Journal of Colloid and Interface Science 351 (2010) 140–150

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Giant biocompatible and biodegradable PEG–PMCL vesicles and microcapsulesby solvent evaporation from double emulsion droplets

Tobias Foster *, Kevin D. Dorfman, H. Ted DavisUniversity of Minnesota, Department of Chemical Engineering and Materials Science, 421 Washington Ave. SE, Minneapolis, MN 55455, USA

a r t i c l e i n f o

Article history:Received 12 March 2010Accepted 7 May 2010Available online 12 May 2010

Keywords:VesicleMicrocapsuleSolvent evaporation methodDouble emulsionEncapsulationWettingDewettingPhase separationBiodegradableBiocompatibleMicrofluidicsFlow focusing

0021-9797/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.jcis.2010.05.020

* Corresponding author. Address: University of CoChemistry, Luxemburger Str. 116, 50939 Cologne, Germ

E-mail address: tobias.foster@uni-koeln.de (T. Fost

a b s t r a c t

We report the formation of giant vesicles and microcapsules having diameters of the order of 250 lmobtained using the biocompatible and biodegradable amphiphilic block copolymer poly(ethylene gly-col)-b-poly(c-methyl e-caprolactone). Using the solvent evaporation method with monodispersewater–oil–water double emulsion droplets, we found either preferentially giant vesicles or preferentiallyspherical microcapsules with thicker walls compared to the vesicle membrane. The morphology dependson the volume of the evaporating oil shell of the double emulsion droplets for a given concentration of theblock copolymer. Accordingly, by using a microfluidic flow focusing device to form the double emulsion,it is possible to direct whether vesicles or microcapsules are preferentially formed, simply by adjustingthe volume of the oil shell through the volumetric flow rates of the device. During the evaporation ofthe organic solvents from the oil shells, rough surface morphologies of the double emulsion droplets wereobserved as intermediate stages in their transformation to either vesicles or microcapsules. These inter-mediate stages are most likely related to a dewetting of the evaporating oil phase from the surface of theinner droplet, resulting in a spot-like dewetting pattern, or a phase separation of the oil phase into poly-mer-rich and polymer-poor domains.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

The encapsulation of active chemical compounds such as drugsis of great importance for various applications in different fieldssuch as the cosmetics industry or the pharmaceutical industry(drug delivery). Among the various means of encapsulating activecompounds [1,2], their deposition in the interior of vesicles [3–5]or hollow microcapsules [6,7] is a frequently chosen technique,in particular for water-soluble compounds. Here we present a ver-satile microfluidic approach to form either vesicles or hollowmicrocapsules of a biocompatible and biodegradable diblockcopolymer, depending on the volume of the block copolymer solu-tion in the oil shells of monodisperse water–oil–water doubleemulsion droplets used in the solvent evaporation method.

Vesicles are closed structures that enclose a volume within afluid membrane of single- or multiple bilayers comprised ofamphiphilic molecules such as lipids, surfactants, or amphiphilicblock copolymers [2,8–10]. Depending on the type of amphiphile,the resulting vesicles are called ‘‘liposomes” (for lipids [8,11,12]),‘‘niosomes” (in the case of non-ionic surfactants [9]) or ‘‘polymer-

ll rights reserved.

logne, Institute for Physicalany. Fax: +49 221 470 5104.

er).

somes” (for amphiphilic block copolymers) [10,13–15]. Vesicles areclassified by their diameters, ranging from small (10–100 nm) tolarge (0.1–1 lm) to giant (1–100 lm) uni- or multilamellarvesicles.

In recent years, polymersomes in particular have received in-creased scientific attention due to improvements in polymer syn-thesis. These modern synthesis techniques allow precise controlover the properties of the amphiphilic block copolymer moleculesthat constitute the polymersomes, such as the molecular weight ofthe copolymer and the ratio of the lengths of the different blocks,thereby tuning the properties of the polymersomes as well. For in-stance, as the thickness of the membrane is increased by increasingthe molecular weight of the amphiphilic block copolymer, themechanical stability of the vesicle membrane is enhanced whileits permeability is reduced [14]. Vesicles formed by amphiphilicmolecules with low molecular weight (like lipids or surfactants)have thin bilayer membranes (thickness ca. 3-5 nm [16,17]) andthus exhibit properties such as membrane fluidity, deformabilityand high permeability [13]. Polymersomes made of high molecularweight block copolymers (molar weights �104-105 g/mol [14])have thicker membranes (over 20 nm [18]) and thus show greaterresistance to deformation and reduced permeability. In addition,the fluidity of the polymersome membrane decreases with increas-ing molecular weight of the amphiphilic block copolymer as the

T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150 141

membrane becomes more viscous due to the entanglement of thepolymer chains [18–21].

Several different techniques for the formation of vesicles havebeen developed [22,23], with the most widely used being the rehy-dration method [24,25] and the so-called electroformation of ves-icles [26,27]. In the rehydration method, dried films of organicsolutions of the amphiphilic molecule are dissolved in water oraqueous solutions. This yields vesicles with typical diameters of5-20 lm [22] by self-assembly of the lamellae of the dried amphi-philic molecules during the rehydration. The electroformation ofvesicles utilizes alternating electric fields to generate vesicles inaqueous solutions of the amphiphilic molecules, with typical diam-eters of 20-50 lm [22]. In an alternative approach, giant vesicleswith diameters of the order of 10-100 lm have been obtained bythe evaporation of the organic solvent from solutions of amphi-philic molecules (lipids or diblock copolymers) that constitutethe oil phase in water–oil–water double emulsions [28–32]. Thisso-called solvent evaporation method yields vesicles as the amphi-philic molecules self-assemble at the interfaces between the oilphase and the two aqueous phases and form a bilayer as the oilphase becomes thinner during the evaporation of its solvent orsolvent mixture (see Fig. 1 for a schematic drawing).

Microcapsules are closed, usually spherical structures that en-close an active material in a liquid or solid core component withina continuous solid-like barrier or shell that is thick (typically1-10 lm [33,34]) compared to the bilayer membrane of a vesicle.The diameters of microcapsules are in the range of 1-1000 lmaccording to the most commonly used definitions [35]. Usuallythe microcapsule barrier consists of amorphous, in some instancesphysically or chemically cross-linked, homo- or copolymer mole-cules instead of being formed by single- or multiple bilayers ofsome amphiphilic component [35–38]. The resulting solid-like nat-ure of the microcapsule barriers leads to their most importantcharacteristic physical property: a shear rigidity that is not foundin fluid membranes of vesicles. Due to their shear rigidity, largedeformations such as the prolate–oblate transition [39] are pre-vented. In addition, microcapsules exhibit qualitatively differentresponses to shear flow when compared to vesicles, like the wrin-kling of the capsule surface [40].

Microcapsules for the encapsulation of water-soluble activecompounds in a polymeric shell are typically formed by phaseseparation of a polymer in a hydrophobic solvent during eitherpolymerization or solvent evaporation from polymer solutions(see [36] and references therein for an overview). In the solventevaporation method, one approach uses water–oil–water doubleemulsions [7,36,41]. In contrast to vesicle formation via amphi-philic molecules, a water-insoluble homopolymer is dissolved inthe volatile organic solvent. The resulting solution is used as a

Fig. 1. Schematic drawing that documents how water–oil–water double emulsion dro

continuous phase in which an aqueous phase that contains thewater-soluble active compound is emulsified. This water-in-oilemulsion is then, in turn, emulsified in an aqueous phase, andthe subsequent evaporation of the organic solvent from thiswater–oil–water double emulsion leads to the precipitation ofthe polymer in the oil phase. Thus a polymer shell is formed thatsurrounds the liquid aqueous core and eventually encapsulatesthe aqueous solution of the active compound. In this method thesize of the resulting capsules is defined by the diameter of the ini-tial emulsion or double emulsion droplets. Values of 0.1-1000 lmcan be achieved easily by typical emulsification methods that uti-lize mechanical agitation.

