Teflon-Coated Silicon Microreactors: Impact on Segmented Liquid−Liquid Multiphase Flows

rXXXX American Chemical Society A dx.doi.org/10.1021/la2004744 | Langmuir XXXX, XXX, 000–000

ARTICLE

pubs.acs.org/Langmuir

Teflon-Coated Silicon Microreactors: Impact on SegmentedLiquid�Liquid Multiphase FlowsSimon Kuhn, Ryan L. Hartman,† Mahmooda Sultana, Kevin D. Nagy, Samuel Marre,‡ and Klavs F. Jensen*

Department of Chemical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts02139, United States

’ INTRODUCTION

Silicon-based microreactors are useful for chemical synthesisin a microfluidic system by enabling glassware-like compatibilityand access to elevated temperature and pressure conditions noteasily accessed in batch.1�4 However, having fluorous hydrophobicsurface properties is also useful in (i) limiting deposition or growthof particulate matter,5�10 (ii) minimizing axial dispersion,8 and(iii) enabling reaction screening.11 Thus, there is an interest inbeing able to combine the advantage of silicon/glass microflui-dics for chemical synthesis with the ability to integrate fluoro-polymer surfaces.

Fluorinated ethylene�propylene (FEP) and poly(tetrafluo-rethylene) (PTFE) Teflon films have previously been employedin membrane valves and pumps in glass microfluidic devices.12,13

Films of FEP and PTFE can also bemodified and bonded to glassand/or poly(dimethylsiloxane) (PDMS).12,13 Chemical inert-ness to nonaqueous solvents makes FEP and PTFE goodalternatives to PDMS components.14 However, modification ofsilicon microfluidic features with fluoropolymers, such as FEPand PTFE, remains a challenge. Capping of Teflon-coated waferswith wafer-etched features only partially modifies a microchannelwith fluoropolymer, and the bond strength of Teflon-coatedwafers also is significantly reduced when compared to anodicbonding of Pyrex to silicon oxide surfaces.

Poly(tetrafluoroethylene-co-2,2-bis(trisfluoromethyl)-4,5-difluoro-1,3-dioxole) (or Teflon AF) exhibits chemical compatibilitysimilar to that of FEP and PTFE.15�17 Cured films of Teflon AFare optically transparent, and the fluoropolymer is soluble in

selected solvents. Microsensors that rely on optical measure-ments of a liquid phase have been fabricated by taking advantageof Teflon AF properties.18,19 Glass transition temperaturesbetween 160 and 240 �C and good solubility in fluorous solventsmake Teflon AF suitable for spin coating on glass and siliconwafers.18�20 A pretreatment with fluorosilane allows the TeflonAF solution to wet these surfaces and improves the adhesion.PDMS and silicon�glass microfluidic channels can be coatedwith Teflon fluoropolymers employing this technique.21�23

Teflon AF (and other fluoropolymers) can also be made porous,which enables transport of gases across films.24�26 Gases that aregenerated in situ can, therefore, also be removed during chemicalreaction.27

Surface wettability and adhesion are central characteristics ofmicrofluidic systems that further motivate fluoropolymer inte-gration. Solids formed during reaction in microsystems can bringabout device clogging. Minimizing particle-to-wall interactionsunder flow reduces plugging or channel constriction.7,8 Forexample, surfaces that are not wetted by reaction solvents enabledroplet-based microfluidics. Liquid slugs in gas�liquid multi-phase flows in hydrophobic channels are effectively separated.28

Furthermore, the communication between the liquid segments isreduced, resulting in no dispersion in the coated microchannels.29

Though silicon and glass surfaces are naturally hydrophilic,

Received: February 6, 2011Revised: March 24, 2011

ABSTRACT: We describe fluoropolymer modification of sili-con microreactors for control of wetting properties in chemicalsynthesis applications and characterize the impact of the coatingon liquid�liquid multiphase flows of solvents and water.Annular flow of nitrogen gas and a Teflon AF (DuPont)dispersion enable controlled evaporation of fluoropolymersolvent, which in turn brings about three-dimensional polymerdeposition on microchannel walls. Consequently, the wettingbehavior is switched from hydrophilic to hydrophobic. Analysisof microreactors reveals that the polymer layer thickness increases down the length of the reactor from ∼1 to ∼13 μm with anaverage thickness of∼7 μm. Similarly, we show that microreactor surfaces can be modified with poly(tetrafluoroethylene) (PTFE).These PTFE-coated microreactors are further characterized by measuring residence time distributions in segmented liquid�liquidmultiphase flows, which display reduced axial dispersion for the coated microreactors. Applying particle image velocimetry, changesin segment shape and velocity fluctuations are observed resulting in reduced axial dispersion. Furthermore, the segment sizedistribution is narrowed for the hydrophobic microreactors, enabling further control of residence distributions for synthesis andscreening applications.

B dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

surface modification with a fluorosilane creates hydrophobic surfacecharacteristics.30 Fluorosilane modification of silicon oxide sur-faces is only temporary, however, as aqueous and organic acidsand bases readily remove such treatments and further react withthese substrates.

In this work, we present amethod of preparing fluoropolymer-modified, silicon-based microreactors subsequent to devicefabrication. Surface modification with fluorosilane facilitates wettingand adhesion of Teflon solutions in the microchannels, which aresubsequently coated with Teflon AF and PTFE using gas�liquidflows. Furthermore, we explore the influence of the surfacewettability on segmented liquid�liquid multiphase flow byresidence time analysis and particle image velocimetry (PIV).The hydrophobic surface decreases communication betweenaqueous segments and thus reduces axial dispersion. Moreinsights into these findings are provided by applying microparticle image velocimetry and laser-induced fluorescence.

