Designer Polymer-Based Microcapsules Made Using Microfluidics

Published: November 27, 2011

r 2011 American Chemical Society 144 dx.doi.org/10.1021/la203088u | Langmuir 2012, 28, 144–152

ARTICLE

pubs.acs.org/Langmuir

Designer Polymer-Based Microcapsules Made Using MicrofluidicsPhilipp W. Chen, Randall M. Erb, and Andr�e R. Studart*

Complex Materials, Department of Materials, ETH Zurich, 8093 Zurich, Switzerland

1. INTRODUCTION

Filled microcapsules play an important role in the delivery andrelease of materials in medicine, cosmetics, food, and even auto-nomic self-healing materials.1�5 The capsule behavior in suchfunctions depends on the mechanical properties and the perme-ability of the shell, which in turn are tunable by the capsule size,shell thickness, and composition. For controlled and specificrelease, encapsulation systems that provide fine control overthese parameters are often desired.

Currently, capsules can be manufactured from a wide range ofmaterials using liquid or solid templates. Pickering or surfactant-stabilized emulsions are convenient liquid templates to producecapsules with polymeric or ceramic shells through interfacial poly-merization,6 self-assembly,7�9 or condensation processes.10 How-ever, the uncontrolled nature of conventional emulsification routesoften results in a wide size range, making it difficult to control theshell properties. Furthermore, the composition of the shell issometimes limited to specific chemistries. Solid templates such ascolloidal latex or silica particles can also be coated by adsorption,11

layer-by-layer (LbL) techniques, and sol�gel chemistry12 to pro-duce core�shell particles from a large variety of materials. Whilemonodisperse sizes can be achieved, the solid inner core must beremoved thermally or by dissolution, requiring a permeable shelland subsequent reloading with the encapsulant. Despite thesenumerous different approaches and materials available to producemicrocapsules, simultaneous control over the capsules’ dimensionsand chemical composition combined with high encapsulationefficiencies is often very difficult to achieve in a one-step process.

Double emulsion templating in microfluidic devices has beenshown to overcome this difficulty by enabling fine control of thetemplate dimensions and offering high flexibility with regard to

the possible materials that can be used to later form the capsuleshell.13�17 In this process, controlled double emulsification isachieved by dripping or jetting an inner fluid into an immisciblesecond fluid, which is in turn engulfed by a third fluid. Capsulesare formed by consolidating the middle phase (second fluid) ofthe resulting double emulsions, which is typically achieved bypolymerization of monomers16�18 or evaporation-induced jam-ming of colloidal particles initially added to the middle. Finecontrol of the fluid flows leads to defined capsule and shelldimensions. This elegant encapsulation approach has been usedto produce monodisperse colloidosomes,19�21 hydrogel cap-sules,22 ceramic capsules,23 and even capsuleswith composite shells.19

However, due to the many parameters affecting the double emul-sification in microcapillary devices and their complex flowpatterns, there are only a few predictive tools and guidelines thatwould enable deliberate design of the dimensions and propertiesof microcapsules produced via this route.

Here, we investigate the influence ofmicrofluidic experimentalparameters on the sizes and shell thicknesses of double emulsiontemplated microcapsules and explore possible polymeric systemsand colloidal particles that can be used within the shell material totailor the capsules’ permeability and mechanical properties.Based on thorough experimental work and theoretical analysis,we present quantitative and qualitative tools to obtain polymer-based capsules with predictable and well-defined size and shellthickness, aswell as tunablemechanical behavior, shellmicrostructure,and shell permeability. The suggested guidelines are expected to assist

Received: August 8, 2011Revised: November 24, 2011

ABSTRACT: Filled microcapsules made from double emulsion templates in micro-fluidic devices are attractive delivery systems for a variety of applications. Themicrofluidic approach allows facile tailoring of the microcapsules through a largenumber of variables, which in turn makes these systemsmore challenging to predict. Toelucidate these dependencies, we start from earlier theoretical predictions for the size ofdouble emulsions and present quantitative design maps that correlate parameters suchas fluid flow rates and device geometry with the size and shell thickness of mono-disperse polymer-based capsules produced in microcapillary devices. The microcap-sules are obtained through in situ photopolymerization of the middle oil phase ofwater-in-oil-in-water double emulsions. Using polymers with selected glass transitiontemperatures as the shell material, we show through single capsule compression testingthat hollow capsules can be prepared with tunable mechanical properties ranging fromelastomeric to brittle. A quantitative statistical analysis of the load at rupture of brittle capsules is also provided to evaluate thevariability of the microfluidic route and assist the design of capsules in applications involving mechanically triggered release. Finally,we demonstrate that the permeability and microstructure of the capsule shell can also be tailored through the addition of cross-linkers and silica nanoparticles in the middle phase of the double emulsion templates.

145 dx.doi.org/10.1021/la203088u |Langmuir 2012, 28, 144–152

Langmuir ARTICLE

the design and preparation of polymer-based microcapsules withdeliberately tailored properties.

