or

he cellmponentsous phasess the firsto liquidned in orderterfacialfunction of

l inlet andvestigated

Journal of Colloid and Interface Science 286 (2005) 158–165www.elsevier.com/locate/jcis

Interfacial stabilization of organic–aqueous two-phase microflows fa miniaturized DNA extraction module

Varun Reddya, Jeffrey D. Zahna,b,∗

a Department of Bioengineering, Materials Research Institute, Pennsylvania State University, 224 Hallowell Building, University Park, PA 16802, USAb Materials Research Institute, Pennsylvania State University, 224 Hallowell Building, University Park, PA 16802, USA

Received 30 April 2004; accepted 9 December 2004

Available online 18 March 2005

Abstract

Organic–aqueous liquid (phenol) extraction is one of many standard techniques to efficiently purify DNA directly from cells. Tcomponents naturally distribute themselves into the two fluid phases in order to minimize interaction energies of the biological cowith the surrounding solvents. The membrane components and protein partition to the interface between the organic and aquewhile the DNA stays in the aqueous phase. The aqueous phase is then removed with a purified DNA sample. This work studiesteps towards miniaturizing this liquid extraction technique in a microfluidic device. The first step is to understand how the twphases behave in microchannels. Due to the interfacial tension between the two liquid phases, novel approaches must be examito obtain interfacial stability under flow conditions. The stability of the organic–aqueous interface is improved by reducing the intension between the two phases by incorporating a surfactant into the aqueous phase. The variation of the interfacial tension as asurfactant concentration is also quantified in this work. This has led to the ability to create stable stratified microflows in both a duathree inlet microfluidic systems. Also, the first step in understanding biological interactions at the organic–aqueous interface is inusing a fluorescently labeled bovine serum albumin protein. 2005 Elsevier Inc. All rights reserved.

Keywords: Two-phase flow; Phenol extraction; Liquid extraction; DNA purification

ch-fora-

de-AS)inemi-s-edof

s-tan-to-a

0

he-x-thentsThenic–ousare

1. Introduction

A major research thrust in microscale science and tenology is the development of autonomous platformsthe extraction, purification and analysis of biological mterial from cells. Micro- and nanofabricated diagnosticvices have been termed micrototal analysis systems (µTbiochips, or more generically “lab on a chip,” and combsensing mechanisms (physical, optical, electrical or checal) with microfluidics. Very complex biomolecule procesing reactions (cell lysis, electrophoresis, etc.) are performautonomously permitting biochemical characterizationcells and biological material.

* Corresponding author. Fax: +1-814-863-0490.E-mail address: jdzbio@engr.psu.edu(J.D. Zahn).

0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2004.12.052

,

DNA (either genomic or plasmid DNA) extraction uing aqueous–organic liquid extraction is one of many sdard techniques commonly performed in biology laboraries [1]. Briefly, the procedure consists of lysing cells inbasic lysis buffer (10 mM Tris, pH 8.0, 0.1 M EDTA, 2µg/ml Pancreatic RNase, 0.5% SDS, and 100 µg/ml pro-teinase K) and adding an equal volume of immiscible pnol:chloroform:isoamyl alcohol (25:24:1 by volume) miture to the aqueous solution. A vortexing step ‘mixes’two phases. With effective mixing, the cell componenaturally distribute themselves into the two phases.membrane components and protein partition to the orgaaqueous interface, while the DNA stays in the aquephase. After the mixing step, the two immiscible phases

allowed to separate and the aqueous phase is removed usinga micropipette tool. The DNA is concentrated by precipita-tion in ethanol and resuspended in an aqueous buffer.

d and

ousas

ted

NAc-

foril-

theraf-

ownd

ses.e tocialer,

s-

ear

inber,let

and

ter

pil-u-both

l-

oc-thanale

hatis

othres.uouaseer

dsta-

re-Thisheac-cial

ctant

cidr as-irst,0 Åext,tho-the

andsub-

(5:1ur-0 nmss oftepl in-gleeath-are0n

etsrillond

cedxy.