For applications in cosmetics and pharmaceuticals, polymer-somes and microcapsules have been formed using differentbiocompatible and biodegradable polymers or blocks in blockcopolymers. For instance, several hydrophilic and hydrophobicpolymers turned out to be highly biocompatible. An example isthe hydrophilic poly(ethylene oxide) (PEO), which is highly solublein water, highly hydrated and well suited to stabilize e.g. micropar-ticles by steric stabilization against aggregation in biological media[42,43]. As hydrophobic polymers, poly(D,L-lactic acid) (PLA) and itscopolymers with glycolic acid [44–46] and poly(caprolactone)(PCL) [47,48] turned out to be both biocompatible and also biode-gradable. PCL has been reported to be more stable with respect tochemical hydrolysis in vivo [49–51], allowing for a slower drugrelease [49].

Many groups have used block copolymers like PEO–PLA [14,52–54] or PEO–PCL [53–55] to obtain polymersomes for the controlledrelease of encapsulated compounds [15,52,53]. In particular, poly-mersomes from PEO–PLA or PEO–PCL were formed by the afore-mentioned rehydration method [52,53,55] or by injecting asolution of the block copolymer in an organic solvent into wateror an aqueous solution [54]. The resulting polymersomes haddiameters of 50–10 lm [52,53] or 70 nm to 50 lm [54], respec-tively, depending on the details of the vesicle formation protocol.Recently, the solvent evaporation method yielded polymersomesof PEO–PLA with diameters from 10-50 lm starting from water–oil–water double emulsions [31]. In addition, microcapsules havebeen obtained by the solvent evaporation method from water–oil–water double emulsions using, for instance, water-insolublehomopolymers like PLA [56,57], copolymers of lactic acid andglycolic acid [58], or diblock copolymers like PEG–PLA [59].

Due to the semicrystalline and/or glassy nature of these hydro-phobic biodegradable polymers PLA (Tg = 57-59 �C [60]) and PCL(Tg = �60 �C, Tm = 63 �C [60]), it is often necessary to facilitate thedissolution and the self-assembly of the respective block copoly-mers by elevating the temperature (>50 �C) or using an organic sol-vent [61]. Contrarily, by using amorphous hydrophobic blocks with

plets (A) exposed to evaporation of the middle fluid solvent (B) yield vesicles (C).

142 T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150

low glass-transition temperatures, the block copolymers self-assemble in aqueous solution; elevated temperatures or organicsolvents are not required to give chain dynamics that aresufficiently fast to reach a local equilibrium [62]. Thus, in a recentapproach [63] to generate polymersomes in water at moderatetemperatures (25-50 �C), the hydrophobic, potentially biocompati-ble and biodegradable [63,64], and amorphous poly(c-methyle-caprolactone) (PMCL, Tg = �60 �C, [65]) has been applied ashydrophobic block in a block copolymer with PEO. It turned outthat depending on the ratio of the relative block sizes, spheres, cyl-inders and vesicles were observed, as well as regions of coexistenceof the mentioned aggregate morphologies. Using the rehydrationmethod, vesicles with diameters of the order of 100-200 nm wereobtained [63].

In this work we show how the solvent evaporation methodstarting from water–oil–water double emulsions can be utilizedto selectively obtain either vesicles or microcapsules, dependingon the volume of the oil shell relative to the volume of the innerdroplets of the double emulsion droplets for a given concentrationof the amphiphilic diblock copolymer. Thus we show how the re-cently introduced protocol to obtain monodisperse polymersomes[28–31] can be extended and generalized to selectively obtaineither vesicles or microcapsules just by choosing the appropriateproperties of the starting double emulsion droplets in the solventevaporation method. By using a microfluidic flow focusing deviceto generate the double emulsion droplets, we easily control thevolume of the oil shell by the volumetric flow rates of the differentfluids. The resulting flexibility to control the properties of theenclosing shell of the vesicle or microcapsule provides us with ameans to optimize the properties of the resulting structure, as re-quired for particular applications. As described above, the mechan-ical stability is expected to be increased by an increased thicknessof the enclosing shell, which in turn reduces the permeability ofthe shell. Thus, for instance, the time-dependent release of anencapsulated drug can be optimized according to the specificrequirements of the drug. In addition to the aspect of mechanicalstability, tuning the mechanical properties of the enclosing shellcould be useful to optimize the interactions of the vesicle or micro-capsule with a given substrate, as the mechanical properties ofcapsules or cells strongly influence their adhesion to surfaces,whereby high energies of mechanical deformation resist deforma-tions that might be necessary for the capsule or cell to adhere to asubstrate [66–69].

We use this approach to generate giant (250 lm diameter) ves-icles and microcapsules formed at room temperature with thebiocompatible and biodegradable amphiphilic diblock copolymerpoly(ethylene glycol)-b-poly(c-methyl e-caprolactone) or PEO–PMCL. Although diameters are required to be below 1-10 lm forthe intravenous injection and below 100-120 lm for the intramus-cular or subcutaneous injection of the microcapsules in drug deliv-ery [35,70,71], there are many different applications for vesicles ormicrocapsules with diameters of the order of 100–500 lm. In

Fig. 2. Chemical structure of the biocompatible and biodegradable amphiphilic diblock cused in the experiments.

addition to the general applications in chemistry and biology thatwere recently reviewed [72,73], specific applications include for in-stance cell encapsulation [74–78], single molecule reactions invesicles [79], protein evolution [80,81], artificial cells and encapsu-lated gene expression systems [82,83] or single-cell studies likeHTS on single cells [84].

2. Experimental section

2.1. Materials

The inner fluid of the water–oil–water double emulsion drop-lets was deionized water (Millipore Direct-Q3 ultrapure watersystem, electrical resistance 18 MX). A solution of 2 wt.% poly(vi-nyl alcohol) (PVA, molar mass 13–23 kg/mol, 87-89 % hydrolyzed,Sigma–Aldrich) in a mixture of deionized water and glycerol(46 wt.%, Sigma–Aldrich, P99.5%) was used as the outer fluid,forming the continuous aqueous phase of the double emulsion.As the volatile oil component, i.e. the middle fluid of the doubleemulsion, we used a 1 wt.% solution of the biocompatible andbiodegradable amphiphilic diblock copolymer poly(ethylene gly-col)-b-poly(c-methyl e-caprolactone) (PEO–PMCL, see Fig. 2, Mn =10.3 kg/mol, wPEO = 0.18 [63]) in a mixture of the solvents toluene(Mallinckrodt Chemicals, 99.5 %) and tetrahydrofurane (THF,Mallinckrodt Chemicals, 99–100%) containing 50 wt.% of THF. Allchemicals were used without further purification. The outer fluidwas filtered before the experiments using a syringe filter with0.45 lm pore size.