2. EXPERIMENTAL PROCEDURES

Device Fabrication. We build glass capillary microfluidic devicesby aligning two cylindrical glass capillary tubes inside a square tube onglass slides, as first outlined by Utada et al.13 (see Figure 1). The twocapillaries are tapered using a P-97 Flaming/Brown micropipet puller(Sutter Instrument, USA). Their tips are then adjusted with a MF-830microforge (Narishige, Japan) or manually polished to about 140 to150 μm diameter for collectors and around 24 μm in the case of emittercapillaries. For the droplet size prediction experiments, we also use largercollectors of 180 μm to facilitate device fabrication. To prevent wettingof the emitter capillary by the aqueous inner fluid and thus ensure ef-ficient dripping, we hydrophobize the capillary tip using a solution ofoctadecyltrimethoxysilane (ODTMS, Acros Organics, Belgium) andbutylamine in toluene.24 Unless noted otherwise, all chemicals are pur-chased from Sigma Aldrich (USA) and used without further purification.The capillaries are spaced 70 to 80 μm apart and fixed using epoxy glue.Preparation of Fluids. As the inner and outer fluids, we use 2 wt %

aqueous solutions of poly(vinyl alcohol) (PVA, Mw 31000�50000,87�89% hydrolyzed). Fluorescein sodium salt is added to the innerphase as a fluorescent contrast material. For the middle phase of themodeling experiments, we use isobornyl acrylate (IBOA, TCI Deutsch-land, Germany) with 4 wt % sorbitan trioleate (Span 85). To makecapsules, we use IBOA, 2-phenoxyethyl acrylate (PEA, TCI Deutsch-land, Germany), 1,6-hexanediol diacrylate (HDDA, TCI Deutschland,Germany), silica particles of 250 and 500 nm diameter (Fiber OpticCenter, USA), Span 85, toluene, hexadecane (ABCR, Germany), andthe photoinitiator Darocur 1173 (BASF, Germany).

To disperse the silica particles, we first hydrophobize them in anethanol solution containingODTMS and butylamine.24 After stirring forone day, the suspension is then centrifuged in a Universal 320 centrifuge(2700 RCF, 15 min, Hettich Laborapparate, Switzerland) and redis-persed in toluene multiple times to wash out residual ODTMS. The re-sulting hydrophobized silica particles are readily dispersible in themonomer solutions, thus enabling the preparation of stable silica/mono-mer suspensions later used as middle fluid. A similar procedure is alsoused tomodify the particles with 3-aminopropyltrimethoxysilane (AcrosOrganics, Belgium). To disperse untreated silica particles in the mono-mer solution, we use 10 wt % Span 85 relative to the silica content. Allsilica/monomer suspensions are dispersed using a Vibra-Cell VCX 130sonicator (Sonics &Materials, USA) for at least 5 min and remain stablefor the duration of the experiments.Double Emulsions and Capsules.Double emulsions are formed

by pumping the three fluids through the microfluidic device using PHD2000 syringe pumps (Harvard Apparatus, USA). The inner fluid ispumped through the inlet capillary into the coflowing middle fluid. Bothphases are flow-focused into the collector capillary together with a thirdouter fluid pumped through the opposite extra-capillary region (Figure 1).Under the right experimental conditions, monodisperse double emul-sions can be obtained. Typically, the flow rates range from 10000 to30000 μL/h, 1000 to 3000 μL/h, and 500 to 1500 μL/h for the outer,middle, and inner fluids, respectively. To make capsules, the doubleemulsions are collected in 15 mL vials containing water and simultaneouslyexposed to UV light from an Omnicure 1000 light source (APM Technica,Switzerland) with an iris opening of 50% for at least 2 min.

Double emulsions and microcapsules are investigated under a DM6000B fluorescence microscope (Leica, Germany) and in a LEO 1530scanning electron microscope (SEM, Carl Zeiss SMT, Germany). Priorto SEM analysis, a 5 nm platinum layer is applied to the capsules using aSCD 050 sputter coater (Bal-Tec, Liechtenstein). To view cross-sectional areas, the capsules are manually ruptured between two glass

slides. For the size analysis of capsules or double emulsions, we measureat least 100 capsules per sample, and we correct for optical aberrationsdue to the different refractive indices of the fluids involved.

To quantify the permeability of our capsules, we collect samples withdefined amounts of dye-filled capsules, which are washed multiple timeswith water. All capsules are produced with the samemicrofluidic device andflow settings. After various storage times, the supernatant liquid is removedand analyzed with a CARY 1E UV/vis spectrophotometer (Varian, USA).From a calibration curve obtained from known concentrations of the samedye, we quantify the average leakage of the capsule sample.

We perform compression tests on individual microcapsules using acustom-made instrument based on the setup proposed by Keller and Sot-tos.25 The compression tester consists of a DC motor actuator and con-troller (M-230.25 DC and C-863Mercury, Physik Instrumente, Germany)to deform individual capsules at a deformation rate of 0.5 μm/s and a 30 gload cell (GSO, Transducer Techniques, USA) tomeasure the applied load.Acquisition of live load and displacement data is performed by a NI-9237DAQ module (National Instruments, USA) and the motor controller,respectively. During the test, the in situ deformation of capsules is followedlaterally with a Z16 APO macroscope (Leica, Germany).