oughn-

on-ion.ra-d as

V. Reddy, J.D. Zahn / Journal of Colloi

To date there has been limited research into autonomsample preparation and DNA purification. Cell lysis hbeen reported using either electrical[2] or physical means[3,4] and on-chip DNA extraction has been demonstraby adsorbing DNA onto silicon pillars[5,6]. Phenol ex-traction may be used to extract genomic and plasmid Dfrom cells within a microfluidic device. After the extration procedure, the DNA can be processed downstreamPCR amplification[7], restriction enzyme digest and caplary electrophoresis[7,8].

Since the phenol and aqueous phases are immiscibleis an interfacial tension between the two phases whichfects the flow profile. The stability of the interface under flconditions is a strong function of the interfacial tension aviscosity ratio[9] between the aqueous and organic phaThe ratio of viscous shear forces to elongate an interfacthe surface tension forces which act to minimize interfaarea between the phases is given by the Capillary numb

Ca = µ2U/σ,

and the viscosity ratio is defined as

λ = µ1/µ2,

whereµ1 andµ2 are respectively the viscosity of the diperse and continuous phase (Pa s),U is the average flowvelocity (m/s) andσ the interfacial tension (N/m). The crit-ical capillary number for droplet breakup in simple shflow as a function of viscosity ratio has been examined[10].Below the critical capillary number, a droplet will maintaa steady-state shape. Above the critical capillary numthe droplet will elongate and eventually break up. Dropdeformation[11] and free surface entrainment tubes[12]have been induced at capillary numbers of order 0.1above. It has been reported that for stokes flow (Re = 0)droplet breakup will not occur if the viscosity ratio is greathan 3.1[13]. The tendency of immiscible liquids to formdynamic droplet emulsion patterns as a function of calary number and flow velocity ratios between a continous and noncontinuous phase has been exploited froma T-section microfluidic channel[14,15] and a symmetricthree-input channel[16] to obtain a tight control of the emusion droplet size distribution.

A stable elongated interface between the two phasescurs when the viscous forces are equal to or greaterthe interfacial forces. Computational studies of a microscsheath flow of two immiscible liquids have concluded tthe critical capillary number for stable stratified flowof order 1[17]. Recently, stable[18,19] and unstable[19]stratified flow profiles have been demonstrated using bmiscible and immiscible organic–aqueous solvent mixtuThese two-phase systems have been used for continliquid extraction techniques of molecules from one phinto another[18,20]. These flows were stabilized by eith

adding microtexturing to the microchannels to guide the liq-uid flow paths[21] so that droplet formation did not occur orby treating the microchannel surface with a self-assembled

Interface Science 286 (2005) 158–165 159

e

s

monolayer (SAM)[22] in order to change the liquid soliinteractions of the organic solvent to increase interfacialbility.

Another approach to increase interfacial stability is byducing the interfacial tension between the two phases.work details a significant improvement in the stability of ttwo-phase stratified flow pattern by incorporating a surftant into the aqueous phase and quantifying the interfatension between the two phases as a function of surfaconcentration in the aqueous phase.

