pubs.acs.org/Langmuir © 2009 American Chemical Society
Synthesizing Microcapsules with Controlled Geometrical and Mechanical Properties with Microfluidic Double Emulsion Technology Yves Hennequin,† Nicolas Pannacci,† Concepcion Pulido de Torres,‡ Georgios Tetradis-Meris,‡ Stephane Chapuliot,§ Elisabeth Bouchaud,§ and Patrick Tabeling*,† † Microfluidics, MEMS & Nanostructures, ESPCI, 10 rue Vauquelin, 75005 Paris, France, ‡Corporate Research, Unilever R&D Colworth, Sharnbrook, Bedford MK44 1LQ, United Kingdom, and §CEA Saclay, F-91191 Gif sur, Yvette, France
Received February 4, 2009. Revised Manuscript Received March 25, 2009 Using lithography-based microfluidic technology, we produce monodisperse single-core microcapsules with UV-cured TPGDA (triprophylene glycol diacrylate) shells. We show that the geometrical and mechanical characteristics of the microcapsules can be predicted on a quantitative basis and tuned by varying the flow conditions. Shell thicknesses are varied by changing the flow rates of the inner or intermediate phases, according to mass conservation constraint. Off-centering of the core with respect to the shell is controlled by varying the shell phase viscosity. The mechanical properties of the capsules can be varied by changing the flow conditions and are quantitatively predicted by a numerical simulation. The simulation moreover provides a correct qualitative description of their rupture. As a whole, the work carried out in the present paper shows, on a quantitative basis, that microfluidic technology allows to finely control the geometrical and mechanical properties of microcapsules generated on chip. The level of control we reach here is not accessible, by far, to conventional technologies. Combined with parallelization, the present work opens routes toward the production of novel families of monodisperse microcapsules with tunable properties.
Microencapsulation is a technique that consists in isolating a dispersed phase from an external medium by engulfing or coating it with a protective material. With the current technologies, the microcontainers that can be produced have sizes typically ranging from 1 μm to 1 mm. The benefits of encapsulation in industrial applications are countless; the food and cosmetic industries use it for masking odors and flavors or for hindering the oxidation of ingredients, thus increasing the shelf life of a number of products; in pharmaceutical applications, microcapsules are used as vectors for drug delivery; in the printing or textile industries, microcapsules are used to release pigments or odors under mechanical stress. Current methods of encapsulation include spray-drying,1 coextrusion,2 interfacial polymerization, or complex coacervation.3 These processes allow to produce capsules under high throughput. Nonetheless, in the current state of the art, the properties of the capsules produced with these processes are determined empirically, and they cannot be monitored. Moreover, the capsule sizes, mechanical properties, and delivery characteristics obtained with these techniques are substantially nonuniform across the capsule population. There is a hope that microfluidic technology improves the situation. One advantage offered by this technology is the possibility to work at the “single capsule level” rather with large populations of objects; another is the outstanding flow control that is achieved.4 By combining these advantages, monodisperse populations of microparticles of various shapes could be synthesized under *Corresponding author. E-mail:
[email protected]. (1) Jafari, S. M.; Assadpoor, E.; He, Y. H.; Bhandari, B. Drying Technol. 2008, 26, 816. (2) Yadav, S. K.; Khilar, K. C.; Suresh, A. K. J. Membr. Sci. 1997, 125, 213. (3) Nihant, N.; Grandfils, C.; Jerome, R.; Teyssie, P. J. Controlled Release 1995, 35, 117. (4) Tabeling, P. An Introduction to Microfluidics; Oxford University Press: New York, 2005.
