., Schlenoff, J. B., Eds.; VCH Verlagsgesellschaft Mbh: Weinheim, Germany, (2) Gordon, V. D.; Xi, C.; Hutchinson, J. W.; Bausch, A. R.; Marquez, M.; Weitz, D. A. J. Am. Chem. Soc. 2004, 126, (3) Lasic, D. D. Liposomes: From Physics to Applications; Elsevier: Amsterdam, (4) Caruso, F.; Caruso, R. A.; Möhwald, H. Science 1998, 282, copolymers, 5 interfacial emulsion polymerization, 6 assembly of polymer micelles by noncovalent or covalent bonds, 7 spontaneous colloidosome assembly of polymers 8 and nanoparticles, 9 etc. In 1998, a novel route was achieved by layerby-layer (LBL 10 ) adsorption of oppositely charged polyelectrolytes on decomposable particles and followed by core removal. 11 Compared with other approaches, this method is quite simple. Capsules are achieved with high yield and tunable size. Moreover, functional groups can be fabricated on the capsule shell by the LBL technique. (5) Discher, B. M.; Won, Y. Y.; Ege, D. S.; Lee, J. C. M.; Bates, F. S.; Discher, D. E.; Hammer, D. A. Science 1999, 284, (6) (a) Okubo, M.; Konishi, Y.; Minami, H. Colloid Polym. Sci. 1998, 276, 638. (b) Jang, J.; Ha, H. Langmuir 2002, 18, (7) (a) Ding, J. F.; Liu, G. J. Chem. Mater. 1998, 10, 537. (b) Huang, H. Y.; Remsen, E. E.; Kowalewski, T.; Wooley, K. L. J. Am. Chem. Soc. 1999, 121, (c) Duan, H. W.; Chen, D. Y.; Jiang, M.; Gan, W. J.; Li, S. J.; Wang, M.; Gong, J. J. Am. Chem. Soc. 2001, 123, (d) Antonietti, M.; Forster, S. AdV. Mater. 2003, 15, (8) Dinsmore, A. D.; Hsu, M. F.; Nikolaides, M. G.; Marquez, M.; Bausch, A. R.; Weitz, D. A. Science 2002, 298, (9) (a) Wong, M. S.; Cha, J. N.; Choi, K.-S.; Deming, T. J.; Stucky, G. D. Nano Lett. 2002, 2, 583. (b) Cha, J. N.; Birkedal, H.; Euliss, L. E.; Bartl, M. H.; Wong, M. S.; Deming, T. J.; Stucky, G. D. J. Am. Chem. Soc. 2003, 125, (10) (a) Decher, G.; Hong, J. D.; Schmitt, J. Thin Solid Films 1992, 210, 831. (b) Lvov Y.; Decher G.; Möhwald, H. Langmuir 1993, 9, 481. (c) Decher, G. Science 1997, 277, (d) Mamedov, A. A. Nat. Mater. 2002, 1, 190. (11) Donath, E.; Sukhorukov, G. B.; Caruso, F.; Davis, S. A.; Möhwald, H. Angew. Chem., Int. Ed. 1998, 37, /cm050103m CCC: $ American Chemical Society Published on Web 04/19/2005

2 2604 Chem. Mater., Vol. 17, No. 10, 2005 Dong et al. Among these capsules, weakly cross-linked melamine formaldehyde (MF) particles have been employed as sacrificial cores, 10 because MF particles can be obtained with a narrow size distribution and have a good acid solubility. The pair of poly(styrene sulfonate) (PSS) and poly(allylamine hydrocloride) (PAH) is selected to fabricate the capsule wall since it produces stable and robust multilayers. 9c After core decomposition in low ph solution, stable capsules (MF capsules) are obtained. 10 Intense research, therefore, focuses on the investigation of their physicochemical properties, like electron 12 and energy transfer, 13 mechanical properties, 14 diffusion 15 and encapsulation properties, 16 optical and spectroscopic ones, 17 etc. To drive these capsules toward smart applications, an essential prerequisite is to first understand their structurepermeability relationships and then to control them. As yet, these relationships are unclear. The permeability of these capsules is different from those of planar polyelectrolyte multilayers (PEMs) prepared by the same LBL technique. For example, the planar PEMs of (PSS/PAH) n, where n is the number of bilayers, exhibit a very dense and robust film: they enable limiting of ion diffusion across the layers and show selective permeability for different ions but have the ability to resist proteins or other macromolecules. 1,10 It had been reported that the penetration of dyes into flat PEMs occurs with very low diffusion coefficient of the order of m 2 /s. 18 However, unlike the planar PEMs, MF capsules of (PSS/PAH) n exhibit a semipermeable behavior since they can exclude PSS oligomers with molecular weight above 5 kda (4200 g/mol and upward), 19 but the diffusion coefficient of dye has been observed to be on the order of m 2 /s. In principle, the semipermeability is a very important feature for these capsules. It allows the efficient loading of the polymer or biomacromolecule inside the capsules so that the polymer inside can be used to encapsulate drugs or to synthesize particles by variation of the environmental conditions in the capsule interior. On the other hand, one can easily switch off their permeability by reducing the mesh size when (12) Dai, Z.; Dahne, L.; Donath, E.; Möhwald, H. Langmuir 2002, 18, (13) Tedeschi, C.; Li, L.; Möhwald, H.; Spitz, C.; von Seggern, D.; Menzel, R.; Kirstein, S. J. Am. Chem. Soc. 2004, 