’EXPERIMENTAL SECTION

Silicon-Fabricated Microreactors. Microchannel devices werefabricated from a double-polished silicon wafer and a Pyrex wafer (both150 mm in diameter and 650 μm thick). The fabrication processinvolved several photolithography steps, deep reactive ion etching ofsilicon, and growth of silicon oxide (0.5 μm). Pyrex anodic bondingcapped silicon oxide features and thus enabled a microfluidic device.Two different devices were fabricated: (1) amicrochannel (0.15� 0.2�150 mm) used to study surface modification and the Teflon AF coatingprocess and (2) a microchannel (0.2� 0.4� 750 mm) used in carryingout thickness experiments. Standard machining techniques were em-ployed to produce a compression chuck (316 stainless steel) for eachmicroreactor, which was cooled (to 20 �C) using a Thermo ScientificNESLAB RTE-7 refrigerating bath. The microreactor inlets and outletswere compressed in this chuck. The temperature in the main channel ofeach device was driven by an Omega 120 V cartridge heater controlledby an Omega CN9000 series PID controller. An aluminum chuck wasdesigned to maintain the process temperature and contact the silicondevice surface.Fluorosilane Modification. Fluoropolymer coating of silicon

microreactors was made possible by first modifying the surface with afluorosilane. The treatment procedure applied involved several steps,modified from a previously discussed procedure.30 Fabricated deviceswere injected with the sequence of solvents (at 30 μL/min) prescribedin Table 1. All liquids were loaded into Hamilton Gastight syringes(10 mL) and delivered using a syringe pump (Harvard Apparatus).

Upon completion of the last deionized wash, silicon microreactors weretransferred to an oven at 114 �C and baked overnight.Fluoropolymer Thickness Approximation. Coating thick-

nesses were approximated by bonding (with high-temperature epoxy)a rectangular Pyrex piece of known dimensions to the face of themicroreactor. Water was injected using a Hamilton Gastight syringe(100 μL) and syringe pump (Harvard Apparatus). The air�watermeniscus was monitored throughout injection into the device. Theedges of the Pyrex piece were used as starting and stopping points forincremental volume measurements. The serpentine geometry of thereactor thus enabled estimation of the volume as a function of the totalreactor length. Experiments were repeated before and after Teflon AFcoating.Reagents and Analytical Procedure. All experiments were

carried out using reagent grade solvents. Fluorosilane solutions wereprepared by volumetric addition of (1H,1H,2H,2H-perfluorodecyl)tri-chlorosilane (30μL) (fromAlfa Aesar) with hexane (15mL) in amoisture-controlled glovebox. Teflon coating experiments were carried out usingTeflon AF 1600 grade (Tg = 160 �C) or PTFE (TE-3859, Tg = 337 �C)obtained from DuPont. Gas�liquid flows were established with indus-trial grade nitrogen gas (Airgas), which was regulated to the desiredpressure for each test. Images of the coating process were captured with ahigh-speed color CCD camera (JAI CV-S3200 series). Hexane andfluorescein dissolved in water were used to demonstrate the switchingbehavior of segments and to approximate the contact angles of differentreactor surfaces. Images used in contact angle estimation were capturedwith a Canon PowerShot S5IS camera fitted on a Diagnostic Instru-ments Leica MZ12 microscope. Toluene and water were used to createthe multiphase flow for the residence time distribution measurementsusing 0.1 wt % sodium benzoate dissolved in water as the pulse-injectedtracer. The absorption of this tracer at a wavelength of 269 nm wasrecorded via anOceanOptics HR2000CG detector as a function of time.This measurement was made possible by a Z-type flow cell downstreamof the microreactor. Particle image velocimetry measurements wererealized using an inverted fluorescence microscope (Zeiss Axiovert 200).The light source was provided by a frequency-doubled Nd:YAG laser(BigSky Ultra CFR, 30mJ, 532 nm), and the images were recorded usinga dual-frame charge-coupled device camera (PCOSensicamQE, 1376�1024 pixels2, 8bit). The same setup was used for laser-induced fluores-cence measurements to investigate the concentration distribution of apulse-injected tracer dye (Rhodamine 6G) in the aqueous phase. All liquidswere loaded into Hamilton Gastight syringes (10 mL) and deliveredusing a syringe pump (Harvard Apparatus).

’RESULTS AND DISCUSSION

Surface Modification and Contact Angle Measurements.Fluorosilane-modified microreactors were coated with TeflonAF by evaporation of fluoropolymer solvent (FC-75 manufac-tured by 3M). Removal of the solvent brought about the depositionof fluoropolymer on microchannel walls. This deposition wasaccomplished in a microfluidic system using a gas�liquid flowconsisting of an inert gas phase (e.g., nitrogen) and Teflon AFsolution. In general, multiphase flows are useful to a broad rangeof microfluidic applications and enable precise control over masstransport between immiscible phases.31�38

Teflon AF solution was first injected into a microreactor(at 30 μL/min). Upon injecting one residence volume of liquid,nitrogen gas was delivered to the microchannel as shown inFigure 1 (1 min). The wetting nature of the microchannelresulted in liquid fluoropolymer flowing along the wall, whereasgas flow took place in the center of the channel. This annular-type flow enabled evaporation of the solvent from the wall layer,

Table 1. Fluorosilane Modification Recipea

step treatment time (min)

1 deionized water 10

2 hydrogen peroxide 15

3 deionized water 15

4 2-propanol 15

5 hexane 15

6 hexane�fluorosilane mixture 30

7 hexane 15

8 2-propanol 15

9 deionized water 15aThe table shows the steps followed to modify silicon microreactorswith (1H,1H,2H,2H-perfluorodecyl)trichlorosilane. All steps were in-jected at 30 μL/min.

C dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

which in turn increased the apparent liquid viscosity. Increasingthe gas flow rate led to the shearing of liquid along the wall asshown by the waves formed in Figure 1 after 5 min. Initially, thenitrogen pressure at the inlet of the microreactor was set to 1 psi,which was then steadily increased to 60 psi, and the reactor outletwas kept at ambient pressure. Prolonged exposure to the gas flowresulted in a thin, smooth fluoropolymer coating on channelsurfaces (see Figure 1 at 15 min).Most of the fluoropolymer solvent was evaporated by flowing

nitrogen under ambient temperature for 4 h. The remainingsolvent was removed by subsequently heating the microreactorabove the solvent boiling point (i.e., >102 �C). The solvent wasevaporated as fresh nitrogen gas flowed continuously over thechannel surfaces for an additional 4 h. To ensure adhesion of thefilm to the microreactor walls, the temperature was increased toabove the glass transition temperature of the polymer (i.e.,165 �C) and held constant for 1 h. The entire process was repeatedtwo more times to ensure all channel surfaces had been coated.Gas�liquid flow was further applied to create PTFE-coated

microreactors. In this case, the crystalline melting point of thefluoropolymer (∼337 �C) was considerably higher than that ofTeflon AF (∼160 �C), which is useful for high-temperatureapplications. The PTFE resin took the form of an aqueous dis-persion (60% solid content) of submicrometer particles rangingin size from ∼0.1 to 0.3 μm.Microscopy images (Figure 2) clearly show the difference

between silicon oxide- and PTFE-coated microreactors duringthe injection of hexane�water segmented flow. It should benoted that the milky color enhanced under the microscope isinvisible to the naked eye. Consequently, the PTFE does not

prevent observation of microreactor flows through the transpar-ent Pyrex top. As an example, a segmented flow consisting ofhexane dispersed in water (with fluorescein as the dye) showsthat the silicon oxide surface is readily wetted by water, causing thewater segments to be connected through thinmenisci (Figure 2, topright). After surface modification with PTFE the organic phasewets the walls, isolating the fluorescently labeled water in individualsegments (Figure 2, right bottom). This ability to switch wettingcharacteristics is particularly advantageous in microscale flows.Further analysis of the contact angles was carried out by analysis

of microscope photographs, analogous to those presented inFigure 2. Contact angles of water and hexane on silicon oxide-,fluorosilane-, Teflon AF-, and PTFE-coated microreactors wereestimated and are reported in Table 2. As shown in this table, thecontact angle of water estimated for silicon oxide was 41� ( 6�.Contact angles vary depending on several factors, such as surfaceroughness, impurities, chemistry, and temperature. The advan-cing contact angle of water on thermally grown silicon oxide (i.e.,5000 Å) has previously been reported39 as 46.7�( 1.6�, which isconsistent with our results. In the case of monolayer fluorosilanemodification, the surface roughness is not expected to signifi-cantly change by comparison to that of thermally grown oxide.Therefore, contact angle measurements are primarily a functionof the surface chemistry. After treatment of the microreactorsurfaces with fluorosilane, Teflon AF, or PTFE, water no longerwets the walls. The hexane contact angle is found in Table 2 andshows a value of 66�( 4� on fluorosilane (i.e., using (1H,1H,2H,2H-perfluorodecyl)trichlorosilane). The wetting nature of Tef-lon AF surfaces is further complicated by the fluoropolymerstructure. It is readily known that Teflon AF films can be madeporous,24,25 which can influence contact angle measurements.Moreover, recent investigations have shown that fluoropolymermaterials can be used to create superhydrophobic surfaces depend-ing on the surface topology.40 Consequently, we anticipate thatfluoropolymer surfaces exhibiting a range of contact angles are

Figure 1. Sequence of images taken during the coating of silicon microchannels with liquid Teflon AF. The images show the addition of inert nitrogenunder low-pressure conditions to establish annular flow (after 1 min), increased nitrogen pressure leading to the shearing of liquid Teflon along the wall(5 min), and the final thin film resulting from high shear and solvent evaporation (15 min).

Figure 2. Microscope photographs of silicon oxide- and PTFE-coatedmicroreactors during two-phase liquid�liquid flow of hexane and water.The water segments are dyed using fluorescein. The images show thatmodifying the surface with a fluoropolymer enables switching of thewetting from hydrophilic to hydrophobic throughout the microreactor.

Table 2. Contact Angle Measurements of Water and Hexaneon Treated Microchannel Surfaces

contact angle ( SDa

surface water hexane

silicon oxide 41 ( 6 nw

(1H,1H,2H,2H-perfluorodecyl)trichlorosilane nwb 66 ( 4

Teflon AF (1600) nw 59 ( 3

PTFE nw 51 ( 2a SD = standard deviation of surface measurements. bNonwetting.

D dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

possible. In our system, the above-mentioned procedure forcreating coated microreactors resulted in a hexane contact angleof 59� ( 3�. Others have reported a hexane contact angle onTeflon AF of 40.3�.41 This difference in wetting nature can beexplained by the porosity of the deposited film, or by an increasein surface roughness. For the PTFE-coated surface a value of thehexane contact angle of 51�( 2� is found, which is in agreementwith the literature.42

