bookchapter microfluidic techniques

21
1 Microfluidic techniques for synthesizing particles Adam R. Abate, Sebastian Seiffert, Andrew S. Utada, Anderson Shum, Rhutesh Shah, Julian Thiele, Wynter J. Duncanson, Alirezza Abbaspourad, Myung Han Lee, Ilke Akartuna, Daeyeon Lee, Assaf Rotem, David A. Weitz a School of Engineering and Applied Sciences/Department of Physics, Harvard University, Cambridge, Massachusetts, USA. b Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, Pennsylvania, USA. Introduction Microfluidic devices are networks of micron scale channels that are integrated together to perform functions. The functions typically performed can be divided into two broad classes. In one class, devices perform chemical and biological assays. Cells, beads, and other reagents are introduced into the device, and the channels merge, mix, and split them, as needed for the reactions. This allows a variety of chemical and biological reactions to be executed with high precision, for combinatorial chemical screens, proteomics studies, genetic sequencing, and directed evolution [13]. In the second class of functions, drop formation devices are used to synthesize microparticles [47]. Solutions or melts of monomers or crosslinkable polymers are introduced into the device, along with an immiscible carrier phase; the devices disperse these solutions into equally sized micro-droplets, which can then be solidified by polymerization, crosslinking, or crystallization, thereby producing solid, monodisperse microparticles. The principal advantage of this microfluidic synthesis is that while the chemical composition of the particles is obtained by selecting which solutions to introduce into the device, the final particle structure is obtained by the fluidics. For example, a device that forms anisotropic particles can do so with a variety of fluids, yielding particles with a range of distinct chemical properties but identical structure. This ability to independently select structure and chemical composition is a significant advance over traditional bulk synthesis approaches, since in these cases the final structure is intrinsically linked to the chemistry in the reactor, for example, as it is particle synthesis by emulsion polymerization [810]. This greatly broadens not only the kinds of materials that can be formed into particles, but also the kinds of structures that can be formed, from simple monodisperse plastic spheres to anisotropic magnetic hydrogels, non-spherical Janus particles, and core-shell capsules of a variety of compositions. In this chapter, we provide an overview of the kinds of particles that can be synthesized with microfluidic techniques. We describe the two dominant types of microfluidic devices that are used to synthesize particles, beginning with glass capillary microfluidics and ending with lithographically fabricated poly(dimethylsiloxane) (PDMS) devices. Glass capillary devices afford several unique advantages for forming particles, including high chemical resistance and an ideal coaxial flow focusing geometry; this enables creation of particles of a wide range of compositions and structures. By contrast, lithographically fabricated PDMS devices do not match capillaries in these regards, but afford other advantages that make them superior for certain applications. This includes the ability to tailor channel networks to overcome specific challenges. As we will show, this allows them to create new kinds of particles. The inherently parallel lithographic fabrication also allows these devices to be replicated in large numbers, making them attractive for large- scale synthesis applications. SECTION 1: Glass capillary microfluidics Single emulsion particle templating The first step to forming particles with microfluidic devices is to form monodisperse populations of micro-droplets. The principle of drop formation in microfluidic devices can be explained using a water faucet as an example. If we turn on a faucet at a low flow rate, water drips out one drop at a time. The drop size is a result of the balance between the surface forces of the hanging drop and its weight, and therefore depends on the surface tension of the fluid and the size of the faucet. Since both the surface tension and the faucet size are constant, all drops dripping from a faucet exhibit a narrow size distribution. However, if we gradually increase the flow rate through the faucet, a thin water stream, or a jet is formed. Although the jet eventually breaks up into drops as well, these have a larger size range [1114]. The same principle can be employed in microfluidic channels that have sizes on the order of tens of micrometers. One significant difference between drop formation at a faucet and in microfluidic devices is that in the former case drops are formed in air, whereas in the latter case drops are formed in another immiscible liquid. Capillary microfluidics presents a way to controllably generate drops of one liquid in another immiscible liquid in devices that consist of coaxial assemblies of glass capillaries. One of the inherent advantages of these devices is that since they are made from glass, their wettability can be easily and precisely controlled by a surface reaction with an appropriate surface modifier. For example, a quick treatment of octadecyltrimethoxysilane will render the glass surface hydrophobic, whereas a treatment of 2-[methoxy(poly- ethyleneoxy)propyl]trimethoxysilane will make the surface more hydrophilic. An additional benefit of using glass is that the devices are both chemically resistant and rigid. Lastly and perhaps most importantly, these devices offer the distinct capability of creating truly three-dimensional flows, which is critical for some of the applications that we describe later. To build these devices, we begin with a circular glass capillary with an outer diameter of 12 mm. This capillary is heated and pulled using a pipette puller to create a tapered geometry that culminates in a fine orifice; this is our “faucet”. The precisely pulled circular capillary is carefully inserted into a square glass capillary to form a simple microfluidic device. We

Upload: lgrosche

Post on 29-Dec-2015

167 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Bookchapter Microfluidic Techniques

1

Microfluidic techniques for synthesizing particles Adam R. Abate, Sebastian Seiffert, Andrew S. Utada, Anderson Shum, Rhutesh Shah, Julian Thiele, Wynter J. Duncanson, Alirezza Abbaspourad, Myung Han Lee, Ilke Akartuna, Daeyeon Lee, Assaf Rotem, David A. Weitz a School of Engineering and Applied Sciences/Department of Physics, Harvard University, Cambridge, Massachusetts, USA.

b Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, Pennsylvania, USA.

Introduction

Microfluidic devices are networks of micron scale channels that are integrated together to perform functions. The functions typically performed can be divided into two broad classes. In one class, devices perform chemical and biological assays. Cells, beads, and other reagents are introduced into the device, and the channels merge, mix, and split them, as needed for the reactions. This allows a variety of chemical and biological reactions to be executed with high precision, for combinatorial chemical screens, proteomics studies, genetic sequencing, and directed evolution [1–3]. In the second class of functions, drop formation devices are used to synthesize microparticles [4–7]. Solutions or melts of monomers or crosslinkable polymers are introduced into the device, along with an immiscible carrier phase; the devices disperse these solutions into equally sized micro-droplets, which can then be solidified by polymerization, crosslinking, or crystallization, thereby producing solid, monodisperse microparticles.

The principal advantage of this microfluidic synthesis is that while the chemical composition of the particles is obtained by selecting which solutions to introduce into the device, the final particle structure is obtained by the fluidics. For example, a device that forms anisotropic particles can do so with a variety of fluids, yielding particles with a range of distinct chemical properties but identical structure. This ability to independently select structure and chemical composition is a significant advance over traditional bulk synthesis approaches, since in these cases the final structure is intrinsically linked to the chemistry in the reactor, for example, as it is particle synthesis by emulsion polymerization [8–10]. This greatly broadens not only the kinds of materials that can be formed into particles, but also the kinds of structures that can be formed, from simple monodisperse plastic spheres to anisotropic magnetic hydrogels, non-spherical Janus particles, and core-shell capsules of a variety of compositions.

In this chapter, we provide an overview of the kinds of particles that can be synthesized with microfluidic techniques. We describe the two dominant types of microfluidic devices that are used to synthesize particles, beginning with glass capillary microfluidics and ending with lithographically fabricated poly(dimethylsiloxane) (PDMS) devices. Glass capillary devices afford several unique advantages for forming particles, including high chemical resistance and an ideal coaxial flow focusing geometry; this enables creation of particles of a wide range of compositions and structures. By contrast, lithographically fabricated PDMS devices do not match capillaries in these regards, but afford other advantages that make them superior for certain applications. This includes the ability to tailor channel networks to overcome specific challenges. As we will show, this

allows them to create new kinds of particles. The inherently parallel lithographic fabrication also allows these devices to be replicated in large numbers, making them attractive for large-scale synthesis applications. SECTION 1: Glass capillary microfluidics Single emulsion particle templating The first step to forming particles with microfluidic devices is to form monodisperse populations of micro-droplets. The principle of drop formation in microfluidic devices can be explained using a water faucet as an example. If we turn on a faucet at a low flow rate, water drips out one drop at a time. The drop size is a result of the balance between the surface forces of the hanging drop and its weight, and therefore depends on the surface tension of the fluid and the size of the faucet. Since both the surface tension and the faucet size are constant, all drops dripping from a faucet exhibit a narrow size distribution. However, if we gradually increase the flow rate through the faucet, a thin water stream, or a jet is formed. Although the jet eventually breaks up into drops as well, these have a larger size range [11–14]. The same principle can be employed in microfluidic channels that have sizes on the order of tens of micrometers. One significant difference between drop formation at a faucet and in microfluidic devices is that in the former case drops are formed in air, whereas in the latter case drops are formed in another immiscible liquid. Capillary microfluidics presents a way to controllably generate drops of one liquid in another immiscible liquid in devices that consist of coaxial assemblies of glass capillaries. One of the inherent advantages of these devices is that since they are made from glass, their wettability can be easily and precisely controlled by a surface reaction with an appropriate surface modifier. For example, a quick treatment of octadecyltrimethoxysilane will render the glass surface hydrophobic, whereas a treatment of 2-[methoxy(poly-ethyleneoxy)propyl]trimethoxysilane will make the surface more hydrophilic. An additional benefit of using glass is that the devices are both chemically resistant and rigid. Lastly and perhaps most importantly, these devices offer the distinct capability of creating truly three-dimensional flows, which is critical for some of the applications that we describe later.

To build these devices, we begin with a circular glass capillary with an outer diameter of 1–2 mm. This capillary is heated and pulled using a pipette puller to create a tapered geometry that culminates in a fine orifice; this is our “faucet”. The precisely pulled circular capillary is carefully inserted into a square glass capillary to form a simple microfluidic device. We

Page 2: Bookchapter Microfluidic Techniques

2

ensure coaxial alignment of the two capillaries by choosing the tubes such that the outer diameter of the circular capillary is the same as the inner dimension of the square capillary. Flowing one fluid inside the circular capillary while flowing a second fluid through the square capillary in the same direction results in a three-dimensional coaxial flow of the two fluids, as illustrated in Fig. 1a. This is known as coflow geometry [15, 16].

Fig 1 Formation of single emulsions in a coflow microfluidic device. (a) Schematic of a coflow microcapillary device for making droplets. Arrows indicate the flow direction of fluids and drops. (b) Image of drop formation at low flow rates (dripping regime). (c) Image of a narrowing jet generated by increasing the flow rate of the continuous fluid above a threshold value, while keeping the flow rate of the dispersed phase constant. (d) Image of a widening jet generated by increasing the flow rate of the dispersed fluid above a threshold value, while keeping the flow rate of the continuous phase constant. (e) Monodisperse droplets formed using a microcapillary device. Parts (b), (c), and (d) are reprinted from [16]. Copyright 2007 American Physical Society.

When both fluids flow at low rates, individual monodisperse drops are formed periodically at the tip of the capillary orifice, in a process termed dripping, shown in Fig. 1b [15–17]. In the dripping regime, the drop at the end of the capillary tube experiences two competing forces: viscous drag pulling it downstream and forces due to surface tension holding it to the capillary. Initially, surface tension dominates but as the attached droplet grows, drag forces eventually become comparable; this is when the droplet breaks off and is carried away by the flow of the continuous phase [17]. If we increase the flow rate of either fluid beyond a certain critical limit, the result is a jet, which is a long stream of the inner fluid with drops forming downstream, as shown in Fig. 1c and 1d. Typically, these drops have a broader size distribution than drops formed from dripping because the point at which the drops separate from the jet changes with each drop.

We observe two distinct dripping-to-jetting transitions when we vary the two flow rates, respectively, which is due to a different balance of forces on the jetting liquid. The first transition is driven by the flow rate of the outer fluid; as it is increased, drops formed at the tip decrease in size until the emerging fluid is finally stretched into a jet. At this point, drop breakup occurs downstream at the end of the thin jet, as shown in Fig. 1c. The second transition is driven by the flow rate of the inner fluid; as it is increased, the dripping drop is pushed downstream and ultimately pinches-off from the end of the resultant jet, shown in Fig. 1d.

Fig 2 Dependence of the transition between dripping and jetting on the capillary number of the outer flow and the Weber number of the inner flow of a coflow microcapillary device. Squares and diamonds: ηin/ηout = 0.01, with slightly different geometries; hexagons and circles: ηin/ηout = 0.1, with slightly different geometries; pentagons: ηin/ηout = 1; triangles, ηin/ηout = 10. Reprinted from [16]. Copyright 2007 American Physical Society.

