Você está na página 1de 29

HIGHLIGHT

Polymer Chemistry in Flow: New Polymers, Beads, Capsules, and Fibers

JEREMY L. STEINBACHER, D. TYLER MCQUADE Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853

Received 8 June 2006; accepted 15 June 2006


DOI: 10.1002/pola.21630 Published online in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: The union between polymer science and microuidics is reviewed. Fluids in microreactors allow the synthesis of a wide range of polymeric materials with unique properties. We begin by discussing the important uid dynamics that dominate the behavior of uids on the micrometer

scale. We then progress through a comprehensive analysis of the polymeric materials synthesized to date. This highlight concludes with an overview of the methods used to make microreactors. We enthusiastically endorse microreactors as a powerful approach to making materials with controlled properties, al-

though we have tried to provide a critical eye to help the nonexpert C 2006 Wiley Periodienter the eld. V
cals, Inc. J Polym Sci Part A: Polym Chem 44: 65056533, 2006

Keywords: colloids; microencapsulation; microuidics; microparticles; particle size distribution

Jeremy L. Steinbacher is currently pursuing a doctorate in the Department of Chemistry and Chemical Biology at Cornell University (Ithaca, NY) under the tutelage of Professor D. Tyler McQuade. His research interests include the development of site-isolated catalysts and the production of materials within microuidic devices. A native of Pennsylvania, he earned his B.A. in Chemistry from Franklin & Marshall College (Lancaster, PA) in 2003.

J. L. STEINBACHER and D. T. MCQUADE

D. Tyler McQuade, Assistant Professor of Chemistry and Chemical Biology at Cornell University, began his faculty position in 2001.

Correspondence to: D. T. McQuade (E-mail: dtm25@ cornell.edu)


Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 44, 65056533 (2006)
C 2006 Wiley Periodicals, Inc. V

6505

6506

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

He is currently a Dreyfus, 3M, Rohm and Haas, Beckman, and New York State Ofce of Science, Technology, and Academic Research Young Investigator and one of the 2004 Massachusetts Institute of Technology Tech Review 100. McQuade was born in Atlanta, GA, and raised in the Santa Cruz Mountains of California. He received a B.S. in Chemistry and a B.S. in Biology from the University of CaliforniaIrvine and a Ph.D. in Chemistry from the University of WisconsinMadison under the guidance of Professor Samuel Gellman. His education was completed by a National Institutes of Health fellowship at the Massachusetts Institute of Technology with Professor Timothy Swager. The McQuade Group is focused on creating synthetic systems that use site isolation (via encapsulation) and selectivity (via recognition) to carry out multistep syntheses with greater efciency. The multidisciplinary environment of the group, which uses tools from biology, chemistry, and materials science, is yielding both polymers and small molecules that will provide the building blocks for the next generation of reagents for sustainable process chemistry. Many of these early innovations nucleated the Systanix enterprise that won Cornells Big Red Ventures 2005 Business Idea Competition.

INTRODUCTION
The continuous synthesis of materials in ow began in the early 1900s in the ber industry. Continuous processing in ow is a mainstay of oil rening and bulk chemical manufacturing. For most of the 20th century, synthesis in ow was relegated to low-margin areas in which the benets of continuous processing were most strongly felt. The view that ow is only useful for commodity products has changed over the last 20 years as two communities have recognized that ow not only provides low-end assembly-line benets but also has value for high-margin products because of the heat-transfer, mixing, and ow-alignment attributes conveyed by microreactors. Academic chemists entered the microreactor area largely through the efforts of George Whitesides and his use of poly(dimethylsiloxane) (PDMS) to create inexpensive microuidic devices and through the microreactor community led by the Institute for Molecular Manufacturing (IMM) in Germany and Yoshidas microreactor initiatives in Japan. The area of microreactors and their impact on chemical and materials synthesis are well reviewed in a wide variety of contexts.1,2 In this highlight, we attempt to examine microreactors through the lens of a polymer scientist describing, by use of examples from the literature, how microreactors can add value to a typical polymer scientists laboratory.

UNIQUE PROPERTIES OF MICROREACTORS


Mixing Stirring in classical reactors, such as round-bottom asks and larger batch reactors, is limited by inhomogeneities in the ow elds created by the stirring bar or impeller, respectively. As a uid approaches the stirrer, convection is induced, resulting in turbulence and chaotic mixing. The shear forces that cause the convection are signicantly dampened away from the stirrer, and the majority of the round-bottom ask or reactor experiences little or no mixing. The idle portions of a reaction environment allow time for unproductive chemistry to take place because of local concentration gradients. If unproductive chemistry is not an issue, longer reaction times may be used to drive reactions to completion, decreasing efciency. Rapid mixing results in faster reaction times and thus cleaner chemistry, although rapid mixing is difcult to achieve in a classic reactor for the reasons outlined previously.1 On the other hand, continuously owing microreactors allow rapid and homogeneous mixing because of their small dimensions. Microreactors can achieve complete mixing in microseconds, whereas classical reactors mix on the timescale of seconds.3 Microreactors achieve this rapid mixing using a variety of stratJournal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6507

Figure 1. (A) IMM interdigital microreactors with characteristic lamellar dimensions on the order of tens of micrometers and mixing times of several hundred milliseconds and (B) schematic of a Velocys stacked microreactor that forces submicrometer droplets of the disperse phase through a porous membrane into owing streams of the continuous phase. Reproduced with per r Mikrotechnik Mainz GmbH and from (B) Chemical Engineering mission from (A) Institut fu Progress, August 2004.

egies. Researchers at IMM in Germany and at Battelle in the United States (Fig. 1) have created devices that mix uids using a multilamellar approach in which layers of uids ranging in thickness from 50 to 200 lm are sandwiched together. The small dimensions allow rapid diffusional mixing to occur in as little as 100 ls. Such rapid mixing occurs because these lamellar systems achieve surface-to-volume ratios of up to 30,000 m2 m3, in contrast to laboratory beakers and batch reactors, which typically have surface-to-volume ratios of 100 and 4 m2 m3, respectively. As discussed later, these surface-to-volume ratios impact thermal and mass transport. Other strategies that complement the lamellar design are twisted or undulating tubes, impinging ows, or static mixers placed in ow.4,5

Thermal Management Heating and cooling a reaction is an important variable that, if left uncontrolled, can lead to either very slow reactions (needing heat) or runaway reactions that can lead to explosions (needing cooling). Also, reactions that offer two potential products from either kinetic or thermodynamic pathways are very sensitive to temperature. Batch reactors often provide broad temperature proles that can allow access to multiple pathways when only one pathway is desired. Figure 2 compares the temperature distributions in batch and microreactors to the kinetic energy needed to access a byproduct-forming pathway.2 Although the batch reactors broad temperature distribution allows the undesired side reaction, the narrow temperature distribution in

Figure 2. (left) Comparison of the ideal temperature distributions for a hypothetical reaction (black) and the actual temperature distributions in a batch reactor (blue) and a microreactor (red) and (right) schematic comparison of these temperature distributions to two product-forming pathways. The batch reactors broad temperature distribution allows the production of undesired product C, but the narrow temperature distribution in the microreactor restricts the reaction to target product B. Reprinted with permission from Schwalbe, T. et al., Org Process Res. Dev. 2004, 8, 440454. Copyright 2004, American Chemical Society. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

6508

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 5. Example of the use of laminar ow to pattern polymeric structures. Here, oppositely charged polyelectrolytes precipitate at the interface of two coowing streams. Reprinted with permission from Kenis, P. J. A. et al., Science 1999, 285, 8385. Copyright 1999, AAAS.

Figure 3. A demonstration of laminar ow provided by owing two dyed solutions together in a Y-shaped junction. The streams experience no convection and mix by diffusion only. Reprinted from Choban, E. R.; Markoski, L. J.; Wieckowski, A.; Kenis, P. J. A. J. Power Sources 2004, 128, 54 60, copyright 2004, with permission from Elsevier.

the microreactor restricts the reaction to the target product. Microreactors achieve such efcient input or removal of heat and nearly constant reaction temperatures because of their high surface-to-volume ratios. Rapid temperature changes6 and heat-exchange coef-

cients up to 25 kW m2 K1 are possible, depending on the materials and heat exchanger used.3 The precise thermal management available in a microreactor enables efcient, high-temperature reactions, such as the conversion of methane to methanol, ne control over the reaction selectivity, and the ability to perform otherwise explosive chemistry. Anyone familiar with Imperial Chemical Industries history of polyethylene research recognizes that exotherms resulting from polymerizations can cause explosions.7 A microreactor can enable highly exothermic polymerizations safely and with a degree of control that limits side reactions such as premature chain termination or chain transfer.

Figure 4. Cartoon illustrating the effects of the channel diameter on the laminar ow. (A) With a large channel, the sh, representing uid streamlines, can swim in any direction without feeling the walls. (B) Shrinking the channel causes the sh to swim mostly in the same direction, but the channel is still large enough to turn (or turbulently mix). (C) When the channel is shrunk further, the sh are now forced to swim in one direction without turning (laminar ow). [Color gure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 6. Simulations of a ow eld in a mixer including the velocity eld and convective diffusion of molecules from the center stream to the outer, sheath streams. Reprinted with permission from Santiago, J.G. Stanford Microuidics Laboratory website. Available at: http://microuidics.stanford.edu/ micromix.htm. Copyright 2006, Stanford Microuidics Laboratory. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6509

Figure 7. Long-strand DNA structures in a microchannel: (C) DNA molecules in ow at 10 lL min1 (the actual ow speed at the focus position is ca. 100 lm s1) and (D) a uorescence intensity distribution map of one DNA molecule from part C. In a nonowing state, these DNA molecules form a coiled structure. However, in ow, they stretch and undulate as though swimming and orient in the same direction as the ow. Bright and dark points can be seen within individual molecules. The presence of these points suggests that well-stretched and nonstretching parts exist within individual molecules. Reprinted from Analytical Biochemistry, Vol. 332, Yamashita, K., Direct Observation of Long-strand DNA, 274279 (2004), with permission from Elsevier.