Further experiments were carried out to investigate the influenceof temperature and chemistry on the fluoropolymer-modifiedmicroreactors. Deionized water was injected into a microreactor(at 30 μL/min), heated to 80.0 �C, and held constant at thattemperature for 4 h. No apparent change in wetting characteristicsor fluoropolymer coating was observed throughout the experiment.Next, a multiphase segmented flow was established in the micro-reactor by injecting toluene and deionized water each at 15 μL/min(i.e., 30 μL/min total). The reactor temperature was increased to80.0 �C and held constant for an additional 4 h. The fluoro-polymer modification upheld the higher temperature conditionssince no change in wetting behavior was observed throughout thecourse of the experiment: water segments remained dispersed ina continuous organic phase.A segmented flow comprised of 2.0 M potassium hydroxide

dispersed in 1,4-dioxane yielded similar results. Exposure of themicroreactor for 6 h (at 20.0 �C and 30 μL/min) had no influenceon the hydrophobic wetting characteristics. It is readily knownthat potassium hydroxide attacks silicon and oxide surfaces, whichultimately leads to device failure.43 The ability to switch the wettingcharacteristics prevented this attack under ambient conditions.Nevertheless, Teflon AF is not entirely inert to chemistry. Theadsorption of hydroxide ions and electrolytes onto and transportof organic molecules through Teflon AF fluoropolymer chainshas been elucidated by the work of others.44�46 Periodic wallcontact by drying out of one phase could eventually lead toexposure. This mechanism has also brought about deposition ofnanoparticles during multiphase flows.36 Cracks in the polymerlayer or diffusion through them could bring about accumulationof compounds on microreactor walls during chemical transfor-mations, which may or may not be desirable. The above-mentionedexperiments were repeated for PTFE-coated microreactors, andthe results were identical.Fluoropolymer Layer Thickness. Microfluidic coatings are

relevant to a broad range of applications, and several techniqueshave been employed to approximate coating thicknesses.47�50 Adiesaw or other cutting methods can be used to reveal channelcross sections visible by microscopy, but with the risk of damagingfragile polymer films. Bulk measurements, such as pressure dropdifferences, can also elucidate coating thicknesses. Estimation ofpressure drop across an entire channel, however, only gives

knowledge of an average thickness with missing informationabout variations. Alternatively, we approximated the Teflon AFcoating thickness as a function of the microreactor length bymeasuring incremental differences in the reactor volume beforeand after coating.One observes in Figure 3a,b an apparent difference in the wall

thickness between the microreactor inlet and outlet after applica-tion of three coatings. This difference was revealed by plottingthe change in reactor volume as a function of the reactor lengthbefore and after fluoropolymer coating (Figure 4). The differ-ence in cumulative reactor volume (before and after coating)increased nonlinearly down the length of themicroreactor, whichcorresponds to an increase in the wall coating thickness. For asquare microchannel, a coating thickness was approximated onthe basis of this change in volume. As shown in Figure 4, thecoating thickness appeared to increase from∼1 to∼13 μm fromthe microreactor inlet to the outlet. These observations are con-sistent with more than one possible coating mechanism. Theshear at the gas�liquid interface could bring about a layer-by-layer coating process. For example, most of the fluoropolymer atthe inlet is transported along the wall and eventually recoatslayers that had already been deposited downstream. This processcould be further driven by differences in evaporation rates. Theannular headspace along the wall is expected to be the leastsaturated with volatile solvent at the beginning and the mostsaturated by the end of the microreactor. Consequently, a phaseequilibrium may exist, and the composition of the solvent in thevapor and liquid phases could significantly change down thelength of the reactor. The average thickness over the entirereactor length was estimated to be∼7 μm. The average thicknesswas later estimated to be 9.0( 3.8 μmby the injection of a tracer(e.g., 0.01 vol % toluene in heptane) into a coated and uncoatedmicroreactor at 50 μL/min. As is evident in Figure 3c, a mixingzone (near the inlet) comprised of a serpentine geometry wasalso modified with the fluoropolymer.Residence Time Distributions. Multiphase transformations,

such as gas�liquid, liquid�liquid, and solid�liquid, and combi-nations thereof, help overcome challenges in achieving effectivecontact between phases in microfluidic systems.35,36 The switch-ing of the wetting characteristics of a microchannel from hydro-philic to hydrophobic would enable broad control over droplet-based microfluidics.7,8 Furthermore, phase switching also has

Figure 3. Microscope images of the microchannels coated with fluoro-polymer. The images show apparent differences in layer thickness from(a) inlet to (b) outlet and (c) themicroreactormixing zone comprised ofa serpentine geometry.

Figure 4. Cumulative differences in microreactor volume before andafter fluoropolymer coating. A coating thickness approximation as afunction of the reactor length is also given.

E dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

implications on how chemical reactions take place in two-phaseliquid�liquid flow. When a reaction is carried out in the con-tinuous phase, the fluid regions between each dispersed segmentare able to communicate with each other through axial dis-persion,51 and partially changing the wetting characteristics helpsto reduce this.29 This dispersion, however, can change the reactantconcentration at any given point in the reactor and thus maskintrinsic kinetic parameters. Switching the segments eliminatescommunication and the need for dispersion models, which hasthe potential to impact the discovery and development of finechemicals.52 For example, researchers have recently shown thatautomated microfluidic platforms can be applied to screen a broadrange of important reactions, such as bicyclo[3.2.1]octanoid scaf-folds.53 Fluoropolymer-coated microreactors could potentiallyimprove the screening rate by taking advantage of multiphase flows.We investigated the residence time distribution (RTD) of a

pulse-injected tracer (sodium benzoate dissolved in water) intoluene�water multiphase flow in oxide- and PTFE-coated micro-reactors by applying UV�vis spectroscopy. The microreactorsused in this study had inner dimensions of 400 μm� 400 μmanda total length of L = 0.75 m. Two different flow rates were studied:(1) 5 μL/min for each phase (total flow rate of 10 μL/min) and(2) 12.5 μL/min (total flow rate of 25 μL/min).The postprocessed RTD curves are depicted as closed symbols