To elucidate these processes, we plot them on a single phase diagram based on the relevant force balance that induces the transition. We find that the behavior of the system is determined by two non-dimensional numbers: the capillary number and the Weber number. The capillary number reflects the balance between the drag of the outer fluid pulling the drop downstream and surface tension forces that resist the flow in the jet as pinch-off occurs. The Weber number reflects a balance between inertial forces of the inner liquid pushing the drop downstream and, again, the surface tension forces resisting the flow. Since both numbers describe a balance between an applied force and surface tension forces, the boundary between dripping and jetting occurs when either number, or their sum, is approximately equal to 1, as shown in Fig. 2.

Fig 3 Schematic of a flow-focusing microcapillary device for making droplets. An alternate geometry for drop formation in capillary devices is the flow-focusing geometry [18, 19]. In contrast to coflow capillary devices, the two fluids are introduced from the two ends of the same square capillary, from opposite directions. The inner fluid is hydrodynamically focused by the outer fluid through the narrow orifice of the tapered round capillary, as

Page 3: Bookchapter Microfluidic Techniques

3

shown in Fig. 3. Under dripping conditions, drop formation occurs as soon as the inner fluid enters the circular orifice, whereas under jetting conditions it occurs further downstream. An advantage of this method is that it allows us to make monodisperse drops with sizes smaller than that of the orifice. This feature is useful for making small droplets (~1–5 μm in diameter), especially those from a particulate suspension, where the particles may clog the orifice in the coflow geometry [20]. The use of a capillary with a larger orifice minimizes the probability of such tip clogging by the suspended particles or any entrapped debris. Templating thermoresponsive microgels from single emulsions

The ability to form droplets with a microfluidic device

allows one, essentially, to structure fluids – to disperse a continuous fluid into a series of equally-sized liquid spheres. These spheres can then be solidified to produce particles of the same size and shape. As an example of this “templating” process, we make monodisperse microgels that consist of a crosslinked network of poly(N-isopropylacrylamide), PNIPAm, swollen with water. Such PNIPAm microgels swell and shrink reversibly in response to changes in temperature. This size change occurs around 32 °C, which is close to the human body temperature. Hence, these microgels are being extensively evaluated for controlled delivery of water soluble drugs. Low polydispersity of PNIPAm microgels is desirable for drug delivery applications as it

could lead to narrow distribution of drug loading levels and uniform release kinetics.

To make these microgels, we use a glass capillary single emulsion device [21, 22]. We begin with an aqueous pre-gel mixture which is emulsified in an oil phase using the device. The monodisperse drops are then gelled just after formation, thereby creating uniform microgels. A schematic of the device employed for fabricating PNIPAm microgels is shown in Fig. 4a. An aqueous phase containing the monomer, N-isopropylacrylamide, a crosslinker, N, N’ methylene-bis-acrylamide (BIS), and a reaction initiator, ammonium persulfate, is introduced from the left end of the left square capillary. An oil, kerosene, pumped in from the right end of the left square capillary, hydrodynamically focuses the aqueous phase into the collection tube where the aqueous phase breaks into monodisperse drops. A reaction initiator, N,N,N’,N’-tetramethyethylenediamine (TEMED), dissolved in kerosene, is introduced into the device through the left end of the right square capillary. TEMED is soluble in both oil as well as water. As a result, it diffuses into the aqueous droplets triggering a redox reaction which causes simultaneous polymerization and gelation of monomers dissolved in the droplets. The resulting monodisperse PNIPAm microgels are subsequently washed and redispersed in water. Since the size of the microgels follows the size of the emulsified drops, it can be tuned by controlling drop size, and hence, by adjusting flow rates. The resultant microgels exhibit excellent thermal response; when heated, their diameter reduces to less than half the original diameter, as shown in Fig. 1b. The size change is reversible and reproducible over several cycles.

Fig 4 (a) Schematic illustration of a capillary-based microfluidic device for fabricating monodisperse PNIPAm microgels. Fluid A is an aqueous suspension containing the monomer, crosslinker, and initiator; fluid B is an oil, and fluid C is the same oil as fluid B but contains a reaction accelerator that is both water- and oil-soluble. The accelerator diffuses into the drops and polymerizes the monomers to form monodisperse microgels. Cross-sectional views at different points along the device length

are shown in the second row. (b) Size change of PNIPAm microgels in water triggered by changing the temperature. The scalebar in Panel b denotes 100 m. (a) Reproduced from [21]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.), (b) Reproduced from [22]. Copyright 2008 The Royal Society of Chemistry.

Functional materials can be embedded inside such

microgels to impart additional properties. This is typically done by suspending such materials in the aqueous pre-gel mixture prior to emulsification. We demonstrate this by fabricating microgels that are complexed with polymer particles, quantum dots, or magnetic nanoparticles, shown in Figs. 5(a–c) [23]. The addition of such materials has no detrimental effect on the thermosensitive behavior of the microgels, as they are not chemically bonded to the polymer network but are only physically trapped within it. Alternatively, microgels with

embedded voids can also be fabricated by embedding polymeric particles inside microgels and subsequently dissolving the embedded particles, shown in Fig 5d [21, 22]. Incorporation of voids improves the kinetics of size change of the microgels since the kinetics is a function of the rate of diffusion of water through the microgels, which increases upon incorporation of voids.

(b)(a)

Page 4: Bookchapter Microfluidic Techniques

4

Fig 5 Microgels with embedded materials. (a) A fluorescence microscope image

of a microgel containing fluorescently labeled 1- m diameter polystyrene particles. (b) A fluorescence microscope image of a microgel containing 19-nm quantum dots. (c) A bright-field microscope image of a microgel containing 10-nm magnetic particles. (d) A bright-field microscope image of a microgel with embedded voids. Reproduced from [21] and [23]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.

Similar microgels can also be generated through

controlled crosslinking of pre-fabricated PNIPAm polymer chains. To demonstrate this concept, we use PNIPAm chains which are functionalized with dimethylmaleimide side groups, a moiety that can be selectively transformed into dimers by UV irradiation [24, 25], thereby crosslinking the chains. We prepare these precursors by copolymerizing N-isopropylacrylamide and a DMMI-functionalized acrylamide-derivative in a free-radical reaction in water, and we control the molecular weight of the resultant copolymers by performing this polymerization in the presence of sodium formate [26]. Once formed, the resultant crosslinkable precursor polymers are used to create pre-microgel droplets in a microfluidic device. We emulsify aqueous precursor solutions with polymer concentrations in the semidilute unentangled regime, an intermediate range above the threshold for coil overlap, c*, yet below the onset of chain entanglement, ce*; these experimental conditions guarantee that a space-filling polymer network can be formed inside each droplet, while the viscosity of the polymer solution is not too high. After forming pre-microgel droplets, microgel particles are obtained by droplet gelation, achieved through photocrosslinking of the precursor polymers in the drops. The advantage of this approach of particle fabrication is that it separates the polymer synthesis from the particle gelation; this allows each to be controlled independently, thus enabling microgel particles to be formed with well-controlled composition and functionality [26]. Templating colloidosomes from single emulsions

Besides just gelling a pre-particle droplet, higher order particles, alternatively known as supraparticles, can be fabricated by controlling the assembly of nanoparticles/colloidal particles within microfluidically generated single emulsion droplets. As an example, we demonstrate the fabrication of monodisperse colloidosomes, microcapsules with a shell composed of tightly packed colloidal particles, using colloidal PNIPAm microgels as building blocks [27]. An aqueous suspension of amine-functionalized sub-micrometer-sized PNIPAm microgels is emulsified in an oil using a single-emulsion microfluidic device. Prior to emulsification, a small amount of glutaraldehyde is added to the aqueous mixture. The colloidal PNIPAm microgels assemble at the oil-

water interface within the emulsion droplets due to the presence of hydrophobic isopropyl groups and hydrophilic acrylamide groups. Glutaraldehyde molecules, owing to their two reactive sites each, serve as connecting links between the amine-functionalized microgels through an amine-aldehyde condensation reaction. Interlinking of the microgels at the oil-water interface results in the formation of colloidosomes. The overall schematic is presented in Fig. 6a. Such colloidosomes exhibit thermosensitive behavior similar to that displayed by their constituent microgels: when the temperature is increased above the phase-transition temperature of PNIPAm, the diameter of the colloidosomes decreases by 42%, which roughly translates to an 80% decrease in volume, as shown in Fig. 6b. Thus, they can be of immense potential in applications that require targeted pulsed-release of active materials. Templating Janus particles from single emulsions

Another class of supraparticles that can be made using

microfluidically generated droplets as templates are Janus particles. Janus particles are biphasic particles with two sides of different composition and functionality. We fabricate monodisperse Janus supraparticles with a PNIPAm microgel-rich side and a polyacrylamide (PAAm) rich side [28]. This is accomplished by emulsifying an aqueous suspension of amine-functionalized microgels using a single-emulsion microfluidic device. Prior to emulsification, we add a small amount of high molecular weight polyacrylic acid (PAAc) to the microgel suspension to induce clustering of microgels by electrostatic interactions between the ammonium ions of the microgels and the carboxyl groups of PAAc [29]. We also dissolve 10 wt% acrylamide in the microgel suspension along with a crosslinker and a photoinitiator, as shown in Fig. 7a. This aqueous mixture is emulsified in an oil and heated at 65 °C in an oven. Upon heating, the weakly associated PNIPAm microgel aggregate shrinks and becomes compacted on one side of the droplets by pushing the acrylamide containing water to the other side, thus forming phase-separated Janus droplets, as shown in Fig. 7b. The acrylamide monomer is then polymerized and cross-linked by exposure to ultraviolet (UV) radiation, forming Janus supraparticles with a PNIPAm microgel-rich side and a PAAm-rich side, as shown in Fig 7c. The functional dichotomy of such Janus particles can be further enhanced by embedding different materials selectively into either side of the particles, as illustrated by the incorporation of magnetic nanoparticles in the microgel-rich side of the particles. Anionic magnetic beads are added to the aqueous mixture of the PNIPAm microgels and other monomers. Since the microgel particles are cationic, the magnetic beads electrostatically bind to the surface of the microgels, and are thus trapped only in the PNIPAm phase of the Janus particles, as shown in Fig. 7d. Such magnetically anisotropic particles can be used to make magnetically actuated displays or other applications that require directional orientation or transportation of particles.

50μm50μm

(a) (b) (c) (d)

Page 5: Bookchapter Microfluidic Techniques

5

Fig 6 (a) Schematic representation of a technique used for fabricating colloidosomes using poly(N-isopropylacrylamide) microgel particles as building blocks and emulsion droplets as templates. (b) Equilibrium size change of PNIPAm colloidosomes and the constituent PNIPAm microgels. Colloidosomes were dispersed in water and heated from 20 to 50 °C in fixed increments of 2 °C and then to 55 °C. Images were captured after allowing the sample to equilibrate for 30 min at each temperature. The sample was then cooled down to 20 °C using the same temperature steps. Size-change data of the constituent PNIPAm microgels over the same temperature range were collected using dynamic light scattering. Adapted and reproduced from [27]. Copyright 2010 American Chemical Society.

Fig 7 (a) Schematic representation of a process for making Janus supraparticles. (b) Aggregation and compaction of PNIPAm microgels on one side of pre-microgel droplets upon heating (c) Fluorescent microscope image and SEM image of a Janus particle formed by photopolymerizing the monomers in the phase separated droplets. The PNIPAm microgels are tagged with Rhodamine B to enhance visual contrast between the two phases. (d) Magnetically anisotropic Janus particles generated by embedding oppositely charged iron oxide particles selectively into the PNIPAm microgel rich phase. The inset shows a magnified image of a single particle with the PNIPAm side attracted to a magnet. Reproduced from [28]. Copyright 2009 Wiley-VCH Verlag GmbH & Co. KGaA.

Templating shell-stabilized bubbles from gas-filled single emulsions

Microfluidic emulsification is not limited to the use of liquid fluids, but also allows gaseous fluids to be used. This enables synthesis of microparticles that encapsulate gas-filled voids, or, in other words, shell-stabilized bubbles. Free gas

bubbles are by far the most effective sound scatterers in liquids [30]. As they oscillate under acoustic pressure waves, they act as a point sound source to enhance contrast in ultrasound imaging [31]. However, the contrast enhancement from free gas bubbles is limited as they are inherently unstable due to high surface tension at the gas-liquid interface [32]. To enhance bubble lifetime or bubble stability, shell coatings of lipids, sugars, or polymers have been used to minimize the dissolution of gas [30, 32]. Traditionally, chaotic, high-shear batch processing methods produce bubbles with broad size distributions and a wide range of shell thicknesses [33]. The thickness and properties of the shell layer directly affect the bubble lifetime and stability, as well as the acoustic response [34]. Thus, control over shelled bubble fabrication allows us to design bubbles for specific applications, as we can precisely tune their properties.

Fig 8 Images showing the textured morphology of AIM oil-shell stabilized bubbles (a) wet with a bright-field microscope and (b) dried with an SEM. The scale bar denotes 100 µm.