Langmuir 2009, 25(14), 7857–7861
excellent control.5-8 Producing homogeneous populations of more complex objects;such as double droplets9-12;and synthesizing monodisperse capsules using interfacial polycondensation13 or photocuring one of the phases forming the compound droplet7,14 have also been demonstrated with microfluidic technology. In the meantime, a number of groups have shown that systems made with glass capillaries, which exploits the physics of miniaturization in a way similar to microfluidic technology, also generate well-controlled droplets, double droplets, and microcapsules.15-17 The glass capillary technique, however, is unable to parallelize large numbers of devices and thereby deliver quantities of industrial interest. By contrast, microfluidic droplet emitters fabricated with photolithographic techniques can easily be parallelized provided the dynamical phenomena that are induced by the parallelization are kept under control.18 (5) Okushima, S.; Nisisako, T.; Torii, T.; Higuchi, T. Langmuir 2004, 20, 9905–9908. (6) Seo, M.; Nie, Z. H.; Xu, S. Q.; Mok, M.; Lewis, P. C.; Graham, R.; Kumacheva, E. Langmuir 2005, 21, 11614–11622. (7) Nie, Z. H.; Xu, S. Q.; Seo, M. S.; Lewis, P. C.; Kumacheva, E. J. Am. Chem. Soc. 2005, 127, 8058–8063. (8) Oh, H. J.; Kim, S. H.; Baek, J. Y.; Seong, G. H.; Lee, S. H. J. Micromech. Microeng. 2006, 16, 285–291. (9) Hwang, D.; Dendukuri, D.; Doyle, P. Lab Chip 2008, 8, 1640–1647. (10) Okushima, S.; Nisisako, T.; Torii Micro Total Anal. Syst. 2005, 258–260. (11) Seo, M.; Paquet, C.; Nie, Z. H..; et al. Soft Matter 2007, 3, 986–922. (12) Pannacci, P.; Bruus, H.; Bartolo, D.; Etchart, I.; Lockhart, T.; Hennequin, Y.; Willaime, H.; Tabeling, P. Phys. Rev. Lett. 2008, 101, 164502. (13) Takeuchi, S.; Garstecki, P.; Weibel, D. B.; Whitesides, G. M. Adv. Mater. 2005, 17, 1067–1072. (14) Oh, H. J.; Kim, S. H.; Baek, J. Y.; Seong, G. H.; Lee, S. H. J. Micromech. Microeng. 2006, 16, 285–291. (15) Utada, A. S.; Lorenceau, E.; Link, D.; Kaplan, P.; Stone, H. A.; Weitz, D. A. Science 2005, 308, 537-54. (16) Loscertales, I. G.; Barrero, A.; Guerrero, I.; Cortijo, R.; Marquez, M.; Ga~nan-Calvo, A. M. Science 2002, 295, 1695–1698. (17) Quevedo, E.; Steinbacher, J.; McQuade, D. T. J. Am. Chem. Soc. 2005, 127, 10498–10499. (18) Barbier, V.; Willaime, H.; Tabeling, P.; Jousse, F. Phys. Rev. 2006, E 74, 046306.
Published on Web 05/01/2009
DOI: 10.1021/la9004449
7857
Article
Figure 1. Principle of the microcapsules production in the microfluidic device. The width and height of the channels before the expansion are 150 and 80 μm, respectively. The width and height of the expanded channel are 300 and 250 μm, respectively. Blue color represents silicon oil (phase 3), green acrylate monomer (phase 2), and red water (phase 1). Photopolymerization is done in the expanded part (the green arrows sketch UV light).
The fact that microfluidic capsules have homogeneous geometrical characteristics does not necessarily imply that their mechanical properties have a comparable level of homogeneity. There can be some randomness generated by the cross-linking process itself, or induced by a strong sensitivity of this process to dissolved gas (such as oxygen or CO2) or liquid/solid impurities, or some variability due to a poor control of the position of the core in the compound droplet as cross-linking takes place. It is thus important to quantify these aspects so as to assess the extent to which the mechanical properties of the capsules can be controlled, an issue which has not been addressed thus far in the literature and which may hinder the success of microfluidic technology in the domain of microcapsule synthesis. This is the goal of the paper. In this paper, we describe the formation of the compound drops in our microfluidic device and analyze the off-centering of the core with respect to the microcapsule center. Then, we achieve the in situ photopolymerization of the liquid shell into a solid one; finally, we compare the mechanical behavior of the synthetised capsules to numerical simulations and conclude that the mechanical properties of the microcapsules can be predicted on a quantitative basis. The three immiscible fluids are driven by syringe pumps. In our experiments, fluid 1 is a mixture of glycerin and water (the core), fluid 2 is tripropylene glycol diacrylate (TPGDA) (the shell), and fluid 3 is silicon oil (the continuous phase) (see Figure 1). The microdevice consists of two consecutive flow focusing junctions which generate double emulsions according to a two-stage process. At low flow rates q1 and q2 (here the indices refer to the fluids), the aqueous phase is symmetrically pinched off at the first junction to form monodisperse aqueous plugs of the monomer phase. At the second junction, the oil phase encapsulates liquids 1 and 2, forming double droplets of aqueous and monomer phases in silicon oil. The droplets further reach a third junction where the channel cross section is expanded by a factor of 6.25. This leads the compound droplets to adopt spherical shapes. Mass conservation forces the droplets to slow down considerably in the large section, reducing the distances between consecutive droplets and thus favoring their coalescence. For this reason, additional oil is supplied to keep the drops well separated in the wide section (see the V-shaped entries in Figure 1). The distance between successive compound drops along with their speed is 7858 DOI: 10.1021/la9004449
Hennequin et al.