126, (14) (a) Gao, C. Y.; Leporatti, S.; Moya, S.; Donath, E.; Möhwald, H. Langmuir 2001, 17, (b) Vinogradova, O. I.; Andrienko, D.; Lulevich, V. V.; Nordschild, S.; Sukhorukov, G. B. Macromolecules 2004, 37, (c) Lulevich, V. V.; Radtchenko, I. L.; Sukhorukov, G. B.; Vinogradova, O. I. J. Phys. Chem. B 2003, 107, (d) Dubreuil, F.; Elsner, N.; Fery, A. Europhys. J. E. 2003, 12, 215. (e) Sukhorukov, G. B.; Shchukin, D. G.; Dong, W. F.; Möhwald, H.; Lulevich, V. V.; Vinogradova, O. I. Macromol. Chem. Phys. 2004, 205, 530. (15) (a) Antipov, A. A.; Sukhorukov, G. B.; Donath, E.; Möhwald, H. J. Phys. Chem. B. 2001, 105, (b) Shi, X.; Caruso, F. Langmuir 2001, 17, (c) Qiu, X.; Leporatti, S.; Donath, E.; Möhwald, H. Langmuir 2001, 17, (16) (a) Radtchenko, I. L.; Sukhorukov, G. B.; Leporatti, S.; Khomutov, G. B.; Donath, E.; Möhwald, H. J. Colloid Interface Sci. 2000, 230, 272. (b) Dahne, L.; Leporatti, S.; Donath, E.; Möhwald, H. J. Am. Chem. Soc. 2001, 123, (c) Gao C. Y.; Donath E.; Möhwald H.; Shen J. C. Angew. Chem., Int. Ed. 2002, 41, (d) Shchukin, D. G.; Dong, W. F.; Sukhorukov, G. B. Macromol. Rapid Commun. 2003, 24, 462. (17) Dong, W. F.; Sukhorukov, G. B.; Möhwald, H. Phys. Chem. Chem. Phys. 2003, 5, (18) Klitzing, R. v.; Möhwald, H. Macromolecules 1996, 29, (19) Sukhorukov, G. B.; Brumen, M.; Donath, E.; Möhwald, H. J. Phys. Chem. B 1999, 103, additional layers adsorb on the capsule surfaces. For instance, the diffusion coefficient of dye decreases to m 2 /s, when four bilayers of PEM are fabricated on these semipermeable capsules. 20 However, until now, little was known about the fundamental principles governing the semipermeability of MF capsules. The reason is two-fold. First, the semi permeability of MF capsules is sensitive to preparation conditions and core quality. Second, the permeability is strongly influenced by shell materials and shell thickness, as well as the core material. Therefore, different preparation routes lead to conflicting conclusions. Furthermore, different techniques for measuring permeability also yield different results. Clearly, capsule semipermeability is a complicated issue. Here, we use the model system of standard pairs of PSS and PAH as the shell and the acid-soluble MF particles as the cores to prepare capsules. The core decomposition is investigated as a function of ph using confocal Raman microscopy. Because the permeability of capsules is dominated by the capsule structure after core removal, we present a detailed investigation on the structure by surface force microscopy and then explore the permeability by confocal laser scanning microscopy. We find that two critical factors control the permeability properties of MF capsules: one is the core decomposition condition (the ph and excess acid), and the other is the PEM wall thickness, which is controlled by the variation of solvent quality (i.e. salt concentration) and the number of bilayers. Finally, the capsules are classified by their permeability associated with the capsule architecture, and a new route of core decomposition is demonstrated to improve the yield of intact capsules. Experimental Section Materials. Poly(styrene sulfonate) sodium salt (PSS, M r 70 kda), poly(allylamine hydrochloride) (PAH, M r 70 kda), FITC (fluorescein isothiocyanate), TRITC (Rhodamine isothiocyanate), TRITC-labeled dextran (4, 66 kda), and sodium chloride were purchased from Aldrich. All chemicals were used as received except for PSS, which was dialyzed before use (M r ) 50 kda cutoff). FITC-PAH was prepared according to the literature The water used in all experiments was purified in a three-stage Milli-Q Plus 185 purification system and had an initial resistivity higher than 18.2 MΩ/cm. Weakly cross-linked melamine formaldehyde particles (MF particles, 5.0 ( 0.2 µm) were obtained from Microparticles GmbH, Germany. The melamine formaldehyde particles are prepared by polymerization from a precondensate solution, by the reaction between melamine (I) and formaldehyde (II) monomers, as shown in Scheme 1. Microcapsule Preparation. Hollow polyelectrolyte microcapsules were made by stepwise adsorption of polyelectrolytes onto a charged colloidal MF template, which was followed by the dissolution of the MF templates. 11,14 Self-assembled films were formed as follows: 1.0 ml of aqueous PSS solution (5 mg/ml, either with 0.5 or 1.0 M NaCl, ph ) 6.5) was added to 1.0 ml of the positively charged MF particle dispersion. The particle concentration was approximately 10 8 particles/ml. The dispersion was (20) Ibarz, G.; Dahne, L.; Donath, E.; Möhwald, H. Chem. Mater. 2002, 14, (b) Ibarz, G.; Dahne, L.; Donath, E.; Möhwald, H. Macromol. Rapid Commun. 2002, 23, 474.