Glass capillary microfluidics can form coated gas

bubbles with controlled size and shell thickness. To form these bubbles, we use a microfluidic device to create an emulsion consisting of a gas bubble surrounded by an oil shell, dispersed in an aqueous carrier phase. Due to the ability to independently control the structure of the bubbles and their chemical composition, we can form these same structures using a variety of gases and materials for the shells. In one

Heat

Cool

Heat

Cool

Temperature (oC)

20 30 40 50 60

Rela

tive d

iam

ete

r d

/d20

0.5

0.6

0.7

0.8

0.9

1.0

ConstituentPNIPAmmicrogels

PNIPAmcolloidosomes

T= 20 °C

T= 55 °C

Temperature (oC)

20 30 40 50 60

Rela

tive d

iam

ete

r d

/d20

0.5

0.6

0.7

0.8

0.9

1.0

ConstituentPNIPAmmicrogels

PNIPAmcolloidosomes

T= 20 °C

T= 55 °C

(a) (b)

H20 +

v

NH2H2N

NH2

NH

2H

2N

NH 3

H 3N

H3 N

UV

OIL OIL

NH2H2N

NH2

NH

2H

2N

NH 3

H 3N

H3 N

NH2

H2 N

NH

2

NH 2

H 2N

NH3H3N

H3 N

+

+

+ Photo initiator

OIL

COOH

COOH

COO-

-OOC

65 °C 65 °C25 °C

H20 +

v

NH2H2N

NH2

NH

2H

2N

NH 3

H 3N

H3 N

NH2H2N

NH2

NH

2H

2N

NH 3

NH 3

H 3N

H 3N

H3 N

H3 N

UV

OIL OIL

NH2H2N

NH2

NH

2H

2N

NH 3

H 3N

H3 N

NH2H2N

NH2

NH

2H

2N

NH 3

NH 3

H 3N

H 3N

H3 N

H3 N

NH2

H2 N

NH

2

NH 2

H 2N

NH3H3N

H3 N

NH2

H2 N

NH

2

NH 2

H 2N

NH3NH3H3NH3N

H3 N

H3 N

+

+

+ Photo initiator

OILOIL

COOH

COOH

COO-

COO-

-OOC-OOC

65 °C 65 °C25 °C

(a)

(b) (c) (d)

Page 6: Bookchapter Microfluidic Techniques

6

system, we use low-melting point AIM or paraffin oils for the shells and air as the inner phase, as shown in Fig. 8. The bubbles are collected in an ice water bath that instantly solidifies the shells. These solid-shelled bubbles exhibit superb stability, making them of potential use for ultrasound imaging and triggered or targeted imaging applications.

Another system that produces stable gas bubbles uses glassy polymers, such as poly(D, L-lactic-co-glycolic acid) (PLGA), polystyrene (PS), and poly(methyl methacrylate) (PMMA) for the shells, and different filling gases (N2, CO2, He, and compressed air) for the cores [35]. In this approach, a polymer is dissolved in the oil phase, and the volatile solvent is removed via evaporation, thereby generating a polymer shell at the air-water interface. The stability of the bubble shell against elastic instability-induced deformation strongly depends on the ratio of shell thickness to bubble radius. Bubbles with smaller size or thicker shells exhibit enhanced stability against deformation. In addition, the stability of bubbles is improved by using polymers with higher stiffness and gas with lower solubility. These observations provide guidance for the generation of stable bubbles with controlled physical properties, for applications in ultrasound imaging and ultrasound-triggered release [36].

Nanoparticles can also be used to produce exceedingly stable “armored” bubbles [37, 38]. To make these bubbles, we disperse hydrophobic silica nanoparticles in a toluene oil phase, and use this as the shell around a gas bubble [39, 40]. We can precisely control the shell thickness and bubble size by varying the flow rates during the microfluidic formation. Upon solvent evaporation, the nanoparticles jam at the bubble surface, providing a highly stable solid shell.

Double and triple emulsion particle templating

As we have seen, single emulsion microfluidic devices afford unparalleled control over size and chemical composition of particles. However, this approach almost always produces spherical drops, because the templating drops are themselves nearly always spheres. While spheres are desirable for certain applications, there are others in which a non-spherical shape is needed, or in which the particles must have voids in them, for example, as compartments into which to load drugs or other active materials. In this section, we describe a microfluidic technique that can create non-spherical particles and core-shell capsules. This is achieved by forming double emulsions – drops with smaller drops contained inside – and then solidifying them to produce compound structures.

Fig 9 Fabrication of double emulsions in microfluidic devices. Schematic of a capillary microfluidic device that combines co-flow and flow focusing.

A multiple emulsion is a drop containing additional, smaller drops in its bulk. These structures have a core-shell architecture, like a capsule, allowing them to be used for the encapsulation and release of materials in cosmetics, drug delivery, and food applications. Accurate control of the size and structure of emulsions is often essential for these applications, because these features directly affect the loading levels and the release kinetics of the encapsulated substances. The high degree of control offered by glass capillary microfluidic devices enables fabrication of multiple emulsions. One design to do this combines both co-flow and flow focusing, as shown in Fig 9 [18]. This device consists of two circular capillaries arranged end-to-end within a square capillary. By ensuring that the inner dimensions of the square capillary are the same as the outside diameters of the round capillaries, we achieve good coaxial alignment. The inner fluid is pumped through the tapered circular capillary while the middle fluid, which is immiscible with the inner and outer fluids, flows through the outer capillary in the same direction. The outermost fluid flows through the outer capillary in the opposite direction and hydrodynamically flow focuses the coaxially flowing stream of the other two fluids, which approach from the opposite end. When the three fluids enter the collection tube, a double emulsion is formed.

Fig 10 (a) Dependence of thread radius Rthread and drop radius Rdrop on the scaled flow rate QOF/Qsum. The open symbols represent the Rthread for different liquids and double emulsions consisting of a single silicon drop surrounded by a liquid shell (3QIF = QMF, triangle). The dashed line represents the predicted Rthread. Rdrop values are represented with solid identical symbols. Half-filled triangles correspond to the radius of the internal droplets of the double emulsions. The solid line represents the predicted Rdrop. (b) The flat velocity profile of the flow as it enters the capillary tube. (c-e) Double emulsion drops

flowing through the collection tube. The ratio of viscosities, MF/ OF ≈ 0.08. The

Page 7: Bookchapter Microfluidic Techniques

7

radii of drops shows that Rdrop = 1.82Rthread, in agreement with what is expected theoretically [44]. Flow rates were controlled with stepper-motor-controlled syringe pumps (Harvard Apparatus). Reprinted from [16]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.

To better control the size and composition of the double emulsions, we examine the physical mechanism of drop formation. At lower flow speeds, double emulsions typically form approximately one tube diameter downstream from the entrance, within the collection tube [18, 41–44]. When drops form here, the size is controlled by the ratio of the flow rates of the combined inner and middle fluids to the outer fluid, which is given by Eq. (1), where Qsum is the sum of the of inner and middle fluid flow rates, QOF is the flow rate of the outer fluid, Rthread is the radius of the fluid thread that breaks into drops, and Rorifice is the radius of the collection tube where the drops are formed.

Qsum / Qof = Rthread2

/ ( Rorifice2 Rthread

2 )

[1]

This equation is valid for plug-flow (Fig. 10b), which is a

reasonable assumption given the proximity of drop formation to the entrance of the collection tube. The experimentally measured diameters of different threads and the corresponding drops that pinch from these threads are shown as the open and closed symbols, respectively, in Fig. 10a. By solving for Rthread/Rorifice in Eq. (1) and plotting it as a function of Qsum/QOF, we can quantitatively predict the experimentally measured values, as shown by the dashed line in Fig. 10a. When a fluid thread breaks, the diameter of the resulting drop is proportional to the diameter of the thread as well as the

viscosity ratio. For a viscosity ratio of MF/ OF = 0.1, the drop diameter is approximately twice the diameter of the thread [13]. We neglect the decrease in viscosity due to the contribution of the inner liquid because its volume fraction in the drop is small. By multiplying the predicted thread diameter by a factor of two, we again see very good agreement between the model and the data. These simple physical arguments highlight the versatility of this method in generating monodisperse double emulsions under different flow conditions (Fig. 10c–e). Templating microshells and capsules from double emulsions

Just like single emulsions, double emulsion can be used to template structures, but these structures can have more elaborate shapes. To illustrate this templating process, we make monodisperse polymer microshells. We use a glass capillary device to form monodisperse drops of a curable precursor solution that consists of a polymerizable or crosslinkable substance. We use thermoresponsive pNIPAM as the matrix polymer to obtain environmentally sensitive microgel particles, which we either fabricate from monomeric NIPAM and BIS or which we gel from pre-fabricated, photocrosslinkable pNIPAM precursors. To form the pre-microgel droplets, we emulsify an aqueous microgel precursor solution in a continuous oil phase. At the moment of their formation, these pre-microgel droplets are loaded with inner droplets of another oil, thereby creating a shell structure, as

shown in Fig. 11a. After droplet formation, the shell is gelled by thermal monomer polymerization of by photochemical polymer-analogous gelation, thereby yielding uniform PNIPAm microshells as shown in Fig. 11b. Operating the device with different flow rates produces shells with two cores, as shown in Fig. 11c.

Fig 11 Microfluidic production of hollow microgel shells. (a) A glass microcapillary device is used to create an oil-water-oil double emulsion with a semidilute solution of crosslinkable pNIPAAm as aqueous phase. Subsequently, these droplets are cured by UV exposure as they flow through a delay capillary a few centimeters downstream (not shown). (b) pNIPAAm microshells obtained from the experiment in Panel A. (c) Double-core microshells obtained upon slight variation of the flow rates in the experiment in Panel A (cf. Ref. [26] for

details). All scale bars denote 200 m. Reproduced from [26]. Copyright 2010 The Royal Society of Chemistry.

Double emulsion-templated vesicles

Vesicles are compartments of fluid enclosed by

bilayers of amphiphilic molecules, such as diblock copolymers, phospholipids, and polypeptides. Their properties, such as permeability and selectivity, are tunable by varying the thickness and chemical composition of the bilayer membranes; thus they are widely used as encapsulating structures of active ingredients for applications ranging from pharmaceuticals to cosmetics. Due to the resemblance of the structures to cell membranes, materials encapsulated in them can potentially be released directly into cells through fusion of the vesicles with the target cells. Recently, polymer vesicles, or polymersomes, have been shown to exhibit better mechanical stability than phospholipid vesicles, while retaining many of their attractive properties such a tunable permeability and targeted release, and are thus a promising alternative. We have developed an alternate approach for fabricating monodisperse vesicles from these polymers, by employing monodisperse double emulsions as templates.

To illustrate this approach, we form monodisperse polymersomes from a biodegradable, biocompatible diblock copolymer, poly(ethylene glycol)-block-poly(lactic acid) (PEG-b-PLA) using a capillary microfluidic device [45]. PEG is a hydrophilic block while PLA is a hydrophobic block. Hence, when PEG-b-PLA is dissolved in the shell phase of a W/O/W double emulsion, the PEG-b-PLA molecules adsorbs at the oil/water interfaces. Upon evaporation of the oil phase, the double emulsion drops undergo a dewetting transition where the shell phase is collected into a single drop of solvent with the excess diblock copolymers, as shown in the series of images in Fig. 12c. The inner droplet is enclosed by a bilayer membrane that consists of the diblocks at the two interfaces. The solvent drop is subsequently evaporated and a polymersome is formed. The resultant vesicles have great size uniformity and

To UV Exposure

Oil

Oil

A B

C

Page 8: Bookchapter Microfluidic Techniques

8

encapsulation efficiency as demonstrated by the encapsulation of a model fluorescent dye shown in Fig. 12b. The actives can be released with a simple osmotic shock, for instance, by increasing the osmolality of the environment and bursting the vesicles by sucking water out, as shown in Fig. 12d. By dissolving functional homopolymers in the shell phase of the double emulsion templates, we can also incorporate homopolymers in the membrane and tune the properties of the polymersomes [45]. This approach for forming polymersomes is general and has also been applied to form polymersomes of poly(normal-butyl acrylate)-block-poly(acrylic acid) (PBA-b-PAA) [46] and polystyrene-block-poly(ethylene oxide), (PS-b-PEO) [47]. Using this double emulsion-templated approach, we have also fabricated monodisperse phospholipid vesicles with high encapsulation efficiency [48].