Figure 2. Relative shell thickness h = (Douter - Dinner)/Douter
against ratio q2/q1. The solid line represents 1 - (1 + q2/q1)-1/3 inferred from mass conservation constraint. In all cases, the outer diameter Douter is 180 ( 20 μm. Disks: q1 = 1 μL/min; squares: q1 = 2 μL/min; triangles: q1 = 4 μL/min. The images show two typical drops for the ratios q2/q1 = 0.25 and 1 corresponding to relative shell thickness h of 7% and 21%.
adjusted by changing the oil flow rate in the additional entries. This quantity is typically set so that the drops are separated by approximately two drop diameters. Our flow geometry provides comfortable ranges of flow rates within which double droplets are generated with an excellent size monodispersivity (∼3%). A detailed analysis of the corresponding flow conditions is out of the scope of this article. In practice, we operated at low capillary numbers Ca = μU/γ, where μ is a fluid viscosity, U a typical velocity, and γ an interfacial tension. We worked with Ca at most ∼0.1, which corresponds to the socalled dripping regime;a regime dominated by absolute instabilities19;and for which excellent size control can be achieved. In practice, we restricted ourselves to flow rates below 4 μL/min. Throughout the experiments, i.e., for the flow conditions we considered, the outer drop diameter Douter is equal to 180 ( 20 μm. Figure 2 shows the relative shell thickness of the core shell droplet defined by h ¼
Douter -Dinner Douter
as a function of the flow rate ratio q1/q2 of the core phase over the intermediate phase. The evolution of h with the flow rate ratio can be well accounted for by invoking mass conservation. The corresponding theoretical curve is given by the formula Douter -Dinner q2 -1=3 ¼ 1- 1 þ h ¼ Douter q1
ð1Þ
which accurately represents the measurements. Our experiments thus demonstrate that control of the shell thickness h is achieved by tuning the flow rate ratio q2/q1. In our experiments, relative shell thicknesses could be varied from 7% up to 50% with excellent control and reproducibility. We now come to the positioning of the core with respect to the compound droplet. As underlined in a recent study,12 after the double droplet is formed, the inner droplet moves relatively to the center of the globule as the double drop progresses downstream. (19) Huerre, P.; Monkewitz, P. A. J. Fluid. Mech. 1985, 159, 151.
Langmuir 2009, 25(14), 7857–7861
Hennequin et al.
Article
Figure 3. Influence of liquid shell phase viscosity on off-centering. The normalized center-to-center distance β = 2δ/(Douter - Dinner) between the outer and inner droplets is plotted as a function of the viscosity of the shell phase μm relative to that of the external fluid, μo. In each case, the double emulsion droplets are flowing into a straight channel at the same velocity of 3 mm/s and at the same flow rates. The full line represents the theoretical line (3) obtained with τ ∼ 60 ms.
This phenomenon is due to the action of recirculating flows that develop inside the compound droplet as it moves downstream, exerting viscous drags that displace the core with respect to the globule. The displacement mechanism must obviously be controlled to be able to produce solid capsules that are fully closed and more generally to obtain objects with predictable geometrical and mechanical properties. The effect of the confinement on the hydrodynamics of compound drops was analyzed for channel restrictions20 and straight channels.12 We shall use the description of ref 12 as a reference for the analysis of our measurements although this paper assumes cores significantly smaller than the microcapsule diameter. The off-centering parameter can be defined as follows: β ¼
2δ Douter -Dinner
ð2Þ
in which δ is the center-to-center distance. According to ref 12, one may expect the following evolution for β: β≈
Uτ ðDouter -Dinner Þð1 þ λÞ
ð3Þ
in which U is the double droplet speed, τ is a fixed time, and λ = μm/μo is the viscosity ratio of the intermediate phase (the monomer) over the external phase (silicon oil). By varying the monomer concentration in the shell, we could vary the viscosity ratio λ by more than 1 order of magnitude. We could thus test formula 3 with U fixed. The result is shown in Figure 3. The plot represents the off-centering β after a time τ following the double droplet formation at the second junction. As expected on physical grounds, the off-centering β is considerably reduced by increasing the monomer viscosity μm. Formula 3 sketched in Figure 3 with τ as the fitting parameter shows trends similar to the experiment. The modest quality of the agreement between the theory and the experiment probably expresses the fact that, in contrast with ref 12, the inner droplet cannot be considered as much smaller than the capsule. The theoretical description of the dynamics of the core in straight microchannels certainly remains to be done. In the rest of the paper, we will use a viscosity ratio λ equal to ∼6 so as to limit the off-centering to less than 20%. (20) Zhou, C.; Yue, P.; Feng, J. J. Int. J. Multiphase Flow 2007, 34, 102–109.