Although vastly improved mixing and thermal management are very useful, these virtues of microreactors add value by making already tractable problems easier to manage. The truly revolutionary advantage that microreactor systems convey for the polymer chemist is precise control over the uid behavior and structure. Fluid Behavior Most synthetic chemists have no formal training in uid dynamics or uid behavior. Unlike our colleagues in physics and engineering, we view uids as media that we do not consider beyond polarity effects and perhaps viscosity. A uid in a round-bottom ask represents a relatively boring variable, yet as uids are forced into small volumes, their physical behavior begins to change. Fluid behavior at the micrometer scale rarely corresponds with our day-to-day intuition. This is nicely illustrated by Whitesidess Science cover8 in 1999, in which seven dyed solutions owed together in a small channel

and Keniss two dyed solutions owing together in a Y-junction9 (Fig. 3). In this particular case, the small channel width forced the uid into a laminar ow regime, and the uids did not mix convectively. Rather, the individual streams owed side by side, mixing only by diffusion at the interface. Engineers, being more practical than the average physical scientist, often reduce basic observations

Figure 8. Formation of droplets from a jet of uid by breakup due to the RayleighPlateau instability. Minimization of the surface area drives breakup when the jet experiences perturbations larger than the circumference of the jet. Reprinted with permission from Anna, S. L. et al., Applied Physics Letters, 2003, 82, 364366. Copyright 2004, American Institute of Physics. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 9. Demonstration of the effect of Ca on the ow regime. (b) At low ow rates, surface tension dominates over viscous forces, and plugs are formed from the combination of both water phases. (c) At an intermediate Ca value, the water streams take turns snapping off droplets. (d,e) At the highest Ca values, viscous forces dominate over surface tension, and the water phases begin to coow with the oil continuous phase. Reprinted with permission from Zheng, B. et al., Anal. Chem. 2004, 76, 49774982. Copyright 2004, American Chemical Society. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

6510

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

down to dimensionless parameters that are empirically derived. Many dimensionless parameters exist, but when we consider reactions in ow, only two, the Reynolds number (Re) and Capillary number (Ca), offer visceral insight. The following discussion is meant only as a primer on uid behavior and simplies certain issues for the sake of clarity. This primer may be followed up by a reading of the authoritative review by Squires and Quake10 on the subject of uids in microuidic devices. Re Turbulent mixing and chaotic mixing are the tools of the average chemist armed with a round-bottom ask. However, moving to channels with length scales on the order of nanometers to millimeters prevents turbulent and chaotic mixing altogether. This transition from turbulent ow to smooth ow occurs when viscous effects begin to dominate over inertial effects in the uid. Representing this ratio of inertial forces to viscous forces, Re is dened as follows: Re
Figure 10. Droplet patterns in rounded channels at different water and oil pressures (psi noted in the gure) and the corresponding phaselike diagram depicting the relationship between the oil and water pressure differences and droplet morphology: ( ) solid water stream, (*) elongated droplets, (~) triple droplet layer, (~) double droplet layer, (n) jointed droplets, and (u) separated droplets. Solid lines are used to dene approximate boundaries between the following droplet states (top to bottom): solid water stream, ribbon layer, pearl necklace, single droplets, and solid oil stream. Reprinted gure with permission from T. Thorsen et al., Phys. Rev. Lett., 2001, 86, 41634166. Copyright 2001 by the American Physical Society.

qvl g

where q is the density, v is the mean velocity, l is a characteristic length scale (e.g., the diameter of a tube), and g is the dynamic viscosity. The exact transition point depends on the given channel geometry, but uids with Re below *2000 ow laminarly, without turbulence. By thinking of a uid as a series of sh moving in a stream, one can get a visceral feel for uid movement on different length scales (assuming that the uid and its velocity are held constant). Figure 4 shows sh swimming in a large channel in which the sh can swim in any direction without feeling the

Figure 11. (left) Schematic of a microuidic device for creating double emulsions and (right) red and blue aqueous droplets contained in larger organic droplets. Reprinted in part with permission from Okushima, S. et al., Langmuir 2004, 20, 99059908. Copyright 2004, American Chemical Society. [Color gure can be viewed in the online issue, which is available at www. interscience.wiley.com.] Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6511

Scheme 1

walls. Shrinking the channel causes the sh to swim in the same direction, but the channel is still large enough to turn (or turbulently mix). If we shrink the channel further, the sh now swim in one direction without turning (laminar ow). The sh represent the smallest turbulent motions available to the uid based on the values of the variables in eq 1. Once the channel dimensions are smaller than this length, the turbulence is damped, and the uid moves nonturbulently with parallel streamlines. Laminar ow provides the polymer chemist with an opportunity. The interface between uids offers the opportunity to perform polymerization at the interface at which the two uids meet. Figure 5 is an early example from the Whitesides laboratory in which a polymer was deposited at the interface between two uids to create a permeable membrane in situ.8 Most microreactors are designed so that the interface of the two uids contacts the ceiling and oor of the channels, and when an interfacial polymerization occurs, the polymer adheres to the walls. This issue can be overcome with axisymmetric or coaxial systems (discussed later), which prevent one phase from touching the walls of the device. The interface between liquids owing laminarly has some unique attributes. A uid jet like the one shown in Figure 6 creates an extensional ow eld in which large macromolecules can be aligned. If two uids owing past each other are moving at different rates, a shear force develops at the interface that can align structures. For instance, Maeda et al.11 showed that DNA can be elongated and aligned at the interface formed by a uid jet (Fig. 7). Depending on the parameters of the system, a liquid jet can transform from a thin nger of uid to droplets (Fig. 8). The process that breaks the stream into droplets is called a RayleighPlateau instability,12,13 and it arises when the jet becomes unstable to perturbations longer than its circumference and breaks up into droplets to minimize the surface area. The droplets are typically monodisperse13 and can be organized downstream into interesting structures. The minimization of the interface is where Re bows out and turns over responsibility to Ca. Ca A surface tension exists at the interface between two uids. In a two-phase, owing system, the balance
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

between this surface tension and viscous forces within the uids dominates the structure of the uids. Thus, Ca is dened as follows: Ca vg c 2

where v is the mean velocity, g is the dynamic viscosity, and c is the interfacial surface tension. Ismagilov et al.15 provided one of the nicest examples demonstrating how variations in Ca impact the geometry that a uid adopts in ow. Figure 9 shows an experiment in which two channels containing dyed water enter a third channel containing a uorinated solvent carrier phase. In Figure 9(b), Ca is low, meaning that surface tension dominates the uid geometry. The aqueous ows entering the channel from the top and bottom are stiff because of the surface tension and do not bend under pressure from the third uid owing orthogonally. The ngers of uid from top and bottom are so stiff that they collide with each other before the pressure from the orthogonal uid snaps the two uids as a single plug. In Figure 9(c), the ow rate has been increased, and this makes the viscous forces more important. Now the nger from the bottom ow is no longer stiff enough to resist the pressure from the orthogonal ow and protrudes from the junction slightly before it snaps off. Figure 9(d,e) shows cases in which the surface ten-

Figure 12. Microuidic system for cation-initiated vinyl ether polymerization. The initiator and monomer mix at M1, react through microtube reactor R1 submerged in a cooling bath, and are quenched by the addition of the solvent at mixer M2. Reprinted with permission from Nagaki, A. et al., J. Am. Chem. Soc. 2004, 126, 1470214703. Copyright 2004, American Chemical Society.

6512

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Scheme 2

sion no longer dominates over viscous forces. We see that the ngers of uid are much more exible and the uid thins until smaller droplets snap off far downstream. This example shows that with small variations in the uid properties and ow rates, one can create a range of monodisperse emulsions with different sizes. These emulsions provide an opportunity to create monodisperse beads or capsules more easily than trying to optimize a batch emulsion polymerization to be monodisperse. The full impact of Ca is best viewed from the perspective of how the emulsion droplets can be organized in ow. Quake et al.16 created monodisperse droplets in which water and oil collide using a simple Tjunction shown in Figure 10. By varying the ratio of the water pressure to the oil pressure and total pressure, they were able to assemble droplets into a wide array of structures. These structures ranged from single monodisperse droplets to beautiful patterns of connected droplet strings (Fig. 10). This is a treasure trove for any polymer scientist trying to create new bers, beads, capsules, or particles of novel shapes. Quake et al.s work has been followed by a wide range of publications detailing other structures that can be formed in ow. For example, Nisisako et al.17 formed monodisperse double emulsions using a tandem T-junction arrangement (Fig. 11). These two examples are just the tip of the Ca iceberg. As we discuss instances in which new materials were produced in microreactors, we will point out how capillary forces played a role.