Fig 12 (a) Schematic illustrating a double-emulsion-templated approach for fabricating polymersomes. (b) Encapsulation of a fluorescent dye in

polymersomes. Scale bar is 100 m. (c) Dewetting transition of a double emulsion drop. Successive images are taken at intervals of 1.8 s, and the scale

bar is 10 m. (d) Shrinkage and breakage of a polymersome after an osmotic

shock. The scale bar is 10 m. Reprinted from [45]. Copyright 2008 American Chemical Society.

Templating multi-core capsules

Double emulsions with a single-phase inner droplet can be used directly for encapsulating actives or indirectly as templates for more sophisticated capsules. An important technical challenge in this area is the co-encapsulation of multiple incompatible actives without cross-contamination. To achieve this, double emulsion drops with multiple separate inner droplets for the different actives are needed. If multiple reactants can be stored separately within the same double emulsion droplets, the desired reactions can be triggered by coalescence of the inner droplets and subsequent mixing of the reactants. Microfluidic devices are apt for fabricating such structures due to their easily customized geometries. We demonstrate this using a modified glass capillary device, which has an injection tube with two separate internal channels, illustrated in Fig. 13a [49]. The model encapsulants, wright stain (blue) and rhodamine B (red) dyes flow separately into the device, and then in separate channels before forming two separate inner droplets without mixing. Thus, the actives in the

separate droplets are not pre-mixed before triggered release. Using such a device, we have fabricated double emulsion droplets with two cores, as shown in Fig. 13b.

To make the encapsulating structures more robust, we use a molten crystallizable oil as the shell phase of the double emulsion; thus the droplets can be cooled below the melting point of the oil to yield solid capsules. By freezing the double emulsion drops quickly to speed up their solidification, the inner droplets remain separate as long as the capsule is solid, as shown in Fig. 13c. Release of the actives from these capsules can be easily triggered by melting the shells. By carefully tuning the morphology of such capsules, it is possible to manipulate the release profile of actives encapsulated in this fashion. If the two inner compartments are far enough apart, the two actives will be released to the outer phase separately, making these capsules ideal for applications that require simultaneous release of incompatible actives. However, if the two inner compartments are close to each other, they will coalesce before release to the surroundings. Thus, these capsules can act as micro-reactors in which mixing of reactants is triggered by heating. By designing devices with more complicated geometries that enable separate injection of multiple inner phases, it is possible to encapsulate more than two incompatible actives or reactants. The ability to fabricate multi-core double emulsions and capsules creates new opportunities for encapsulation-related applications in personal care products, food and beverages as well as cosmetics.

Fig 13 (a) Schematic of a capillary microfluidic device for generating double-core double emulsions. (b) Double emulsion drops with two inner drops

encapsulating a red and a blue dye separately. The scale bar is 400 m. (c) Wax capsules with two compartments formed by freezing double emulsion drops

with a crystallizable shell phase. The scale bar denotes 400 m. Reprinted from [49]. Copyright 2010 American Chemical Society.

Making triple emulsions with glass capillary microfluidics So far we have described the formation of double emulsions by concurrent coflow and flow-focusing emulsification. To complete this picture, we now introduce another method to form multiple emulsions: sequential coflow emulsification. As we will show, this technique can be extended

(a

)

(b

)

(c)

Diblock copolymer

solven

t

Solvent

Removal

Double Emulsion Drop Vesicle

t=0 min t=8 min t=11 min t=12 min t=13 min

(a

)

(b

)

(c)

(d

)

Page 9: Bookchapter Microfluidic Techniques

9

indefinitely, to produce even higher-order emulsion droplets. The devices consist of a coflow junction followed by additional coflow junctions. They are composed of the injection tube (a cylindrical capillary with a tapered end), and the transition tube (a second cylindrical capillary with an inner diameter D2), as shown in Fig. 14a. Images of each stage of emulsification are shown in Fig. 14b and 14c. Using the exquisite control over the number and size of the internal droplets of the double emulsions, we are able to vary the number and size, respectively, of the inner droplets of the double emulsions, as shown in Fig. 14d and 14e.

Fig 14 Capillary microfluidic device and the formation of precisely controlled monodisperse double emulsions. (a) Schematic diagram of the device geometry. The outer fluid must be immiscible with the middle fluid and the middle fluid must be immiscible with the inner fluid. (b) and (c) High-speed optical micrographs of the first (b) and second (c) emulsification stages. (d) Optical micrographs of monodisperse double emulsions containing a controlled number of monodisperse single emulsions. (e) Optical micrographs of monodisperse double emulsions showing controlled increase of the diameter of the inner droplets while the number is constant. All double emulsions were made in the same device and with the same fluids. The flow rates of the inner, middle, and outer fluids in (b) and (c) are Q1 = 350, Q2 = 2000, and Q3 = 5000

L/hr, respectively. In (d), the flow rates of middle and outer fluids are fixed at

Q2 = 2000 and Q3 = 5000 L/hr, and those of the inner fluid are Q1 = 20, 55, 70,

85, 150, 200, 225, and 240 L/hr, from the left top to the right bottom. In (e), the variation ranges for the flow rates of the inner, middle, and outer fluids are

Q1 ≈ 20–600, Q2 ≈ 1600–5000, and Q3 ≈ 2000–8000 L/hr (in each case, Q3 is

larger than Q2). All scale bars are 200 m. Reproduced from [50]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.

This method for fabricating multiple emulsions has the advantage that it can be extended very easily, thus enabling us

to generate additional hierarchical levels of multiple emulsions. We illustrate this concept by fabricating monodisperse triple emulsions, which consist of water-in-oil-in-water-in-oil (W/O/W/O) drops. This preparation is accomplished by adding a second transition tube at the outlet of the first, and injecting the outermost fluid to flow coaxially around this tube to form the third level of emulsification at its outlet, as shown in Fig. 15a. The individual steps of drop formation leading to the triple emulsions are shown in Fig. 15(b–d). Although there are large deformations of the droplets during their formation as they flow through the tapered regions of the capillaries, the emulsification process remains stable. Again, both the diameter and the number of the individual drops at every level can be precisely controlled, as illustrated by the series of triple emulsions with one to seven innermost drops, and one to three middle drops in each outer drop (Fig. 15e). In all experiments, the variance of the diameters of triple emulsions is less than 1.5%. Furthermore, the technique can clearly be sequentially scaled to even higher levels of emulsification if desired; for example, quadruple emulsions could be made by adding an additional stage.

Fig 15 Generation of highly controlled monodisperse triple emulsions. (a) Schematic diagram of the extended capillary microfluidic device for generating triple emulsions. (b)–(d) High-speed optical micro- graphs displaying the first (b), second (c), and third (d) emulsification stages. The flow rates of the inner, middle (I), middle (II), and outer fluids in (b)–(d) are Q1 = 50, Q2 = 500, Q3 =

2000, and Q4 = 5000 L/hr. (e) Optical micrographs of triple emulsions that contain a controlled number of inner and middle droplets. The variation ranges for the flow rates are Q1 ≈ 5–100, Q2 ≈ 200–1000, Q3 ≈ 2000–3500, and Q4 =

5000 L/hr. (f) Schematic diagram detailing an alternate method for generating triple emulsions where the middle fluid (II) is injected from the entry side of the first square tube, leading to flow-focusing of the first middle fluid into the

Page 10: Bookchapter Microfluidic Techniques

10

transition capillary. (g) and (h) High-speed optical micrographs showing the formation of double emulsions in a one-step process in the transition capillary (g) and the subsequent formation of triple emulsions in the collection capillary (h). (i) and (j) Optical micrographs of triple emulsions that contain a different number of double emulsions. The variation ranges for the flow rates in (g)–(j) are Q1 = 50–200, Q2 = 1600–2500, Q3 = 4000–8000, and Q4 = 5000–12000

L/hr. The scale bar in all images is 200 m. Reproduced from [50]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.

The ability to precisely control the formation of multiple emulsions offers new opportunities to engineer novel materials. To illustrate this potential, we use the triple emulsion structure to fabricate monodisperse microcapsules made with a thermo-sensitive, poly(N-isopropylacrylamide) hydrogel that can be used for controlled release of active substances [50]. Since each fluid layer is sandwiched between two immiscible fluids, we are able to perform a polymerization reaction in a specific layer. In this case, we form a W/O/W/O triple emulsion and add monomer, crosslinker, and initiator in the outer aqueous layer. We also add an accelerator to the inner oil phase, where it diffuses into the outer aqueous shell and speeds polymerization of the hydrogel. The result is a microcapsule consisting of a shell of thermosensitive hydrogel that encapsulates an oil drop containing several water drops, all in a continuous oil phase, as shown in Fig. 16a. Upon heating from 25 °C to 50 °C, the thermosensitive hydrogel rapidly shrinks by expelling water; however, because of the incompressibility of the inner oil, the hydrogel shell breaks, providing spontaneous, pulsed release of the innermost water droplets into the continuous oil phase, as shown in Fig. 16b-e. This structure has a Trojan-horse like behavior, protecting the inner most water droplets in the hydrogel shell until their temperature-induced release. This experiment demonstrates the utility of our technique to generate highly controlled capsules with multiple internal volumes that remain separate from each other; it also highlights the potential of this microfluidic device to create highly engineered structures for controlled release of active substances. Further refinements could adjust the thickness of the layers and the number of droplets, thus enabling fine control over diffusion of compounds contained within the innermost droplets, which would facilitate their highly controlled release.

Fig 16 Temperature-sensitive hydrogel microcapsule for pulsed release. (a) Optical micrograph of a microcapsule with a shell comprised of a thermosensitive hydrogel containing aqueous droplets dispersed in oil. Upon increase of the temperature, the hydrogel shell shrinks by expelling water. This capsule was generated from a triple emulsion, where the continuous liquid is oil, the hydrogel shell is aqueous, the inner middle fluid is also oil, and the innermost droplets are aqueous. (b)–(e) Optical micrograph time series showing the forced expulsion of the oil and water droplets contained within the micro-capsule when the temperature is rapidly increased from 25 to 50°C. The time series begins once the temperature reaches 50°C. The extra layer surrounding the microcapsule in (b)–(e) is water that is squeezed out from the hydrogel shell as it shrinks. The coalescence of the expelled inner oil with the outer oil cannot be resolved, since both liquids have the same index of

refraction. The scale bar is 200 m. Reproduced from [50]. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA.

SECTION 2: PDMS microfluidics Introduction

The microfluidic devices described to this point are built by aligning and gluing glass microcapillary tubes together. Although quite simple, this process enables the creation of devices with superb flow properties, particular for forming double emulsions. Nevertheless, a drawback to these devices is that the manual fabrication makes it difficult to construct more than a few devices at a time. Moreover, the nature of the fabrication, the shaping of the capillary tips by heating and stretching them, is limited in precision and reproducibility. The process therefore remains challenging for producing large number of identical devices.

In this section, we describe an alternative process that sacrifices the idyllic flow conditions in microcapillaries to achieve greater control and reproducibility. This process utilizes photolithography in poly(dimethylsiloxane) (PDMS) [51] and is thus inherently parallel and precise, allowing fabrication of devices down to a micron in width. This allows very small channels to be made, for the synthesis of small particles. It also affords greater flexibility when designing devices, to tailor the channels to overcome specific challenges. As we show, this allows creation of new kinds of structures. Fabricating PDMS devices using photolithography

The general concept in photolithography is to fabricate three-dimensional microfluidic devices from drawings of the channels. This works by first printing a to-scale picture of the device on transparency plastic, and then transferring this image to a substrate using the photolithographic process. For the process we describe, the device drawings, called photomasks, are all black except for the regions that are to become the channels; these regions are left transparent, to allow light to pass through. To create a device from the photomask, a silicon wafer, usually a few inches in diameter, is coated with photoresist, normally SU8. This is achieved by pouring a glob of the resist on the wafer and spinning it at a high speed. Depending on the viscosity of the photoresist and the spin rate and duration, the thickness of the coating can be controlled to micron precision. This is important because this thickness determines the height of the final microfluidic channels. The coated wafer is then heated to evaporate solvent, and cooled to solidify the coating. At this point, the coating is solid but fragile and can easily be removed with a solvent. The photomask is placed on top of the coated wafer and the two are exposed to ultra-violet (UV) light. The light passes through the transparent regions of the photomask but is blocked by the black parts. The photons that penetrate create radicals in the photoresist that crosslink these regions, hardening them. The wafer is then heated for a few minutes and placed in a solvent wash to removes the unexposed,

Page 11: Bookchapter Microfluidic Techniques

11

uncrosslinked photoresist. This leaves behind the crosslinked parts, which are an inverse of the photomask. This creates “positive” (i.e. protruding) channels where the light exposed the resist.