3. RESULTS AND DISCUSSION

Formation of Double Emulsions and Microcapsules. Theformation of double emulsions in a typical microcapillary deviceis shown in Figure 1. The water-in-oil-in-water droplets areformed at a rate of approximately 15000/min. All emulsionsremain stable for several minutes or more, if a surface-activepolymer such as partially hydrolyzed PVA is used. In contrast,conventional surfactants like sorbitan monoleate (Span 80) orpolysorbate 20 (Tween 20) are not as effective in stabilizing thesystem. The efficient stabilization of droplets with PVA is likelydue to steric hindrance imposed by the polymer adsorbed at theliquid interface,26 which prevents close approximation betweenadjacent droplets. The relatively high shear conditions also favorinterfacial layers formed by larger and thus less displaceablemolecules. The addition of nanoparticles into the middle phase isfound to enhance the emulsion stability to several hours. This isin agreement with our previous studies on the formation ofcolloidosomes from double emulsions21 and it supports the ideathat steric forces provide a very effective mechanism to stabilizedroplets made in microfluidic devices.Double emulsions containing PVA in the aqueous fluids are

stable enough to enable successful polymerization of the mono-mer middle phase. By this means, emulsions are converted intomicrocapsules with virtually perfect yields (>99%) withoutleakage of the inner phase (see Figure 2a).The accurate control over the emulsifying shear forces enabled

by the microfluidic approach results in highly monodisperse

Figure 1. Picture of an actual glass capillary microfluidic device indica-ting the ports of the three fluid flows Qo, Qm, and Qi. The inset sche-matically shows the formation of double emulsions at the capillaryjunction. The red dotted lines represent (1) inner thread diameter dth,i,(2) the outer thread size dth,o, (3) the collector diameter C, and (4) theeffective collector size Ceff.


Langmuir ARTICLE

emulsions and capsules. For all samples, the inner diameter di andouter diameter do feature a polydispersity of 5% or less (seeFigure 2b). Monodisperse double emulsions and capsules withsizes ranging from 70 to 250 μm and shell thicknesses rangingfrom 7 to 50 μm were prepared in the microcapillary devices, asillustrated by the representative images shown in Figure 2c.Size Control. The outer diameter of the precursor emulsions

do and thus the capsule size can be varied within a broad range bychanging the size of the collector capillary C (see Figure 3a). Forone specific device with constant C, we find that adjusting theouter, middle, and inner fluid flow rates (Qo, Qm, and Qi,respectively) allows for fine control of the capsule dimensions(see Figure 3b and Figure 5a).The strong dependence of the outer droplet size do on the

device dimensions and flow rates can be understood with thehelp of an analytical model describing the sizes of drippingdroplets made in microfluidic devices.27 The droplet size isdetermined by the interplay between the shear force of the sur-rounding fluid promoting droplet break-up and the interfacialtension of the fluids that keeps the droplet attached to the inletcapillary or the main fluid body. Dripping occurs when the ratio

between these forces, known as the capillary number, reaches acritical value Cacrit. For double emulsions, there are two possiblesequences of dripping events, depending on whether the inner orthe outer droplet reaches Cacrit first. In our experiments, weobserve and consider only the case where the capillary number ofthe outer droplet reaches the critical value. The shear force isapproximated as a Stokes drag force, while the interfacial tensiontakes into account the break-ups of both the inner and themiddlefluids.27 The outer droplet size do can then be obtained by solvingthe following quartic equation:

do4 þ ΓðQo þ Q�Þdo3 � ½C2 þ Γdth, oðQo þ Q�Þ�do2

� ΓQ�C2do þ ΓQ�C2dth, o ¼ 0 ð1Þ

where

Γ ¼ 12μπCacritðdth, oγo þ dth, iγiÞ

ð2Þ

dth, o ¼ C

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiQm þ Qi

Qo þ Qm þ Qi

rð3Þ

dth, i ¼ dth, o

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiQi

Qm þ Qi

rð4Þ

Figure 3. (a) Overview of capsule sizes produced by different devicesusing various types of fluids. Each column of data of one given colorrepresents specific collector diameter and a practical sample of capsulesproduced by adjusting the flow rates of the fluids. (b) Influence of outerflow rate on dimensions of double emulsions produced with a 180 μmcollector (Ceff = 194 μm), a 22 μm emitter and an intercapillary distanceof 85 μm. The other flow rates are kept constant atQm = 3000 μL/h andQi = 1500 μL/h. The symbols are experimental data and the linesrepresent model predictions.

Figure 2. (a) Fluorescence micrograph of microcapsules showing thedye-filled inner capsule compartments. The scale bar represents 200 μm.(b) Size distribution for the inner and outer diameters of a representativebatch of microcapsules (2 μm binning). (c) Variety of capsules madeusing microfluidics. The scale bar is 250 μm and applies to all samples.