3 Influence of Shell Structure on Capsules Chem. Mater., Vol. 17, No. 10, Scheme 1. Chemical Structures of (I) Melamine, (II) Formaldehyde, (III) the Methylolmelamine Structure by the Ether Bridge, (IV) the Methylmelamine Structure by the Methyl Bridge, and (V) MF Particles occasionally stirred during the adsorption time of 20 min at 283 K. Low temperatures were used to minimize the postpolymerization of MF particles, which leads to increased stability of the core toward acid. The dispersion was centrifuged at 1000g for 5 min at 283 K, the supernatant was decanted, 2.0 ml of water was added, and the particles were redispersed by gentle shaking. The centrifugation/ wash/redispersion cycle was repeated an additional 3-4 times to ensure removal of free polyelectrolyte from solution. In all cases the polyelectrolyte concentration is much higher than that required for saturation coverage of the particles. This was repeated until the desired numbers of polyelectrolyte layers were reached. The MF particle core was then dissolved using dilute HCl (ph ) 1.0) solution for 20 min. The dissolution product and HCl were removed by centrifuging the dispersion at 3000g for 10 min at room temperature. The capsules were washed using centrifugation/wash/ redispersing cycles until near neutral ph was established. Measurements. Scanning Force Microscopy (SFM). SFM images were recorded in air at room temperature (20-25 C) using a Nanoscope III Multimode SFM (Digital Instrument Inc., Santa Barbara, CA). Silicon nitride (Si 3 N 4 ) cantilevers with a force constant of 0.58 N/m (Digital Instrument) were used for contact SFM. Silicon tips (Olympus and Nanotips, DI) with a resonance frequency of 300 khz and a spring constant of 40 N/m were employed for tapping mode SFM imaging. The contact force between the tip and the sample was kept as low as possible (<10 nn), and images were acquired in constant force mode (height mode) at a scan rate of Hz. The samples were prepared by applying a drop of the capsule solution onto freshly cleaved mica or onto glass cover slips. Images were sampled from different areas of the sample. SFM images were processed by using Nanoscope software and Image PC software (version beta 2, Scion Corp.). Confocal Laser Scanning Microscopy (CLSM). The size and permeability of individual capsules were determined by an inverse Leica TCS NT confocal laser scanning microscope (CLSM, Leica, Germany) with either a 100 or a 40 oil immersion objective with a numerical aperture of 1.0. Four different probe materials are used to the permeability measurement. They are TRITC, Fitc- PAH, and Tritc-dextran with differing molecular weights. (The GPC chromatograms of the two different dextrans have no overlap region, which ensures the accuracy of mesh size measurement.) Before measurement, the probe solution is mixed with the capsule dispersion with a 1:1 volume ratio (for example, 2 mg/ml dextran solution), and it takes 24 h for the incubation to ensure equilibrium. Confocal Raman Microscopy. Raman spectra and Raman imaging of capsules were taken in water at room temperature using a Confocal Raman microscope (CRM200, Witec) equipped with a piezo scanner (P-500, Physik Instrumente) and high NA microscope objectives (60 NA ) 0.80 or 100 oil NA ) 1.25, Nikon). In a typical experiment, circularly polarized laser light (diode pumped green laser, λ ) 532 nm, CrystaLaser) was focused down to the located material with diffraction limited spot size (diameter approximately l/2). The spectra were taken with an air-cooled CCD (PI-MAX, Princeton Instruments) behind a grating (600 g/mm) spectrograph (Acton) with a resolution of 1 cm -1. Pulsed Field Gradient Nuclear Magnetic Resonance (PFG- NMR). The NMR spectrometer used here for diffusion measurements was a Bruker Avance DMX400 wide-bore spectrometer with a field gradient coil capable of reaching a 12 T/m magnetic field gradient. The pulse program used was the stimulated echo sequence described by Stejskal and Tanner, 21 [π/2-τ-π/2-t-π/2-τ-echo], containing magnetic field gradient pulses. The diffusion measurement was performed by varying the gradient pulse strength, and from the resulting monoexponential echo decay the diffusion coefficient was extracted. A detailed description of the measurement is described in a recent paper. 22 Results and Discussion Mechanism and Optimization of Core Removal. One of the most common substrates used to synthesize hollow polyelectrolyte multilayer capsules consists of melamine resin microparticles, which are a network of melamine formaldehyde (MF) cross-linked by ether linkages (III) or methylene bridges (IV), as shown in Scheme 1. After adsorption of the polymer onto the (MF), the core is dissolved by tuning the external solution. Although the ether linkages are decomposed by acid hydrolysis, the methylene bridges remain (21) Tanner, J. E.; Stejskal E. O. J. Chem. Phys. 1968, 49, (22) Adalsteinsson, T.; Dong, W. F.; Schönhoff, M. J. Phys. Chem. 2004, 108,