Langmuir 2009, 25(14), 7857–7861
Figure 4. Images of photocured capsules, 180 μm in diameter. (A)
SEM image of a cut-open capsule with a ∼8 μm thick shell. (B) A population of capsules showing monodispersivity. (C) Manually crushed capsules suggesting cracks are initiated at the poles.
We now turn to the photopolymerization of the double droplets, aiming at obtaining microcapsules with controlled characteristics. We use the microscope fluorescence mercury lamp to provide the necessary UV radiation through a 10 objective (numerical aperture 0.25). The polymerization is performed in situ the microsystem, right after the channel expansion (see Figure 1). The exposure time is typically of the order of ∼1 s, during which the polymerization proceeds. The conversion of the acrylate monomer is fast, and the shells readily solidify around the liquid core during the UV exposure. The solid capsules are further collected at the outlet of the microsystem and washed several times in order to remove unreacted monomer. Figure 4A shows an image of a solid capsule obtained in our system, and Figure 4B demonstrates size monodispersivity of such capsules. In practice, we estimate the size dispersivity we could obtain on the order of 5%. DOI: 10.1021/la9004449
7859
Article
Figure 5. Measured force-displacement curve for single capsules with shell thicknesses h = 0.2 (squares) and h = 0.3 (disks), corresponding to flow rate ratios q2/q1 of the middle to the inner phase equal to 1 and 2, respectively. The measured force is normalized by the cross-sectional area of the capsules πR2. The dashed lines are the numerical simulation described in the text.
We further probed the mechanical properties of individual capsules by compressing them between two flat plates. In the test we carried out, the displacement is imposed while the load applied on the capsule is measured. Figure 5 shows the measured stressstrain curves for capsules with dimensionless shell thicknesses h (i.e., the shell thickess divided by the capsule radius; see formula 1) of 0.2 and 0.3. In this plot, F is the applied force and Δ is the plate displacement. As expected, the capsules with thicker shells are harder. Interestingly, the response of the capsules seems linear up to large relative displacement Δ/R of ∼15-20%. We used the homemade CEA Cast3 M finite element code21 to reproduce this behavior. The numerical calculation is axisymmetric and is based on four-node linear elements. By symmetry, we restrict ourselves to the upper quarter of the space. In the simulation, the microcapsule is assumed to be compressed between two infinitely rigid plane surfaces. The contact between the platen and the capsule is frictionless (in practice, this is achieved by maintaining the nodes located at the north pole of the sphere below the platen level during the deformation process). Throughout the calculation, we consider the liquid volume enclosed by the capsule is conserved; in practice, this is achieved by increasing the internal pressure at each step of the simulation so as to satisfy this constraint. In our procedure, we impose the platen displacement and calculate stresses and sphere deformations. The total load exerted by the platen onto the capsule is an output of the computation. Physically, it results from a combination of the capsule mechanical resistance and the internal liquid pressure. The mechanical behavior of UV-cured TPGDA is known in the literature.22,23 It is characterized by a Young modulus Ecap = 1.0 GPa in the elastic range and a saturation level located at σy ∼ 20 MPa for large deformations. We could reproduce the behavior of our capsules by using an elastic perfectly plastic model with 1 GPa as the Young modulus and a yield stress equal to 25 MPa. Figure 6 shows the distribution of the principal stress in the microcapsule for h = 0.3 and for a relative displacement Δ/R = 0.07. We can see that the maximum stresses are located at the north pole of the capsule, which suggests that capsule rupture (21) Cast3M F.E. software, CEA, www-cast3m.cea.fr, 2007. (22) Olivier, A.; Pakula, T.; Best, A.; Benmouna, M.; Coqueret, X.; Maschke, U. Mol. Cryst. Liq. Cryst. 2004, 412, 461–467. (23) Olivier, A.; Benkhaled, L.; Kada, F.; Berrayah, A.; Pakula, T.; Ewen, B.; Maschke, U. http://frsl06.physik.uni freiburg.de/DFKG/archive/2005/pdf/p53. pdf.
7860 DOI: 10.1021/la9004449
Hennequin et al.