2. Experimental procedures

2.1. Fabrication

Microchannels were fabricated by using hydrofluoric awet etching of glass with a chrome–gold masking layedescribed in the literature[8]. Briefly, to create glass capillary microchannels a wet etching approach is used. Fa masking material consisting of 250 Å chrome and 400gold (Cr/Au) was sputtered onto a clean glass slide. Na photoresist is deposited to a thickness of 2 µm and ligraphically patterned. After the photoresist is patterned,pattern is transferred into the metal layer by using a goldchrome etchant (Transene Corporation). Next, the glassstrate is etched in an agitated 5:1 buffered hydrofluoricBHF) acid solution to a depth of 30 µm. The channel sface roughness was determined to be between 5 and 3along the length of the channel with an average roughne20 nm as determined by profilometry (Tencor Inc. Alphas200). Two different geometries have been tested, a dualet consisting of two converging flow channels into a sindaughter branch and a three-inlet channel to create a shing flow profile. The final daughter channel dimensions80 µm× 30 µm× 2 mm for the dual inlet geometry and 20µm× 30 µm× 2 cm for the three-inlet geometry, as showin Fig. 1. After the channels are formed, inlets and outlare created by using a drill with a silicon carbide tipped dbit. The channels are closed by thermally bonding a secglass substrate to the channels in a furnace at 600◦C. The in-terconnecting tubing with a Luer connection is then plainto the device and glued into place using a two-part epo

2.2. Experimental procedure

The phenol and aqueous phases were co-infused thrthe microfluidic structure using a syringe pump (KD scietific Inc.) at a variety of flow rates from 0.1 to 10 µl/min.A glass syringe (Whatman Laboratory Products) is cnected to the microfluidic channel through a luer connectPhenol:chloroform:isoamyl alcohol at a 25:24:1 volumetio was obtained through a vendor (USB Corp.) and use

received. The viscosity of the phenol:chloroform:isoamyl al-cohol mixture was determined to be 3.21× 10−3 Pa s. Theaqueous phase consisted of deionized water (18 M� cm)

160 V. Reddy, J.D. Zahn / Journal of Colloid and Interface Science 286 (2005) 158–165

metr

al-own0.2,ol

asehy-r-mi-

ing ad.)isthertedrce

l to

i-rho-

entso-

n toas

Cl,

n-tion

mi-ddual

Fig. 1. (Top) Schematic of the dual inlet geo

with the addition of sodium dodecyl sulfate (SDS) (Cbiochem Inc., 98.64% pure, used as-received) at a knconcentration of 0.01, 0.025, 0.05, 0.075, 0.10, 0.15,0.5% by weight to volume. For flow visualization the phenphase is labeled with a lipophilic dye, DiI (1,1′-dioctadecyl-3,3,3′,3′tetramethylindocarbocyanine perchlorate) for phcontrast of the organic phase during flow. Due to thedrophobic nature of DiI, it stays partitioned within the oganic phase which appears bright using epifluorescentcroscopy with a TRITC filter cube (Olympus IX71).

The interfacial tension measurements were made usCIT 100 model Camtel interfacial tensiometer (Camtel Ltusing the Du Nouy Ring technique. The Du Nouy ringwetted in the heavy (phenol) phase and pulled throughorganic–aqueous interface while measuring the force exeon the ring. The maximum force is assumed to be the forequired to break the lamella and is directly proportionathe interfacial tension.

Finally, in order to conduct a first study of biologcal interactions at the organic–aqueous interface, a

damine labeled bovine serum albumin (BSA) was pre-pared. (5-6)-Carboxytetramethyl rhodamine, succimidyl es-ter (5(6)-TAMRA,SE) was reacted with BSA in BRB80

y. (Bottom) Schematic of the three-inlet geometry.

buffer (80 mM PIPES, 1 mM MgCl2, 1 mM EGTA, pH 6.8)for 30 min on a lab rotator. This crosslinks the fluorescmolecule to the protein through the succimidyl ester. Thelution was then filtered through a chromatography columremove the free rhodamine from the solution. The BSA wdiluted in phosphate buffered saline (PBS) (150 mM Na8.4 mM Na2HPO4·7H2O, 1.8 mM NaH2PO4·H2O, pH 7.5)with 0.5% (w/v) SDS (to be consistent with the SDS cocentration used in the biological protocol) to a concentraof 0.5 mg/ml.