in Figure 5. One can observe that for a constant total flow rate(10μL/min for Figure 5a,c; 25 μL/min for Figure 5b,d) the RTDcurves are much narrower for the Teflon-coated microreactors.In addition, the mean residence time (given in Table 3) is increasedfor these cases. Both results are explained by the eliminated me-niscus in the case of the hydrophobic surface and the correspond-ing reduction in communication between the water segments(Figure 6). The oxide microreactor is wetted by the water phase,forming a meniscus that allows communication between theindividual water segments. Toluene is the dispersed phase and isnot in contact with the wall. In the case for the Teflon-coatedmicroreactor (Figure 6, bottom), the toluene phase is now the

wall-wetting fluid, and hence, the communication between thewater segments is significantly reduced, resulting in narrower res-idence time distributions.To further analyze the experimental RTDdata, we apply a one-

dimensional axial dispersion model of the form54,55

EðθÞ ¼ 1ffiffiffiffiffiffiffiffiffiffiffiffiffi4πθ

DuL

r exp �ð1� θÞ2

4θDuL

2664

3775 ð1Þ

where D/uL is the vessel dispersion number, with the dispersioncoefficientD, the superficial velocity of the water segments u, andthe length of the reactor L. The nondimensional time θ is

Figure 5. Residence time distributions for the silicon oxide (a, b) and PTFE (c, d) coatedmicroreactors. Results are shown for a total flow rate of 10 and25 μL/min, respectively. The solid lines represent the fit of the axial dispersion model (eq 1).

Table 3. Mean Residence Times and Dispersion Coefficientsfor Silicon- and PTFE-Coated Microreactors

flow rate (μL/min) residence time (s) dispersion coefficient (m2/s)

10 (silicon) 1725( 12 2.62� 10�6

10 (PTFE) 1774( 25 1.87� 10�6

25 (silicon) 668( 2 6.79� 10�6

25 (PTFE) 693( 8 4.91� 10�6

Figure 6. Sketch of the switching phase behavior depending on thewettability of the surface. The top row depicts the situation for the oxidemicroreactor and the bottom row for the PTFE-coated microreactor.

F dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

obtained by dividing the measurement time with the mean res-idence time:

θ ¼ tτ

ð2Þ

For a given flow rate and microreactor geometry, the onlyparameter in eq 1 is the dispersion coefficient D, and thus, usingthis value, the equation can be fitted to the experimental RTDcurves (shown as solid lines in Figure 5). The resulting dispersioncoefficients are tabulated in Table 3. For both flow rates, adecrease of the dispersion coefficient of 40% is observed for thePTFE-coated microreactor, demonstrating conclusively that thephase switching behavior in multiphase flows reduces the com-munication between the water segments and therefore also theaxial dispersion.The scalar dispersion in the aqueous phase is further investi-

gated by laser-induced fluorescence. Similar to the RTD mea-surements, we use a pulse-injected tracer dye (aqueous solutionof Rhodamine 6G) and record its concentration at two positionsclose to the inlet and the outlet of the microreactor. Figure 7(left) depicts the same water segment after entering the siliconoxide microreactor (top row) and just before leaving it (bottomrow). A decrease in the fluorescent intensity is observed, whichmeans that the tracer concentration of this water segment isreduced due to axial dispersion. In contrast, no reduction offluorescence intensity and therefore concentration between thereactor inlet and outlet is observed in the water segment for thePTFE microreactor (Figure 7, right). These results highlight thesuppressed communication between the aqueous segments forthe hydrophobic surface.Particle ImageVelocimetry.We appliedmicro particle image

velocimetry56,57 to investigate the spatial variation of the waterphase velocity. For the oxide microreactor, the water phase alwayswets the wall (Figure 8). Consequently, the toluene segmentshave less cross-sectional area to flow through the microreactorand therefore travel faster compared to the water phase. Thisphenomenon becomes visible in the water phase surrounding theaccelerated toluene segments as increased velocity fluctuations,which enhance the communication between the water segmentsand therefore also the axial dispersion. The same observationscan bemade for the higher total flow rate of 25 μL/min (Figure 8,

bottom), except that the acceleration of the water phase adjacentto the toluene segment is more pronounced, which results inadditional axial dispersion.Figure 9 shows the velocity fields for the hydrophobic micro-

reactor in which the toluene phase now wets the wall of themicroreactor, reducing the communication between the watersegments. An interesting behavior is observed for the velocityfield in the water phase. The maximum values for the streamwisevelocity are found close to the walls and not in the center of themicrochannel. In addition, the wall is still wetted by a toluenelayer with a layer thickness estimated to be on the order of 20 μm.Due to the difference in dynamic viscosity (toluene, 0.6 mPa s at20 �C;water, 1.0mPa s at 20 �C), the toluene phase flows at fastervelocities at the same pressure drop. Since the wall is covered withtoluene, the water phase adjacent to it will also be accelerated.In addition, we used the PIV images to address the segment

size distribution for both microreactors. The results in Table 4show that the toluene segments have a larger size in the oxide

Figure 7. Rhodamine 6G dyed water segments for the siliconmicroreactor (left) and PTFEmicroreactor (right). The top row shows the water segmentat the inlet of the reactor, and the bottom row corresponds to the outlet. A change in fluorescent intensity, i.e., concentration, is observed for the siliconmicroreactor (wetting) (left) due to exchange with neighboring segments, whereas no change in fluorescent intensity is observed for the PTFE reactor(nonwetting) (right) as the water segments do not communicate.

Figure 8. Velocity vector field around a toluene segment for thehydrophilic microreactor at a total flow rate of 10 μL/min (top) and25 μL/min (bottom). The contours represent the levels of the rootmean square of the streamwise velocity component. The flow directionis from left to right.

G dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

reactor, but they also exhibit a larger standard deviation. For thePTFE-coated microreactor the segment size distribution is muchnarrower. One also observes in Table 4 that increasing the flowrate directly influenced the segment size, which is consistent withthe work of others.58

’CONCLUSIONS

Surface modification using multiphase flow enabled fluoro-polymer�silicon hybrid microreactors. Such devices were modifiedsubsequent to microreactor fabrication. A fluorosilane modifica-tion enabled microreactor surfaces wetted by Teflon AF andPTFE dispersions. Subsequent coating of the surfaces was madepossible by establishing annular-type flow of nitrogen gas andfluoropolymer dispersed liquids. Evaporation of solvent broughtabout the deposition of fluoropolymer on the microchannelwalls, which took a form of the channel geometry. Consequently,the wetting characteristics of the microreactors were switchedfrom hydrophilic to hydrophobic.

Analysis of Teflon AF-coated microreactors revealed that thepolymer coating thickness increased down the length of a reactorfor the case of multiple coatings. Polymer thickness approxima-tions gave values in the range of 1�13 μm. Microreactor wettingcharacteristics were preserved upon exposure to organic solventsand water at higher temperatures and a potassium hydroxide seg-mented flow at ambient temperature conditions. The lessonslearned were further exploited to engineer PTFE-coated micro-reactors. Residence time distribution measurements showed adecrease of the dispersion coefficient of 40% in the hydrophobicmicroreactor. Laser-induced fluorescence measurements confirmed

that the concentration of the aqueous segments does not changewhile they travel through the PTFE-coated microreactor. Usingparticle image velocimetry, the phase switching behavior and thechange of segment shape were elucidated, and the results showthat the increase in axial dispersion can be related to streamwisevelocity fluctuations induced by the toluene segments. Further-more, the segment size distribution is narrower in the case ofhydrophobic microreactor surfaces. Both findings allow furthercontrol of multiphase flows for applications by minimizing particu-late deposition on surfaces and enabling reaction screening.

’AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected]. Phone: (617) 253-4589. Fax: (617)298-8992.

Present Addresses†Department of Chemical & Biological Engineering, Universityof Alabama, Box 870203, Tuscaloosa, AL 35487.‡CNRS, Universit�e de Bordeaux, ICMCB, 87 avenue du docteurAlbert Schweitzer, 33608 Pessac, France.

’ACKNOWLEDGMENT

We thank the Novartis-MIT Center for Continuous Manu-facturing, the U.S. National Science Foundation (Grant CHE-650714189), and the U.S. Army Research Office through theInstitute for Soldier Nanotechnology (Grant DAAD-19-02-00020). S.K. acknowledges funding from the Swiss NationalScience Foundation (SNF).

’REFERENCES

(1) Jensen, K. F. Silicon-based microchemical systems: Character-istics and applications. MRS Bull. 2006, 31 (2), 101–107.

(2) Hartman, R. L.; Jensen, K. F. Microchemical systems for con-tinuous-flow synthesis. Lab Chip 2009, 9, 2495–2507.

(3) Marre, S.; Adamo, A.; Basak, S.; Aymonier, C.; Jensen, K. F. Designand packaging of microreactors for high pressure and high temperatureapplications. Ind. Eng. Chem. Res. 2010, 49 (22), 11310–11320.

(4) Marre, S.; Jensen, K. F. Synthesis of micro and nanostructures inmicrofluidic systems. Chem. Soc. Rev. 2010, 39 (3), 1183–1202.

(5) Shestopalov, I.; Tice, J. D.; Ismagilov, R. F. Multi-step synthesisof nanoparticles performed on millisecond time scale in a microfluidicdroplet-based system. Lab Chip 2004, 4 (4), 316–321.

(6) Hartman, R. L.; Naber, J. R.; Zaborenko, N.; Buchwald, S. L.;Jensen, K. F. Overcoming the challenges of solid bridging and constric-tion during Pd-catalyzed C-N bond formation in microreactors. Org.Process Res. Dev. 2010, 14, 1347–1357.

(7) Poe, S. L.; Cummings, M. A.; Haaf, M. R.; McQuade, D. T.Solving the clogging problem: Precipitate-forming reactions in flow.Angew. Chem., Int. Ed. 2006, 45 (10), 1544–1548.

(8) Song, H.; Chen, D. L.; Ismagilov, R. F. Reactions in droplets inmicroflulidic channels. Angew. Chem., Int. Ed. 2006, 45 (44), 7336–7356.

(9) Song, H.; Tice, J. D.; Ismagilov, R. F. A microfluidic system forcontrolling reaction networks in time. Angew. Chem., Int. Ed. 2003, 42(7), 768–772.

(10) No€el, T.; Naber, J. R.; Hartman, R. L.; Mcmullen, J. P.; Jensen,K. F.; Buchwald, S. L. Palladium-catalyzed amination reactions in flow:Overcoming the challenges of clogging via acoustic irradiation.Chem Sci.2011, 2, 287–290.

(11) Kreutz, J. E.; Shukhaev, A.; Du, W.; Druskin, S.; Daugulis, O.;Ismagilov, R. F. Evolution of catalysts directed by genetic algorithms in aplug-based microfluidic device tested with oxidation of methane byoxygen. J. Am. Chem. Soc. 2010, 132 (9), 3128–3132.

Figure 9. Velocity vector field in the water segment surrounded by thetoluene phase for the hydrophobic microreactor at a total flow rate of10 μL/min (top) and 25 μL/min (bottom). The contours represent thelevels of the root mean square of the streamwise velocity component.The flow direction is from left to right.