4 2606 Chem. Mater., Vol. 17, No. 10, 2005 Dong et al. Figure 1. Raman spectrum of the MF core as function of ph value at (a) ph ) 7, (b) ph ) 2, and (c) ph ) 1.0 of HCl acid solution (with MF particles dissolved in ph ) 1.0 acid and dried solutions measured) and Raman spectrum of the capsule interior after the core decomposition at (d) ph ) 2, (e) ph ) 1.0, and (f) ph ) 0.5 of PEM-coated MF particles. intact. Thus, the core is decomposed into oligo- as well as polymeric units. Here confocal Raman microscopy (CRM) is used to study the chemical structure of the (MF) core in-situ. Figure 1a shows a typical Raman spectrum of a suspension of (MF) particles (V): 23 (1) The sharp and strong Raman active mode at 981 cm -1 is attributed to the ring breathing in the melamine group. (2) The peak at 671 cm -1 is that of the melamine monomer (I) which broadens upon condensation. (3) The shoulder at 906 cm -1 is from the C-O-C symmetric stretch in the ether bridge which disappears after acid hydrolysis. (4) The bands at 1394, 1453, and 1555 cm -1 are assumed to be the different ring breathing modes in the methylolmelamine. First consider the decomposition of naked MF microparticles by acid hydrolysis in aqueous solution. Suspensions of MF particles were exposed to hydrochloric acid solutions of varying ph s, and the Raman spectrum of the suspension was recorded at equilibrium. The spectrum of MF particles in (ph ) 2.0) solution is shown in Figure 1b. It can be readily observed that there is no appreciable difference between this and the reference spectrum in Figure 1a indicating the stability of the particles under these conditions. When the MF particles are dispersed at ph ) 1.0, the Raman signal is attenuated greatly due to rapid particle decomposition. (The minimum ph for MF decomposition was found to be 1.7.) (23) (a) Meier, R. J.; Maple, J. R.; Hwang, M.-J.; Hagler, A. T. J. Phys. Chem. 1995, 99, (b) Meier, R. J.; Tiller, A.; Vanhommerig, S. A. M. J. Phys. Chem. 1995, 99, To obtain a measurable spectrum, a drop of solution was evaporated to form a solid film, the spectrum of which is shown in Figure 1c. The typical change in spectrum is the band at 906 cm -1, which disappears as expected due to the acid hydrolysis of the ether band. Also the band at 1453 cm -1 (the NCH 2 O entity) in the methylolmelamine (III) has also decreased on hydrolysis. Meanwhile several new Raman active modes in the methyl-bridged melamine structure become more intense, such as at 1031 cm -1 (tentatively attributed to the motion of the triazine ring), at 1435 cm -1 (the CH 2 vibration in NCH 2 N), and at 1513 cm -1 (the CH 2 vibration of the methyl bond). Additional bonds can be observed in the range between 1339 and 1537 cm -1. The diameter of the MF degradation product (i.e. MF oligomers) is estimated to about 2.5 nm by an analytical ultracentrifuge measurement, which is in agreement with previous studies. 24 This suggests that the degradation product (MF oligomers) are on average composed of melamine formaldehyde polymers with a degree of polymerization (n ) 60). The acid hydrolysis decomposition of MF resin becomes a more complex function of ph in the presence of encapsulating polyelectrolyte multilayers. For simplicity, this discussion is restricted to results obtained when four bilayers, (PSS/PAH) n)4, were fabricated by the layer-by-layer technique onto MF particles at a salt concentration C NaCl ) 0.25 M. The Raman spectrum of the capsule interior was obtained using confocal imaging at high magnification (i.e. 100 ) so that the diameter of the laser (D l 1 µm) was smaller than the capsule diameter (D c 5 µm). As with the naked MF, the encapsulated MF cannot be dissolved for a ph value greater than ph ) 2.0, as can be seen by comparison of the spectrum shown in Figure 1d with those in Figure 1b (naked particles at ph ) 2.0) and Figure 1a (naked particles at ph ) 7.0) When the acidity was increased to ph ) 1.0, the turbidity of the suspension decreased rapidly suggesting decomposition of the MF core. The absence of Raman active modes in the capsule interior, as shown in Figure 1e, signifies complete core removal. Additional spectra obtained by scanning across the capsule wall confirmed the absence of MF oligomer in the polymer matrix. If core removal proceeds at (ph ) ), strong Raman bands can be observed within the capsule, as in Figure 1f. By comparison with Figure 1c, it can be concluded that the fraction of Raman activity by MF oligomers has increased, prompting the following question: why there are so many oligomers inside the capsules? As discussed previously, the MF oligomers are formed due to the hydrolysis of the ether bridges of MF particles. The hydrolysis product remains active at low ph and can easily form supramolecular structures due to hydrogen bonding. The same effect can be observed in the decomposition of naked particles. (After exposure to low ph, the particles are degraded into oligomers of size 3 nm. By incubation of the decomposition product for 1 week at room temperature, the oligomers self-aggregate (24) (a) Gao, C. Y.; Moya, S.; Donath, E.; Möhwald, H. Macromol. Mater. Eng. 2001, 286, 355. (b) Gao, C. Y.; Moya, S.; Donath, E.; Möhwald, H. Macromol. Chem. Phys. 2002, 203, 953.