3. Experimental results

The phenol and water phases were infused into thecrochannels as described.Fig. 2 shows a stable stratifieco-infusion of the organic and aqueous phase within ainlet channel 80 µm wide×30 µm deep×2 mm long daugh-ter branch at a total flow rate of 10 µl/min (5 µl/min from

each inlet). The phenol phase is dyed with the lipophilic dyeand appears bright under epifluorescent microscopy. The av-erage linear flow velocity for this system is 6.94 cm/s.

d and

asenin

chnd-

the

hea

sta-

be

eapil-redin-ore

r.ainhase.tionaletantly-le

at-ow

%a-DS

apil-Thisin-bee-

V. Reddy, J.D. Zahn / Journal of Colloi

Without the addition of detergent, the flow profile wvery difficult to control due to the interfacial tension betwethe two immiscible fluid phases. The interfacial tensionthe absence of surfactant was determined to be 2.80 mN/m.At a flow rate of 10 µl/min (6.94 cm/s linear velocity),a slug flow profile was produced in which slugs of eaphase would alternatively flow through the channel. Depeing on the fluid velocity these slugs could break off frominlet very slowly (at a total flow rate of 2 µl/min) or veryrapidly (at a total flow rate of 10 µl/min) where they couldhardly be seen using standard video equipment (Fig. 3). Inthese experiments, the phenol is contributing a higher s

Fig. 2. Co-infusion of water and phenol:chloroform solution showing a

Fig. 3. (a)–(c) Three video frames showing the water phase breaking into abetween frames is 16.67 ms. (d) Many alternating slugs at high velocity (tot

Interface Science 286 (2005) 158–165 161

r

due to its higher viscosity and is therefore considered tothe continuous phase. For a total flow rate of 10 µl/min (lin-ear velocity of 6.94 cm/s), the viscosity ratio is 3.21, thReynolds number for the phenol phase is 0.83 and the clary number is 0.0796 which is much lower than is requifor a stable stratified co-infusion. The slug flow profile mimizes the interfacial interaction area and is therefore mstable than the stratified flow at this low capillary numbe

In order to minimize the interfacial tension thus to obta stable stratified flow, SDS was added to the aqueous pSince this work attempts to miniaturize the phenol extracprocedure with little modification to the existing macroscprotocols, a 0.5% SDS solution was chosen. This surfacconcentration is the same as is required in the cellularsis buffer used in[1]. At a 0.5% SDS concentration, a stabstratified flow profile was obtained. This stratified flow ptern would also recover, almost immediately, when the flwere subjected to velocity perturbations.

Although a stratified flow profile was obtained with 0.5SDS, it was important to quantify the variation of interfcial tension between the two phases as a function of Sconcentration in the aqueous phase to determine the clary number dependency on surfactant concentration.leads to the quantification of the value of the maximumterfacial tension with which a stratified flow pattern canobtained.Fig. 4 shows interfacial tension between the ph

nol and water phases as a function of surfactant concentra-

ofns,

ble stratified flow profile. The phenol is dyed with a lipophilic dye and isbright under epifluorescent microscopy. tion. At low SDS concentrations an interfacial tension

2.80 mN/m is determined. At higher SDS concentratio

slug of fluid during a slug flow profile. The total flow rate is 2 µl/min. The timeal flow rate of 10 µl/min).

and

ur-Si-of

ce.riseol ister-ionnt

tenitiveion,eseten-tra-

wasm-

was-outer-thetra-

%uce

thethetem

ultsred

ta-n.

he-omesoc-

tjec-.70and

ac-ere,

eledtant.