SYNTHESIS OF MATERIALS IN FLOW


Now that we have covered some valuable attributes of microreactors, we will highlight where they have been applied to create four basic polymeric materials: linear polymers, beads and other solids, microcapsules, and nally, bers. Linear Polymers Life without linear polymers would not be as rich, safe, or hospitable. Despite the innite value that linear polymers have provided, polymer chemists continue to develop means of synthesizing old polymers with

Figure 13. Microreactor system used in the polymerization of Ne-benzyloxycarbonyl-L-lysine, alanine, leucine, or NCA. The system consists of a PDMS micromixer and poly(tetrauoroethylene) microtubes. The mixer and reaction region were kept at 30 8C in a water bath. From T. Honda et al., Lab on a Chip 2005, 5, 812818. Reproduced by permission of The Royal Society of Chemistry. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6513

Figure 14. (top) Microuidic device fabricated with a rapid prototyping technique based on an optical thiolene resin. The reagents are introduced on the left, active mixing is induced with a stir ea, and the reaction proceeds in the wavy channel. (bottom) Size exclusion chromatograms of polymers synthesized in the microuidic device at various ow rates. Increasing the ow rates results in shorter residence times and lower molecular weight polymers (thin lines). The molecular weight is further tunable by the reduction of the amount of the initiator (thick line). Reprinted in part with permission from Wu, T. et al., J. Am. Chem. Soc. 2004, 126, 98809881. Copyright 2004, American Chemical Society.

greater control over the molecular weight and polydispersity index (PDI) as well as discovering and characterizing new linear polymers. Microreactors are emerging as tools to achieve both better syntheses and the discovery of new polymers. Cationic, ring-opening, and free-radical polymerizations are three types of polymerization mechanisms that have beneted from the rapid mixing and precise temperature control found in a microreactor. Yoshida et al.18 were among the rst to show that microreactors can yield improved polymerizations. They showed that N-methoxycarbonyl- N -(trimethylsilylmethyl)butylamine decomposes to a cation (2) that is irreversibly formed and can be collected (Scheme 1). By rapidly mixing cation 2 with a vinyl ether in an IMM microreactor, they found that the resulting cationic polymerization provided PDIs superior to those of polymers prepared in a batch reactor (Fig. 12). The authors demonstrated that rapid mixing controls the PDI by adjusting the ow rates. Mixing in a IMM microreactor increases with the ow rate, and as Yoshida et al. lowJournal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

ered the ow rates, the resulting polymers exhibited broadened PDIs. Thus, more efcient mixing at a high ow rate yields the narrowest PDIs. Iwasaki and Yoshida19 went on to show that microreactors are also useful in radical polymerization, yielding lower PDIs than those observed in batch reactions. The authors noted that highly exothermic polymerizations show improved control in the microreactors relative to the batch because microreactors dissipate heat much more rapidly. Not all radical reactions in ow provide signicant improvements in PDI. Jones et al.,20,21 using reversible additionfragmentation chain-transfer polymerization, and Cunningham et al.,22 using nitroxide-mediated radical polymerizations, observed similar or higher PDIs in tubular reactions than in batch reactors. In both of these cases, however, mixing was performed before introduction into the microreactor, and the reactions were performed in microemulsions that owed through the microreactors. Jones et al. suggested that PDI erosion was due to axial dispersion of the plugs and thus a distribution of residence times. As the plugs became large, the PDIs approached the batch values, and this was consistent with axial dispersion being the problem. Maeda et al.23 showed that the benets of microreactors are not limited to cationic and radical polymerizations by creating homopolymers of Ne-benzyloxycarbonyl-L-lysine, alanine, leucine, or glutamic acid (NCA). Reactions were run in batches and microreactors by combining NCA and triethylamine (Scheme 2). Instead of an IMM microreactor, a PDMS multilayered system was used (Fig. 13).24 In all cases, the microreactor provided slightly higher molecular weights and much better PDIs. Again, the NCA polymerization was strongly affected by mixing and potentially by temper-

Figure 15. Schematic of microchannel-conned surface-initiated polymerization (lSIP). The base of the device is a silicon wafer coated with a self-assembled monolayer of an initiator for atom-transfer radical polymerization. A microchannel is formed by the bonding of a molded slab of PDMS to the substrate. Polymer brushes with a thickness gradient are formed by the monomer owing from the inlet, across the channel, and out the other end. Reprinted with permission from Xu, C. et al., Macromolecules 2005, 38, 68. Copyright 2005, American Chemical Society.

6514

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

methacrylate) on-chip with excellent control over the conversion, PDI, and block ratios. Beers et al.,27 in addition to their fundamental work with linear polymer synthesis in ow, made surfacegrafted polymer brushes with thickness gradients. Gradients are produced by the monomer owing slowly over a surface activated for atom transfer radical polymerization (Fig. 15). The surface at the inlet receives much more monomer than that near the outlet, so the brushes have a linear gradient of heights moving from the inlet

Figure 16. (left) Schematics of the emulsication device made from etched silicon. A monomer solution is forced through the microchannels into the well area, which contains the continuous phase. Monodisperse droplets of the monomer snap off because of interfacial tension. (right) The droplets are later thermally polymerized to form crosslinked beads. Particles of different sizes have been prepared from two devices with smaller (top) or larger (bottom) microchannels. Reprinted with permission from Sugiura, S. et al., Ind. Eng. Chem. Res. 2002, 41, 40434047. Copyright 2002, American Chemical Society.

ature control, although no experiments directly tested this hypothesis. The authors noted that their device could produce *15 g day1. If the authors numbered up their system to 1000 microreactors, they could produce 5 Mg year1. Beers et al.25 showed that microreactors are useful for systematically varying the molecular weight of both homopolymers and block copolymers. They developed microuidic methods to carry out controlled radical polymerizations in which variables such as the monomer ratios or monomer/initiator ratio are altered to produce gradients of molecular weights or monomer ratios in copolymers. The homopolymer synthesis was achieved with a two-channel microreactor with the two channels meeting in a reservoir containing a magnetic ea stirring bar for active mixing. The mixing chamber emptied into a much longer microchannel reactor (Fig. 14, top). One inlet channel contained the monomer and catalyst, and the other contained the initiator. Both were dissolved in 50:50 water/methanol. The initiator/ monomer ratios were varied by changes in the relative concentrations, and the polymerization at a xed initiator/monomer ratio was then monitored as a function of the ow rate. The molecular weight decreased as a function of the ow rate, which was inversely proportional to the residence time (Fig. 14, bottom). The PDIs under all conditions were very good, ranging from 1.19 to 1.32. Using a similar approach, but with three channels instead of two, Beers et al.26 were able to create poly(ethylene oxide)-b-poly(2-hydroxypropyl

Figure 17. Patterns of droplet formation observed in the Tjunction (left) and pocket (right) when the ow rate of the continuous phase (Qc; aqueous) was varied at a xed dispersed phase (monomer) ow rate of 0.1 mL h1: (a,b) Qc 0.5 mL h1, (c,d) Qc 1.0 mL h1, (e,f) Qc 2.0 mL h1, (g,h) Qc 4.0 mL h1, (i,j) Qc 18.0 mL h1, and (k,l) Qc 22.0 mL h1. Reprinted from Chemical Engineering Journal, Vol. 101, Nisisako, T., Novel Microreactors for Functional Polymer Beads, 2329, Copyright 2004 with permission from Elsevier. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6515

Figure 18. Microchannel geometry used to create plugs and disks: (a) a schematic of a channel with plug and disk creation zones marked, (b) an optical micrograph of polymerized plugs in the 200-lm section of the channel (38-lm height), and (c) an optical micrograph of polymerized disks in the 200-lm section of the channel (16-lm height). Reprinted with permission from Dendukuri, D. et al., Langmuir 2005, 21, 21132116. Copyright 2005, American Chemical Society.

end to the outlet end. Within the microchannel, ow is laminar, and little diffusive mixing occurs on the timescale of the polymerization rate. This is an example in which a specic polymer morphology is achieved within a microreactor that would be difcult to form otherwise. Beads, Disks, and Other Solid Materials Soon after Quakes seminal microuidic emulsion manuscript15 was published, reports describing the formation of solid, monodisperse microspheres began to appear. The rst work that we have identied was from Nakajimas laboratory.28,29 He and his coworkers created divinylbenzene-containing droplets in a novel microchannel emulsication device (Fig. 16, left) using a continuous phase of aqueous poly(vinyl alcohol). The droplets were then thermally polymerized to yield highly crosslinked beads (Fig. 16, right). Nakajimas approach of forming monomer droplets rst and polymerizing them later ensured that the emulsion apparatus did not get clogged by the solid material. Particle sizes were several to tens of micrometers with diameter coefcients of variation (CVs) of less than 5%. The approach is noteworthy for its novelty, but the droplet
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 19. (a) Schematic of the ow-focusing geometry used in the microuidic droplet generator. The two immiscible liquids, A and B, are forced into the narrow orice, in which the inner liquid core breaks to release monodisperse droplets in the outlet channel. (b,c) Schematics of devices used for producing photochemically and thermally solidied particles. The channels used for photochemical crosslinking are elongated to allow longer durations of exposure of the droplets to UV light. (df) Representations of the shapes of droplets in the microuidic channels. If the volume of the droplet exceeds that of the largest sphere that can be accommodated in the channel, the droplet is deformed into (e) a disk or (f) a rod. From S. Xu et al., Angew Chem Int Ed Engl 2005, 44, 724728; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission.

6516

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 20. Microscopy images of poly(TPGDA) particles produced with photoinitiated polymerization, including (a) microspheres, (b) a crystal of microspheres, (c) rods, (d) disks, and (e) ellipsoids, and optical microscopy images of (f) agarose disks and (g) bismuth alloy ellipsoids produced with thermal solidication. From S. Xu et al., Angew Chem Int Ed Engl 2005, 44, 724 728; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission.

size depends only very weakly on the applied pressure, determined instead by the interfacial tension and the size of the channel. Thus, to change the nal particle size, one must adjust the surface tension of the continuous phase, which may be inconvenient or impossible, or fabricate a new device, which requires lengthy lithography and silicon etching. A few years after Nakajimas report, Nisisako lead a series of publications creating polymer beads in ow. Unlike Nakajima, Nisisako30 used a T-junction approach to droplet formation. As shown in Figure 17, two uids collided at a T-junction, just like Quakes experiment, and the resulting droplets were subsequently photopolymerized downstream. The monomer was 1,6-hexanediol diacrylate and was initiated with Darocur 1173. Nisisako showed that the ow rate could be used to tune the droplet sizes (30120 lm) and thus the nal bead size after polymerization. The method produced beads with diameter CVs less than 2%, although micrometer-sized satellite droplets were also produced at the T-junction. Nisisako17 also made a device with sequential T-junctions to make double emulsions, although these emulsions were not later used to template solid particles.