Langmuir ARTICLE

Q� ¼ Qm

1� Qi

Qi þ Qm

� �2=3ð5Þ

Here, dth,o is the diameter of the thread formed between thetwo inner fluids and the outer flow-focusing fluid, whereas dth,i isthe size of the thread formed between the inner and the middlefluid (see Figure 1). Both are derived using mass balanceassumptions. μ is the viscosity of the outer fluid, γo and γi arethe interfacial tensions between the outer and middle andbetween the middle and inner fluids, respectively. Since we usetapered collector capillaries, the double emulsions may forminside the capillary rather than at its tip. In this case,C needs to bereplaced by the effective diameter Ceff at that location, which istypically 10�15% larger than the original C. In a previous study,we showed that Cacrit = 0.1 is a good choice for the systemsinvestigated here. We also measured μ = 0.0016 mPa 3 s and γo =γi = 27.9 mN/m.27

For the experimental conditions used in this study, we find thatthe thread diameter dth,o is typically smaller but still comparable tothe resulting outer droplet size do. Based on eq 1, we find that thediameter of the collector capillary serves as the main parameter forcoarse tuning of the droplet sizes. Indeed, we observe this strongrelation between do andC if the experimental results from a series of11 different devices are plotted in one single graph (Figure 3a).Finer tuning of the outer droplet diameter can then be achieved in

one given device by controlling the flow rate of the outer fluid,as exemplified in Figure 3b. The resulting diameter do is welldescribed with our predictive model, as shown by the theo-retical fitting curve in Figure 3b. Our results verify that themodel is a useful tool to find out the outer flow rate requiredto accurately tune the capsule size.While the diameter do is determined by shear and interfacial

forces acting on the outer droplet, the size of the inner droplets ofthe double emulsions di is governed by a simple mass balance thatleads to the following equation:

di ¼ do

1 þ Qm

Qi

� �1=3ð6Þ

Such a linear relation between the inner and outer diameters isindeed observed experimentally, as shown in Figure 4 for arepresentative flow rate ratio Qm/Qi of 2. The linear coefficient of0.71 obtained from our experimental data is close to the theoretical

value of 0.69 expected from eq 6, confirming the applicability of thisrelation as a tool for predicting the inner droplet diameter.Shell Thickness Control. The shell thickness τ of microcap-

sules can be varied by changing the ratio of flow rates of the middleand inner fluids, as depicted in Figure 5a. This dependence can beunderstood by rearranging the mass balance eq 6, resulting in thefollowing dimensionless and explicit relation for the relative shellthickness s:

s ¼ 2τdo

¼ 1� 1 þ Qm

Qi

� ��1=3" #

ð7Þ

The experimental results shown in Figure 5b reveal that eq 7describes very well the dependence of the shell thickness on theratio of the two inner fluids, Qm/Qi, in good agreement withprevious studies on similar double emulsion systems.17,19 Inaddition, Figure 5a also shows that eqs 1 and 6 are also effective inpredicting droplet diameters when the flow rate of the inner fluidis varied over a broad range. Effective capillary diameters in the

Figure 4. Linear correlation observed between the inner droplet size diand the outer droplet diameter do for a fixed flow rate ratioQm/Qi of 2, inagreement with eq 6.

Figure 5. (a) Shell thickness τ of double emulsions produced with a140μmcollector. The outer flow rate and the sumof themiddle and innerflow rates are kept constant at 15000 μL/h and 4000 μL/h, respectively.The solid lines represent the predicted sizes calculated from eqs 1 and 6,with the theoretical shell thickness defined as the difference between theouter and inner radii. (b) Calculated (solid line) and experimental(symbols) relative shell thickness as a function of the flow ratio betweenthe middle and the inner fluid. (c) SEM micrograph of the full cross-section of a ruptured microcapsule. The scale bar represents 20 μm.


Langmuir ARTICLE

range 1.08C < Ceff < 1.2C were measured experimentally to ensureaccurate predictions.Pannacci et al. found that, in double emulsion templates, the

inner droplet moves relative to the center of the outer one becauseof recirculating flows, while the double emulsion moves down-stream before it is polymerized.28 A measure for the resulting off-centeredness is given by the ratio of the center-to-center distanceΔ of the inner and outer droplets to the maximum off-centered-ness, i.e., the difference between the outer and inner droplet radii.This off-center parameter δ is thus defined as

δ ¼ 2Δdo � di

ð8Þ

In our experiments, we observe average off-center parameters inthe range of 6% to 11%, which indicates a good homogeneity com-pared to similar capsules produced with microfluidic techniques.28