mi-thing

SAline

res-einface.lim-for

162 V. Reddy, J.D. Zahn / Journal of Colloid

a significant drop of interfacial tension to a minimum sface tension of 0.108 mN/m is observed at a 0.10% SDsolution. This minimum is assumed to be the critical mcelle concentration (cmc) resulting from the adsorptiondodecanol (a hydrolysis product of SDS) to the interfaA further increase in the SDS concentration shows ain the interfacial tension measurement as the dodecansolubilized within micelles and SDS replaces it on the inface. At a SDS concentration of 0.5% the interfacial tensis measured to be 0.1 mN/m. However, this measuremeis expected to have some measurement error since thesiometer is not temperature-controlled and not very sensat such low interfacial tension measurements. In additthe interfacial tension is very sensitive to impurities. Theffects may account for the apparent rise in interfacialsion after the cmc and slight drop at higher SDS concentions. However, since a 0.5% (w/v) SDS concentrationused in all flow experiments, this value is in capillary nuber calculation. By comparison, the system in[19] used awater/tetradecane co-infusion and the interfacial tensionseen to drop from 56 to 3 mN/m when SPAN 80 surfactant was added with a critical micelle concentration of ab0.02% (w/w). This report also shows a slight rise in intfacial tension just after the cmc with a further drop tominimum interfacial tension at higher surfactant concentions.

Fig. 4. Phenol:chloroform:isoamyl (25:24:1)—water interfacial tension as

Fig. 5. (Left) Three-inlet co-infusion of phenol and water with 0.5% SDS.corresponding to a capillary number of 0.56. (Right) Outlet region of the samdevice. Overhead light is also transmitted to show the microchannel feature

Interface Science 286 (2005) 158–165

-

A stable stratified co-infusion of water containing 0.5SDS and phenol from a three-inlet flow system to proda sheathing flow profile is shown inFig. 5. This flow pro-file is stable along the entire length of the device (>2 cm).Here the water phase is on the sides of the channel withphenol phase forming a liquid membrane in the center ofchannel. The flow rates for the water/phenol/water sysare 2.5, 1.25, 2.5 µl/min for a total flow rate of 6.25 µl/min.This corresponds to an average velocity of 1.74 cm/s and aReynolds number of 0.20 for the phenol phase. This resin a capillary number of 0.56 which is of order 1 as requiby [15].

As the flow rate is reduced, the flow transitions from a sble stratified profile to a protruding jet with droplet ejectioAs shown inFig. 6, as the flow rate is decreased the pnol phase becomes progressively thinner and the jet becmore tortuous. The flow rates at which droplet ejectioncurs for the water/phenol/water system are 1, 0.5, 1 µl/minfor a total flow rate of 2.5 µl/min. At this flow rate, the jepenetrates 3 mm into the flow channel before droplet etion occurs. This corresponds to an average velocity of 0cm/s, a Reynolds number of 0.08 for the phenol phasea transition capillary number of 0.224.

Finally, the first attempt to examine the biological intertions at the organic aqueous interface was undertaken. Hthe lipophilic dye was not used, and fluorescent dye labBSA was added to the aqueous phase containing surfacNow, the aqueous flow is bright under epifluorescentcroscopy and the organic phase appears dark. The sheaflow pattern, however, is the same as the setup inFig. 5 fora water/phenol/water liquid membrane. As shown inFig. 7,as the protein solution flows through the device, the Bprecipitates at the organic–aqueous interface to create aof protein precipitate which appears bright under epifluocent microscopy. This shows the ability to remove protfrom an aqueous solution at the organic–aqueous interHowever, complete removal was not possible due to theited surface area of the interface. Further work is required

complete removal of BSA from the aqueous solution. Com-

rfa-ue.

a function of sodium docecyl sulfate (SDS) surfactant concentration. Thecritical micelle concentration is estimated at a concentration of 0.1%.

plete removal will be accomplished by increasing the intecial area through an electrohydrodynamic mixing techniq

The flow rates for the water/phenol/water co-infusion are 2.5/1.25/2.5 µl/mine device 2 cm downstream showing a stable stratified profile for the length of the

s.

Thereaks off.

V. Reddy, J.D. Zahn / Journal of Colloid and Interface Science 286 (2005) 158–165 163

Fig. 6. (a)–(d) Four video frames showing a three-inlet co-infusion of phenol and water with 0.5% SDS with droplet ejection from the stratified jet.timebetween frames is 16.67 ms. The dashed lines show the channel boundaries. Frame 3 shows the jet thinning and necking just before the droplet bTheflow rates for the water/phenol/water co-infusion are 1/0.5/1 µl/min corresponding to a capillary number of 0.224.

and een

isanictiono-er-out

thetingandro-the

owd a

Fig. 7. Co-infusion of undyed phenol and water containing 0.5% SDSprecipitating along the organic–aqueous interface.