2.2. Microfluidic device and experiments

Water–oil–water double emulsions were formed in a microflu-idic flow focusing device [28,29] that consists of two glassmicrocapillaries with spherical cross section (World PrecisionInstruments, outer diameter 1 mm, inner diameter 0.75 mm)nested in a glass tube with square cross section (VitroCom, innerdimensions 1 mm) (see [28,29] for a schematic drawing of this set-up). The tip of one of the glass microcapillaries, the so-called injec-tion tube, had a tip diameter of 50-70 lm and was obtained using apipet puller (Narishige PC-10). The inner fluid of the double emul-sion was injected through this injection tube into a stream of themiddle fluid (volatile oil component) that was pumped throughthe outer coaxial region of the injection tube and the square tube,yielding a coaxial flow of the inner fluid within the middle fluid(Fig. 3A). This coaxial flow was injected into a stream of the outerfluid that was pumped from the opposite direction through theouter coaxial region of the other glass capillary, the so-called exittube, and the square tube. The resulting coaxial flow of the innerfluid within the middle fluid within the outer fluid enters the exittube (opening diameter �400 lm) where it eventually breaks upinto individual double emulsion droplets when the appropriate

opolymer poly(ethylene glycol)-b-poly(c–methyl e–caprolactone) (PEO–PMCL) [63]

Fig. 3. (A) Double emulsion formation in a microfluidic flow focusing device by the break-up of a coaxial flow of the aqueous inner fluid in the volatile middle fluid (oil phase)in the aqueous outer fluid. (B) Bright field image of the resulting highly monodisperse double emulsion droplets after their collection on a glass microscope slide.

T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150 143

volumetric flow rates of the three fluids are applied (dripping re-gime of double emulsion formation [29]). For the dimensions ofthe microfluidic device used, typical volumetric flow rates werein the range of 0.4–1.2 ml/h, 1–2 ml/h, and 3.5–5.5 ml/h for the in-ner-, middle- and outer fluid, respectively, to yield double emul-sion droplets by the dripping mechanism. In the experimentsshown here, each double emulsion droplet contained one singledroplet of the inner fluid surrounded by a shell of the middle fluid.Typically, immediately after the double emulsion droplets wereformed in the microfluidic device, the radius of the inner dropletwas in the range of 280–370 lm, while the radius of the doubleemulsion droplet was in the range of 350–500 lm. The volumeof the middle fluid shell relative to the volume of the inner dropletwas increased by higher volumetric flow rates of the middle fluidand/or lower volumetric flow rates of the inner fluid. Samples ofthe double emulsion were collected in the form of droplets withdiameters of several millimeters in a PDMS reservoir (depth 0.5–1 mm, dimensions approximately 0.5 � 0.5 cm) on a glass micro-scope slide. After the collection of the double emulsion, thereservoir was covered with a glass coverslip, ensuring a sufficientlyslow evaporation of the volatile solvent of the middle fluid. Theformation of the double emulsion droplets in the microfluidicdevice and the solvent evaporation were both observed by brightfield video microscopy, using an inverted microscope (Leica DMI4000, 5� or 20� objective) and a CCD-camera (Photometrics CoolSnap EZ). All experiments were conducted at room temperature.

3. Results and discussion

3.1. Formation of monodisperse double emulsion droplets

Monodisperse water–oil–water double emulsion droplets wereformed in a microfluidic flow focusing device by the break-up of acoaxial flow of the inner fluid (water) within the middle fluid with-in the outer fluid of the double emulsion (see Fig. 3A). The middlefluid is the volatile oil phase that surrounds a single inner fluiddroplet in each resulting double emulsion droplet, and consists ofa solution of the biocompatible and biodegradable amphiphilic di-block copolymer poly(ethylene glycol)-b-poly(c-methyl e-capro-lactone) (PEO–PMCL, see Fig. 2) in a mixture of toluene andtetrahydrofurane (THF). The outer fluid constitutes the continuousaqueous phase of the resulting double emulsion and is a solution ofpoly(vinyl alcohol) (PVA) in a mixture of water and glycerol. Fol-lowing references [30–32] PVA is added to the outer fluid as itwas reported to increase the stability of the double emulsion drop-lets with respect to coalescence. According to different recent pub-lications [29,85] the double emulsion droplets are formed in thisso-called dripping regime when viscous drag forces in the coaxialflow exceed surface tension forces that prevent the rupture ofthe coaxial flow. Initially, when the volume of the droplet that isformed in the device is small, the viscous drag forces from the flow

of the middle fluid on the inner fluid and of the outer fluid on themiddle fluid are small compared to the surface tension forces.However, as the volume of the droplet grows the viscous dragforces increase and pull the growing droplet downstream so thateventually a double emulsion droplet is pinched off when the dragand the surface tension forces become comparable. Thereby, theinterfacial tensions at the two water–oil interfaces are governedby the amount of THF in the middle fluid and the concentrationof PEO–PMCL and PVA in the middle fluid and the outer fluid,respectively. Due to the aforementioned competition of drag andsurface tension forces in the course of droplet formation, reducedinterfacial tensions decrease the resistance against rupture of thecoaxial flow and accordingly lead to dripping at reduced volumet-ric flow rates. In the same fashion, the volumetric flow rates of thedifferent fluids control the frequency of droplet generation for a gi-ven geometry of the microfluidic device and given interfacialtensions.

Samples of the resulting double emulsion were collected in res-ervoirs on glass microscope slides. After the collection of the sam-ple the reservoirs were covered with glass coverslips to ensure asufficiently slow evaporation of the volatile solvents of the oil com-ponent. The representative image in Fig. 3B, taken 5 min after thecollection of the double emulsion, shows that the double emulsiondroplets are very monodisperse with respect to the size of both theinner droplet and the middle fluid shell. The inner droplets inFig. 3B have diameters of the order of 370 lm, while the doubleemulsion droplets have diameters of the order of 440 lm. Thesediameters could be varied by changing the applied volumetric flowrates of the fluids that constitute the double emulsions [29]. Sincethe solvents toluene and THF of the middle fluid are highly volatilethey begin to evaporate after the collection of the double emulsion.Thereby it turned out that a sufficiently slow evaporation of themiddle fluid solvents, ensured by covering the reservoir, is crucialto avoid (i) the coalescence of the inner droplet with the continu-ous aqueous phase formed by the outer fluid, and (ii) the rupture ofthe forming vesicles during the solvent evaporation, as reportedpreviously in other publications [31,32].

3.2. Solvent evaporation and dewetting transition

Several stages of solvent evaporation from the double emulsiondroplets were observed after the collection of the double emulsionsamples, the first of which is shown in the representative time ser-ies given in Fig. 4. Initially, as the evaporation of the middle fluidsolvents proceeds, the volume of the middle fluid shells of the dou-ble emulsion droplets decreases continuously with time, as it isclearly visible in Fig. 4. In addition, water of the inner and the outerfluid is expected to diffuse into the middle fluid shell. However, inthe early stages of solvent evaporation we did not observe anyscattering of incident light from the middle fluid shells as it wouldbe expected in the case of the formation of small water-in-oil

Fig. 4. Sequence of bright field images that document how the solvent evaporation from the double emulsion droplets proceeds and the middle fluid shells become thinnerwith time. Time intervals between the images AB, BC, CD, DE and EF are 22, 24, 9, 7 and about 60 min, respectively.

144 T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150

emulsion droplets within the middle fluid shell due to the entry oflarger volumes of water (contrarily to the situation in later stagesof solvent evaporation, see below). Most likely the mass fractionof 1 wt.% of the amphiphilic diblock copolymer PEO–PMCL is notsufficient to stabilize such emulsion droplets.