Doyle et al.,31 using the same T-junction approach as Nisisako, created solid disk and plug-shaped particles. As shown in Figure 18, snap-off at the junction yielded plugs containing Norland optical adhesive (NOA) 60, a photopolymerizable resin. When an ultraviolet (UV) beam was shone on the plug in the narrow channel, a plug structure was captured, but if the plug was allowed to expand laterally and then polymerized, a disk was formed. Because of the small dimensions of the plugs and disks, the photopolymerization occurred in less than 1 ls, using only 3 J cm2 of energy. A few weeks before Doyle et al.s work,31 the team of Whitesides and Kumacheva32 reported a ow-focusing method to make a variety of shapes out of photopolymerized tripropylene glycol diacrylate (TPGDA), dimethacrylate oxypropyldimethylsiloxane, divinylbenzene, ethylene glycol diacrylate, and pentaerythritol triacrylate. Figure 19 shows a schematic of the owfocusing device and the various shapes realized with channels of different geometries. Figure 19(b) also illustrates the wavy channel used to increase the residence time while the droplets and other structures are photopolymerized.

Figure 21. Optical microscopy images of a lattice of dimethacrylate oxypropyldimethylsiloxane disks (a) before and (b) after in situ photopolymerization and (c) typical SEM image of the solidied disks. The scale bar is 200 lm. Reprinted with permission from Seo, M. et al., Langmuir 2005, 21, 47734775. Copyright 2005, American Chemical Society. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6517

Figure 22. (a) Schematic of the production of droplets in a microuidic ow-focusing device by a laminar coow of (A) silicone oil, (B) monomer, and (C) aqueous phases. The orice has a rectangular shape with a width and height of 60 and 200 lm, respectively. (b) Schematic of the wavy channel used for the photopolymerization of the monomer in the coreshell droplets. (c) Optical microscopy image of core shell droplets owing in the wavy channel used for in situ photopolymerization. The scale bar is 200 lm. (d) Photograph of a polyurethane microuidic system. Reprinted with permission from Nie, Z. H. et al., J. Am. Chem. Soc. 2005, 127, 80588063. Copyright 2005, American Chemical Society. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 23. Ternary phase-like diagram of the hydrodynamic conditions used in the production of droplets with various morphologies. The axes of the diagram correspond to the ow rates of the water (Qw), monomer (Qm), and oil phases (Qo), normalized to the total ow rate. The letters correspond to droplet shapes subsequently polymerized to the microparticles shown in Figure 24. Reprinted with permission from Nie, Z. H. et al., J. Am. Chem. Soc. 2005, 127, 8058 8063. Copyright 2005, American Chemical Society.

The combination of ow focusing and alteration of channel dimensions allowed the capture of microspheres, rods, disks, and ellipsoids (Fig. 20). Channel clogging was prevented because the continuous phase of aqueous 2 wt % sodium dodecyl sulfate (SDS) prevented the droplets from contacting the walls. The authors mentioned that up to 250 particles could be formed per second and 510% shrinkage was observed after polymerization. Also, the authors were able to control the size by changing the ow rate and observed diameter CVs smaller than 1.5%. In addition to the excellent range of structures, the team of Whitesides and Kumacheva32 was able to incorporate dyes, quantum dots, and liquid crystals into the microparticles. Using essentially the same type of ow-focusing approach, Kumacheva et al.33 organized discoid droplets of silicone oil or dimethacrylate oxypropyldimethylsiloxane into two-dimensional lattices [Fig. 21(a)]. The lattices were then polymerized by a photoinitiated radical polymerization. As shown in Figure 21(b), the polymerized disks shrink by 7%. The shrinkage is important because if the discs increased in size, the channels would clog. Figure 21(c) shows a scanning electron microscopy (SEM) image of the monodisperse discs after polymerization.
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Kumacheva et al.34 followed up this work by creating a microuidic device that ow-focused three uids to create droplet-in-droplet structures. In this system, three immiscible liquids are used: aqueous SDS (liquid C), monomers (liquid B), and silicon oil (liquid A; Fig. 22). After the photopolymerization of the two

Figure 24. (ae) SEM images of polymer microparticles obtained by the polymerization of TPGDA in droplets obtained in regimes AD in Figure 23 after the removal of the silicone oil cores and (f) cross section of a polyTPGDA particle with three cores obtained by the polymerization of coreshell droplets with three cores. All scale bars are 40 lm. Reprinted with permission from Nie, Z. H. et al., J. Am. Chem. Soc. 2005, 127, 80588063. Copyright 2005, American Chemical Society.

6518

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 25. Schematic illustration of the process for fabricating the apparatus used to manufacture microparticles: (A) xing the acryl anchor, bonding the glass pipette onto the anchor, pouring the PDMS prepolymer, and curing on the hot plate; (B) taking out the cured PDMS and cutting; (C) pulling out the micropipette; (D) puncturing the hole for sheath ow; (E) exposing to oxygen plasma and bonding the pulled micropipette; and (F) inserting and bonding the outlet pipette and shielding. An expanded illustration of the microuidic nozzle and reagents used to fabricate the microparticles is shown on the bottom. Reprinted in part with permission from Jeong, W. J. et al., Langmuir 2005, 21, 37383741. Copyright 2005, American Chemical Society. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

monomers, TPGDA or ethylene glycol dimethacrylate, the droplet-in-droplet structures were captured, yielding novel bead structures. Subtle changes in the ratios of the three solution ow rates allowed Kumacheva et al.34 to precisely control the fascinating structures produced in their device. They mapped out a ternary phase-like diagram in which the axes reected the effect of the ow rates of the aqueous, monomer, and oil phases on the structure of the formed particle (Fig. 23).

The letters in the pseudo-phase diagram correspond to the letters in Figure 24, which shows SEM images of the various microcapsule morphologies. Also, in a separate report, Kumacheva et al.35 demonstrated the creation of TPGDA-co-acrylic acid copolymer beads. Seong et al.,36 using a ow-focusing device (Fig. 25), created hydrogel beads impregnated with glucose oxidase and horseradish peroxidase. They formed the beads by ow-focusing a solution of the enzymes toJournal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6519

Figure 26. SEM micrograph of microparticles polymerized from droplets formed at a monomer ow rate of 1 lL min1 and a continuous phase ow rate of 225 lL min1. These particles have diameters of 90 lm with CVs less than 2%. Reprinted in part with permission from Jeong, W. J. et al., Langmuir 2005, 21, 37383741. Copyright 2005, American Chemical Society.

gether with 85 wt % 4-hydroxybutyl acrylate, 11 wt % acrylic acid, 3 wt % 2,2-dimethyoxy-2-phenyl-acetophenone (the initiator), and 1 wt % ethylene glycol dimethyl acrylate. The monomer/initiator/enzyme droplets were then polymerized via a photoinitiated radical polymerization (Fig. 26). The device produced monodisperse beads at rates up to 2600 per minute with a 1 lL min1 disperse phase ow rate. The enzyme hydrogel beads were then packed into a nozzle and reagents, which yielded a uorescent product after the sequential action of the two enzymes, were passed through the packed beads. A uorescent product was observed, indicating that both encapsulated enzymes were active, although the activity of the enzymes was not quantied. To the best of our knowledge, this

Figure 28. Electrical actuation of biphasic, electrically anisotropic spheres: (a) color switching test using prepared particles [a voltage of 100 V was applied between the 0.4-mm gap of two electrode panels (40 mm 40 mm) at a switching frequency of 1 Hz] and (b) magnied top view of the display panel in part a. From T. Nisisako et al., Adv. Mater. 2006, 18, 11521156; ' 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley. com.]

Figure 27. Formation of a bicolored droplet at the sheath ow junction. To capture clear images of the color edge, white pigments were not used in this case. Reprinted from Chemical Engineering Journal, Vol. 101, Nisisako, T., Novel Microreactors for Functional Polymer Beads, 23 29. Copyright 2004 with permission from Elsevier. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

report represents the rst instance of active enzymeencapsulated beads formed in ow. Using a ow-focusing device, De Smedt et al.37 synthesized biodegradable microgels by forming dextran hydroxyethyl methacrylate/Irgacure 2959 droplets and initiating subsequent polymerization by passing the droplets over a UV beam. Monodisperse beads were formed with a diameter of *10 6 0.3 lm. Also, the authors explored the use of enzyme-containing beads in controlled-release experiments. In two separate reports,30,38 Nisisako and coworkers pushed complex bead creation forward by creating biphasic, Janus-faced droplets and solid beads. Figure 27 illustrates the formation of Janus droplets by two solutions owing laminarly into a ow-focusing device.30 The bicolored droplets contain either carbon black or titanium dioxide to provide black and white hemispheres, respectively. The droplets contain a photoinitiator and 1,6-hexanediol diacrylate or isobornyl acrylate to effect polymerization. After photocuring the droplets into beads, Nisisako et al.38 then used these Janus beads as the active elements in electroactive, switchable surfaces. The switching is due to the different charging properties exhibited by carbon black and titanium dioxide. Figure 28 shows beads on top of a display panel. Depending on the bias of the surface, the Janus-faced particles display their white or black sides. Figure 28(a) portrays the display panel with a given bias and also 0.5 s after the bias was reversed, whereas Figure 28(b) is a blowup of the black and

6520

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 29. (a) Schematic depicting the experimental setup used in the study. A mask containing the desired features is inserted in the eld-stop plane of the microscope. The monomer stream ows through the all-PDMS device in the direction of the horizontal arrow. Particles are polymerized by a mask-dened UV-light beam emanating from the objective and then advected within the unpolymerized monomer stream. The side view of the polymerized particles can be seen in the inset shown on the right. Also shown is the unpolymerized oxygen inhibition layer that allows the particles to ow easily after being formed. (b) Bright-eld microscopy image (x y plane) of an array of cuboids moving through the unpolymerized monomer. (c) Cross-sectional view of the cuboids seen in part b upon collection in a droplet that has turned most particles on their sides. Reprinted with permission from Macmillan Publishers, Ltd: Nature Materials, 5, 365369, 2006.

white portions of the panel. The size control and low dispersity of these beads could make this type of chromic display sharper than displays made with other methods. The authors also demonstrated a new device design for making large quantities (tons per year) of these beads. Very recently, Doyle et al.39 introduced a revolutionary technique for fabricating irregularly shaped microparticles in a ow device. The authors employed microscope projection photolithography to photopolymerize oligomeric diacrylate monomers owing in a PDMS microuidic device (Fig. 29). Because the polymerization is faster than 100 ms, the particles do not move appreciably during the patterning, minimizing the distortion of the shapes. Doyle et al. were able to

create a variety of shapes, including triangles, cuboids, cylinders, and various irregular forms, with a spatial resolution of around 3 lm (Fig. 30). The key to the technique lies in an oxygen-rich zone near the PDMS walls, formed as O2 diffuses from the elastomer into the monomer stream. The oxygen in this zone, roughly 2.5 lm thick, inhibits polymerization and ensures that no polymer forms in contact with the channel. In addition to novel microparticle shapes, the authors created chemically biphasic particles by using coowing laminar streams of monomer, one of which contained a rhodamine-labeled crosslinker, and forming particles across this interface. The method presented by Doyle et al. promises to yield many exciting applications in the future.
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6521

Figure 30. SEM images of particles produced using microscope projection photolithography. The scale bar in all cases is 10 lm. (ac) Flat, polygonal structures that were formed in a 20lm-high channel, (d) colloidal cuboid that was formed in a 9.6-lm-high channel, (e,f) high-aspect-ratio structures with different cross sections that were formed in a 38-lm-high channel, and (gi) curved particles that were all formed in a 20-lm-high channel. The inset in each panel shows the transparency mask feature that was used to make the corresponding particles. Reprinted with permission from Macmillan Publishers, Ltd: Nature Materials, 5, 365369, 2006.