This is confirmed by SEM micrographs of the cross sections ofmanually ruptured capsules, as exemplified in Figure 5c.Minimum shell thickness can be achieved by increasing the

flow rate of the outer fluid (decreasing do) while keeping theother flow rates constant, and by increasing the inner flow raterelative to the middle flow (minimizing Qm/Qi), as predicted byeq 7. To evaluate the lower limit for the shell thickness, weattempt to maximize the flow rates of the outer and inner fluids.For the fluids, surface active species and device geometry used inthis study, we are able to make stable capsules with minimumshell thicknesses in the range 7�10 μm. Double emulsions withsmaller thicknesses are more challenging to produce, as thethinner middle phase makes themmore susceptible to rupture inthe high-shear microfluidic environment.This limitation can be overcome by adding another solvent to

the middle fluid that is slightly soluble in the outer fluid, leading tothinner shells upon solvent evaporation after the polymerization

step. The mutual solubility of the solvent destabilizes the doubleemulsions to the extent that PVA surfactant alone is not sufficient toprevent coalescence of the droplets. Thus, stabilizing particles needto be added as well. Additionally, this technique also offers a way toovercome the high viscosity of middle phase suspensions with highvolume fraction of particles, thus enabling the preparation of capsuleshells with increased particle content.For the double emulsion system investigated here, toluene is

used as the additional evaporative solvent due to its miscibility inthe middle oil phase and slight solubility in the outer aqueousphase.21,29,30 To demonstrate our ability to obtain thinner shellsthrough solvent removal, we use a middle oil phase consisting of45 vol% toluene, 40 vol% IBOA, and 15 vol% silica nanoparticles.After polymerization under UV exposure, the capsules arecollected in open containers and evaporation of toluene proceedsat room temperature under slight agitation to prevent adhesionof the toluene-swollen shells. After 6 days of evaporation, theshell thickness indeed shrinks from an initial value of 17.4 μmdown to 11.4 μm, corresponding to a volume reduction of 43%(Figure 6a). Both the final shell thickness and the volumereduction agree well with the theoretically expected values of11.0 μm and 45%, respectively. In other samples, starting outfrom thinner original shells, we obtain final thicknesses as low as5 μm. SEMmicrographs of the dried capsules reveal a dense shell,as depicted in Figure 6b, indicating that the removal of thesolvent does not cause any porosity within the shell. Currentresearch is dedicated to exploring the lower limit in shellthickness achievable with this method.Design Maps. In view of their accurate description of the

experimental droplet size data (Figures 3, 4, and 5), one can use

Figure 6. (a) Thickness of capsules with toluene-swollen polymer shellsas a function of evaporation time. The dashed line represents thetheoretical shell thickness expected after complete removal of thesolvent. The inset shows capsules before (left) and after 144 h (right)of drying. (b) SEMmicrograph of amanually ruptured capsule shell aftersolvent removal. The scale bar represents 1 μm.

Figure 7. Examples of design maps to predict (a) the outer droplet sizeand (b) the thickness of the middle phase of double emulsion based oneqs 1 and 6. In (a), the outer droplet sizes for varying Qo and C areobtained with Qm = 3000 μL/h and Qi = 1500 μL/h. Experimental datafrom Figure 3b are included in the map to confirm the close agreementbetween experiments and theory. In (b), Qo and the sum of Qm and Qi

are kept constant. The dashed lines represent effective collector dia-meters Ceff = 1.2C.


Langmuir ARTICLE

eqs 1 and 6 to build design diagrams that explicitly correlate themajor microfluidic parameters with their effect on the size of theouter and inner droplets produced in the glass capillary devices.Examples of such design maps for the fluids and surface-activespecies used in this study are shown in Figure 7. Given the vari-ability of the effective collector diameter Ceff, at which thedroplets break off, these design maps should either incorporateCeff directly (Figure 7a) or present a range of predictions forpractical values ofCeff (Figure 7b) to be used as guidelines for thedesign of capsule dimensions.Permeability. The permeability of the capsule shells changes

upon addition of variable concentrations of a suitable cross-linkeror plasticizer of the shell polymer. This is illustrated here using1,6-hexanediol diacrylate (HDDA) as cross-linker and hexade-cane as a plasticizer of the shell polymer poly(isobornyl acrylate).We observe that neat poly(IBOA) capsules display leakage of thelow molecular weight (Mw 376 g/mol) fluorescent dye, whichstrongly increases with the addition of plasticizer (Figure 8).Adding small amounts of cross-linker decreases the permeabilityof the shell. Capsules with 0.3 wt %HDDA display extremely lowleakage of less than 0.1% of the dye even after a week, which ishalf the leakage of neat samples, whereas adding 0.5 wt %HDDAto the monomer phase completely seals the shell. This effect islikely caused by a reduction of the pore size of the polymernetwork due to the cross-linking and by a more pronounceddensification of the shell in the presence of the cross-linker.Polymeric shells become denser upon addition of the cross-linkerdue to the higher polymerization shrinkage of pure HDDA(17%) as compared to that of pure IBOA (11%).31 Conversely,plasticizing the polymer shell increases its pore size and thus itspermeability.Mechanical Properties. The use of double emulsions as tem-

plates allows for the preparation of capsules with a wide variety ofshell materials with corresponding influence on the capsules’final properties. Taking advantage of the flexibility of the method,we fabricate capsules containing polymers with very differentglass transition temperatures as shell material. For example, ifpoly(IBOA) with a glass transition temperature (Tg) of around52 �C is used to form the shell, rather stiff and brittle capsules areobtained. In contrast, poly(2-phenoxyethyl acrylate) (poly(PEA)),which has a Tg of only 5 �C, results in shells exhibiting uniqueelastomeric behavior. Monomer blends can also be used to even-tually cover the entire spectrum of mechanical properties betweenthese extreme cases.The strong effect of the shell polymer on the mechanical