Table 4. Segment Size Distribution for Silicon- and PTFE-Coated Microreactors

flow rate (μL/min) slug length (mm)

10 (silicon) 27.7( 11.2

10 (PTFE) 13.4( 0.4

25 (silicon) 15.3( 8.1

25 (PTFE) 7.8( 0.6

H dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

(12) Willis, P. A.; Hunt, B. D.; White, V. E.; Lee, M. C.; Ikeda, M.; Bae,S.; Pelletier, M. J.; Grunthaner, F. J. Monolithic Teflon (R) membranevalves and pumps for harsh chemical and low-temperature use. Lab Chip2007, 7 (11), 1469–1474.(13) Grover, W. H.; von Muhlen, M. G.; Manalis, S. R. Teflon films

for chemically-inert microfluidic valves and pumps. Lab Chip 2008, 8(6), 913–918.(14) Lee, J. N.; Park, C.; Whitesides, G. M. Solvent compatibility of

poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem. 2003,75 (23), 6544–6554.(15) Tamejiro, H.; Kanie, K. Organofluorine Compounds: Chemistry

and Applications; Springer: New York, 2000; p 272.(16) Resnick, P. R. The preparation and properties of a new family of

amorphous fluoropolymers—Teflon-Af. Abstr. Pap.—Am. Chem. Soc.1990, 199, 107.(17) Hung, M. H.; Resnick, P. R. Functional fluoropolymers.

U.S. Patent 5,354,910, 1994.(18) Datta, A.; Eom, I. Y.; Dhar, A.; Kuban, P.; Manor, R.; Ahmad, I.;

Gangopadhyay, S.; Dallas, T.; Holtz, M.; Temkin, F.; Dasgupta, P. K.Microfabrication and characterization of Teflon AF-coated liquid corewaveguide channels in silicon. IEEE Sens. J. 2003, 3 (6), 788–795.(19) Manor, R.; Datta, A.; Ahmad, I.; Holtz, M.; Gangopadhyay, S.;

Dallas, T. Microfabrication and characterization of liquid core wave-guide glass channels coated with Teflon AF. IEEE Sens. J. 2003, 3 (6),687–692.(20) Wu, C. W.; Gong, G. C. Fabrication of PDMS-based nitrite

sensors using Teflon AF coating microchannels. IEEE Sens. J. 2008, 8(5�6), 465–469.(21) Cho, S. H.; Godin, J.; Lo, Y.-H. Optofluidic waveguides in

Teflon AF-coated PDMSmicrofluidic channels. IEEE Photonics Technol.Lett. 2009, 21 (15), 1057–1059.(22) Krishnan, M.; Burke, D. T.; Burns, M. A. Polymerase chain

reaction in high surface-to-volume ratio SiO2 microstructures. Anal.Chem. 2004, 76 (22), 6588–6593.(23) Dress, P.; Franke, H. A cylindrical liquid-core waveguide. Appl.

Phys. B: Lasers Opt. 1996, 63 (1), 12–19.(24) Rudel, M.; Kruse, J.; Ratzke, K.; Faupel, F.; Yampolskii, Y. P.;

Shantarovich, V. P.; Dlubek, G. Temperature dependence of positronannihilation lifetimes in high permeability polymers: Amorphous Te-flons AF. Macromolecules 2008, 41 (3), 788–795.(25) Zhao, H.; Zhang, J.; Wu, N. Q.; Zhang, X.; Crowley, K.; Weber,

S. G. Transport of organic solutes through amorphous Teflon AF films.J. Am. Chem. Soc. 2005, 127 (43), 15112–15119.(26) O’Brien, M.; Baxendale, I. R.; Ley, S. V. Flow ozonolysis using a

semipermeable Teflon AF-2400 membrane to effect gas-liquid contact.Org. Lett. 2010, 12 (7), 1596–1598.(27) Sahoo, H. R.; Kralj, J. G.; Jensen, K. F. Multistep continuous-

flow microchemical synthesis involving multiple reactions and separa-tions. Angew. Chem., Int. Ed. 2007, 46 (30), 5704–5708.(28) Cubaud, T.; Ulmanella, U.; Ho, C.-M. Two-phase flow in

microchannels with surface modifications. Fluid Dyn. Res. 2006, 38 (11),772–786.(29) Oskooei, S. A. K.; Sinton, D. Partial wetting gas-liquid segmen-

ted flow microreactor. Lab Chip 2010, 10, 1732 –1734.(30) Srinivasan, U.; Houston, M. R.; Howe, R. T.; Maboudian, R.

Alkyltrichlorosilane-based self-assembled monolayer films for stictionreduction in siliconmicromachines. J. Microelectromech. Syst. 1998, 7 (2),252–260.(31) Wootton, R. C. R.; deMello, A. J. Continuous laminar evapora-

tion: Micron-scale distillation. Chem. Commun. 2004, 3, 266–267.(32) Timmer, B. H.; van Delft, K. M.; Olthuis, W.; Bergveld, P.;

van denBerg, A.Micro-evaporation electrolyte concentrator. Sens. Actuators,B 2003, 91 (1�3), 342–346.(33) Kralj, J. G.; Sahoo, H. R.; Jensen, K. F. Integrated continuous

microfluidic liquid-liquid extraction. Lab Chip 2007, 7 (2), 256–263.(34) Gunther, A.; Khan, S. A.; Thalmann, M.; Trachsel, F.; Jensen,

K. F. Transport and reaction in microscale segmented gas-liquid flow.Lab Chip 2004, 4 (4), 278–286.

(35) Gunther, A.; Jhunjhunwala, M.; Thalmann, M.; Schmidt, M. A.;Jensen, K. F. Micromixing of miscible liquids in segmented gas-liquidflow. Langmuir 2005, 21 (4), 1547–1555.

(36) Gunther, A.; Jensen, K. F. Multiphase microfluidics: from flowcharacteristics to chemical and materials synthesis. Lab Chip 2006, 6(12), 1487–1503.