As time progresses, the shells of middle fluid become thinnerand thinner. For the given initial volume of the middle fluid shellsof the double emulsion droplets shown in Fig. 4A, almost all sol-vent has evaporated after about the first 60 min of the evaporation(the time intervals between images A and B, B and C, and C and Dare 22, 24 and 9 min, respectively). Some of the double emulsiondroplets that are visible in Fig. 4C, D and E have much thicker shellsof middle fluid than the majority of the other double emulsiondroplets. This is related to the fact that some of the double emul-sion droplets do not reside at the interface of the continuous aque-ous phase and air after the collection of the double emulsion, butare instead below the top layer of double emulsion droplets thatreside at this interface. This can be clearly seen in Fig. 4A, as indi-cated by the white arrow. The middle fluid shell of submergeddroplets thins more slowly than their counterparts at the liquid–air interface. Whereas the middle fluid can evaporate from the lat-ter droplets, the middle fluid shell of a submerged droplet onlythins through diffusion of the solvent into the continuous aqueousphase. Accordingly, for a given time of evaporation the middle fluidshells in these droplets are larger compared to the double emulsiondroplets that reside at the surface of the continuous aqueousphase. In addition, some emulsion droplets of middle fluid withoutany inner aqueous droplet can be observed as well in the images inFig. 4, for instance as indicated by the black arrow in the right halfof Fig. 4A. These middle fluid droplets were formed by coalescenceof the inner droplet with the outer fluid. This process was observedoccasionally in the course of the experiments, particularly immedi-ately after the collection of the double emulsion (in the first 5 minof solvent evaporation) or when the solvent evaporation wasconducted without the aforementioned covering of the reservoir.

In the final part of the first stage of the solvent evaporation pro-cess shown in Fig. 4, the surfaces of some of the double emulsiondroplets start to become rough and uneven, indicated by the arrowin image E, taken 7 min after image D was taken (about 62 min of

solvent evaporation). This rough surface morphology was observedfor all double emulsion droplets at the end of the first stage of sol-vent evaporation, as shown in image F, taken about 60 min afterimage E was taken and after about 120 min of solvent evaporation.(Additional representative images of double emulsion dropletsshowing this rough surface morphology are shown below inFigs. 5J and 7A).

A series of images that documents the formation of this roughsurface morphology in greater detail is provided in Fig. 5. Fig. 5Ashows a double emulsion droplet in a stage of solvent evapora-tion just before the rough morphology of the droplet surface ap-pears, as it is visible in the last image Fig. 5J (taken 6 min and20 s after the image shown in Fig. 5A). It is clearly visible thatinitially (Fig. 5A) the inner droplet is partially engulfed by theevaporating middle fluid, with an excess-volume of the middlefluid attached to a part of the surface of the inner droplet, asindicated by the white arrow in Fig. 5A. This can be interpretedas a partial dewetting of the evaporating middle fluid from thesurface of the inner droplet.

Such a partial dewetting during the solvent evaporation fromdouble emulsion droplets has been previously reported in experi-ments on the solvent evaporation from double emulsion dropletsthat were stabilized by PEG–PLA diblock copolymers havinghydrophobic PLA-blocks with much higher molar weight thanthe hydrophilic block [31]. The partial wetting is associated witha spreading coefficient, S [86],

S ¼ cIF—OF � ðcIF—MF þ cMF—OFÞ ð1Þ

that is only slightly negative [31,87]. In the latter, cIF–OF, cIF–MF andcMF–OF are the interfacial tensions of the inner fluid (IF) and the out-er fluid (OF), the inner fluid and the middle fluid (MF), and themiddle fluid and the outer fluid, respectively. In fact, when weuse the Young-equation with the approximations applied in [30]we can rewrite equation (1) (assuming cIF–MF = cMF–OF) and expressthe spreading coefficient using the contact angle h between the tan-gents at (i) the IF–OF-interface and (ii) the MF–OF-interface in thethree fluid contact point:

S ¼ 2cIF—MF � ðcos h� 1Þ ð2Þ

Fig. 5. Sequence of bright field images that document how the solvent evaporation yields double emulsion droplets with a surface morphology that is uneven and roughon the length scale of observation. The time intervals between the images are: for AB 150 s, for BC, CD, DE 10 s, for EF, FG 15 s, for GH 40 s, for HI 25 s, and for IJ 45 s,respectively.

Fig. 6. Schematic drawing that describes a possible mechanism for the formation of the rough surface morphologies of the double emulsion droplets during the solventevaporation. Initially the number density of diblock copolymer molecules residing in the IF–MF/MF–OF interfaces is equal for the excess-volume (interfacial tension ce) andthe thin layer (interfacial tension cl) of middle fluid, leading to equal interfacial tensions (A): cl = ce (assuming cIF–MF � cMF–OF [30]). As the solvent evaporation continues,excess diblock copolymer accumulates in the excess-volume of middle fluid (B), leading to the condition cl > ce. These gradients in interfacial tension lead to Marangoniflows on the surface of the inner droplet (B) with a mass transport from the excess-volume of middle fluid to the thin film. These flows cause local fluctuations in the thicknessof the thin film (B) that were suggested to cause dewetting in thin fluid films [92], leading to spot-like dewetting patterns of polymer-rich (bilayer) and polymer-poordomains (C).

T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150 145

We obtain S = �0.1 ± 0.1 mN/m with h = 10 ± 5� (from Fig. 5A)and an estimated value of cIF–MF = 3 ± 2 mN/m [30] for the interfacialtension of the inner fluid and the middle fluid. A low value of cIF–

MF � 1 mN/m is expected particularly for the final stage of solventevaporation characterized by a high concentration of diblockcopolymer in the middle fluid (>>1 wt.%) and a high numberdensity of diblock copolymer molecules residing in the IF–MF/MF–OF-interface [30]. Apparently, cIF-MF is sufficiently reduced by thePEG–PMCL block copolymer so that the middle fluid wets the innerdroplet sufficiently well to prevent a more complete dewetting. Sucha more complete dewetting would lead to an ‘‘acorn-like” structureof a middle fluid droplet attached to the inner droplet, as reported inprevious publications [30,31]. It was reported further that aftermore complete solvent evaporation these droplets of excess middlefluid separated occasionally from the vesicles that were finallyformed from the inner droplets [30,31]. Instead, we observed thatno further dewetting of the evaporating middle fluid from the sur-face of the inner droplet occurs with further evaporation. Rather,the excess-volume of middle fluid shrinks and becomes thinner sothat a larger part of the surface of the inner droplet is still covered

with a thin layer of middle fluid, as indicated by the white arrowin Fig. 5B and C (the time intervals between the images in Fig. 5Aand B, and B and C are 150 s and 10 s, respectively).

In addition, as it is clearly visible from Fig. 5A as well, the ex-cess-volume of the middle fluid appears dark (indicated by thewhite arrow in Fig. 5A). We interpret this observation as a phaseseparation of the middle fluid into polymer-rich and polymer-poorphases, forming domains that scatter light and thus lead to a darkappearance of the phase-separated regions. This phase separationis most likely induced by (i) the increase of the PEG–PMCL concen-tration in the middle fluid due to the evaporation of the solvent, (ii)the non-solvent water (for PEG–PMCL) diffusing into the evaporat-ing middle fluid shell, and (iii) the presumably lower solvent qual-ity of toluene for the block copolymer compared to THF: toluene isa near-theta solvent for poly(caprolactone) [88]; since THF evapo-rates much more readily compared to toluene, the solvent mixtureof the middle fluid is expected to be enriched with toluene in thisstage of solvent evaporation.

Interestingly, however, the other part of the inner dropletsurface that is not covered by the excess-volume of middle fluid

Fig. 7. Bright field microscopy images that show the formation of vesicles in the solvent evaporation process: (A) rough surface morphology after the evaporation of thesolvent from the double emulsion droplets shown in Fig. 3B. (B) and (C) Vesicles and (D) a microcapsule formed after the solvent evaporation has continued for more time.

146 T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150

(indicated by the black arrow in Fig. 5C) also becomes dark as thesolvent evaporation proceeds further, as can be observed in imagesFig. 5D–J (for Fig. 5 the time intervals between the images are: forC–D, D–E 10 s, for E–F, F–G 15 s, for G–H 40 s, for H–I 25 s, and forI–J 45 s, respectively). First visible in the image in Fig. 5D, the edgeof the droplet becomes dark, with a dark front highlighted by thearrow in Fig. 5D appearing and moving from right to left (compareFig. 5D and E, and see video in the supporting information) untilthe entire surface of the inner droplet shows a clearly heteroge-neous surface morphology (see the dark spots in image Fig. 5I).As the evaporation proceeds further, these heterogeneities leadto the aforementioned rough surface morphology, as can be seenin Fig. 5J (and in Figs. 4F and 7A as well).