behavior ofmicrocapsules is illustratedby the load versus displacement

curves shown inFigure 9a.The results are obtained fromsingle capsulecompression tests of three different types of capsules: (1) anelastomeric capsule consisting of a 50/50 blend of PEA and IBOAwith a shell thickness of14.4(0.5μm,(2) abrittle capsule comprisingIBOA only (shell thickness: 15.1( 1.6 μm), and (3) a brittle capsulecontaining 90 vol% IBOA and 10 vol% hydrophobized silica nano-particles (shell thickness: 15.3 ( 1.1 μm). All capsules are around130 μm in diameter.The brittle shells display a quasi-linear load response with an

average fracture load of 71 ( 11 mN and rupture at about 23 (3.8% deformation with respect to their diameter. The incorpora-tion of fully hydrophobized silica particles into the brittle shellleads to embrittlement, as indicated by the smaller deformation(15 ( 3.6%) and load at rupture (58 ( 10 mN) of capsulesexhibiting particle-laden shells. This effect might be explained bythe weak interfacial bonding between the shell polymer and thesilica particles, which effectively work as stress-concentratingdefects within the brittle polymeric matrix. The rubbery capsulesshow much lower stiffness and can be deformed to over 70% with-out displaying catastrophic rupture. At such high deformations,the microcapsules eventually form a solid, flattened disk.

Figure 8. Dye leakage of neat poly(IBOA) and capsules plasticized withhexadecane (HD) obtained from UV/vis spectroscopy of the surround-ing supernatant fluid after various storage times.

Figure 9. (a) Typical load�displacement curves of brittle (red), rubbery(green), and particle-filled brittle (blue) capsules. The inlet shows thefracture mode of a 130 μm particle-laden capsule under verticalcompression. (b) Weibull distribution obtained from 20 brittle ([)and 23 particle-laden (9) capsules. (c) Compression tests of 9 elasto-meric capsules.


Langmuir ARTICLE

The increased contact area of this disk and the strain hardeningdue to chain orientation within the elastomeric shell lead to thestrong increase of the loading force shown in Figure 9c, at whichpoint the compression test is discontinued.The rupture of brittle capsules occurs through the propagation

of a longitudinal crack along the polymeric shell, as depicted inthe inset in Figure 9a. Due to their brittle nature, the maximumload applied to the IBOA-based capsules can be described usingWeibull statistics,32 as shown in Figure 9b. According toWeibull,the probability of failure P of brittle materials as a function of theexternally applied load F can be described by the following equa-tion:

PðFÞ ¼ 1� exp � FF0

� �m" #

ð9Þ

where m is the Weibull modulus and Fo is the characteristic load.The Weibull modulus m is given by the slope of the Weibull

plot in Figure 9b and is indicative of the reliability of the processused to produce the microcapsules. The Weibull moduli of 5.9and 6.7 obtained for poly(IBOA) capsules with and withoutparticles, respectively, are slightly lower but yet on the same orderof magnitude of those typically observed for conventional cera-mic manufacturing processes.33 Quantification of such variabilitythrough the Weibull modulus m and the characteristic load Foallows for the determination of the threshold load needed torupture a given percentage of microcapsules uponmechanical sti-mulus. This indicates a reasonably small variability of theinvestigated encapsulation method.The remarkably distinct mechanical response of the capsules

achieved by selecting polymers with tailored glass transitiontemperatures has important practical implications. Capsulesexhibiting brittle fracture are stiffer but can be ruptured at verylow strains, making them suitable for situations where the releaseof encapsulants is desired under low external strain. The releaseof a fluorescent dye used as model encapsulant upon manualrupture of a representative brittle capsule is illustrated in the toprow of Figure 10. Conversely, capsules exhibiting an elastomericshell might be of benefit in applications that require efficientprotection of the encapsulant. In this case, the elastomeric natureof the shell allows for extensive and partially reversible deforma-tion of the capsule without release of the encapsulated material,as shown in the bottom row of Figure 10.Shell Microstructure. In addition to decreasing the strength

of brittle capsules, colloidal particles can also be used to produceshells with unique microstructures. Depending on the particles’

surface chemistry, both homogeneous and layered structures canbe obtained. Full hydrophobization of the silica particles withoctadecyltrimethoxysilane (ODTMS) leads to a homogeneousdispersion within the capsule shell (Figure 11a). In contrast,untreated silica particles also form a monolayer on both the innerand outer shell surfaces, as shown in Figure 11b and c. Partiallyimmersed particles can be obtained for example through silaniza-tion with 3-aminopropyltrimethoxysilane (Figure 11d).The contact angle of the silica particles at the water/monomer

interface can be tailored through the degree or type of silaniza-tion, enabling the coverage of the entire spectrum from particlescompletely immersed in the monomer phase to interfaciallyadsorbed particles positioned mostly in the aqueous phase. Atruly layered structure with a virtually particle-free shell can alsobe obtained by adding a particle content to the middle fluid just

Figure 10. Manual deformation sequences of brittle poly(IBOA) (top row) and elastomeric poly(PEA) (bottom row) microcapsules filled withfluorescent dye. The top left figure shows how the capsules are compressed between a pair of tweezers. All scale bars represent 100 μm.