Capsules The transition from creating solid beads to hollow capsules requires changing from a polymerization within the droplet to either the polymerization of double emulsions or an interfacial polymerization. This change sounds relatively simple, but most devices would rapidly clog if an interfacial polymerization were performed within them. Two almost simultaneous reports from the Weitz and Whitesides groups describe the use of radially symmetric ow-focusing devices to create microcapsules in ow. These devices create droplets that are protected from the device walls by a coaxial sheath of uid, unlike device schemes such as T-junctions or planar ow-focusing devices. Weitzs device40 (Fig. 31) consists of nested glass capillaries, allowing two coaxial uids to collide, and a third, outer uid provides the ow focusing. All three are then forced into a small orice leading to an exit tube. This device creates double emulsions with a great deal of control over the size and number of internal droplets. To create capsules, the Weitz group uses a 70:30 mixture of NOA and acetone as the outer layer of the double emulsion (i.e., the middle layer of the originally focused phases). The double emulsion passes
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

through a UV beam, forming a shell of crosslinked NOA [Fig. 32(B,C)]. The authors noted that droplet production is up to an astounding 5000 per second. Weitz et al.41 also formed polymerosomes by using water-inoil-in-water emulsions with a diblock copolymer, poly(n-butyl acrylate)-b-poly(acrylic acid), dissolved in the middle organic phase. Upon drying, a exible, permeable polymer shell remains [Fig. 32(DF)]. The Whitesides device42 also accomplishes droplet formation by focusing a coaxial ow through a small orice (Fig. 33). The Whitesides design uses an insulated optical ber cut with a scalpel as a template (Fig. 34). The ber is exposed by the insulation being pulled back slightly, and a large segment of the ber, including the exposed ber section, is encased in PDMS. Once the PDMS is fully cured, all the ber components are removed from the PDMS. Tubes are placed in the voids left by the insulated portions. The small orice in the center of the device shown in Figure 33 was formed by the exposed portion of the optical ber. The Whitesides group created nylon-6,6 capsules by using aqueous 1,6-diaminohexane as the disperse phase and introducing adipoyl chloride through a third inlet, as shown in Scheme 3). Presumably, they started the interfacial polymerization downstream of where the

6522

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 31. Microcapillary geometry for generating double emulsions from coaxial jets. Part A is a schematic of the microcapillary uidic device. The geometry requires the outer uid to be immiscible with the middle uid and the middle uid to be in turn immiscible with the inner uid. The typical inner dimension of the square tube is 1 mm; this matches the outer diameter of the untapered regions of the collection tube and the injection tube. Typical inner diameters of the tapered end of the injection tube and the orice of the collection tube are 1050 and 50500 lm, respectively. The thickness of the coating uid on each drop can vary from extremely thin (<3 lm), as in part B, to signicantly thicker. Parts F and G show double emulsions containing many internal drops with different size and number distributions. Part H shows double emulsion drops, each containing a single internal droplet, owing in the collection tube. From Utada, A. S. et al., SCIENCE 308: 537541 (2005). Reprinted with permission from AAAS.

droplets form, rather than including the acid chloride directly in the continuous phase, to prevent clogging at the narrow orice. The axisymmetric device yielded capsules with monodisperse diameter CVs and allowed the formation of capsules with interesting features such as encapsulated magnetic particles (Fig. 35). At the same time that the Weitz and Whitesides publications appeared, Ganan-Calvo et al.43 introduced a ow-focusing device that his group calls an atomizer. The device, as shown in Figure 36, relies on an outer uid to focus one or two coaxial phases as the three uids leave the device. In some cases, the focusing uid was actually air. With the compound atomizer, poly(D,L-lactic acid-co-glycolic acid) (PLGA) microcapsules containing encapsulated gentamycin, an antibiotic, were formed. The process started with PLGA dissolved in an organic solvent (outside layer of the coaxial uids) and gentamycin in an aqueous solution (core of the coaxial ow). After breakup, the organic solvent evaporated, leaving a PLGA shell surrounding the gentamycin solution. Besides evaporative coacervations, Ganan-Calvo used three immiscible uids to create multicored capsules out of the photopolymer SK9 (Fig. 37). The same authors recently published a report

using the same device to create monodisperse polystyrene microcapsules and dye-labeled microcapsules, although both of these results were mentioned in the original report.44 A few months after the reports by Weitz, Whitesides, and Ganan-Calvo, we introduced a very simple microuidic device45 consisting of laboratory tubing and small-gauge needles that form T-junctions in the middle of the tube channel. As shown in Figure 38, a continuous phase is pumped through 1/16-in. poly(vinyl chloride) (PVC) tubing, and a 154-lm-i.d. needle introduces a disperse phase to the continuous ow. Polyamide capsules are formed by the inclusion of poly (ethylene imine) in the continuous phase and by the addition of sebacoyl chloride and 1,3,5-benzene tricarboxylic acid chloride in a chloroform/cyclohexane disperse phase. The interfacial polymerization begins immediately as the two phases meet, yielding capsules with diameter CVs ranging from 3.3 to 8.6%. As in the ow-focusing cases, the capsule sizes can be easily varied from 865 to 313 lm by the alteration of the ow rate (Fig. 39). The McQuade system is very easy to set up and operate, providing the opportunity to create many differJournal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6523

Figure 32. Coreshell structures fabricated from double emulsions generated in the microcapillary device. (A) Optical photomicrograph of the water-in-oil-in-water double emulsion precursor to solid spheres. The oil consists of 70% NOA and 30% acetone. (B) Optical photomicrograph of rigid shells made by the crosslinking of the NOA by exposure to UV light. (C) SEM image of the shells shown in part B after they have been mechanically crushed. The scale bar in part C also applies to parts A and B. (D) Bright-eld photomicrograph of the water-in-oil-in-water double emulsion precursor to a polymer vesicle. The oil phase consists of a mixture of toluene and tetrahydrofuran (70/30 v/v) with dissolved diblock copolymer (poly(butyl acrylate)-b-poly(acrylic acid)) at 2% (w/v). (E) Phase-contrast image of the diblock copolymer vesicle after evaporation of the organic solvents. (F) Phase-contrast image of the deated vesicle after osmotic stress is applied through the addition of 0.1 M sucrose to the outer uid. The scale bar in part F also applies to parts D and E. From Utada, A. S. et al., SCIENCE 308: 537541 (2005). Reprinted with permission from AAAS.

ent materials in ow. As an example to underscore this fact, the McQuade group46 created hierarchical capsules composed of oligomeric and crystalline diphenylsilanediol. We formed these spiny capsules by using a continuous phase of 70% aqueous glycerol and snapping off monodisperse droplets of dichlorodiphenylsilane. The chlorosilane quickly hydrolyzed to diphenylsilanediol, which then oligomerized and crystallized. The oligomers formed the amorphous shell, and the crystals were the spines (Fig. 40). Fibers All the previous examples of preparing solid materials in ow used the droplet pseudophase to template the nal materials. Yet, as discussed previously, coowing laminar streams are also attainable within a microuidic device, and a number of groups have created materials shaped by these extended boundaries to form elongated membranes and bers. The rst synthesis of such a structure in ow was provided by the Whitesides laboratory.8 The lament formation is shown in
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 41, in which a solution of polystyrene sulfonate and hexadimethrine bromide are collided under laminar ow conditions. The two oppositely charged polymers precipitate at the interface to yield a ber. Because the

Figure 33. Three-dimensional, axisymmetric ow-focusing channel. The channel is composed of a cylindrical tube with a narrow cross section halfway down its length. The narrow region serves as the orice in which uid is focused and breaks into aqueous droplets. In this geometry, wetting problems are avoided because the aqueous phase does not make contact with the walls. From S. Takeuchi et al., Adv Mater 2005, 17, 10671072; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission.