We suggest three possible explanations for this observationthat, to our knowledge, has not been reported before in the courseof solvent evaporation from water–oil–water double emulsions.The first two explanations assume that the part of the surface ofthe inner droplet that is indicated by the black arrow in Fig. 5C iscovered with a thin layer of middle fluid yielding a bilayer mem-brane that is swollen with middle fluid, as depicted in the drawingin Fig. 6A.

Hayward et al. [30] suggested that such a middle fluid-swol-len bilayer coexists with an excess-volume of middle fluid dur-ing the formation of polymersomes by solvent evaporationfrom double emulsion droplets. Such a state could be stabilizedby a sufficiently high number density of diblock copolymermolecules residing in the IF-MF/MF-OF interfaces (initially:cIF–MF � cMF–OF [30]).

In the first scenario, after the nucleation of phase separation inthe excess-volume of middle fluid (indicated by the white arrow inFig. 5A), the phase-separated domains grow, forming a front be-tween the homogeneous and the phase-separated region. Thisfront moves over the surface of the double emulsion droplet asthe phase separation proceeds into the region of the middlefluid-swollen bilayer. When the phase separation is complete thesurface of the entire double emulsion droplet is covered by thephase separated middle fluid. The domains of different polymercontent scatter light and result in the heterogeneous surface mor-phology visible in Fig. 5I. As the evaporation proceeds further, thisheterogeneous distribution of block copolymer (i.e. domains thatare rich in block copolymer, and domains poor in block copolymer)vitrifies due to the continued increase of polymer concentrationand thus viscosity of the polymer solution [37,38,89]. The hetero-geneous polymer density turns into the aforementioned rough sur-face morphology observed in the final images in Figs. 4 and 5 (seealso Figs. 7A and 8A). This first mechanism is similar to ‘‘spinodaldewetting” that can cause such patterns as well [90,91].

A second explanation for the origin of the rough surface mor-phology is based on the observation of spot-like dewetting pat-terns found in a thin film of a first fluid coated on top of a thinfilm of a second fluid that is not wetted by the first fluid [92].The model for the formation of these spot-like dewetting patternsassumes that local fluctuations in the composition of the film of thefirst fluid lead to gradients in composition and thus interfacial ten-sion [92]. These gradients cause a Marangoni flow that creates amass transport in the plane of the surface of the first fluid. This

Fig. 8. Bright field images that document the solvent evaporation from (A) water–oil–water double emulsions with oil shells that are thick compared to the oil shells in thepreviously shown experiments. Accordingly mainly microcapsules are formed, as can be seen in (B) and in larger magnification in (C). A representative microcapsule with anasymmetric distribution of polymer in the capsule shell is highlighted by an arrow in (B). In addition, some vesicles are found as well, as indicated by the black arrow in (D),while a representative microcapsule is highlighted by the white arrow.

T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150 147

surface flow is visible as a ‘‘dewetting front” that moves over thesurface as the flow leads to fluctuations of the film thickness ofthe first fluid, inducing a local dewetting under formation of theaforementioned spot-like dewetting patterns [92].

To explain our observation by this mechanism we assume againthat the surface of the inner droplet indicated by the black arrow inFig. 5C is covered with a middle fluid-swollen bilayer, as describedabove (see Fig. 6A) and suggested by [30]. This state is expected tobe a metastable state with respect to dewetting, since dewetting ofthe middle fluid from the diblock copolymer layer that covers theinner droplet would yield a bare bilayer membrane (withoutmiddle fluid swelling) that is expected to have a further reducedinterfacial energy compared to the middle fluid-swollen bilayer.Thus, this metastable state is expected to be associated with a neg-ative spreading coefficient S as well.

As further solvent evaporation leads to differences in the inter-facial concentration of diblock copolymer molecules in the excess-volume of middle fluid and in the thin layer of middle fluid (seeFig. 6B), gradients in composition and thus in the interfacial ten-sion arise at the contact line that separates the thin layer fromthe excess-volume. Due to the accumulation of excess polymer inthe excess-volume of middle fluid we expect that transient non-equilibrium states occur in which the interfacial tensions cIF–

MF � cMF–OF � ce are smaller at the interfaces of the excess-volumeof middle fluid with the inner and the outer fluid when comparedto the respective interfacial tensions at the interfaces of the thinlayer of middle fluid, cIF-MF � cMF–OF � cl: cl > ce (Fig. 6B). Such gra-dients in interfacial tension lead to a Marangoni flow that creates amass transport from the excess-volume of middle fluid to the thinlayer (Fig. 6B); this flow occurs in the thin middle fluid film on the(fluid) surface of the inner droplet. In fact, the movie in thesupporting information indeed shows a flow and flow vorticesoccurring from the excess-volume of middle fluid in the directionof the thin layer.

In the same fashion as the Marangoni flow in the surface planeof a non-wetting first fluid coated on a layer of a second fluid cre-ates fluctuations in the film thickness that induce local dewetting[92], the Marangoni flow occurring in the middle fluid film thatcovers the surface of the inner droplet is expected to cause fluctu-ations in the thickness of the middle fluid film (as indicated in theschematic drawing in Fig. 6B), leading to a local dewetting of themiddle fluid from the inner droplet surface due to the negativespreading coefficient S. Accompanied with the Marangoni flow thatinduces local dewetting, a dewetting front is expected to moveover the surface of the inner droplet, similar to the mechanismsuggested in [92]. This local dewetting in turn leads to the forma-tion of spot-like dewetting patterns, i.e. polymer-poor and poly-mer-rich domains (see Fig. 6C) that cause the heterogeneouscomposition of the surface of the inner droplet and the rough sur-face morphology, as discussed above. It is possible that the poly-mer-rich domains are in fact domains where diblock copolymerbilayers are found that are not swollen with middle fluid solvent,as indicated in Fig. 6C. To summarize, this second mechanism sug-gests that the Marangoni flow described here induces dewetting ofa thin middle fluid film that remains on the surface of the innerdroplet in a metastable state with respect to dewetting duringsolvent evaporation. We note that Marangoni flows have been sug-gested to explain the pattern formation (into polymer-rich- andpolymer-poor domains) in polymeric fluids that phase separatedue to solvent evaporation [89,93].

In our third explanation, we assume that the excess-volume ofmiddle fluid (white arrow in Fig. 5C) coexists with a bare blockcopolymer bilayer (black arrow in Fig. 5C) that is not swollen withmiddle fluid. As in the scenario discussed above, Marangoni flowsare expected to arise from gradients in the interfacial tensioncaused by gradients in the interfacial composition [94,95]. Thesegradients are expected to occur near the location of the contact line(dashed black arrow in Fig. 5C) at which the excess-volume of

148 T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150

middle fluid (white arrow in Fig. 5C) meets the other part of thesurface of the inner droplet that is not covered by this excess-vol-ume (black arrow in Fig. 5C). Associated with the resultingMarangoni flow, a front of middle fluid (arrow in Fig. 5D) movesover the surface of the inner droplet (compare Fig. 5D and E) untilthe entire surface is covered with the phase separated middle fluid.The presence of such Marangoni flows is clear in the moving frontand vortices present in the video of the process (supporting infor-mation). The phase separated state of the middle fluid that finallycovers the entire droplet surface leads to polymer-poor and poly-mer-rich domains on the droplet surface causing the observedrough surface morphology. According to this mechanism, theinitially observed dewetting of the evaporating middle fluid fromthe inner droplet surface, leading to the formation of the excess-volume of middle fluid, creates the gradients in composition andthus interfacial tension that result in a Marangoni flow that tendsto evenly distribute the evaporating middle fluid on the surface ofthe inner droplet. Contrarily to the second suggested mechanism,the Marangoni flow described here does not induce dewetting ofa thin middle fluid film that covers the inner droplet surface, butevenly distributes the phase separated middle fluid on the surfaceof the inner droplet.