To mold a microfluidic device from this “master,” another polymer, poly(dimethylsiloxane) (PDMS), is poured over it. The PDMS hardens when baked in an oven, becoming a clear, rubbery material. The edges of the solid PDMS are sliced with a scalpel and the slab is peeled from the master. The resulting block is imprinted with negative features of the SU8 channels, creating the PDMS microchannels. Holes are punched in the block where it is to be interfaced with tubing and the imprinted side is bonded to a flat substrate – either another block of PDMS or a glass slide – enclosing and sealing the channels. The bonding is achieved by oxidizing the surfaces of the blocks using oxygen plasma etching. Silanol groups generated by the plasma undergo condensation reactions with silanols on the other block, forming covalent Si–O–Si linkages that create an irreversible seal. At this point fluids can be pumped through the channels to use the device; however, often additional processing must occur to control the surface properties of the channels. For example, to create emulsions, the wettability of the device must be controlled using a chemical treatment. PDMS is hydrophobic by default, enabling the production of water-in-oil emulsions with move solvents. The devices can also be treated with Aquapel, a commercial glass treatment that fluorinates the surfaces, to make them even more hydrophobic, for the emulsification of other kinds of solvents. Other coatings can also be deposited, including sol-gel and polymer coatings, to make the devices more chemically resilient. As will be shown in later sections, this ability to functionalized PDMS channels is essential when forming structures with the devices, especially those templated from multiple emulsions. Single emulsion drop formation

The first drop formation geometry developed for PDMS microfluidics was a so-called “T-junction” drop maker [52]. In this device, two channels intersect to form a T shape; to make drops, the dispersed phase is injected from a side channel and the continuous phase from a vertical channel; drops form where these two channels intersect, as shown in Fig. 17a. Interestingly, in these devices the physics by which the drops form changes as a function of flow rate. At low flow rates viscous forces are small compared to surface tension (Ca < 0.01), so that drops form as a consequence of plugging and squeezing [52]: the inflating tip of the dispersed phase blocks the downstream nozzle, constricting the path of the continuous phase, which continues to be pumped in; this causes a pressure rise in the continuous phase that, in turn, squeezes on the dispersed phase, eventually pinching off a drop. At high flow rates when viscous forces become comparable to surface tension (Ca > 0.01), shearing forces start to play a role. As the dispersed phase bulges into the nozzle, it is sheared by the drag of the continuous phase. When this shear force is equal to the tensile force adhering the drop to the inlet, the interface is

stretched, shearing off a drop. In fact, and in contrast to the unbounded flows of capillary co-flow drop makers, even at high Ca plugging effects still likely play a role in these devices, due to the highly confined nature of the flows. Whichever drop formation mechanism used, monodisperse drops can be formed with a size that can be controlled by adjusting flow rates.

Fig 17 Schematics of two most common drop formation geometries for PDMS microfluidic devices. (a) In a T-junction geometry the dispersed phase is injected from one channel (left) and the continuous phase from another channel (top). The drops form in the nozzle channel, where these two fluids meet. (b) In a flow-focus geometry the dispersed phase is injected from a central channel (top) and the continuous phase from the two side channels (left and right). In both cases, monodisperse drops can be formed with a size that can be adjusted by changing flow rates.

Another way to produce drops is to use a so-called

“flow-focus” geometry [53]. This geometry has many variants, but they all share the same basic structure, in which two channels intersect to form a four-way cross, as shown in Fig. 17b. In this geometry, the dispersed phase is injected into the central inlet and the continuous phase into the two side inlets, as shown in Fig. 17b. These fluids meet in the nozzle where, again, drops are formed. As with T-junction drop formation the physics of the process depends on flow rates, dominated by plugging and squeezing at low flow rates and shearing at high flow rates. Again, drop size can be controlled by adjusting flow rates.

In practice, T-junction and flow-focus drop makers behave quite similarly, though there are subtle differences between them, especially with respect to the size and production rate of the drops for a given set of flow conditions. Another difference is that T-junctions tend to yield more monodisperse drops at low flow rates, due to the enhanced flow stability of having only a single continuous phase inlet [54]. Flow-focus junctions yield better emulsions at higher flow rates, since the centered position of the dispersed phase tends to enable dripping at higher flow speeds. Nevertheless, in most cases either geometry will suffice, and the choice is a matter of personal preference. Valve-based drop formation

As described in previous sections, drop size depends

on flow conditions, specifically, on the ratio of the dispersed-to-continuous phase flow rates. However, there is another parameter on which drop size depends, which is the diameter

Page 12: Bookchapter Microfluidic Techniques

12

of the drop formation orifice: the larger the orifice, generally, the larger the drops. Nevertheless, this parameter is rarely used as a means to adjust drop size, since doing so requires fabrication of different devices. There is a technique, however, that can adjust orifice size without having to fabricate additional devices, which is called valve-based flow focusing [55]. In this technique, single-layer membrane valves [56] are used to constrict the size of the drop formation nozzle, making it larger or smaller as needed, to vary the size of the drops, as shown in Fig 18a. This allows drop size to be adjusted in real time, without changing flow rates, as shown in Fig 18b-c. This is useful when using drop makers in combination with other components, like additional drop makers, injection channels, delay lines, or sorters, in which the behavior of the devices can be very sensitive to changes in flow rate.

Fig 18 Valve-based flow focusing drop formation. (a) Schematic of valve-based flow focusing device; water is injected downward into the central inlet and oil from the two side inlets. These fluids are injected into the nozzle, where drops are formed. The drop size depends on the nozzle dimensions, which can be adjusted using the single layer membrane valves; this allows the drop size to be adjusted without changing flow rates. (b) Formation of large drops with the valves un-actuated and (c) formation of smaller drops by constricting the nozzle with the valves. (d) Sequence of images for different actuation states of the

valves. The scale bars denote 100 m. Reproduced from [55]. Copyright 2009 American Institute of Physics

The dominance of wetting in PDMS devices

Whichever drop formation geometry used, the wettability properties of the channels are of critical importance in determining the types of drops formed – for example, whether oil drops in water or water drops in oil are formed. This is in contrast to glass capillary devices which, due to their circular coaxial geometry, can form drops of either type, relatively independent of channel wettability. The reason for this difference has to do with the way in which the fluids are injected into PDMS drop makers. When the dispersed phase enters such a device, it is initially in contact with the upper and lower surfaces of the channels. To be formed into a drop, it must be lifted off these surfaces and surrounded by the continuous phase. Whether this happens depends on the wettability properties of the channels. If, for example, the

channels are hydrophobic, water will be lifted off the walls by the oil, resulting in water drops. By contrast, if the channels are hydrophilic then the water clings to the walls, lifting off the oil, producing oil drops. Wetting is of such importance in these devices that even if the inlets for the oil and water are switched, drops of the same type are formed. For example, if water is injected into the oil inlet and oil into the water inlet of a hydrophobic flow-focus device, water drops will still be formed, but they will drip from the two side channels, as though the device were two opposing T-junctions. For these reasons, the wetting of PDMS devices must be controlled carefully when forming emulsions, and especially so when forming multiple emulsions.

Despite the usual dominance of wetting in these devices, there are strategies that can be used to relax these constraints. Using a combination of geometrical control and shear, it is possible to form either type of emulsion in a device with fixed wettability. These PDMS devices mimic the superior flow properties of glass capillary devices, by exploiting three-dimensional channel graduations. Essentially, they combine the best attributes of glass capillaries with the reproducibility and scalability of lithographically fabricated devices, and are the subject of the final section of this chapter. Applications of PDMS single emulsification Close packed encapsulation

There are occasions when using microfluidics in which it is necessary to encapsulate objects inside of drops. For example, when performing biological assays, cells and beads must be encapsulated in drops; when creating double or higher order multiple emulsions, drops must be encapsulated in drops. However, a unique challenge when encapsulating objects in drops is that while the drop formation is controlled and periodic, the in-flow of the objects is normally random. This circumstance makes it difficult to synchronize drop formation with the objects; generally, this is not attempted and the objects are encapsulated at random, thereby loading the objects according to Poisson statistics and leading to a large number of improperly filled drops and only a few containing single-objects.

One strategy to beat Poisson statistics is to order the objects so that they enter the drop maker at more regular intervals. This can be accomplished using inertial particle ordering [57, 58]. In this technique, the particles are flowed at high velocities down a long channel such that the Reynolds number approaches one. At these velocities, inertial effects become important, creating forces between the particles that cause them to order. However, a challenge with this technique is that the high velocities required can preclude stable drop formation, instead leading to jetting and polydisperse drops. Another challenge is that it can be difficult to order all the particles; instead normally there are stretches of ordered particles separated by empty gaps of continuous phase. In these regions, empty drops are formed, lowering overall encapsulation efficiency.

Page 13: Bookchapter Microfluidic Techniques

13

Our approach for beating Poisson statistics is to use mechanical forces to order the particles [59]. This is achieved by injecting the particles into a device at very high volume fraction. Due to the high volume fraction, the particles pack together, and since they are monodisperse, they spontaneously order into a crystalline array, since this reduces the overall compressive force on each particle by its neighbors; an image of packed, ordered particles is shown in Fig 19a. Since the ordered array moves at constant velocity, the particles enter the drop maker at regular intervals, as shown in Fig 19b and inset into Fig 19c. The particle frequency can be synchronized with the drop formation by adjusting particle velocity, to encapsulate exactly one particle per drop, as shown in Fig 20a–b. By increasing the velocity of the particles, their rate with respect to the drop rate can be increased, to encapsulate two, three or even four particles per drop. The encapsulation efficiency with this technique is far better than can be achieved with random encapsulation, as shown in Fig. 20c.

Fig 19 (a) Polyacrylamide gel particles injected at high volume fraction into the

inlet of a microfluidic drop maker. The particles are about 25 m in diameter. Because the particles are monodisperse and packed together, they naturally order into a regular lattice; this introduces them into the lower junction at a regular frequency. (b) Intensity time traces or particles moving through the boxed region shown in the image. (c) Power spectrum of the intensity time trace; there is a tall peak at the frequency of the particle flow, due to the regular flow of the particles. Reproduced from [59]. Copyright 2009 The Royal Society of Chemistry.

Close-packed particle encapsulation for the formation of core-shell microgels

The ability to encapsulate controlled numbers of microgels into drops offers a powerful means to template multi-layered microgel particles. To realize this concept, we create monodisperse hydrogel particles, and then we wrap these particles in an aqueous polymer shells using a microfluidic device. The device consists of two cross-junctions in series, as sketched in Fig. 21a [60]. In the first junction we add a semidilute, aqueous solution of crosslinkable pNIPAAm chains as shell phase. In the second junction we add oil to form bi-

layered pre-microgel drops. We then lock in these structures by crosslinking the pNIPAAm chains in the shell. The resultant particles consist of a hydrophilic polymer core nested in a hydrophilic polymer shell, both crosslinked and swollen in water, but both formed from different macromolecular precursors.

To demonstrate the utility of this technique, we use a shell phase that is tagged with a green fluorescent tracer polymer, along with red-tagged core microgel particles; this allows us to visualize the formation of core-shell structures which exhibit a well-controlled number of core particles in each shell and a well-controlled shell-thickness, as shown in Fig. 21b-c. There is also no interpenetration of the shell material into the core, as evidenced by the middle and lower row of micrographs in Fig. 21b, which show separate visualizations of the green-labeled shells and the red-labeled cores of the microgels in the upper row. To substantiate this finding, spatially resolved profiles of the fluorescence intensity across the micrographs in the left column of Fig. 21b are plotted in Fig. 21d. These profiles show that the red-tagged core material and the green-tagged shell material are well separated.

Fig 20 (a) Encapsulation of single particles using close-packed ordering; in this example, the particles are nearly the same size as the drop formation nozzle, allowing them to perturb the flows, to trigger drop formation. This allows every drop to be encapsulated with exactly one particle. (b) Encapsulation of single particles in drops with a wide nozzle, in which drop formation is not triggered. In this example, slight differences in the frequency of the drop formation and particle flow can lead to occasional empty drops. (c) Encapsulation of particles in which the particle flow is not ordered; in this example, a random number of particles is encapsulated per drop. Reproduced from [59]. Copyright 2009 The Royal Society of Chemistry.

In an implementation of this approach, we incorporate non-thermo-responsive polyacrylamide (pAAm) particles (not labeled) into thermo-responsive pNIPAAm shells (green fluorescently labeled) [60]. The behavior of these core-shell particles upon increase of the temperature to 35 °C is visualized in Fig. 22a and detailed in Fig. 22b: while the shell collapses due to the volume phase transition of pNIPAAm, the core remains unaffected. Due to this selective sensitivity, these particles are applicable for encapsulation and controlled release purposes: when the pNIPAAm shell is swollen, it is porous and permeable,

Page 14: Bookchapter Microfluidic Techniques

14

whereas it becomes non-porous and impermeable when it collapses. By contrast, the pAAm core remains unaffected by temperature, providing stability of shape. Thus, when the shells are swollen, the particles can be loaded with hydrophilic low molecular weight or mesoscopic additives. Upon increase of the temperature, the thermo-responsive shell collapses and encapsulates these actives in the pAAm core. Then, all

surrounding feed material can be removed and the loaded particles can be stored at elevated temperatures. However, as soon as the temperature is decreased, the actives are rapidly released. This application is substantiated in Fig. 22c, which shows a sequence of images from an experiment where RITC-tagged dextran (M = 10,000 g mol

–1) is released from pAAm–

pNIPAAm microgels.