Figure 11. SEM micrographs showing fracture and outer surfaces ofcapsules with (a) fully hydrophobized, (b) untreated, and (d) amine-modified silica particles. Also shown is the inner surface of a capsule withuntreated silica (c), where the particles are covered by dried PVAsurfactant. All scale bars represent 1 μm.


Langmuir ARTICLE

enough to form a particle monolayer at the water/monomerinterfaces. Such structures coupled with the rich chemistry of silanesshould allow for specific surface modifications of the microcap-sules in targeted applications, while the adhesion and mechanicalstability of the particles on the shell surface can be improvedthrough controlled positioning of the particles at the liquid�polymer interface.

4. CONCLUSIONS

The size, shell thickness, permeability, mechanics, and micro-structure of monodisperse capsules made in microfluidic devicescan be designed and tuned within a broad range. Based on ourexperimental data and on theoretical predictions, we find that thedimensions of the microcapsules can be coarsely tuned with thediameter of the collector capillary and finely tuned by varying theflow rates. For one given device, we offer quantitative relationsbetween the outer and inner diameters of the double emulsionsand the flow rate of the outer fluid, allowing for fine-tuning of thecapsule size. The shell thickness can similarly be predicted andcontrolled through the flow rate ratio of the middle and innerfluids. The permeability of the resulting polymer-based micro-capsules is found to depend on the cross-linking or swelling of thepolymeric shell. Using hydrophobic monomers that form poly-mers of different glass transition temperatures enables tuning ofthe mechanical response of the capsules from elastomeric tobrittle. The strength of brittle capsules can be quantitatively des-cribed using Weibull statistics and is shown to decrease uponaddition of silica nanoparticles into the polymeric shell. In addi-tion to homogeneous composite shells, silica nanoparticles canbe surface tailored to obtain unique heterogeneous multilayeredshells. We expect that the quantitative and qualitative relationsput forward in this study will serve as guidelines for the designand fabrication of polymer-based microcapsules for a variety ofestablished and emerging applications.

’ACKNOWLEDGMENT

We thank Rebecca Huber and Elias Rehmann for their con-tributions to the experimental part of this study and ETH for thefinancial support through the ETHIIRA grant number 0-20676-10.

’REFERENCES

(1) Park, J. K.; Chang, H. N. Microencapsulation of microbial cells.Biotechnol. Adv. 2000, 18 (4), 303–319.(2) Tsuji, K. Microencapsulation of pesticides and their improved

handling safety. J. Microencapsul. 2001, 18 (2), 137–147.(3) Gouin, S. Microencapsulation: industrial appraisal of existing

technologies and trends. Trends Food Sci. Technol. 2004, 15 (7�8),330–347.(4) Shahidi, F.; Han, X.-Q. Encapsulation of food ingredients. Crit.

Rev. Food Sci. Nutr. 1993, 33 (6), 501–547.(5) White, S.; Sottos, N.; Geubelle, P.;Moore, J.; Kessler,M.; Sriram,

S.; Brown, E.; Viswanathan, S. Autonomic healing of polymer compo-sites. Nature 2001, 409 (6822), 794–797.(6) Brown, E.; Kessler, M.; Sottos, N.; White, S. In situ poly(urea-

formaldehyde) microencapsulation of dicyclopentadiene. J. Microencap-sul. 2003, 20 (6), 719–730.(7) Akartuna, I.; Tervoort, E.; Studart, A. R.; Gauckler, L. J. General

Route for the Assembly of Functional Inorganic Capsules. Langmuir2009, 25 (21), 12419–12424.(8) Akartuna, I.; Studart, A. R.; Tervoort, E.; Gonzenbach, U. T.;

Gauckler, L. J. Stabilization of Oil-in-Water Emulsions by Colloidal

Particles Modified with Short Amphiphiles. Langmuir 2008, 24 (14),7161–7168.

(9) Duan, H.; Wang, D.; Sobal, N.; Giersig, M.; Kurth, D.; Mohwald,H. Magnetic colloidosomes derived from nanoparticle interfacial self-assembly. Nano Lett. 2005, 5 (5), 949–952.

(10) O’Sullivan, M.; Zhang, Z.; Vincent, B. Silica-Shell/Oil-CoreMicrocapsules with Controlled Shell Thickness and Their BreakageStress. Langmuir 2009, 25 (14), 7962–7966.

(11) Caruso, F.; Caruso, R.; Mohwald, H. Nanoengineering ofInorganic and Hybrid Hollow Spheres by Colloidal Templating. Science1998, 282 (5391), 1111–1114.

(12) Katagiri, K.; Koumoto, K.; Iseya, S.; Sakai, M.; Matsuda, A.;Caruso, F. Tunable UV-Responsive Organic-Inorganic Hybrid CapsulesChem. Mater. 2009, 21 (2), 195–197.

(13) Utada, A.; Lorenceau, E.; Link, D.; Kaplan, P.; Stone, H.; Weitz,D. Monodisperse Double Emulsions Generated from a MicrocapillaryDevice. Science 2005, 308 (5721), 537–541.