6524

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 34. (a) Schematic depicting the fabrication process of the ow-focusing device. The insulation surrounding an optical ber was cut with a scalpel, and the ends were pulled to expose the ber. The ber was embedded in a block of PDMS. After the PDMS had cured, the ber was removed by being pulled through the end of the PDMS block. Two glass capillaries were inserted as an inlet and outlet. (b) Image of the nal device. From S. Takeuchi et al., Adv Mater 2005, 17, 10671072; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission.

that the polymeric material formed in the channels is stuck there permanently. Beebe, with Lee and others,48 continued to create microbers and hollow tubes in ow by using a coaxial microuidic device. A coaxial setup ensures that no part of the ber touches the device wall; otherwise, the polymerized material adheres to the walls, causing the device to clog. They achieved efcient ber formation by using the same ow-focusing device that Beebe developed with Seong and Lee to make enzyme-impregnated microgels. Figure 43(a) illustrates the device in which a sample uid ows through a capillary surrounded by a sheath uid. This approach provides a thin sample stream whose diameter can be varied by changes in the sheath ow rate. By lling the sample stream with a photopolymerizable cocktail [4-hydroxybutyl acrylate (85%), acrylic acid (11%), ethylene glycol diacrylate (1%), and 2,20 -dimethyoxy2-phenyl-acetonephenone (3%)], they could cure the thin stream by a UV beam downstream [Fig. 43(a)]. In the same publication, the authors created hollow microtubes by increasing the complexity of their system. They introduced a second core phase, as shown in Figure 43(b). Using an inert inner core solution and the same photocurable outer core solution, they created hollow bers. In addition to characterizing the bers and tubes, the authors impregnated the tubes with enzymes, created biphasic bers, and wove the bers into a responsive swatch.

CONCLUSIONS
The use of microreactors and microuidic devices for the production of polymers with controlled properties and morphologies appears to have a bright future. The small dimensions of these microdevices provide benets

ribbon is trapped within the device, no further characterization or discussion is provided. Beebe et al.47 recognized the value of depositing a lm within a microuidic device and reported the formation of a nylon membrane by colliding two uids at a cross-junction. The authors designed their cross-junction so that the aqueous phase entering from the north would run along a path to the west by coating the north inlet and west outlet with a hydrophilic surface (Fig. 42). Conversely, the south inlet and east outlet were coated with a hydrophobic surface to direct an organic phase. By putting 1,6-diaminohexane into the aqueous phase and adipoyl chloride into the organic phase, they deposited a nylon membrane at the interface at which the two uids met on their respective wetting pathways. The membrane was determined to have a pore size smaller than 200 nm. This example and the Whitesides example are interesting and provide useful membranes, but they suffer from the limitation

Scheme 3. From S. Takeuchi et al., Adv Mater 2005, 17, 10671072; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6525

Figure 35. (a) Collection of nylon-6,6-coated aqueous droplets, (b) coated droplet containing 50-nm-diameter magnetic particles in an applied magnetic eld (the particles are aligned with respect to the magnetic eld), and (c,d) distributions of the diameter of the droplets before and after polymerization, respectively. From S. Takeuchi et al., Adv Mater 2005, 17, 10671072; ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission.

Figure 36. Flow-focusing atomizer: (a) a simple jet with (1) a focusing uid, (2) a focused uid, and (3) a meniscus and (b) a compound atomizer with two concentric needles with (1) a focusing uid, (2) focused uids (core uid and shell uid), and (3) a compound meniscus. From L. Martin-Banderas et al., Small 2005, 1, 688692. ' 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

6526

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

Figure 37. (left) In-ight photograph of the breakup of a coaxial jet of blue ink surrounded by a photopolymer (SK9) and (right) optical microscopy image of the multicored microcapsules produced after UV curing. From L. MartinBanderas et al., Small 2005, 1, 688692. ' 2005 WileyVCH Verlag GmbH & Co. KGaA, Weinheim, reproduced by permission. [Color gure can be viewed in the online issue, which is available at www. interscience.wiley.com.]

in thermal and mass transport as well as unique control over the uid structure. Enhanced transport characteristics allow more precise synthesis of soluble polymers that can also be produced in traditional batch reactors. Perhaps more importantly, the interesting and unique uid morphologies available within microuidic devices open the door to particle shapes not accessible through other means. Solid, hollow, and multicored, asymmetric, and irregularly shaped polymer microparticles as well as membranes and bers can all be produced with microuidic devices. These particles are generally monodisperse, and the sizes may be tuned from a few micrometers to hundreds of micrometers, depending on the specic device geometry and ow parameters. Novel microparticles may have applications as functional colloids, in photonics, and in the encapsulation of materials

Figure 39. Light microscopy images of polyamide capsules formed with a constant organic ow rate and an increasing aqueous ow rate: (A) 2.00, (B) 11.0, (C) 13, and (D) 25 mL min1. Reprinted with permission from Quevedo, E. et al., J. Am. Chem. Soc. 2005, 127, 1049810499. Copyright 2005, American Chemical Society.

for catalysis and controlled delivery. The process of rapid prototyping and the emergence of simple, homemade devices that take only minutes or hours to construct have enabled many groups to explore this exciting eld. Indeed, the eld is currently limited only by the imagination of those seeking to push the bounds of polymer chemistry and particle synthesis.

BRIEF OVERVIEW OF MICROREACTOR FABRICATION


To a polymer chemist accustomed to the world of stirred reaction vessels, the vast array of microuidic device geometries and materials may seem daunting. To help would-be microreactor users, we are providing a brief overview of device fabrication. From hard lithographic techniques to soft lithography and in-housebuilt devices, a concise description is given, and readers may follow the appropriate citations for more details. The foundations of microuidic fabrication techniques lie in the well-established eld of semiconductor microelectronics. For instance, the microfabrication tools of lithography, resist layers, and wet and dry etching have been instrumental in the fabrication of many microuidic systems. These methods have been used to create microuidic systems in both direct and indirect ways. In the direct method, the microfabricated component is used directly as a channel in a device.
Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 38. Schematic of the simplied microuidic device. The continuous phase ows through PVC tubing, and a disperse phase is introduced through small-gauge needles. Both phases are driven by syringe pumps. Reprinted with permission from Quevedo, E. et al., J. Am. Chem. Soc. 2005, 127, 1049810499. Copyright 2005, American Chemical Society. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

HIGHLIGHT

6527

Figure 40. SEM images of organosilane microcapsules produced in a microuidic device: (A) a population of typical, spheroidal microcapsules; (B) a higher magnication image of the spiny surface, and (C) a microcapsule fragment showing the inner, amorphous shell and the outer, spiny layer. Reprinted with permission from Steinbacher, J. L. et al., J. Am. Chem. Soc. 2006 in press. Copyright 2006, American Chemical Society.

Direct methods have the disadvantage that each completed fabrication yields only one device. However, if the fabrication is fast, this can allow for rapid prototyping of many designs. On the other hand, indirect methods use the microfabricated component as a master to transfer the design to a secondary material in a replication step. This molded material is then used in the nal device. Indirect methods have the advantage of needing only one master to create many (up to hundreds) of nal devices.

Master Production Most device fabrication strategies begin with a lithography step. Lithography is the process of using a mask that contains a desired pattern to expose specied areas of a photosensitive material to X-ray or UV radiation.49 The mask pattern is thereby transferred to the underlying material (Fig. 44). Usually, the photosensitive material is a polymeric resist layer coated as a thin lm on a substrate. After exposure, the resist is then developed, and a washing step removes either the exposed portion (positive tone) or the unexposed por-

tion (negative tone). At this point, various etching techniques may be employed to remove substrate material that is no longer shielded by the resist, reproducing the original mask pattern (or its negative) in the substrate. The most common substrate for microuidic master production is elemental silicon. Silicon may be removed by two general methods, wet or dry etching. Wet etching,50 by far the more widely used and less expensive, is further divided into isotropic or anisotropic methods. In isotropic etching, aqueous strong acids such as HF and HNO3 remove material without preference to any crystallographic direction. Thus, rounded channels are produced, and mask undercutting may be a concern (Fig. 45, left). On the other hand, alkaline anisotropic etching conditions lead to the removal of certain crystallographic planes over others, etching at an angle of 54.78 from the (100) direction in silicon. In this case, trapezoidal features are formed, and they limit the depth

Figure 41. Schematic and optical micrograph of a polymeric structure deposited on glass at the laminar ow interface of 0.005% aqueous solutions of poly (sodium 4-styrenesulfonate) and hexadimethrine bromide. Reprinted with permission from Kenis, P. J. A. et al., SCIENCE 1999, 285, 8385. Copyright 1999, AAAS. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 42. Fabrication of a semipermeable polyamide membrane: (a) a schematic illustration of the surface-patterned channel and (b) a schematic illustration of the polymer membrane fabricated inside the channel by interfacial polymerization. Reprinted in part with permission from Zhao, B. et al., J. Am. Chem. Soc. 2002, 124, 52845285. Copyright, 2002 American Chemical Society. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

6528

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

metal, often nickel or a nickel alloy.53 After the substrate is removed, the metal electroform may be used as a master in later replication steps instead of the silicon master. A prominent method that avoids the use of silicon altogether is the so-called LIGA [lithographie (lithography) galvanoformung (electroplating) abformung (molding)] process.54 In the LIGA process, a polymeric substrate, usually poly(methyl methacrylate) (PMMA), is ablated with either X-rays or UV lasers through a patterned mask (Fig. 46). High aspect ratios and very low surface roughness are achieved. After patterning, the resulting structures are electroplated with a metal, the polymeric substrate is removed, and the metalized master is used in subsequent replication steps. A simpler method of producing masters is rapid prototyping, introduced by Whitesides et al.55 in the mid-1990s. With high-resolution laser printers, patterns

Figure 43. (a) Schematic of the basic apparatus for fabricating microbers. (b) By the addition of an additional stage, the apparatus is capable of producing microtubes. From T. Honda et al., Lab on a Chip 2005, 5, 812818. Reproduced by permission of The Royal Society of Chemistry.

and aspect ratio of features that may be fabricated (Fig. 45, middle). Conversely, dry etching techniques51 employ low-temperature plasmas to chemically and physically etch materials. Reactive-ion etching (RIE) employs reactive gasses, such as SF6, which form molecular ions and radicals in the plasma, which then react with substrate silicon to form volatile species. Alternatively, ion-beam etching (IBE) uses inert gasses, resulting in physical etching. When proper inhibitors are used, dry etching processes result in large aspect ratios with vertical features (Fig. 45, right). For certain applications, the disadvantages of silicon, namely its brittleness and potentially poor release properties,52 may preclude its use as a master. Thus, a patterned silicon substrate may be electroplated with a

Figure 44. Schematic of the lithography process. A mask containing the desired pattern is placed between a UV or Xray beam and a photosensitive resist layer spun on top of the substrate. After exposure, the resist is developed, and a washing step removes either the exposed portion (positive tone, right) or the unexposed portion (negative tone, left). An etching step removes substrate material no longer shielded by the resist, which is removed, reproducing the original mask pattern (or its negative) in the substrate. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6529

Figure 45. Schematic of etching processes. Isotropic etching, using acidic media, produces rounded channels and possibly undercut masks. Alkaline anisotropic etching yields straight but trapezoidal channels. Dry etching techniques, such as RIE and IBE, produce vertical walls with high aspect ratios. [Color gure can be viewed in the online issue, which is available at www. interscience.wiley.com.]

are printed directly on transparencies with black ink and used as photolithographic masks with resolutions of roughly 20 lm. Although this resolution is not as ne as those of traditional lithographic masks, which are less than 1 lm, the cost is roughly $1 per square inch versus roughly $1500 per square inch for chromium-on-glass masks. In addition to the obvious cost advantage, patterns can be designed and printed in house rather than contracting custom fabricators, who may require weeks to months. Once the mask has been printed, a master is made via lithography of negative tone resists such as Microposit 1813 or SU-8, which are durable polymer resists. These rapidly prototyped, polymeric masters are then used as masters in further replication techniques.