5 Influence of Shell Structure on Capsules Chem. Mater., Vol. 17, No. 10, into complexes with an average size of 160 nm, as measured by analytic ultracentrifugation.) In addition, it also has been reported that melamine can be hydrolyzed to ammeline. 24 Therefore, if the ph is too low (ph < ), a second reaction (acid-catalyzed repolymerization) occurs in parallel with hydrolysis and results in the formation of C-N-C linkages. Since, the second reaction becomes increasingly favorable during the hydrolysis, the reaction equilibrium favors the formation of an MF gel although some of the MF oligomers escape from the shell. Both of these mechanisms underscore the role of confinement of the hydrolysis product by the polyelectrolyte shell on the core release process and suggest an optimum window (1.0 < ph < 2.0) for core removal. Capsule Stability and Yield. The formation of hollow capsules after core removal is a function of the number of polyelectrolyte (PSS/PAH) bilayers and salt concentration and, hence, the structure and thickness of the shell. 10 This results from a competition between the osmotic stress from in the capsule interior during decomposition and the mechanical resistance conferred by the polymer shell. For capsule stability, the elastic modulus must be of the same order of magnitude as the stress in the capsule wall. Typically, removal of the MF core results in a low yield of stable capsules when the number of bilayers, n, is less than 4. When (n g 4), intact capsules can be observed. This dependence of stability on the number of bilayers suggests that there is a transition in the structure (and mechanical properties) of the shell as function of its thickness. In this work, the shell thickness of capsules was measured by surface force microscopy (SFM) using a standard method introduced elsewhere The capsules are first dried on mica, and then the minimum height in a fold-free region is measured. Figure 2a,b shows a typical top view SFM images of capsules with n ) 4 and n ) 9atC NaCl ) 0.25 M, respectively. In an area free of folds, the minimum thickness h SFM thickness, min is regarded as twice the wall thickness. Figure 2e shows the increase of h SFM thickness, min as a function of the number of bilayers (n) atc NaCl )0.25 and 0.5 M, respectively. The average thickness h single bilayer then is found by normalizing h SFM thickness, min by twice the number of bilayers. Figure 2f shows the dependence of h single bilayer on the bilayer number and salt concentration. If the film growth were linear, a constant value in the thickness per bilayer would be expected; i.e., the thickness/bilayer would be independent of the total thickness. However, the plateau in the thickness as a function of layer number indicates the presence of a phase transition between expanded and condensed bilayers at both salt concentrations. The onset of this transition occurs earlier for a higher salt concentration, which also implies that this transition depends on the total film thickness. This suggests that h singlebilayer can be used to evaluate the capsule stability. In general, when the capsule wall is too thin (h < 10 nm), it cannot sustain such a high osmotic pressure, and then the capsule shell is destroyed. As the thickness increases between (12 < h < 16 nm), a transition in the shell structure, as evidenced in thickness of a single bilayer, occurs. Thereafter, the capsule stability and yield is increased. Figure 2. SFM images of capsules: (a) four bilayers at C NaCl ) 0.25 M; (b) nine bilayers at C NaCl ) 0.25 M after removing the core by HCl with ph ) 1.0. The SFM thickness of the capsules is measured by the difference in the heights corresponding to the arrow in (a) or (b), where the heights along the line are shown in (c) and (d). (e) The SFM thickness and (f) the average single bilayer thickness of the capsules are shown as a function of the number of bilayers at different salt concentrations of C NaCl ) 0.25 M (h, j) and C NaCl ) 0.5 M (g, i) (see the text). Capsule Integrity. It is important to differentiate the capsule stability and yield from capsule integrity. The presence of stable capsules does not imply that all capsules are bounded by a continuous surface. A method for the identification of intact and broken capsules by osmotic deformation has been developed previously. 