(14) Chu, L.; Utada, A.; Shah, R.; Kim, J.; Weitz, D. Controllablemonodisperse multiple emulsions. Angew. Chem., Int. Ed. 2007, 46 (47),8970–8974.

(15) Shah, R. K.; Shum, H. C.; Rowat, A. C.; Agresti, D. L. J.; Utada,A. S.; Chu, L.-Y.; Kim, J.-W.; Fernandez-Nieves, A.; Martinez, C. J.;Weitz, D. A. Designer emulsions usingmicrofluidics.Mater. Today 2008,11 (4), 18–27.

(16) Oh, H.-J.; Kim, S.-H.; Baek, J.-Y.; Seong, G.-H.; Lee, S.-H.Hydrodynamic micro-encapsulation of aqueous fluids and cells via ’onthe fly’ photopolymerization. J. Micromech. Microeng. 2006, 16 (2),285–291.

(17) Hennequin, Y.; Pannacci, N.; Torres, C. n. P. d.; Tetradis-Meris, G.; Chapuliot, S.; Bouchaud, E.; Tabeling, P. SynthesizingMicrocapsules with Controlled Geometrical and Mechanical Propertieswith Microfluidic Double Emulsion Technology. Langmuir 2009, 25(14), 7857–7861.

(18) Hwang, D. K.; Dendukuri, D.; Doyle, P. S. Microfluidic-basedsynthesis of non-spherical magnetic hydrogel microparticles. Lab Chip2008, 8 (10), 1640–1647.

(19) Lee, D.; Weitz, D. Double Emulsion-Templated NanoparticleColloidosomes with Selective Permeability. Adv. Mater. 2008, 20 (18),3498–3503.

(20) Dinsmore, A.; Hsu, M. F.; Nikolaides, M.; Marquez, M.;Bausch, A.; Weitz, D. Colloidosomes: Selectively Permeable CapsulesComposed of Colloidal Particles. Science 2002, 298 (5595), 1006–1009.

(21) Sander, J. S.; Studart, A. R. Monodisperse Functional Colloido-somes with Tailored Nanoparticle Shells. Langmuir 2011, 27 (7), 3301–3307.

(22) Shah, R. K.; Kim, J.-W.; Agresti, J. J.; Weitz, D. A.; Chu, L.-Y.Fabrication of monodisperse thermosensitive microgels and gel capsulesin microfluidic devices. Soft Matter 2008, 4 (12), 2303–2309.

(23) Ye, C.; Chen, A.; Colombo, P.; Martinez, C. Ceramic micro-particles and capsules via microfluidic processing of a preceramicpolymer. J. R. Soc. Interface 2010, 7 (Suppl 4), S461–S473.

(24) Mooney, J. F.; Hunt, A. J.; McIntosh, J. R.; Liberko, C. A.;Walba, D. M.; Rogers, C. T. Patterning of functional antibodies andother proteins by photolithography of silane monolayers. Proc. Natl.Acad. Sci. U.S.A. 1996, 93 (22), 12287–12291.

(25) Keller, M. W.; Sottos, N. R. Mechanical Properties of Micro-capsules Used in a Self-Healing Polymer. Exp. Mech. 2006, 46 (6),725–733.

(26) Lankveld, J. M. G.; Lyklema, J. Adsorption of polyvinyl alcoholon the paraffin-water interface. I. Interfacial tension as a function of timeand concentration. J. Colloid Interface Sci. 1972, 41 (3), 454–465.

(27) Erb, R. M.; Obrist, D.; Chen, P. W.; Studer, J.; Studart, A. R.Predicting sizes of droplets made by microfluidic flow-induced dripping.Soft Matter 2011, 7 (19), 8757–8761.

(28) Pannacci, N.; Bruus, H.; Bartolo, D.; Etchart, I.; Lockhart, T.;Hennequin, Y.; Willaime, H.; Tabeling, P. Equilibrium and Nonequili-brium States in Microfluidic Double Emulsions. Phys. Rev. Lett. 2008,101 (16), 164502.


Langmuir ARTICLE

(29) Lorenceau, E.; Utada, A. S.; Link, D. R.; Cristobal, G.; Joanicot,M.; Weitz, D. A. Generation of Polymerosomes from Double-Emulsions.Langmuir 2005, 21 (20), 9183–9186.(30) Shum, H. C.; Lee, D.; Yoon, I.; Kodger, T.; Weitz, D. A. Double

Emulsion Templated Monodisperse Phospholipid Vesicles. Langmuir2008, 24 (15), 7651–7653.(31) Shrinkage of UV Monomers. Sartomer Application Bulletin

2005.(32) Weibull, W. A statistical theory on the strength of materials.

Ingeni€ors Vetanskaps Akademiens Handlingar: 1939; Vol. 151.(33) Gauckler, L. J.; Graule, T.; Baader, F. Ceramic forming using

enzyme catalyzed reactions. Mater. Chem. Phys. 1999, 61 (1), 78–102.

Documents

Designer Polymer-Based Microcapsules Made Using Microfluidics