Indirect Fabrication Once a suitable master has been produced, the replication of the pattern into the device material is achieved by one of several methods. Similarly to its ubiquitous use for creating macroscopic objects, injection molding has been used to create microchannels. Injection molding is the process of forcing a polymer above its melting temperature (Tm) into a mold under high pressure and then cooling it below the Tm and releasing the mold. McCormick et al.56 used a silicon master to create nickel electroforms that templated channels into injection-molded acrylic, whereas Arnold et al.57 used laser-ablated masters to create injection-molded PMMA channels and gratings. More commonly, imprinting methods are used in which structures on a master are pushed under pressure into a polymer substrate, transferring the pattern to the polymer (Fig. 47).58 Xu et al.59 performed the room-temperature imprinting of PMMA, copolyester, polycarbonate, and polystyrene using silicon masters, creating channels

Figure 46. UV laser micromachining process. A UV excimer laser pulse rapidly breaks chemical bonds within a restricted volume to cause a miniexplosion and ejection of ablated material. Reprinted with permission from Roberts, M. A. et al., Anal. Chem. 1997, 69, 20352042. Copyright 1997, American Chemical Society. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Figure 47. Schematic of the embossing process. A silicon or metal template is pushed under pressure into a slab of a polymer. The pattern on the template is transferred to the polymer, which constitutes the nal channel material after sealing to an appropriate base. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

6530

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

cal samples, although it swells in many organic solvents.64 Direct Fabrication Direct fabrication methods do not rely on master patterns created in prior steps; instead, the patterned material is used directly in the nal device. Microuidic devices may be fabricated directly from the photosensitive glass FOTURAN65 or polymer materials66 patterned via lithography steps similar to those described previously for mask preparation. The polymer resist SU-8 is an especially versatile material for microelectromechanical systems67,68 and microuidic channels,69 supporting aspect ratios as high as 10:1 and displaying exceptional

Figure 48. (A) SEM image and (B) prolometer cross section of an embossed PMMA device. Reprinted with permission from Xu, J. D. et al., Anal. Chem. 2000, 72, 19301933. Copyright 2000, American Chemical Society.

90 lm wide and 3040 lm deep (Fig. 48). Imprinting may also be performed above the glass-transition temperature of the polymer, in which case it is called hot embossing. Ueno et al.60 created 100-lm-wide channels in polystyrene by imprinting with a silicon master at 108 8C, whereas Chou et al.61 created 60-nm-wide channels using PMMA hot-embossed at 200 8C. An indirect duplication method that has gained widespread use in the academic world is soft lithography,62 a collective set of methods that uses patterned elastomers as masks, stamps, or molds. Soft lithography is often paired with masters fabricated with rapid prototyping, as described previously. To create microuidic devices, a liquid prepolymer mixture, often PDMS, is poured onto a patterned master and then cured at a mildly elevated temperature (Fig. 49).63 The solid elastomeric stamp is then peeled off the master, and the channels are sealed by the bonding of the stamp to a wide variety of materials, including silicon, silica, silicon nitride, glass, polystyrene, and polyethylene.53 PDMS is an ideal material for many microuidic applications because of its optical transparency down to 300 nm, the opportunities for surface functionalization, and its compatibility with biologi-

Figure 49. Scheme describing replica molding of microuidic devices in the soft lithography methodology: (A) a master is fabricated by rapid prototyping; (B) posts are placed on the master to dene reservoirs; (C) the prepolymer is cast on the master and cured; (D) the PDMS replica is removed from the master; and (E) exposing the replica and an appropriate material to an air plasma and placing the two surfaces in conformal contact make a tight, irreversible seal. Reprinted with permission from Duffy, D. C. et al., Anal. Chem. 1998, 70, 49744984. Copyright 1998, American Chemical Society. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6531

mechanical durability (Fig. 50). Moreover, LIGA, discussed previously, may be employed to directly fabricate microuidic devices. For instance, Roberts et al.70 used an excimer laser operating at 193 nm to ablate channels into a variety of polymers, including polystyrene, polycarbonate, cellulose acetate, and poly(ethylene terephthalate). Pethig et al.71 used an excimer laser operating at 248 nm in a multistep process to create a prototype cell-handling device. Even higher energy lithography has employed X-ray72,73 and synchrotron74 radiation to directly pattern microuidic channels into PMMA. Although the resolution and aspect ratio are excellent, access to suitable equipment has limited the use of such procedures in the academic community. Microuidic devices may also be fabricated directly by milling techniques, which obviate the need for lithography altogether. Vasile et al.75 used focused ion beams to create smooth, deep channels in PMMA. In the same publication, they employed ion beams to create steel micromilling tools that were then used to mill channels in PMMA with a mechanical, high-precision milling machine.76 Other researchers have used me-

Figure 51. SEM image of a micromachined copper foil. The grooves are 70 lm deep and 100 lm wide. Copyright 2001 from Microstructure Devices for Applications in Thermal and Chemical Process Engineering by K. Schubert, J. Brandner, M. Fichtner, G. Linder, U. Schygulla, and A. Wenka. Reproduced by permission of Taylor & Francis Group, LLC., http://www.taylorandfrancis.com

chanical micromilling techniques to create structures in steel and brass with 50-lm resolution77 and microgrooves and microchannels in stainless steel, aluminum, titanium, silver, and other metals with 8-lm resolution (Fig. 51).78 A recently introduced technique uses commercially available cutting plotters, traditionally used by the graphics arts industry, to cut patterns into a variety of polymeric materials.79 Termed xurography, it has been used to realize a variety of shapes and twoand three-dimensional microuidic channel geometries with a resolution of roughly 35 lm. A trend in the last several years has been the creation of microuidic devices by unconventional, inhouse techniques that do not require the use of photolithography or complex instrumentation. We have discussed these devices previously, but they bear mentioning again. Jeong et al.36 invented a technique that uses a small glass capillary to template a channel in a PDMS slab. The capillary is later removed, and pulled pipettes are then inserted, forming core annular ow and layered channels.48 Similarly, Takeuchi et al.42 used a cut optical ber as a template for a ow-focusing device in PDMS. Utada et al.40 used pulled pipettes nested into square capillaries to achieve coaxial alignment of a ow-focusing device. Finally, we developed a simple microuidic device consisting of laboratory tubing and small-gauge needles that requires only minutes to construct.45

REFERENCES AND NOTES


Figure 50. SEM micrographs of SU-8 resist structures formed by near-UV photolithography. The minimum feature size of 16 lm is resolved in 160-lm-thick SU-8. Reprinted with permission from K. Y. Lee et al., J. Vac. Sci. Technol. B. 1995, 13, 30123016. Copyright 1995, American Institute of Physics. Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola 1. Hessel, V.; Lob, P.; Lowe, H. Curr Org Chem 2005, 9, 765. 2. Schwalbe, T.; Autze, V.; Hohmann, M.; Stirner, W. Org Process Res Dev 2004, 8, 440. 3. Jahnisch, K.; Hessel, V.; Lowe, H.; Baerns, M. Angew Chem Int Ed 2004, 43, 406.

6532

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 44 (2006)