4. Discussion

This work has shown that hydrostatic flow controlable to promote an elongated interface between the organd aqueous phases for efficient on-chip phenol extracof DNA. The ability to stabilize an organic–aqueous cinfusion through microfluidic channels is achieved by lowing the interfacial tension between the two phases. With

the addition of surfactant, the capillary number was suffi-ciently low so that droplet breakup and a slug flow profilewere seen. This is because the shear forces acting tangen

0.5 mg/ml rhodamine labeled bovine serum albumin (BSA). The BSA is s

tially to the interface were not sufficiently strong to keepinterface extended compared to the interfacial force acnormally to the interface to cause the interface to curveproduce droplets. This droplet ejection and slug flow pfile was shown to be highly erratic and uncontrolled ingeometries examined.

The addition of surfactant allowed a stable stratified flprofile to be achieved. The addition of 0.5% SDS allowe

-

stable stratified flow to be obtained at a capillary number of0.56. This stable stratification may be thought of as an ex-tremely elongated droplet in a shearing flow which has not

and

at ables ationionau-ersingw-yerter

micycan

ual

wfi-testhe

tantowwithtedthe

tionheseionsisem-rted

pro-thenpidsion.nic–tein

on,oftenicalionsle-eenbility

difore

at itav-

s uneousble

re-lete

ter-

ur-tingn-ses.

wn-eing

lar

sek ish In-

b-ring

EE

.A.B

Bi-

e-gy,

h.

c.

R.stem

1

3.2)

.

54.Rev.

ula-

nal.nd

l.

164 V. Reddy, J.D. Zahn / Journal of Colloid

broken up along the length of the channel. However,critical capillary number of 0.224, the jet becomes unstaand droplet ejection occurs. A stable stratified profile inecessary requirement to design an efficient DNA extracand purification geometry for integration with PCR reactchambers and capillary electrophoresis columns for antonomous µTAS system. The addition of 0.5% SDS lowthe interfacial tension significantly as seen by determinthe effect of SDS concentration on interfacial tension. Hoever, in the three-inlet system, the width of the phenol lais a strong function of the flow rate ratio between the ou(water) and inner (phenol) flow rates as in hydrodynafocusing[23]. At a higher flow ratio the phenol layer maalso become sufficiently thinned so that droplet breakupoccur by a jet (Rayleigh–Plateau) instability with eventdroplet formation, but this has not been observed.

The addition of surfactant is critical not only for flostability, but also to improve biological interactions for efcient DNA extraction. The surfactant molecule accumulaat the interface and lowers the surface energy betweentwo-phases to lower the interfacial tension. The surfacreaches an equilibrium distribution at the interface. At lconcentrations of SDS, the interface is not saturatedsurfactant. At the critical micelle concentration a saturamonolayer of surfactant is formed on the interface andinterfacial tension cannot be further lowered by the addiof excess surfactant. The SDS concentration used in texperiments is well above the critical micelle concentratto be consistent with the biological protocol. In cellular lythese micelles solubilize the hydrophobic proteins and mbrane components of the cell so that they may be transpoto the organic–aqueous interface during the extractioncedure. At the interface, these biological componentsdisplace the surfactant molecules but the proteins and listill act as surfactants to maintain a lower interfacial tensIn these studies BSA was seen to precipitate at the orgaaqueous interface showing the ability to remove the prousing a miniaturized liquid extraction technique.