3.3. Formation of vesicles or microcapsules from double emulsiondroplets

After the solvent evaporation has continued for several morehours, the rough double emulsion droplets evolve into vesicleswith surfaces that are smooth on the length scale of observation.This transformation is shown in the images presented in Fig. 7.Fig. 7A shows the double emulsion droplets with rough surfacemorphology as they were obtained by solvent evaporation fromthe collected double emulsion droplets shown in Fig. 3B, with a to-tal solvent evaporation time (time after collection of the doubleemulsion in the reservoir) of approximately 110 min. With diame-ters in the range of 320 lm, the droplets shown in Fig. 7A are sig-nificantly smaller than the inner droplets at the beginning of thesolvent evaporation (Fig. 3B), which have diameters around370 lm. Accordingly, the middle fluid shells are permeable towater, as expected due to the finite solubility of water in toluene.Thereby the diffusion of water out of the inner droplets is drivenby the different osmolarities of the inner fluid (water) and the out-er fluid (solution of poly(vinyl alcohol) in a mixture of water andglycerol). This difference in the osmolarities is also regarded asthe origin of the slight deformation of some of the double emulsiondroplets in this stage of solvent evaporation, as it is visible inFig. 7A. The stronger deformation of one single double emulsiondroplet visible at the bottom of Fig. 7A we attribute to the coales-cence of the inner droplet with the continuous aqueous phase(outer fluid) in this late stage of solvent evaporation, leading to acomplete collapse of the highly viscous shell of polymer-richmiddle fluid instead of the immediate transition to a middle fluiddroplet.

As the solvent evaporation was continued for 25 more hours,the surfaces of the droplets shown in Fig. 7A became smooth onthe length scale of observation, as it is clearly visible in Fig. 7Band the close-up shown in Fig. 7C. The close-up (Fig. 7C) provesthat indeed vesicles (polymersomes) have been formed with diam-eters of the order of 240–250 lm. Furthermore, both images Fig. 7Band C show a great number of small spherical structures, as indi-cated by the white arrows in Fig. 7B. These were formed by thecomplete evaporation of solvent from the emulsion dropletsformed by the middle fluid that do not contain any inner aqueousdroplet, as mentioned above, and accordingly are PEO–PMCL-gelparticles.

The formation of vesicles with smooth surfaces from the drop-lets having a rough surface morphology is most likely related to afinite concentration of solvent that has remained in the evaporat-ing middle fluid that forms the surfaces with rough morphology.Thus, the fluidity of this middle fluid is high enough to ensuresubstantial reorganization and structural rearrangement of thecopolymer molecules, leading to the observed smooth surfacemorphologies. This contrasts with the complete vitrification orsolidification of the polymer that is expected for a complete evap-oration of the solvent [37,38,89]. The driving force for the transi-tion to a smoother surface is the reduction of the overallinterfacial area that in turn reduces the interfacial energy of thedroplet. For solvents like toluene, the formation of smooth surfacesfrom thin films of evaporating polymer solutions has been reportedbefore [89], while rough surfaces were reported to be vitrifiedwhen more volatile solvents like THF were used. These differencesin surface roughness were attributed to the differences in solventvolatility. Thereby the use of very volatile solvents prevents thefilm surface from becoming smooth before it is completely vitrifieddue to complete solvent evaporation, yielding a dried state havinga very long viscoelastic relaxation time.

In addition to the vesicles visible in Fig. 7B and C, we observedthat the solvent evaporation from the double emulsion dropletsshown in Fig. 3B yielded a few microcapsules with thicker wallsas well, one of which is shown in the image in Fig. 7D. These cap-sules typically showed an asymmetric distribution of block copoly-mer in the capsule wall, i.e. with locally thicker patches of polymerin the wall, as indicated by the arrows in Fig. 7D. We attribute theformation of such capsules to deviations of the volume of the mid-dle fluid shell in the initial double emulsion from its average valuethat yielded polymersomes in the solvent evaporation.

To prove this interpretation we investigated the influence ofthe volume of the middle fluid shell in the double emulsiondroplets on the solvent evaporation process and the propertiesof the resulting product. To this end, we used double emulsiondroplets with thicker middle fluid shells in the solvent evapora-tion process, using the same concentration of diblock copolymerof 1 wt.% in the middle fluid as in the previous experiment. Theresulting monodisperse double emulsion droplets are shown inFig. 8A, taken 5 min after the collection of the double emulsionin the reservoir. The diameters of the inner droplets are in therange of 340 lm, whereby the double emulsion droplets havediameters around 460 lm. Accordingly the ratio of the volumeof middle fluid to the volume of inner fluid is larger in the dou-ble emulsion droplets shown in Fig. 8A compared to the dropletsshown in Fig. 3B.

As an intermediate stage of the solvent evaporation we againobtained droplets with rough surface morphology (after about80 min of solvent evaporation, data not shown). After approxi-mately 13.5 more hours of evaporation, the rough surface mor-phology of the droplets has turned into a surface morphologythat is smooth on the length scale of observation, as can be seenin Fig. 8B. The resulting microcapsules have diameters around260–270 lm and thicker walls compared to the vesicles shownin Fig. 7B and C. The solid-like nature of the capsule wall arisesas a vitrified physical gel of entangled copolymer moleculesformed as the evaporation of solvent proceeds. As in the previousexperiment, the microcapsule diameter is much lower than thediameter of the inner droplets in the initial double emulsion, dueto the permeability of the middle fluid shell for inner fluid wateras discussed above. Some of the microcapsules have an asymmetricdistribution of block copolymer in the capsule wall, as highlightedby the arrow in Fig. 8B, while the other microcapsules appear tohave a more homogeneous distribution of polymer as indicatedby the uniform thicknesses of the microcapsule walls. Two repre-sentative capsules with such asymmetric distributions of block

T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150 149

copolymer in the capsule wall are shown with larger magnificationin Fig. 8C.

In addition, we observed a small number of vesicles (polymer-somes) in coexistence with the microcapsules, as shown inFig. 8D. Here a representative vesicle is highlighted by the black ar-row and a representative microcapsule is highlighted by the whitearrow. These vesicles have diameters of the order of 250 lm thatare comparable with the diameters of the microcapsules, and showdistinct patches of spot-patterns of excess block copolymer. Thesepatches are larger and much more visible compared to the respec-tive patches found on the surfaces of the vesicles shown in Fig. 7Band c. The patches are formed because a larger number of polymermolecules are deposited onto the surface of each inner dropletwhen the solvent completely evaporates from larger volumes ofmiddle fluid per double emulsion droplet, as reported before[30]. However, as a comparison of Figs. 7 and 8 clearly indicates,a larger volume of middle fluid per double emulsion dropletpreferentially yields microcapsules, while smaller volumes of mid-dle fluid preferentially yields vesicles during solvent evaporation.