(37) Boyd, D. A.; Adleman, J. R.; Goodwin, D. G.; Psagtis, D.Chemical separations by bubble-assisted interphase mass-transfer. Anal.Chem. 2008, 80 (7), 2452–2456.

(38) Yen, B. K. H.; Gunther, A.; Schmidt, M. A.; Jensen, K. F.;Bawendi, M. G. A microfabricated gas-liquid segmented flow reactor forhigh-temperature synthesis: The case of CdSe quantum dots. Angew.Chem., Int. Ed. 2005, 44 (34), 5447–5451.

(39) Thomas, R. R.; Kaufman, F. B.; Kirleis, J. T.; Belsky, R. A.Wettability of polished silicon oxide surfaces. J. Electrochem. Soc. 1996,143 (2), 643–648.

(40) Li, X. M.; Reinhoudt, D.; Crego-Calama, M. What do we needfor a superhydrophobic surface? A review on the recent progress in thepreparation of superhydrophobic surfaces. Chem. Soc. Rev. 2007, 36 (8),1350–1368.

(41) Tavana, H.; Simon, F.; Grundke, K.; Kwok, D. Y.; Hair, M. L.;Neumann, A. W. Interpretation of contact angle measurements on twodifferent fluoropolymers for the determination of solid surface tension.J. Colloid Interface Sci. 2005, 291, 497–506.

(42) Hu, P.; Adamson, A. W. Adsorption and contact angle studiesII. Water and organic substances on polished polytetrafluoroethylene.J. Colloid Interface Sci. 1977, 59, 605–614.

(43) Sato, K.; Shikida, M.; Matsushima, Y.; Yamashiro, T.; Asaumi,K.; Iriye, Y.; Yamamoto, M. Characterization of orientation-dependentetching properties of single-crystal silicon: Effects of KOH concentra-tion. Sens. Actuators, A 1998, 64 (1), 87–93.

(44) Drechsler, A.; Petong, N.; Zhang, J.; Kwok, D. Y.; Grundke,K. Force measurements between Teflon AF and colloidal silicaparticles in electrolyte solutions. Colloids Surf., A 2004, 250 (1�3),357–366.

(45) Welzel, P. B.; Rauwolf, C.; Yudin, O.; Grundke, K. Influence ofaqueous electrolytes on the wetting behavior of hydrophobic solidpolymers—Low-rate dynamic liquid/fluid contact angle measurementsusing axisymmetric drop shape analysis. J. Colloid Interface Sci. 2002, 251(1), 101–108.

(46) Zimmermann, R.; Dukhin, S.; Werner, C. Electrokinetic mea-surements reveal interfacial charge at polymer films caused by simpleelectrolyte ions. J. Phys. Chem. B 2001, 105 (36), 8544–8549.

(47) Abate, A. R.; Krummel, A. T.; Lee, D.; Marquez, M.; Holtze, C.;Weitz, D. A. Photoreactive coating for high-contrast spatial patterning ofmicrofluidic device wettability. Lab Chip 2008, 8 (12), 2157–2160.

(48) Abate, A. R.; Lee, D.; Do, T.; Holtze, C.; Weitz, D. A. Glasscoating for PDMS microfluidic channels by sol-gel methods. Lab Chip2008, 8 (4), 516–518.

(49) Rebrov, E. V.; Seijger, G. B. F.; Calis, H. P. A.; de Croon, M. H.J. M.; van den Bleek, C. M.; Schouten, J. C. The preparation of highlyordered single layer ZSM-5 coating on prefabricated stainless steelmicrochannels. Appl. Catal., A 2001, 206 (1), 125–143.

(50) Conant, T.; Karim, A.; Rogers, S.; Samms, S.; Randolph, G.;Datye, A. Wall coating behavior of catalyst slurries in non-porousceramic microstructures. Chem. Eng. Sci. 2006, 61 (17), 5678–5685.

(51) Trachsel, F.; Gunther, A.; Khan, S.; Jensen, K. F. Measurementof residence time distribution in microfluidic systems. Chem. Eng. Sci.2005, 60 (21), 5729–5737.

(52) Fogler, H. S. Elements of Chemical Reaction Engineering, 4th ed.;Prentice Hall: Upper Saddle River, NJ, 2005.

(53) Goodell, J. R.; McMullen, J. P.; Zaborenko, N.; Maloney, J. R.;Ho, C. X.; Jensen, K. F.; Porco, J. A.; Beeler, A. B. Development of anautomated microfluidic reaction platform for multidimensional screen-ing: Reaction discovery employing bicyclo[3.2.1]octanoid scaffolds.J. Org. Chem. 2009, 74 (16), 6169–6180.

(54) Levenspiel, O. Chemical Reaction Engineering, 3rd ed.; Wiley:New York, 1999.

I dx.doi.org/10.1021/la2004744 |Langmuir XXXX, XXX, 000–000

Langmuir ARTICLE

(55) Levenspiel, O.; Smith, W. K. Notes on the diffusion-type modelfor the longitudinal mixing of fluids in flow. Chem. Eng. Sci. 1957,6, 227–233.(56) Meinhart, C. D.; Wereley, S. T.; Santiago, J. G. PIV measure-

ments of a microchannel flow. Exp. Fluids 1999, 27 (5), 414–419.(57) Santiago, J. G.; Wereley, S. T.; Meinhart, C. D.; Beebe, D. J.;

Adrian, R. J. A particle image velocimetry system for microfluidics. Exp.Fluids 1998, 25 (4), 316–319.(58) Kulkarni, A. A.; Kalyani, V. S. Two-phase flow in minichannels:

Hydrodynamics, pressure drop, and residence time distribution. Ind.Eng. Chem. Res. 2009, 48 (17), 8193–8204.