4. Bayer, T.; Himmler, K. Chem Eng Technol 2005, 28, 285. 5. Hessel, V.; Lowe, H.; Stange, T. Lab Chip 2002, 2, 14N. 6. Sue, K.; Murata, K.; Kimura, K.; Arai, K. Green Chem 2003, 5, 659. 7. Kendall, D. P. J Press Vessel TechnolTrans ASME 2000, 122, 229. 8. Kenis, P. J. A.; Ismagilov, R. F.; Whitesides, G. M. Science 1999, 285, 83. 9. Choban, E. R.; Markoski, L. J.; Wieckowski, A.; Kenis, P. J. A. J. Power Sources 2004, 128, 5460. 10. Squires, T. M.; Quake, S. R. Rev Mod Phys 2005, 77, 977. 11. Yamashita, K.; Yamaguchi, Y.; Miyazaki, M.; Nakamura, H.; Shimizu, H.; Maeda, H. Anal Biochem 2004, 332, 274. 12. Rayleigh, L. Proc R Soc London 1879, 29, 71. 13. Plateau, J. Acad Sci Bruxelles Mem 1849, 23, 5. 14. Jillavenkatesa, A.; Dapkunas, S. J.; Lum, L. H. National Institute of Standards and Technology Special Publication 9601; U.S. Government Printing Ofce: Washington, DC, 2001; p 149. 15. Zheng, B.; Tice, J. D.; Ismagilov, R. F. Anal Chem 2004, 76, 4977. 16. Thorsen, T.; Roberts, R. W.; Arnold, F. H.; Quake, S. R. Phys Rev Lett 2001, 86, 4163. 17. Okushima, S.; Nisisako, T.; Torii, T.; Higuchi, T. Langmuir 2004, 20, 9905. 18. Nagaki, A.; Kawamura, K.; Suga, S.; Ando, T.; Sawamoto, M.; Yoshida, J. J Am Chem Soc 2004, 126, 14702. 19. Iwasaki, T.; Yoshida, J. Macromolecules 2005, 38, 1159. 20. Russum, J. P.; Jones, C. W.; Schork, F. J. Ind Eng Chem Res 2005, 44, 2484. 21. Russum, J. P.; Jones, C. W.; Schork, F. J. AIChE J 2006, 52, 1566. 22. Enright, T. E.; Cunningham, M. F.; Keoshkerian, B. Macromol Rapid Commun 2005, 26, 221. 23. Honda, T.; Miyazaki, M.; Nakamura, H.; Maeda, H. Lab Chip 2005, 5, 812. 24. Yamaguchi, Y.; Ogino, K.; Yamashita, K.; Maeda, H. J Chem Eng Jpn 2004, 37, 1265. 25. Wu, T.; Mei, Y.; Cabral, J. T.; Xu, C.; Beers, K. L. J Am Chem Soc 2004, 126, 9880. 26. Wu, T.; Mei, Y.; Xu, C.; Byrd, H. C. M.; Beers, K. L. Macromol Rapid Commun 2005, 26, 1037. 27. Xu, C.; Wu, T.; Drain, C. M.; Batteas, J. D.; Beers, K. L. Macromolecules 2005, 38, 6. 28. Sugiura, S.; Nakajima, M.; Itou, H.; Seki, M. Macromol Rapid Commun 2001, 22, 773. 29. Sugiura, S.; Nakajima, M.; Seki, M. Ind Eng Chem Res 2002, 41, 4043. 30. Nisisako, T.; Torii, T.; Higuchi, T. Chem Eng J 2004, 101, 23. 31. Dendukuri, D.; Tsoi, K.; Hatton, T. A.; Doyle, P. S. Langmuir 2005, 21, 2113. 32. Xu, S. Q.; Nie, Z. H.; Seo, M.; Lewis, P.; Kumacheva, E.; Stone, H. A.; Garstecki, P.; Weibel, D. B.;

33. 34. 35.

36.

37.

38. 39. 40. 41.

42. 43.

44.

45. 46.

47. 48. 49. 50.

51. 52. 53. 54. 55. 56.

Gitlin, I.; Whitesides, G. M. Angew Chem Int Ed 2005, 44, 724. Seo, M.; Nie, Z. H.; Xu, S. Q.; Lewis, P. C.; Kumacheva, E. Langmuir 2005, 21, 4773. Nie, Z. H.; Xu, S. Q.; Seo, M.; Lewis, P. C.; Kumacheva, E. J Am Chem Soc 2005, 127, 8058. Lewis, P. C.; Graham, R. R.; Nie, Z. H.; Xu, S. Q.; Seo, M.; Kumacheva, E. Macromolecules 2005, 38, 4536. Jeong, W. J.; Kim, J. Y.; Choo, J.; Lee, E. K.; Han, C. S.; Beebe, D. J.; Seong, G. H.; Lee, S. H. Langmuir 2005, 21, 3738. De Geest, B. G.; Urbanski, J. P.; Thorsen, T.; Demeester, J.; De Smedt, S. C. Langmuir 2005, 21, 10275. Nisisako, T.; Torii, T.; Takahashi, T.; Takizawa, Y. Adv Mater 2006, 18, 1152. Dendukuri, D.; Pregibon, D. C.; Collins, J.; Hatton, T. A.; Doyle, P. S. Nat Mater 2006, 5, 365. Utada, A. S.; Lorenceau, E.; Link, D. R.; Kaplan, P. D.; Weitz, D. A. Science 2005, 308, 537. Lorenceau, E.; Utada, A. S.; Link, D. R.; Cristobal, G.; Joanicot, M.; Weitz, D. A. Langmuir 2005, 21, 9183. Takeuchi, S.; Garstecki, P.; Weibel, D. B.; Whitesides, G. M. Adv Mater 2005, 17, 1067. Martin-Banderas, L.; Flores-Mosquera, M.; RiescoChueca, P.; Rodriguez-Gil, A.; Cebolla, A.; Chavez, S.; Ganan-Calvo, A. M. Small 2005, 1, 688. Martin-Banderas, L.; Rodriguez-Gil, A.; Cebolla, A.; Chavez, S.; Berdun-Alvarez, T.; Garcia, J. M. F.; Flores-Mosquera, M.; Ganan-Calvo, A. M. Adv Mater 2006, 18, 559. Quevedo, E.; Steinbacher, J.; McQuade, D. T. J Am Chem Soc 2005, 127, 10498. Steinbacher, J. L.; Moy, R. W. Y.; Price, K. E.; Cummings, M. A.; Roychowdury, C.; Buffy, J. J.; Olbricht, W. L.; Haaf, M.; McQuade, D. T. J Am Chem Soc 2006, 128, 9442. Zhao, B.; Viernes, N. O. L.; Moore, J. S.; Beebe, D. J. J Am Chem Soc 2002, 124, 5284. Jeong, W.; Kim, J.; Kim, S.; Lee, S.; Mensing, G.; Beebe, D. J. Lab Chip 2004, 4, 576. Madou, M. Fundamentals of Microfabrication; CRC: Boca Raton, FL, 1997. Kern, W.; Deckert, C. A. In Thin Film Processes; Vossen, J. L.; Werner, K., Eds.; Academic: New York, 1978; p 401. Jansen, H.; Gardeniers, H.; deBoer, M.; Elwenspoek, M.; Fluitman, J. J Micromech Microeng 1996, 6, 14. Becker, H.; Gartner, C. Electrophoresis 2000, 21, 12. Becker, H.; Locascio, L. E. Talanta 2002, 56, 267. Ehrfeld, W.; Lehr, H. Radiat Phys Chem 1995, 45, 349. Qin, D.; Xia, Y. N.; Whitesides, G. M. Adv Mater 1996, 8, 917. McCormick, R. M.; Nelson, R. J.; Alonso-Amigo, M. G.; Benvegnu, J.; Hooper, H. H. Anal Chem 1997, 69, 2626.

Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

HIGHLIGHT

6533

57. Arnold, J.; Dasbach, U.; Ehrfeld, W.; Hesch, K.; Lowe, H. Appl Surf Sci 1995, 86, 251. 58. Martynova, L.; Locascio, L. E.; Gaitan, M.; Kramer, G. W.; Christensen, R. G.; MacCrehan, W. A. Anal Chem 1997, 69, 4783. 59. Xu, J. D.; Locascio, L.; Gaitan, M.; Lee, C. S. Anal Chem 2000, 72, 1930. 60. Ueno, K.; Kitagawa, F.; Kim, H. B.; Tokunaga, T.; Matsuo, S.; Misawa, H.; Kitamura, N. Chem Lett 2000, 858. 61. Chou, S. Y.; Krauss, P. R.; Renstrom, P. J. Appl Phys Lett 1995, 67, 3114. 62. Xia, Y. N.; Whitesides, G. M. Angew Chem Int Ed 1998, 37, 551. 63. Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974. 64. Lee, J. N.; Park, C.; Whitesides, G. M. Anal Chem 2003, 75, 6544. 65. Dietrich, T. R.; Freitag, A.; Scholz, R. Chem Eng Technol 2005, 28, 477. 66. Lorenz, H.; Despont, M.; Fahrni, N.; LaBianca, N.; Renaud, P.; Vettiger, P. J Micromech Microeng 1997, 7, 121. 67. Lorenz, H.; Despont, M.; Fahrni, N.; Brugger, J.; Vettiger, P.; Renaud, P. Sens Actuators A 1998, 64, 33. 68. Lee, K. Y.; LaBianca, N.; Rishton, S. A.; Zolgharnain, S.; Gelorme, J. D.; Shaw, J.; Chang, T. H. P. J Vac Sci Technol B 1995, 13, 3012.

69. Renaud, P.; van Lintel, H.; Heuschkel, M.; Guerin, L. In Proceedings of Micro-TAS 98, Banff, Canada, 1998. 70. Roberts, M. A.; Rossier, J. S.; Bercier, P.; Girault, H. Anal Chem 1997, 69, 2035. 71. Pethig, R.; Burt, J. P. H.; Parton, A.; Rizvi, N.; Talary, M. S.; Tame, J. A. J Micromech Microeng 1998, 8, 57. 72. Ford, S. M.; Kar, B.; McWhorter, S.; Davies, J.; Soper, S. A.; Klopf, M.; Calderon, G.; Saile, V. J Microcolumn Sep 1998, 10, 413. 73. Ford, S. M.; Davies, J.; Kar, B.; Qi, S. D.; McWhorter, S.; Soper, S. A.; Malek, C. K. J Biomech EngTrans ASME 1999, 121, 13. 74. Ehrfeld, W.; Munchmeyer, D. Nucl Instrum Methods Phys Res Sect A 1991, 303, 523. 75. Vasile, M. J.; Friedrich, C. R.; Kikkeri, B.; McElhannon, R. Precis EngJ Am Soc Precis Eng 1996, 19, 180. 76. Friedrich, C. R.; Vasile, M. J. J Microelectromech Syst 1996, 5, 33. 77. Kussul, E.; Baidyk, T.; Ruiz-Huerta, L.; CaballeroRuiz, A.; Velasco, G.; Kasatkina, L. J Micromech Microeng 2002, 12, 795. 78. Schubert, K.; Brandner, J.; Fichtner, M.; Linder, G.; Schygulla, U.; Wenka, A. Microscale Thermophys Eng 2001, 5, 17. 79. Bartholomeusz, D. A.; Boutte, R. W.; Andrade, J. D. J Microelectromech Syst 2005, 14, 1364.

Journal of Polymer Science: Part A: Polymer Chemistry DOI 10.1002/pola

Você também pode gostar