4. Conclusion

We showed how to apply the solvent evaporation method start-ing from water–oil–water double emulsions to obtain either giantvesicles or microcapsules with diameters of the order of 250 lm,formed by the biocompatible and biodegradable amphiphilic blockcopolymer poly(ethylene glycol)-b-poly(c-methyl e-caprolactone)[63]. The preferential formation of either giant vesicles or micro-capsules depends on the volume of the middle fluid shell perdouble emulsion droplet, providing a simple means to controland tune the properties of the resulting structure just by the volu-metric flow rate in the microfluidic formation process of the doubleemulsion. Accordingly it has been shown how the advantages ofmicrofluidic double emulsion formation, namely the high mono-dispersity of the double emulsion droplets and the precise controlof the relative volumes of the inner droplet and the middle fluidshell, can be utilized to tune the properties of the aggregatesresulting from solvent evaporation. Thus we have extended andgeneralized the recently introduced protocol to obtain monodis-perse polymersomes [28–31] by showing how to selectively obtaineither vesicles or microcapsules just by choosing the appropriateproperties of the starting double emulsion droplets in the solventevaporation method. In the course of solvent evaporation interme-diate stages are observed in which the double emulsion dropletsexhibit a rough surface morphology; this morphology is mostlikely related to a phase separation of the middle fluid in the courseof solvent evaporation or a dewetting of the middle fluid from thesurface of the inner droplet.

In conclusion, this versatile approach of vesicle or microcapsuleformation yields the flexibility to continuously vary the propertiesof the resulting structure just by continuously varying thevolumetric flow rates in double emulsion formation. Accordingly,this method will allow easily executable screening studies for theoptimal thickness of the enclosing shell of the resulting vesiclesor microcapsules, an approach that is particularly useful for appli-cations of the structures in for instance oral drug delivery, cellencapsulation [74–78], protein evolution [80,81], artificial cellsand encapsulated gene expression systems [82,83] or single-cellstudies like HTS on single cells [84].

Acknowledgments

We thank Marc Hillmyer, Matt Petersen (both University ofMinnesota) and the Hillmyer lab for the generous provision ofthe block copolymer used here and for stimulating discussions,

as well as Alon McCormick for stimulating discussions. We greatlyacknowledge financial support by the Regents ProfessorshipProgram of the University of Minnesota. One of us (T.F.) is gratefulfor the financial support by the Alexander-von-Humboldt-founda-tion (Bonn, Germany, Feodor-Lynen Program).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.jcis.2010.05.020.

References

[1] M.R. Rosen, Delivery System Handbook for Personal Care and CosmeticProducts: Technology, Applications, and Formulations, William Andrew, Inc.,Norwich, NY, USA, 2005.

[2] D. Lensen, D.M. Vriezema, J.C.M.V. Hest, Macromol. Biosci. 8 (2008) 991.[3] E. Soussan, S. Cassel, M. Blanzat, I. Rico-Lattes, Angew. Chem. Int. Edit. 48

(2009) 274.[4] R. Langer, Science 249 (1990) 1527.[5] D.A. Christian, S. Cai, D.M. Bowen, Y. Kim, J.D. Pajerowski, D.E. Discher, Eur. J.

Pharm. Biopharm. 71 (2009) 463.[6] S. Benita, Microencapsulation: Methods and Industrial Applications, CRC Press,

Boca Raton, FL, USA, 2006.[7] M. Li, O. Rouaud, D. Poncelet, Int. J. Pharm. 363 (2008) 26.[8] M. Rosoff, Vesicles, Marcel Dekker, Inc., New York, 1996.[9] I.F. Uchegbu, Synthetic Surfactant Vesicles: Niosomes and Other Non-

Phospholipid Vesicular Systems, CRC Press, Boca Raton, FL, USA, 2000.[10] A. Napoli, D. Sebök, A. Senti, W. Meier, in: M. Lazzari, G. Liu, S. Lecommandoux

(Eds.), Block Copolymers in Nanoscience, Wiley-VCH, Weinheim, 2006.[11] A.D. Bangham, M.M. Standish, J.C. Watkins, J. Mol. Biol. 13 (1965) 238.[12] A.D. Bangham, Chem. Phys. Lipids 64 (1993) 275.[13] B.M. Discher, Y.Y. Won, D.S. Ege, J.C.M. Lee, F.S. Bates, D.E. Discher,

D.A. Hammer, Science 284 (1999) 1143.[14] D.E. Discher, A. Eisenberg, Science 297 (2002) 967.[15] D.E. Discher, F. Ahmed, Annu. Rev. Biomed. Eng. 8 (2006) 323.[16] D. Marsh, CRC Handbook of Lipid Bilayers, CRC Press, Boca Raton, Fla, 1990.[17] R. Lipowsky, E. Sackmann, Structure and Dynamics of Membranes, Elsevier,

Amsterdam, Nl, 1995.[18] H. Bermudez, A.K. Brannan, D.A. Hammer, F.S. Bates, D.E. Discher,

Macromolecules 35 (2002) 8203.[19] H. Aranda-Espinoza, H. Bermudez, F.S. Bates, D.E. Discher, Phys. Rev. Lett. 87

(2001) 208301.[20] J.C.M. Lee, M. Santore, F.S. Bates, D.E. Discher, Macromolecules 35 (2002) 323.[21] R. Dimova, U. Seifert, B. Pouligny, S. Forster, H.G. Dobereiner, Eur. Phys. J. E 7

(2002) 241.[22] R. Dimova, S. Aranda, N. Bezlyepkina, V. Nikolov, K.A. Riske, R. Lipowsky,

J. Phys.: Condens. Matter 18 (2006) S1151.[23] P. Walde, S. Ichikawa, Biomol. Eng. 18 (2001) 143.[24] J.P. Reeves, R.M. Dowben, J. Cell. Physiol. 73 (1969) 49.[25] D. Needham, E. Evans, Biochemistry 27 (1988) 8261.[26] M.I. Angelova, D.S. Dimitrov, Faraday Discuss. (1986) 303.[27] M.I. Angelova, S. Soleau, P. Meleard, J.F. Faucon, P. Bothorel, Prog. Colloid.

Polym. Sci. 89 (1992) 127.[28] E. Lorenceau, A.S. Utada, D.R. Link, G. Cristobal, M. Joanicot, D.A. Weitz,

Langmuir 21 (2005) 9183.[29] A.S. Utada, E. Lorenceau, D.R. Link, P.D. Kaplan, H.A. Stone, D.A. Weitz, Science

308 (2005) 537.[30] R.C. Hayward, A.S. Utada, N. Dan, D.A. Weitz, Langmuir 22 (2006) 4457.[31] H.C. Shum, J.W. Kim, D.A. Weitz, J. Am. Chem. Soc. 130 (2008) 9543.[32] H.C. Shum, D. Lee, I. Yoon, T. Kodger, D.A. Weitz, Langmuir 24 (2008) 7651.[33] F. Gao, Z.-G. Su, P. Wang, G.-H. Ma, Langmuir 25 (2009) 3832.[34] L. Liu, J.-P. Yang, X.-J. Ju, R. Xie, L. Yang, B. Liang, L.-Y. Chu, J. Colloid Interface

Sci. 336 (2009) 100.[35] C. Thies, in: S. Benita (Ed.), Microencapsulation: Methods and Industrial

Applications, Informa Health Care, New York, NY, USA, 1996.[36] E. Kvitnitsky, N. Lerner, Y.E. Shapiro, in: M.R. Rosen (Ed.), Delivery System

Handbook for Personal Care and Cosmetic Products: Technology, Applications,and Formulations, William Andrew, Inc., Norwich, NY, USA, 2005.

[37] P. vandeWitte, P.J. Dijkstra, J.W.A. vandenBerg, J. Feijen, J. Membr. Sci. 117(1996) 1.

[38] P. vandeWitte, H. Esselbrugge, P.J. Dijkstra, J.W.A. vandenBerg, J. Feijen,J. Membr. Sci. 113 (1996) 223.