Fig 21 Microfluidic fabrication of microgel capsules that consist of two miscible yet distinct layers. (a) Schematic of a microfluidic device forming aqueous pNIPAAm droplets that are loaded with a well-defined number of pre-fabricated microgel particles of a similar material, pNIPAAm or polyacrylamide. Subsequent droplet gelation leads to microgels with a distinct core-shell architecture. (b, c) The flow rates of the inner particle phase (red-tagged pNIPAAm), the middle polymer phase (green-tagged pNIPAAm), and the outer oil phase control the number of core particles in each shell (b) as well as the shell thickness (c). Pictures in the upper row of Panel B show an overlay of the micrographs in the middle and lower row, which depict separate visualizations of the green-tagged pNIPAAm shell and the red-tagged pNIPAAm core. (d) Spatially resolved intensity profiles of the red and green fluorescence in the single-core particle shown in Panel B, evidencing only very little

interpenetration of its two phases. The scale bar denotes 100 m and applies to all micrographs in Panel B and C. Reprinted from [60]. Copyright 2010 American Chemical Society.

Double emulsification with PDMS devices

As with glass capillary devices, double emulsions can be formed in PDMS devices, though the devices look quite different and, in fact, operate on different principles. A double emulsion consists of at least one liquid droplet encapsulated in another droplet of an immiscible phase. To create such a structure with a microfluidic device requires emulsification of the inner phase followed by emulsification of the outer phase. In PDMS devices, this can be achieved by using two drop makers in series [61–63]. The first drop maker produces the inner drops, which are fed into the inlet of the second drop maker, which produces the outer drops. This type of double emulsification, dubbed “two-step” formation, can be achieved using either cascading T-junctions or flow-focus drop makers. Generally, serial flow-focus devices are used, because the symmetric injection of the middle and continuous phases helps prevent wetting of the drops on the channels, making the formation more robust. In addition to geometrical considerations like this, it is also necessary to control the wetting properties of the devices; however, in contrast to single emulsion formation in which only uniform wetting is required, in double and higher order multiple emulsion formation spatially patterned wettability is required.

Spatially-controlling wettability in PDMS devices

In PDMS devices channel wettability determines the

type of emulsion formed: oil-in-water emulsions are formed in hydrophilic channels, while the inverse, water-in-oil emulsions, are formed in hydrophobic channels. This fact, coupled with the fact double emulsions consist of drops of at least two immiscible phases, means forming them in PDMS devices requires channels with at least two distinct wettabilities. For example, to form an oil-in-water-in-oil double emulsion, the oil drops must be formed in a hydrophilic part of the device and the water drops in a hydrophobic part.

The need to spatially control the wettability in PDMS devices has stimulated development techniques to spatially modify device surface properties. One technique is to use a photo-patterning approach, in which the devices are filled with a solution that only reacts with the channels under exposure to intense UV light [63]. Since the UV beam can be shaped using holes, slits, and lenses, this approach allows for the creation of complex wettability patterns. However, in the case of double emulsion devices only a very simple wettability pattern is needed, and the added complication of having to align the photo-pattern with the microchannels can make this approach unattractive in these instances. For the simple patterns needed for these devices, a different approach is available that trades versatility in the pattern shape for simplicity in the treatment process. This technique, which is called flow patterning [64],

0

1

-120 -60 0 60 120

No

rm. In

ten

sity

r ( m)

To

UV

Exp

osu

re

Oil

pNIPAAm

Microgel

Particles

A B

C

D

Page 15: Bookchapter Microfluidic Techniques

15

patterns wettability by introducing a reactive solution into select portions of a device and preventing it from going using the flow of an inert blocker solution, as depicted in Fig 23. The reactive solution changes the wettability only in the treated

regions, while the untreated areas retain their default wettability. By controlling how the solutions are injected, wettability patterns for W/O/W or O/W/O double emulsions can be created, as shown in Fig 1a–b and 1c–d, respectively.

Fig 22 Thermo-responsive behavior and controlled release application of pAAm–pNIPAAm core-shell microgels. (a) Fluorescence images (left column) and bright field

micrographs (right column) of microgels consisting of a 60 m untagged pAAm core encapsulated in a green-tagged pNIPAAm shell. At ambient temperatures (upper row), the shell is swollen, whereas it collapses at elevated temperatures (lower row). By contrast, the core dimension remains unaffected by the same changes of temperature. (b) Detailed plot of the particle-diameter, d, as a function of temperature, T. Dark blue circles represent the diameter of the entire particle, i.e., pAAm core plus pNIPAAm shell, whereas light blue squares represent only the pAAm core. The dotted lines are guides to the eye. (c) Controlled release of RITC-dextran (M = 10,000 g mol–1) from the particles in Panel A. In the course of the first ten seconds of this experiment, the temperature remains above 33°C, and the particles remain sealed (left three pictures). As the temperature decreases, a spontaneous release of the active incorporated in the particles is triggered by the swelling of the

pNIPAAm shells. All scalebars denote 100 m. Reprinted from [60]. Copyright 2010 American Chemical Society.

Two-step versus one-step double emulsification

Double emulsions can be formed in PDMS devices in two different processes that depend on flow conditions. In the first process, inner drops form in the first drop maker and are encapsulated in the outer drops in the second drop maker. Because the inner and outer drops are formed in two locations in two subsequent steps, this process is dubbed “two step” formation, and is illustrated in Fig 24a. In the second process, no drops are formed in the first junction; instead, the flow rates are set such that a jet of the inner phase is surrounded by the middle phase. These two phases are then surrounded by the continuous phase, yielding a coaxial nested jet in the second junction, as shown in Fig 24b. There, the double jet becomes unstable due to the widening channel, pinching into double emulsions in a single step, as shown to the right in Fig 24b and 25a. Because all jets pinch at the same location and time, this process is dubbed, “one step” formation.

These two processes have different limitations and advantages. Whereas two-step formation can produce multi-core double emulsions containing several inner drops, one-step formation can only produce single-core double emulsions. However, whereas two-step formation is limited to flow rates that enable dripping in the first junction, one-step formation can exceed these flow rates, operating far into the jetting regime. This allows creation of double emulsions with a greater range of shell thicknesses, as shown in Fig 25b. A specific advantage of this is that it allows formation of double

emulsions with exceedingly thin shells. These are ideal for encapsulation because they require minimal volume of shell material per unit of material encapsulated.

Fig 23 Schematic of flow patterning used to spatially control the wetting properties of the microfluidic devices. To make a water-in-oil-in-water double emulsion, the first drop maker junction must be hydrophobic and the second hydrophilic, (a). To create this wettability pattern, an inert solution is injected into the inner phase inlet and a reactive solution into the outlet of the device. The inert solution blocks the reactive solution, confining it to the lower portion

30 C

35 C

A

0

60

120

25 27 29 31 33 35

d(

m)

T ( C)

B

Shell

Core

0

C

t (s)10 205 15

Page 16: Bookchapter Microfluidic Techniques

16

of the device (b). To create oil-in-water-in-oil double emulsion the inverse pattern is needed, (c), which is achieved by switching the inlets into which the reactive and inert solutions, (d). Reproduced from [64]. Copyright 2010 The Royal Society of Chemistry.

Fig 24 Formation of double emulsions using a two-step process, in which the inner drops are formed in the first junction and then encapsulated in the outer drops in the second junction, (a). By increasing the flow rate of the inner phase to cause it to jet, a coaxial jet of the inner inside the middle phase is created in the second junction, which pinches into double emulsions in a one-step process, (b). Reproduced from [65]. Copyright 2010 The Royal Society of Chemistry.

Fig 25 (a) Pinch off location of the inner and outer drops for double emulsions, as a function of the inner-to-middle phase flow rate ratio. At low ratios the inner phase is in the dripping regime, causing drops to form in the first junction as well as in the second junction. As the flow rate ratio is increased, the inner phase jets into the second junction; in this case the inner and middle phases pinch off at nearly the same place. (b) Shell thickness as a function of flow rate ratio. As the flow rate ratio is increased the shell thickness decreases, because a larger fraction of the double emulsion consists of the inner droplet. Reproduced from [65]. Copyright 2010 The Royal Society of Chemistry.

Applications of PDMS double emulsification Templating Janus particles from double emulsions

As described in the section on capillary microfluidics, Janus particles are biphasic structures that have two halves, often with distinct chemical properties. For example, a Janus particle can be composed of an organophilic material on one side and a hydrophilic material on the other. Due to the mixed, spatially configured chemical properties of these particles, they have many uses in pharmaceutics and in industrial processes, as carriers for drugs and other actives, and stabilizers for emulsions. Janus particles can be formed with controlled properties using microfluidic single emulsification. By exploiting the inability of flows to mix quickly at these scales, drops can be formed of two distinct fluids localized in the two halves. The drops can then be solidified, to lock in this biphasic structure. However, a limitation of single emulsion formation is that, because it is necessary to first form drops, only fluid combinations that can both be emulsified by the same continuous phase can be used. This, in general, requires miscible fluids that, because of their miscibility, tend to have similar chemical composition; it is not possible to create single emulsions from immiscible fluids, preventing synthesis of Janus particles with very dissimilar halves.

Fig 26 Creation of organophilic-hydrophilic Janus particles using double emulsions as templates. (a) Oil-in-water-in-oil double emulsions are first formed from two monomer solutions using a double flow-focus junction device, (b). The drops are solidified while in the device to lock in the Janus structure, by solidifying the monomer solutions using photo-polymerization. Reproduced from [66]. Copyright 2009 American Chemical Society.

In this section, we describe a microfluidic approach to

circumvent this constraint, enabling synthesis of Janus particles with two very dissimilar halves [66]. Rather than using a single emulsion strategy, we adopt a double emulsion strategy. We encapsulate an organophilic (oily) drop in a hydrophilic (aqueous) drop, as shown in Fig 26a. Both drops contain polymerizable monomers, allowing them to be solidified upon UV irradiation. By flowing these double emulsions through a confining channel, we induce circulation in their interior that

Page 17: Bookchapter Microfluidic Techniques

17

causes the inner drop to migrate towards the back interface, producing double emulsion droplets with two distinct halves. This biphasic structure is maintained provided the double emulsion flows through the constricting channel. To lock in this structure, we solidify both drops by exposing them to an intense UV beam, as illustrated in Fig 26b. This creates the desired Janus particles in which the two halves have very dissimilar chemical properties.

Fig 27 Electron micrograph (upper right) of Janus particles formed from double emulsion templates. To visualize the internal Janus structure of the particles, we image them with a confocal microscope at several different focal depths. The resulting slices are reconstructed into a three-dimensional image of the Janus particle, shown for different rotation angles in the smaller images. Reproduced from [66]. Copyright 2009 American Chemical Society.

To confirm the Janus structure of the particles, we

image them with a confocal fluorescence microscope, providing a high-resolution image of the particle interior. To enable fluorescence imaging, we infuse the hydrophilic halves of the particles with fluorescein dye, which freely diffuses into these parts because of the high solubility; the dye does not go into the organophilic half, because of its low solubility in oil. This dyes only the hydrophilic portions of the particles, causing them to appear bright in the confocal images while the organophilic parts appear dark; this provides a high contrast image of the internal Janus structure, as shown in the small images in Fig. 27. Because of the small size of the particles, it is difficult to spatially resolve the details of the Janus structure. To see this structure with greater clarity, we image a Janus particle with a scanning electron microscope. The resulting images show a peanut-shaped particle with two distinct halves, as shown in Fig. 27.

Templating magnetic hydrogel particles with anisotropic structure

Magnetic microparticles are useful for contrast imaging in the body, because their interaction with magnetic fields makes it possible to detect them through tissue. They are

also useful for targeted drug delivery applications, because their ability to be manipulated by magnetic fields allows them to be guided to target sites in the body. However, to be useful, in addition to being magnetically responsive, the particles must be biocompatible, and have the ability to take up drugs and other active reagents. In this section, we describe the synthesis of such biocompatible magnetic microparticles using PDMS double emulsion templating [67]. We encapsulate a small magnetic bead in a biocompatible hydrogel shell. Because the bead is surrounded by the shell, the toxicity of the iron oxide nanoparticles that make it magnetic is reduced. Moreover, the hydrogel material can be chosen to maximize the absorption of drug compounds. To make these structures, we create a double emulsion consisting of a ferrofluidic/organopholic monomer for the inner droplet surrounded by an aqueous monomer outer droplet. Due to the immiscibility of these ingredients, the magnetic bead is entirely encapsulated within the shell of the hydrogel, as shown in Fig. 28.