24 When subjected to a solution of high molecular weight polyelectrolyte having an osmotic pressure much greater than that of the capsule interior, intact capsules deform and broken capsules do not. There are two requisites for the probe polymer; it should not adsorb to the capsule exterior and should not permeate though the capsule wall. Therefore, capsule integrity, i.e., fraction of unbroken capsules, was studied using polycation-terminated capsules and dye-labeled polyallyamine hydrochloride, M r ) 70 kda, as a probe. (By CLSM measurement, FITC-PAH is impermeable to the intact capsules. The results are similar to Figure 5d but are not shown here.) Figure 3 shows capsule integrity as a function of the number of bilayers (n) and salt concentration. For n ) 5at C NaCl ) 0.25 M, the unbroken capsule ratio is about 96% and decreases monotonically as a function of layer number. Similar results are found at C NaCl ) 0.5 M. Because the increase of polylectrolyte multilayer thickness as a function of salt concentration is well established, 10,25 these data suggest that the number of broken capsules increases with h.

6 2608 Chem. Mater., Vol. 17, No. 10, 2005 Dong et al. Scheme 2. Capsule Type Evaluated by the Capsule Porosity and Integrity a Figure 3. Fraction of unbroken capsules as a function of the number of layers at different salt concentrations (C NaCl ) 0.25 M and C NaCl ) 0.5 M). In an earlier study, a higher capsule integrity of approximately 97% for n ) 5 and C NaCl ) 0.25 M was reported. 24 These differences arise because of differences in preparation conditions. In the previous study, the synthesis was performed at room temperature using the filtration method. The capsules in this study were prepared using the centrifugation method 24 at a relatively low temperature (283 K) to avoid the repolymerization of the MF oligomers. Clearly, there is a tradeoff between capsule stability and capsule integrity. Below the critical thickness required for mechanical stability, core release results in destruction of the shell. As the shell thickness (12 < h < 16 nm) increases, either by bilayer number (n) or salt concentration, stable capsules, denoted here as type II, can be observed. Further increases in thickness (h > 16 nm) result in capsules of high stability but include both intact (type III) and broken (type IV) capsules. The capsule types are shown in Scheme 2 as a function of shell thickness. The porosity of each capsule type is discussed below. Capsule Mesh Size. Capsule permeability is governed by a combination of hydrodynamic sieving, electrostatic interactions, and Donnan effects 19 in the polymeric shell. By measurement of the equilibrium distribution of fluorescently labeled probe materials of varying size across intact capsules, bounding values of the average capsule mesh size, ξ, can be determined. The transport rates of different probe materials are the subject of a future publication. The hydrodynamic radius, R H, of each probe can be calculated from its bulk diffusion coefficient according to the Stokes-Einstein relation: D ) kt (1) 6πηR H The determination of the diffusion coefficient, D, by PFG NMR with stimulated echo is described in detail elsewhere. 22 Values of D and R H for each probe in this study are listed in Table 1. Consider first the low molecular weight dye TRITC. Figure 4a-c shows that permeation of this probe occurs in both capsule types II and III, independent of shell thickness. This a The curve describes the experimentally derived single bilayer thickness versus the AFM shell thickness of the PEM. There are at least four different capsule types of different structures and permeabilities along the curves, including type II (intact, mesoporous capsule), type III (intact, microporous capsule), and type IV (broken). Table 1. Diffusion Coefficients and Hydrodynamic Radii of Probe Materials Used in This Work material diffusion coeff (10 10 m 2 /s) suggests that the mesh size of these capsules must be at least the hydrodynamic radius of TRITC to neglect the electrostatic repulsion of this probe from the capsule shell. The insitu measurement of the diffusion of organic penetrants into multilayers of (PSS/PAH) synthesized on planar substrates have yielded mesh sizes much smaller, i.e., ξ ) 0.5 nm. 26 Thus, it can be inferred that MF core release results in pore formation in the capsule wall. The use of a neutral analyte eliminates electrostatic effects and allows exclusive focus on mesh size-dependent permeability as a function of shell thickness. For this purpose, two different polysaccharides (dextran) with molecular weights of 4.4 and 66 kda were used. Capsules of differing thickness were prepared by variation of the number of bilayers at a constant salt concentration (C NaCl ) 0.25 M). Figure 4d shows that only type II capsules (3 < n < 6), are permeable to 4 kda dextran. As thickness increases, the mesh size is reduced, and the probe is excluded. The presence of both capsule types III and IV can be confirmed in Figure 4e,f by increasing the concentration of probe material to induce osmotic deformation. Similar results for two different molecular weight species of dextran are shown in Figure 4g-i. Some conclusions regarding the porosity as a function of film thickness can be made. The osmotic deformation of type II capsules by polyallylamine hydrochloride places an upper bound on the mesh size of at least several multiples of the (25) Bosio, V.; Dubreuil, F.; Bogdanovic, G.; Fery, A. Colloids Surf., A 2004, 243, 147. (26) Liu, X.; Bruening, M. L. Chem. Mater. 2004, 16, 351. hydrodynamic radius (nm) TRITC kda TRITC-dextran kda TRITC-dextran kda PAH