[39] H. Noguchi, G. Gompper, Phys. Rev. Lett. 93 (2004) 258102.[40] A. Walter, H. Rehage, H. Leonhard, Colloids Surf. A 183 (2001) 123.[41] Y. Ogawa, M. Yamamoto, H. Okada, T. Yashiki, T. Shimamoto, Chem. Pharm.

Bull. 36 (1988) 1095.[42] V.P. Torchilin, V.S. Trubetskoy, Adv. Drug Deliver. Rev. 16 (1995) 141.[43] G.S. Kwon, Crit. Rev. Ther. Drug Carrier Syst. 15 (1998) 481.[44] A.M. Reed, D.K. Gilding, Polymer 22 (1981) 494.[45] K. Yamaguchi, J.M. Anderson, J. Controlled Release 24 (1993) 81.[46] R.P. Batycky, J. Hanes, R. Langer, D.A. Edwards, J. Pharm. Sci. 86 (1997) 1464.

150 T. Foster et al. / Journal of Colloid and Interface Science 351 (2010) 140–150

[47] J.S. Chawla, M.M. Amiji, Int. J. Pharm. 249 (2002) 127.[48] C.G. Pitt, in: R.S. Langer (Ed.), Biodegradable Polymers as Drug Delivery

Systems, Informa Health Care, 1990.[49] Y. Lemmouchi, E. Schacht, P. Kageruka, R. De Deken, B. Diarra, O. Diall,

S. Geerts, Biomaterials 19 (1998) 1827.[50] J.Z. Bei, J.M. Li, Z.F. Wang, J.C. Le, S.G. Wang, Polym. Adv. Technol. 8 (1997) 693.[51] S.M. Li, J.L. Espartero, P. Foch, M. Vert, J. Biomater. Sci. Polym. Ed. 8 (1996) 165.[52] F. Ahmed, A. Hategan, D.E. Discher, B.M. Discher, Langmuir 19 (2003) 6505.[53] F. Ahmed, D.E. Discher, J. Controlled Release 96 (2004) 37.[54] F.H. Meng, C. Hiemstra, G.H.M. Engbers, J. Feijen, Macromolecules 36 (2003)

3004.[55] P.P. Ghoroghchian, G.Z. Li, D.H. Levine, K.P. Davis, F.S. Bates, D.A. Hammer,

M.J. Therien, Macromolecules 39 (2006) 1673.[56] M.F. Zambaux, F. Bonneaux, R. Gref, P. Maincent, E. Dellacherie, M.J. Alonso,

P. Labrude, C. Vigneron, J. Controlled Release 50 (1998) 31.[57] N. Nihant, C. Schugens, C. Grandfils, R. Jerome, P. Teyssie, J. Colloid Interface

Sci. 173 (1995) 55.[58] G. Crotts, T.G. Park, J. Controlled Release 35 (1995) 91.[59] M. Tobio, R. Gref, A. Sanchez, R. Langer, M.J. Alonso, Pharm. Res. 15 (1998) 270.[60] D.H. Lewis, in: M. Chasin, R.S. Langer (Eds.), Biodegradable Polymers as Drug

Delivery Systems, Informa Health Care, 1990.[61] P. Munk, in: S.E. Webber, P. Munk, Z. Tuzar (Eds.), Solvents and Self-

Organization of Polymers, Kluwer Academic Publishers, Dordrecht, 1996, p. 19.[62] S. Jain, F.S. Bates, Macromolecules 37 (2004) 1511.[63] J.A. Zupancich, F.S. Bates, M.A. Hillmyer, Macromolecules 39 (2006) 4286.[64] P.P. Ghoroghchian, J.J. Lin, A.K. Brannan, P.R. Frail, F.S. Bates, M.J. Therien,

D.A. Hammer, Soft Matter 2 (2006) 973.[65] M. Trollsas, M.A. Kelly, H. Claesson, R. Siemens, J.L. Hedrick, Macromolecules

32 (1999) 4917.[66] B.M. Gumbiner, Cell 84 (1996) 345.[67] D.H. Boal, Mechanics of the Cell, Cambridge University Press, 2002.[68] U. Seifert, R. Lipowsky, Phys. Rev. A 42 (1990) 4768.[69] U. Schwarz, Soft Matter 3 (2007) 263.[70] T.R. Tice, R.M. Gilley, J. Controlled Release 2 (1985) 343.[71] C. Thies, in: M. Rosoff (Ed.), Controlled Release of Drugs: Polymers and

Aggregate Systems, John Wiley & Sons, New York, 1989.

[72] D.T. Chiu, R.M. Lorenz, G.D.M. Jeffries, Anal. Chem. 81 (2009) 5111.[73] A. Huebner, S. Sharma, M. Srisa-Art, F. Hollfelder, J.B. Edel, A.J. Demello, Lab

Chip 8 (2008) 1244.[74] F. Lim, A.M. Sun, Science 210 (1980) 908.[75] H. Uludag, P. De Vos, P.A. Tresco, Adv. Drug Deliver. Rev. 42 (2000) 29.[76] G. Orive, A.R. Gascon, R.M. Hernandez, M. Igartua, J.L. Pedraz, Trends

Pharmacol. Sci. 24 (2003) 207.[77] G. Orive, R.M. Hernandez, A.R. Gascon, R. Calafiore, T.M.S. Chang, P.D. Vos, G.

Hortelano, D. Hunkeler, I. Lacik, A.M.J. Shapiro, J.L. Pedraz, Nat. Med. 9 (2003)104.

[78] W.M. Kühtreiber, R.P. Lanza, W.L. Chick, Cell Encapsulation Technology andTherapeutics, Birkhäuser, Boston, 1999.

[79] P.Y. Bolinger, D. Stamou, H. Vogel, Angew. Chem. Int. Edit. 47 (2008) 5544.[80] D.S. Tawfik, A.D. Griffiths, Nat. Biotechnol. 16 (1998) 652.[81] V.W. Cornish, Nature 440 (2006) 156.[82] V. Noireaux, A. Libchaber, Natl. Acad. Sci. USA 101 (2004) 17669.[83] V. Noireaux, R. Bar-Ziv, J. Godefroy, H. Salman, A. Libchaber, Phys. Biol. 2

(2005) P1.[84] E. Brouzes, M. Medkova, N. Savenelli, D. Marran, M. Twardowski,

J.B. Hutchison, J.M. Rothberg, D.R. Link, N. Perrimon, M.L. Samuels, Natl.Acad. Sci. USA 106 (2009) 14195.

[85] A.S. Utada, A. Fernandez-Nieves, H.A. Stone, D.A. Weitz, Phys. Rev. Lett. 99(2007) 094502.

[86] W.D. Harkins, A. Feldman, J. Am. Chem. Soc. 44 (1922) 2665.[87] S. Torza, S.G. Mason, J. Colloid Interface Sci. 33 (1970) 67.[88] D. Lee, Y.C. Gong, I. Teraoka, Macromolecules 35 (2002) 7093.[89] K.E. Strawhecker, S.K. Kumar, J.F. Douglas, A. Karim, Macromolecules 34 (2001)

4669.[90] F. Brochard-Wyart, P. Martin, C. Redon, Langmuir 9 (1993) 3682.[91] R. Xie, A. Karim, J.F. Douglas, C.C. Han, R.A. Weiss, Phys. Rev. Lett. 81 (1998)

1251.[92] R. Yerushalmi-Rozen, T. Kerle, J. Klein, Science 285 (1999) 1254.[93] Z. Mitov, E. Kumacheva, Phys. Rev. Lett. 81 (1998) 3427.[94] L.E. Scriven, C.V. Sternling, Nature 187 (1960) 186.[95] X. Fanton, A.M. Cazabat, Langmuir 14 (1998) 2554.