Fig 28 Schematic of a microfluidic double emulsion device used to form magnetic hydrogels with anisotropic structure. The inner droplet of the double emulsion is composed of an organophilic monomer/ferrofluid mixture, while the outer droplet is composed of a hydrophilic monomer. The double emulsions are solidified by polymerizing the monomer solution, producing hydrogels with magnetic beads inside. The number of beads can be controlled by adjusting flow rates, to encapsulate one or several cores per double emulsion. Reproduced from [67]. Copyright 2009 Wiley-VCH Verlag GmbH & Co. KGaA.

Fig 29 (a) Velocity field of fluid stirred up by rotating a magnetic hydrogel using a magnetic field. The velocity field is averaged over a full rotation cycle. Due to the anisotropic structure of the particle, the rotation is eccentric. (b) Fluid speed averaged over all rotation angles, as a function of the distance from the center of the rotation. The flow disturbance peaks at a distance equal to about one particle diameter. Reproduced from [67]. Copyright 2009 Wiley-VCH Verlag GmbH & Co. KGaA.

The resulting particles can be infused with compounds

and manipulated with a magnetic field. To demonstrate the magnetic responsiveness, we place a particle in an aqueous

fluid containing 2 m polystyrene beads. By rotating the

Page 18: Bookchapter Microfluidic Techniques

18

magnetic field, we rotate the bead, as shown in Fig. 29. The rotation of the particle disturbs the fluid around it, the motion of which we record by tracking the polystyrene tracer particles. The disturbance is small at the center of the rotation and grows to a maximum at a fixed distance radially outward, as shown by the velocity field summed over a full rotation cycle in Fig. 29a. To quantify the size of the disturbance, we compute the velocity as a function of the distance away from the center, averaged over all angles, Fig. 29b. The disturbance peaks at a distance equal to the particle diameter and falls off steadily for greater distances, as shown in Fig. 29b. This demonstrates that the stirring of the fluid is localized close to the particle, as expected due to the laminar flow properties at this scale.

Fig 30 Formation of copolymer-stabilized W/O/W double emulsions using a conventional microfluidic device with two cross junctions for injecting premixed organic solvents (left), and using a novel device design allowing separate injection of organic solvents (right). The microfluidic devices are sol-gel-coated to increase the resistance of the channel walls against organic solvents. The sol-gel coating in the upper half of the device is hydrophobic, while the coating in the lower part is rendered hydrophilic due to functionalization by grafted poly(acrylic acid). (a, b) Formation of PEG-b-PLA stabilized W/O/W double emulsions using premixed mixtures of toluene and chloroform. (a) Most of the diblock copolymer forms a precipitate after the more volatile chloroform starts to evaporate in the microfluidic device. The precipitates adhere to the surface of the channels, building up a thick layer. (b) Some of the precipitates are observed in the shell phase of the double-emulsion drops formed. Since the organic solvent phase is depleted of the copolymer before the formation of double emulsions occurs, the two interfaces of the shell inside the double emulsions are not sufficiently stabilized. Thus the double-emulsion droplets burst downstream. (c) Novel microfluidic device forming PEG-b-PLA stabilized W/O/W double emulsions. To maintain the stability of the polymersomes, the osmolarity of the inner and outer phase of the double emulsion templates is balanced by adding glucose to the inner phase and polyvinyl alcohol (PVA) to the outer phase. The non-Newtonian nature of the PVA solution causes the middle phase to develop a tail, which initially connects the double emulsions. However, the jet breaks up into stable double emulsion droplets approximately 1 mm downstream in the outlet channel. Scale bar for all panels denotes 100 µm. Reprinted from [68]. Copyright 2010 Wiley-VCH Verlag GmbH & Co. KGaA.

Templating polymersomes from double emulsions

One of the principal advantages of PDMS-based microfluidics is that the lithographic fabrication allows construction of sophisticated channel networks and channels with very fine details. This can be exploited to tailor the devices

to overcome specific challenges, an example of which is the subject of this section. As described in a previous section, glass capillary devices can form monodisperse polymersomes: vesicles composed of a diblock copolymer membrane that encapsulates an active material. Due to their extreme mechanical stability and ability to be ruptured by osmotic shock or changes in temperature, they are attractive for applications that require efficient encapsulation and controlled release. including of drugs, agricultural products, and cosmetics. However, glass capillary devices cannot form them in large quantities, because to do so would require large numbers of identical devices – a feat that is not possible with the existing fabrication techniques.

Fig 31 Serial drop maker arrays used to form multiple emulsions. A single emulsion droplet takes one drop maker to create (a), whereas double, triple, quadruple, and quintuple emulsions take two, three, four, and five drop makers in series to create, (b), (c), (d), and (e), respectively. The wettability of the multiple emulsion devices has been patterned so that it inverts between hydrophilic and hydrophobic from one junction to the next. Reprinted from [69]. Copyright 2009 Verlag GmbH & Co. KGaA.

Lithographically fabricated devices, by contrast, like

PDMS devices, are much easier to fabricate with identical properties and in large numbers. However, due to the amphiphilic properties of the diblock copolymers and their interaction with organic solvents needed to emulsify them, forming double emulsions with these solutions is extremely difficult, especially in PDMS devices. The challenge is that for the double emulsions to dewet to birth the polymersomes, the diblock-containing solvent must be close to the point of precipitation for the diblock copolymers. When such a solution is injected into a PDMS device the diblocks immediately precipitate on the channel walls, fouling them, changing their wetting, and preventing the formation of double emulsions, as shown in a failing device to the left in Fig. 30. To overcome this problem, we exploit the customizability of PDMS devices [68]. We separate the inlets into which the good solvent saturated with diblock copolymers and the unsaturated poor solvent are injected. If these two solvents were pre-mixed before reaching the drop-making junctions, the diblocks would immediately precipitate. However, by injecting them into different inlets, they do not mix until after encapsulation in a double emulsion, as shown in Fig. 30, right half.

The geometry also allows us to

control the flow rate of each solvent independently, and hence, to tune the ratio of the two solvents. This allows us to obtain

Page 19: Bookchapter Microfluidic Techniques

19

solvent ratios that normally could not be used in polymersome synthesis because of precipitation, an ability that is important for forming vesicles form other copolymers.

Fig 32 Optical microscope image of quintuple emulsions formed with serial drop formation device. The outer diameters of the quintuple emulsions are

120 m. Reprinted from [69]. Copyright 2009 Verlag GmbH & Co. KGaA.

Forming high-order multiple emulsions serial drop maker arrays

One of the clearest advantages of PDMS-based

microfluidics is the ability to customize the devices, to construct sophisticated channel networks. In this section, we describe an application that exploits this ability, scaling up device complexity to form high-order multiple emulsions.

A high-order multiple emulsion is the logical continuation of a double emulsion to larger numbers of nested drops: A triple emulsion is a double emulsion encapsulated within a drop, while a quadruple emulsion is merely a triple emulsion inside yet another drop. To create these structures requires a microfluidic device having multiple drop makers aligned in series [69]. To illustrate this, we form emulsions of different orders with serial drop formation devices. We begin by forming single emulsions using a single drop maker, shown in Fig. 31a. To form double emulsions, we add another drop maker; we also pattern wettability, so that the inner drops are encapsulated within larger drops of an immiscible phase, as shown in Fig. 31b. To make triple or quadruple emulsions, we simply add third and fourth junctions, as shown in Fig. 31c–d, respectively. To form quintuple emulsions, we add a fifth junction, encapsulating quadruple emulsions in yet another drop, as shown in Fig. 31e. Because the drops in each stage are monodisperse and the formation is synchronized, the final quintuple emulsions are identical, as shown in Fig. 32. Relaxing the constraints of wetting: double emulsification through geometrical control In contrast to PDMS devices in which wettability properties are crucial when forming emulsions, glass capillary microfluidic devices can form emulsions of almost any

immiscible fluids, independent of how they wet the device surfaces. This difference stems from the distinctive geometrical properties of these devices: Whereas in PDMS devices all fluids are initially in contact with the channels and must be lifted off the surface by preferential wetting, in capillary devices the axially symmetric injection into the orifice imposes a topology on the phases, such that the middle phase never contacts the channel surfaces. This allows formation of emulsions of either type, independent of channel wetting. The goal of this section is to describe strategies to relax the constraints of wetting in PDMS devices [70] by mimicking the three-dimensional sheath flow of capillaries.

Fig 33 The formation of a double emulsion (water-in-oil-in-water (W/O/W)) in PDMS devices with one and two thicknesses (a), (b) Single layered PDMS device that cannot form double emulsions, because the middle oil phase wets the hydrophobic walls better than the continuous phase, making oil drops energetically favorable. (c) Two-thickness device with the same hydrophobic wetting properties that can easily form double emulsions, because the three-dimensional channel graduations reduce the importance of wetting. Water-in-oil drops form at the first junction and oil-in-water form at the geometrically controlled junction. (d), (e) Top and Side views of a similar device with two thicknesses, demonstrating how the continuous aqueous phase shields the middle phase from wetting the channels. Reproduced from [70]. Copyright Assaf Rotem.

We describe three realizations of geometrically

controlled double emulsification in PDMS devices. In the first, we use a two-step, multi-thickness device. For each junction, the inlet channel leading into the nozzle is smaller in cross-sectional size than the nozzle, and centered vertically and horizontally with respect to the nozzle. This is the best geometry for forming double emulsions of any composition without having to control wetting, but the multiple height steps needed make fabrication somewhat labor intensive. A simpler version of the device that works nearly as well but is easier to fabricate is to use a two-step, two-thickness approach. In this approach, the wetting of the device is of a uniform composition such that one type of emulsion is always preferred; for example, the device could be uniformly hydrophobic so that it easily forms water-in-oil emulsions. This is used in the first junction to form the inner drops and then a step is used in the second junction to form the outer drops, as shown in Fig 33. A third implementation is to use a one-step, two thickness device. For example, all fluids can be injected into a single junction, where there is a single height step. This produces a coaxial jet of the inner and middle phases, which is sheared

Page 20: Bookchapter Microfluidic Techniques

20

into double emulsions by a sheath flow from the continuous phase. This produces double emulsion with very thin shells,

with volume fractions of 1:25 shell/inner phase, as shown in Fig. 34.

Fig 34 Microfluidic devices with two different hights for the formation of double emulsions in one step. W/O/W double emulsions are formed with different volume fractions, from 1:1 (inner:shell) volume fraction in the left image to 25:1 (inner:shell) volume fraction on the right. Reproduced from [70]. Copyright Assaf Rotem.

Fig 35 Fabrication process for two thickness emulsification devices. (a) A two-height master is prepared using photolithography. (b) and (c) Two replicates of the PDMS device are cured and then bonded facing each other using plasma bonding. (d) The protruding “key” on one PDMS block slides into the sunken “lock” on the other half, mechanically aligning the two blocks so that the channel features are aligned. To lubricate the blocks and allow the key to explore the lock, a droplet of water is placed on the freshly plasma oxidized surfaces. The water retains the silanol bonds critical for covalent bonding of the blocks, which occurs after the water diffuses away into the blocks while in an oven set to 65°C. (e) An image of a sample device including a channel and a lock and key. Reproduced from [70]. Copyright Assaf Rotem.