7 Influence of Shell Structure on Capsules Chem. Mater., Vol. 17, No. 10, Figure 4. CLSM images of MF capsules of PSS/PAH by the different probe material: (a-c) FITC; (d-f) TRITC-labeled 4 kda dextran; (g-i) TRITClabeled 66 kda dextran as a function of the number of bilayers. (The scale bar is 10 µm, and samples were measured in neutral water.) Indicated are also the types of capsules belonging to the different observations. hydrodynamic radius (R H ) 7.7 nm) of the polymer. However, passage of 66 kda dextran (R H ) 9.0 nm) occurs freely. This suggests that type II capsules are mesoporous, i.e., possess an average pore size between 2 and 50 nm. As previously mentioned, these pores are the result of the burst release of MF oligomers having an average size of 2-3 nm. Because of the large mesh size of the capsule, MF oligomer is released to below the detectable limit of either Raman or IR spectra. It has been shown that the pore size of the type II can be reduced (or healed) by additional layer adsorption and other methods. 20 Therefore, the permeability of type II capsules can be controlled easily and will be studied in further work. Type III capsules are impermeable to 4 kda dextran (R H ) 2.1 nm) but permeable for TRITC (R H ) 0.76 nm). This indicates a microporous, i.e., average pore size less than 2 nm, shell architechture. The smaller mesh size is the result of entrainment of released MF oligomers by the multilayers during the core release. Prior work concludes that as much as 20% (w/w) of MF released remains in the shell. 28 Most existing studies of capsules prepared on sacrificial MF particles 14,24 are type III. Since the capsule shell thickness is around 16 nm, which is in the transition region between types II and III, the macroscopic properties, viz. permeability, (27) Khoury, C.; Adalsteinsson, T.; Johnson, B.; Crone, W. C.; Beebe, D. J. Biomed. MicrodeVices 2004, 5, 35. (28) Moya, S.; Dahne, L.; Voigt, A.; Leporatti, S.; Donath, E.; Möhwald, H. Colloids Surf., A 2001, 183, 27. Figure 5. CLSM image of (a) MF capsules with n ) 4, (b) n ) 6, and (c) n ) 9atC NaCl ) 0.25 M, where the probe material is TRITC-labeled dextran with 4 kda molecular weight. (d) Image of MF capsule (n ) 9atC NaCl ) 0.25 M) with probe material of FITC-PAH. All capsules were prepared by the controlled core decomposition method. (The bar is 5 µm with samples measured in neutral water.) Indicated are also the types of capsules belonging to the different observations. elasticity, etc., depend strongly on the environment (temperature or solvent quality, etc.) and decomposition conditions. Paradigms for Control of Capsule Integrity. Although type II capsules have high integrity, their stability limits the yield and ultimately their application. Type III capsules have high stability, but their integrity is low. Thus, improvement

8 2610 Chem. Mater., Vol. 17, No. 10, 2005 Dong et al. Figure 6. Raman spectrum of (a) MF capsules with n ) 9 (normal method, C NaCl ) 0.25 M), (b) MF capsule with n ) 9 (the controlled core decomposition method, C NaCl ) 0.25 M), and (c) MF capsule with n ) 6 (the controlled core decomposition method, C NaCl ) 0.25 M). of the integrity, i.e., decrease of the fraction of capsules ruptured during core release, is required to drive these materials to application. To suppress the fraction of type IV capsules, controlled core decomposition is proposed. Normally, a particle concentration 0.5% (w/w) is used during core release in acid solution (ph ) 1.0). Low particle concentration (or the excess of acid) ensures a rapid and complete dissolution of MF); however, it increases the maximum osmotic pressure that a capsule is exposed to during release. By increase of the particle concentration or decrease of the excess acid, the kinetic rate of decomposition is slowed, thus decreasing the number of oligomers in the capsule at any time. Under these conditions, the driving force for diffusion of oligomer out of the capsule would remain approximately the same, because the capsule interior would remain essentially saturated with oligomer. To demonstrate the efficacy of this method, core release of MF through PSS/PAH multilayers (n ) 4-9, C NaCl ) 0.25 M) at elevated particle concentration 2% (w/w) was performed. Under these conditions, high capsule integrity (>96%) was measured for all values of n, indicating significant improvement. The measurement of pore size as a function of the number of bilayers of these capsules is shown in Figure 5. Capsules of n ) 4 are freely permeable to 4 kda dextran (Figure 5a), but capsules with n ) 6 and 9 are impermeable (Figure 5b,c). In addition, FITC-PAH permeation through the capsules in Figure 5d is inhibited. Since all capsules are intact, this indicates that the capsules n > 4 are type III according to the Scheme 2. Moreover, Figure 5b,c shows that the average diameters of the capsules n ) 6 and 9 are larger than those of capsules prepared by uncontrolled release because of the hindered release of MF oligomer. The MF oligomers inside the capsules were investigated by confocal Raman spectroscopy. Figure 6 shows the Raman spectra of air-dried capsules. Figure 6a shows the spectra of MF capsules (n ) 9) which are prepared by the normal

9 Influence of Shell Structure on Capsules Chem. Mater., Vol. 17, No. 10, method. Figure 6b shows the results on MF capsules with the same number of bilayers but prepared by the controlled core decomposition method. In a comparison of these two spectra, a Raman band at 981 cm -1 can be easily distinguished in Figure 6b. Even when the number of bilayers (n) is reduced to 6, the same peak is still found in the capsules, as shown in Figure 6c, but as expected the intensity is lower as compared to Figure 6b. This controlled core decomposition results in more MF oligomers inside the capsules. These oligomers can improve the capsule integrity, but they also influence the permeability of the capsules via a strong decrease in the mesh size of the capsules. The presence of MF oligomer is demonstrated by analysis of the enlarged Raman spectrum (from 800 to 1350 cm -1 ), as shown in Figure 6a-c. Shown are four different peaks: 981 cm -1 attributed to MF oligomers and 1030, 1129, and 1194 cm -1 due to PSS. 17 Obviously, the peak of MF oligomer in Figure 6b is higher than that in Figure 6a. After normalization of the peak area of oligomer by the area of one of the PSS peaks, i.e., A 981 /A 1129, the relative percentages of MF can be calculated: 9.5% in Figure 5a, 68.8% in Figure 5b, and 5.3% in Figure 5c, respectively. These values demonstrate that, for the same number of bilayers, the MF that remains in capsules made by controlled core decomposition is about seven times higher in concentration than those of capsules made by the normal way. In summary, controlled core decomposition is a viable alternative to reduce the high broken ratio of capsules when the wall thickness is increased. The capsules prepared by this method are all type III; however, this method does not change the critical multilayer thickness for capsule preparation. Conclusion In this work, the core removal mechanism and the structure-property relationships governing capsule stability and yield, integrity, and mesh size are studied. A model system of poly(styrene sulfonate) (PSS) and poly(allylamine hydrochloride) (PAH) composite capsules prepared on melamine formaldehyde (MF) templates was investigated by confocal Raman microscopy, surface force microscopy, and confocal laser scanning microscopy. The mechanism of MF core decomposition as a function of ph was studied, and the effect of confinement on the hydrolysis of MF into oligomeric units was demonstrated. At low ph value, MF particles convert to MF oligomers with destruction of the ether bond, but in confined systems they can repolymerize into larger subunits. It was shown that control of ph is crucial for obtaining hollow polyelectrolyte capsules. The capsule structure is dominated by two critical factors: the wall thickness and the core decomposition condition. After optimization of the core removal conditions, capsule permeability is shown to be strongly dependent on the thickness. It is concluded that the critical multilayer thickness for capsule preparation is about 10 nm. Above this minimum, capsules of three different structures are obtained as a function of wall thickness, as shown in Scheme 2. In addition, the mesh size of intact capsules is also shown to decrease from meso- to microporous as thickness increases. To improve the capsule integrity, a new mode of core decomposition is proposed and verified by control of the acid excess during core release. By control of the acid amount, the capsule integrity of thicker capsules (at n ) 9) increases to 96% from a previous value of 50%. Acknowledgment. We gratefully thank Prof. Changyou Gao for helpful discussions and Anneliese Heilig for SFM measurements. The work has been supported by the Alexander v. Humboldt foundation via a Sofja Kovaleskaja prize to G.B.S. and a Fellowship to J.K.F. CM050103M