Figure 6. Distribution of the principal stress in the microcapsule, for h = 0.3 and a relative displacement Δ/R = 0.07. The figure shows the existence of large compressive and tensile stresses in the north polar region, suggesting the capsule will likely develop fractures in that region as load increases.
is likely to appear in this region. In Figure 6, the internal pressure we may estimate with the numerical model is on the order of 3 MPa. The numerical calculations of the load-displacement behavior are compared to the experiment in Figure 5. Remarkably, we find good agreement between the calculation and the experiment, using material properties consistent with the literature21,22 throughout a range of deformations extending up to 20%. For larger deformations, the experimental load tends to level off while the numerics keeps raising up. The leveling off of the deformation with the pressure probably marks the development of cracks in the structure. In fact, we observed that a capsule subjected to loads on the order of 50 mN breaks up, as shown in Figure 4C. In the experiments we made, the shells break with their two hemispheres tearing apart, thus releasing their content. Interestingly, in all cases, the fracture is initiated in the polar regions, i.e., where the principal stresses reach the largest levels. The numerical simulation thus also indicates how microcapsules are expected to break up. To summarize, we have shown that by using lithography-based microfluidic technology, monodisperse single-core microcapsules with controlled geometrical and mechanical properties can be produced. Shell thicknesses can be varied from a few percent up to 50% just by changing flow conditions. The mechanical response of the microcapsules can be controlled by changing the flow rate ratio of the phases that are driven through the chip. The agreement we obtain between the simulation and the experiment concerning the mechanical properties of the capsules is a signature of the excellent control we have on the capsule geometry. As a whole, the level of control we reach here cannot be obtained with current technologies. Combined with the possibility of parallelizing the synthesis process, the present work opens routes toward the production of novel microcapsules with tunable mechanical properties.
Materials and Methods Fabrication of Microdevices. The fabrication of the microdevice is made according to standard soft lithography techniques. A ∼75 μm layer of SU8 is first spin-coated on a silicon wafer and insulated with the entire microdevice pattern. Before dissolving the uncured resin, a second ∼220 μm thick layer is again spincoated on the previous one. At this point, only the region corresponding to the outlet channel is then cured in order to produce a device with channels of two different heights (see Figure 1). The wafer is then washed with the appropriated developer, cleaned with isopropanol, and dried. The heights of Langmuir 2009, 25(14), 7857–7861
Hennequin et al. the different channel are then measured with a stylus profiler (Dektak). A mixture of PDMS elastomer and curing agent (Sylgard 184, Dow Corning) in proportion 10:1 is poured on the mold and degassed under vacuum until no bubbles remained visible. The PDMS is then cured for 45 min at 70 °C in order to make a replica. Inlets and outlets holes are punched with a blunt needle, and the device is immediately closed with a glass slide freshly spin-coated with a 100 μm PDMS layer which has been cured for 25-30 min only. Once assembled, the device is left for at least 2 h in at 70 °C in order to strengthen the adhesion of the two partially cured PDMS sides. Fluids Composition. For the inner phase we use aqueous solutions (Milli-Q) of glycerol (Sigma), up to 65% w/w. The middle fluid consists of pure or mixtures of the acrylate monomers tripropylene glycol diacrylate (TPGDA), trimethylolpropane triacrylate (TMPTA), and pentaerythritol triacrylate (PETA) (all from Sigma and used without purification). We typically use a 30:70 w/w mixture of TPGDA and PETA with a viscosity μ2 = 130 mPa 3 s. The external fluid consists of silicone oil (CRC industries) with a viscosity μ3 = 20 cP. Polymerization. For the polymerization we add 4-8% w/w of the photoinitiators Darocure 1173 (Ciba) or 1-hydroxycyclohexyl phenyl ketone (Sigma) to the monomer. The shells
Langmuir 2009, 25(14), 7857–7861
Article are cured in situ the reaction channel, using a 10 objective and the microscope’s fluorescent lamp with a 340-380 nm UV filter. Compression Test. A force transducer (400A, Aurora Scientific. Inc.) is mounted on the stage of an inverted microscope stage with the force probe aligned with the optical axis. The motion of the transducer is computer-controlled with a stepper motor to submicron accuracy. Microcapsules are pipets, deposited on a glass slides and placed in the sample holder of the microscope. Capsules are aligned with the probe, and the compression test is started. At each compression step (∼0.5-1 ms) the force applied on the probe is recorded, and an image is taken with a CCD camera until the transducer reaches its maximum load of 50 mN. For each sample, we take into account the compliance of the probe (0.1 μm/mN) and average the force-displacement curves measured for ∼10 microcapsules.
Acknowledgment. We acknowledge the members of MMN for numerous discussions, in particular F. Monti for the realization of the SEM pictures. We are grateful to D. Barthes Biesel and E. Lac for enligthening discussions. The work was supported by ESPCI, CNRS Region Ile de France, and Unilever.
DOI: 10.1021/la9004449
7861