To fabricate these multi-height devices, we use multi-layer

lithography to create multi-height channels, which are then aligned and bonded to other multi-height channels in a second block of PDMS. This allows construction of devices in which the channel steps are symmetric in the vertical direction, stepping both up and down where the channels widen. To make the multi-layer channels, we use standard fabrication techniques in which a silicon wafer is coated with photoresist and exposed to UV light through different photomasks in several iterations. However, a more challenging aspect of the fabrication is aligning and bonding the two PDMS channel slabs together. To do this, we use a simple approach in which matching physical locks and keys mechanically align the PDMS blocks, illustrated in Fig. 35. One of the PDMS blocks has a hole (the lock) of a particular shape and at a specific location; the other block has a protrusion (the key) of the same shape and at the same location. When the two faces of the PDMS are brought into contact, the key slides into the lock, as shown in Fig. 35 d; friction holds the blocks in place, maintaining the alignments through the bonding process. To allow the keys to explore the locks, we use water as a lubricating fluid. After aligning the blocks, the device is placed in an oven set to 65 °C for 1 hr; during this process the water evaporates, allowing the silanol groups in both blocks to react, creating an

irreversible seal. An especially useful property of using mechanical alignment is that the device remains well aligned even during the heating step, when thermal expansion of the PDMS could break the alignment. The friction of the locks and keys maintains the alignment through the heating process. Conclusions

The principal advantage of microfluidic synthesis is the ability to independently choose the chemical composition and structure of the particles formed. While the structure of the particles is determined by the flow properties in the devices, chemical composition is determined by selecting which fluids to introduce into the device. This enables synthesis of a range of particles with distinct composition and structure, including polymer spheres, Janus particles, nested multiple emulsions, and core-shell capsules. Many of these particles have uses in commercial applications, including therapeutics, medical imaging, crop protection, and cosmetics. However, a challenge in commercializing these techniques is to increase production yields: current systems typically produce only microliters-to-milliliters of product per hour. One strategy to increase yield is to parallelize the devices. Rather than a single particle synthesis device operating at a time, thousands or hundreds-of-thousands could be used in parallel. This should increase production rates to the levels needed for commercial applications. In this respect PDMS devices hold the greatest potential, because they can be replicated in large numbers with precision. However, an open challenge to achieving controlled parallel production is the development of a uniform fluid distribution system. Available methods utilizing bifurcating channel networks lack the uniformity needed, since the compliance of the devices, compressibility of the solutions, and large hydrodynamic resistance of the dispensing channels lead to long-lived pressure oscillations and uneven flow, resulting in polydisperse particles. Nevertheless, while a known and challenging problem, development of such a system is critical for these techniques to achieve their potential in particle synthesis applications, and will likely comprise the next stages of microfluidic research. Acknowledgments

We thank Jinwoong Kim, Bingjie Sun, and Christian Holtze for their help. This work was supported by the NSF (DMR-0602684), the Harvard MRSEC (DMR-0820484), and the Massachusetts Life Sciences Center. S.S. is a research fellow of the German Academy

Page 21: Bookchapter Microfluidic Techniques

21

of Sciences Leopoldina (BMBF-LPD 9901/8-186). J.T. received funding from the Fund of the Chemical Industry (Germany) and the German Academic Exchange Service. W.J.D. was supported by the Advanced Energy Consortium; member companies include BP America Inc., Baker Hughes Inc., Conoco-Phillips, Halliburton Energy Services Inc., Marathon Oil Corp., Occidental Oil and Gas, Petrobras, Schlumberger, Shell, and Total. D.L. and M.H.L. acknowledge support by the PENN MRSEC (DMR-0520020) and the University Research Foundation of the University of Pennsylvania.

References [1] H. Mao, T. Yang, P. S. Cremer, J. Am. Chem. Soc. 124, 4432–4435 (2002). [2] N. Lion, T. C. Rohner, L. Dayon, I. L. Arnaud, E. Damoc, N. Youhnovski, Z.-Y. Wu, C. Roussel, J. Josserand, H. Jensen, J. S. Rossier, M. Przybylski, H. H. Girault, Electrophoresis 24, 3533–3562 (2003). [3] J. J. Agresti, E. Antipov, A. R. Abate, K. Ahn, A. C. Rowat, J. C. Baret, M. Marquez, A. M. Klibanov, A. D. Griffiths, D. A. Weitz, Proc. Natl. Acad. Sci. U.S.A. 107, 4004–4009 (2010). [4] E. Tumarkin, E. Kumacheva, Chem. Soc. Rev. 38, 2161–2168 (2009). [5] Z. Nie, S. Xu, M. Seo, P. C. Lewis, E. Kumacheva, J. Am. Chem. Soc. 127, 8058–8063 (2005). [6] S.-Y. Teh, R. Lin, L.-H. Hung, A. P. Lee, Lab Chip 8, 198–220 (2008). [7] D. Dendukuri, P. S. Doyle, Adv. Mater. 21, 1–16 (2009). [8] F. Tiarks, K. Landfester, M. Antonietti, Langmuir 17, 908–918 (2001). [9] F. Tiarks, K. Landfester, M. Antonietti, Langmuir 17, 5775–5780 (2001). [10] J. M. Asua, Prog. Polym. Sci. 27, 1283–1346 (2002). [11] T. R. Powers, D. F. Zhang, R. E. Goldstein, H. A. Stone, Phys. Fluids 10, 1052–1057 (1998). [12] L. Rayleigh, Proc. R. Soc. Lond. 29, 71–79 (1879). [13] S. Tomotika, Proc. Roy. Soc. Lond. A 150, 322–337 (1935). [14] J. Plateau, Acad. Sci. Bruxelles Mem. 23, 5 (1849). [15] C. Cramer, P. Fischer, E. J. Windhab, Chem. Eng. Sci. 59, 3045–3056 (2004). [16] A. Utada, A. Fernandez-Nieves, H. Stone, D. A. Weitz, Phys. Rev. Lett. 99, 94502 (2007). [17] P. B. Umbanhowar, V. Prasad, D. A. Weitz, Langmuir 16, 347–351 (2000). [18] A. S. Utada, E. Lorenceau, D. R. Link, P. D. Kaplan, H. A. Stone, D. A. Weitz, Science 308, 537–541 (2005). [19] A. Gañán-Calvo, J. Gordillo, Phys. Rev. Lett. 87, 274501 (2001). [20] H. Wyss, D. Blair, J. Morris, H. Stone, D. A. Weitz, Phys. Rev. E 74, 61402 (2006). [21] L.-Y. Chu, J.-W. Kim, R. K. Shah, D. A. Weitz, Adv. Funct. Mater. 17, 3499–3504 (2007). [22] R. K. Shah, J.-W. Kim, J. J. Agresti, D. A. Weitz, L.-Y. Chu, Soft Matter 4, 2303–2309 (2008). [23] J.-W. Kim, A. S. Utada, A. Fernández-Nieves, Z. B. Hu, D. A. Weitz, Angew. Chem. Int. Ed. 46, 1819–1822 (2007). [24] S. Seiffert, W. Oppermann, K. Saalwaechter, Polymer 48, 5599–5611 (2007). [25] X. Yu, C. Corten, H. Goerner, T. Wolff, D. Kuckling, J. Photochem. Photobiol A 198, 34–44 (2008). [26] S. Seiffert, D. A. Weitz, Soft Matter 6, 3184–3190 (2010). [27] R. K. Shah, J.-W. Kim, D. A. Weitz, Langmuir 26, 1561–1565 (2010). [28] R. K. Shah, J.-W. Kim, D. A. Weitz, Adv. Mater. 21, 1949–1953 (2009). [29] E. C. Cho, J.-W. Kim, A. Fernandez-Nieves, D. A. Weitz, Nano Lett. 8, 168–172 (2008). [30] J. Ophir, K. J. Parker, Ultrasound Med. Biol. 15, 319–332 (1989). [31] D. B. Khismatullin “Gas microbubbles and their use in medicine”. In: A. A. Doinikov (Editor) “Bubble and Particle Dynamics in Acoustic Fields: Modern Trends and Applications”, Research Signpost, Kerala 2005. pp. 231–289. [32] K. Ferrara, R. Pollard, M. Borden Ann. Rev. Biomed. Eng. 9, 415–447 (2007). [33] J. A. Feshitan, C. C. Chen, J. J. Kwan, M. A. Borden J. Colloid Interf. Sci. 329, 316–324 (2009). [34] M. McCulloch, C. Gresser, S. Moos, J. Odabashian, S. Jasper, J. Bednarz, P. Burgess, D. Carney, V. Moore, E. Sisk, A. Waggoner, S. Witt, D. Adams. J. Am. Soc. Echocardiogr. 13, 959–967 (2000). [35] M. H. Lee, D. Lee, Soft Matter 6, 4326–4330 (2010).

[36] E. C. Unger, T. Porter, W. Culp, R. Labell, T. Matsunaga, R. Zutshi, Adv. Drug Deliv. Rev. 56, 1291–1314 (2004). [37] A. B. Subramaniam, M. Abkarian, L. Mahadevan, H. A. Stone, Nature 438, 930 (2005). [38] J. I. Park, Z. Nie, A. Kumachev, A. I. Abdelrahman, B. R. Binks, H. A. Stone, E. Kumacheva, Angew. Chem. Int. Ed. 48, 5300–5304 (2009). [39] D. Lee, D. A. Weitz, Adv. Mater. 20, 3498–3503 (2008). [40] M. H. Lee, V. Prasad, D. Lee, Langmuir 26, 2227–2230 (2010). [41] B. Ambravaneswaran, H. J. Subramani, S. D. Phillips, O. A. Basaran, Phys. Rev. Lett. 93, 034501 (2004) [42] C. Clanet, J. C. Lasheras, J. Fluid Mech. 383, 307–326 (1999). [43] W. M. Deen “Analysis of Transport Phenomena”, Oxford Press, Oxford 1998. [44] D. R. Link, E. Grasland-Mongrain, A. Duri, F. Sarrazin, Z. Cheng, G. Cristobal, M. Marquez, D. A. Weitz, Angew. Chem. Int. Ed. 45, 2556–2560 (2006); Angew. Chem. 2006, 118, 2618–2622. [45] H. C. Shum, J. W. Kim, D. A. Weitz, J. Am. Chem. So. 130, 9543–9549 (2008). [46] E. Lorenceau, A. S. Utada, D. R. Link, G. Cristobal, M. Joanicot, D. A. Weitz, Langmuir 21, 9183–9186 (2005). [47] R. C. Hayward, A. S. Utada, N. Dan, D. A. Weitz, Langmuir 22, 4457–4461 (2006). [48] H. C. Shum, D. Lee, I. Yoon, T. Kodger, D. A. Weitz, Langmuir 24, 7651–7653 (2008). [49] B. Sun, H. C. Shum, C. Holtze, David A. Weitz, Appl. Mat. Interf., DOI: 10.1021/am100860b (2010). [50] L.-Y. Chu, A. S. Utada, R. K. Shah, J.-W. Kim, D. A. Weitz, Angew. Chem. Int. Ed. 46, 8970–8974 (2007). [51] Y. Xia, G. M. Whitesides, Angew. Chem. Int. Ed. 37, 550–575 (1998). [52] P. Garstecki, M. J. Fuerstman, H. A. Stone, G. M. Whitesides, Lab Chip 6, 437–446 (2006). [53] T. Ward, M. Faivre, M. Abkarian, H. A. Stone, Electrophoresis 26, 3716–3724 (2005). [54] A. R. Abate, A. Poitzsch, Y. Hwang, J. Lee, J. Czerwinska, D. A. Weitz, Phys. Rev. E, 80, 026310 (2009). [55] A. R. Abate, M. B. Romanowsky, J. J. Agresti, D. A. Weitz, Appl. Phys. Lett. 94, 023503 (2009). [56] A. R. Abate,D. A. Weitz, Appl. Phys. Lett. 92, 243509 (2008). [57] D. Di Carlo, D. Irimia, R. G. Tompkins, M. Toner, Proc. Natl. Acad. Sci. U.S.A. 104, 18892–18197 (2007). [58] J. F. Edd, D. Di Carlo, K. J. Humphry, S. Köster, D. Irimia, D. A. Weitz, M. Toner, Lab Chip 8, 1262–1264 (2008). [59] A. R. Abate, C. H. Chen, J. J. Agresti, D. A. Weitz, Lab Chip 9, 2628–2631 (2009). [60] S. Seiffert, J. Thiele, A. R. Abate, D. A. Weitz, J. Am. Chem. Soc. 132, 6606–6609 (2010). [61] T. Nisisako, S. Okushima, T. Torii, Soft Matter 1, 23–27 (2005). [62] S. Okushima, T. Nisisako, T. Torii, T. Higuchi, Langmuir 20, 9905–9908 (2004). [63] A. R. Abate, A. T. Krummel, D. Lee, M. Marquez, C. Holtze, D. A. Weitz, Lab Chip 8, 2157–2160 (2008). [64] A. R. Abate, J. Thiele, M. Weinhart, D. A. Weitz, Lab Chip 10, 1774–1776 (2010). [65] Adam R. Abate, Julian Thiele, David A. Weitz, Lab Chip, DOI: 10.1039/c0lc00236d (2010). [66] C. H. Chen, R. K. Shah, A. R. Abate, D. A. Weitz, Langmuir 25 4320–4323 (2009). [67] C. H. Chen, A. R. Abate, D. Lee, E. M. Terentjev, D. A. Weitz, Adv. Mater. 21, 3201–3204 (2009). [68] J. Thiele, A. R. Abate, H. C. Shum, S. Bachtler, S. Förster, D. A. Weitz, Small 6, 1723–1727 (2010). [69] A. R. Abate, D. A. Weitz, Small 5, 2030–2032 (2009). [70] Controlling drop formation in microfluidic devices, A. Rotem, A. R. Abate, D. A. Weitz, in preparation.