In macroscale DNA extraction using phenol extractithe interface between the phenol and water phases isseen to become diffuse with the accumulation of biologmaterial. The interfacial tension at high SDS concentratis already very low and the high level of amphiphilic mocules at the interface act to stabilize the interaction betwthe organic and aqueous solvents and increase the misciof the two phases. Thus, as the interface becomes morefuse the interfacial thickness increases which allows mprotein, lipid and surfactant to accumulate there so thcan be efficiently removed from the aqueous solution leing only a purified DNA solution.

These studies represent the first step needed towardderstanding two-phase flows using a phenol and aqusolution for an efficient DNA extraction procedure. A sta

stratified flow profile was achieved by lowering the inter-facial tension between the organic and aqueous phases. Iaddition, a marker BSA protein was shown to precipitate

Interface Science 286 (2005) 158–165

-

-

at the organic aqueous interface indicating the ability ofmoving protein from the aqueous phase. However, compremoval of protein was not possible due to the limited infacial area in the channel.

Future work will focus upon trying to increase the sface area over which the extraction takes place by creaan electrohydrodynamic instability by exploiting the coductivity gradient between the aqueous and organic phaTechniques for removing the organic phase prior to dostream DNA processing of the aqueous phase is also binvestigated. Finally, on-chip DNA extraction from a cellulysate will be realized.

Acknowledgments

The authors would like to thank Erwin Vogler for uof his tensiometer and for helpful discussions. This worsupported by a Penn State University Materials Researcstitute Seed Grant.

References

[1] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning: A Laoratory Manual, Cold Spring Harbor Laboratory Press, Cold SpHarbor, NY, 1989, vols. 1, 2, 3.

[2] S.W. Lee, H. Yowanto, Y.C. Tai, in: Proceedings 11th Annual IEInternational MEMS-98 Conference, 1998, pp. 443–447.

[3] M.T. Taylor, G.T.A. Kovacs, P. Belgrader, R. Joshi, S. Sakai, MNorthrup, K.E. Petersen, in: 2nd Annual International IEEEE-EMSpecial Topic Conference on Microtechnologies in Medicine andology, 2002, pp. 551–555.

[4] D. Di Carlo, L.P. Lee, in: 2nd Annual International IEEE-EMB Spcial Topic Conference on Microtechnologies in Medicine and Biolo2002, pp. 527–530.

[5] L.A. Christel, K. Petersen, W. McMillan, M.A. Northrup, J. BiomecEng. 121 (1999) 22–27.

[6] W.A. McMillan, K.E. Petersen, L.A. Christel, M.A. Northrup, in: Proof the 8th International Symposium on Microbial Ecology, 1998.

[7] R.A. Mathies, E.T. Lagally, T. Kamei, W.H. Grover, C.N. Liu, J.Scherer, R.A. Street, in: Solid State Sensor and Actuator MicrosyWorkshop, Hilton Head Island, SC, 2002, pp. 112–117.

[8] P.C. Simpson, A.T. Woolley, R.T. Mathies, Biomed. Microdev.(1998) 7–25.

[9] G.I. Taylor, Proc. R. Soc. London Ser. A 146 (858) (1934) 501–52[10] C.L. Tucker III, P. Moldenaers, Annu. Rev. Fluid Mech. 34 (200

177–210.[11] W.L. Olbricht, D.M. Kung, Phys. Fluids A 4 (7) (1992) 1347–1354[12] S.T. Thoroddsen, Y.K. Tan, Phys. Fluids 16 (2) (2004) L13–L16.[13] Y. Renardy, D.B. Khismatullin, Phys. Fluids 15 (5) (2003) 1351–13[14] T. Thoresen, R.W. Roberts, F.H. Arnold, S.R. Quake, Phys.

Lett. 86 (18) (2001) 4163–4166.[15] G.E. Karniadakis, A. Beskok, Microflows: Fundamentals and Sim

tion, Springer-Verlag, New York, 2001.[16] Y.C. Tan, J. Collins, A.P. Lee, in: Proceedings of the 7th Internatio

Conference on Micro Total Analysis Systems, 2003, pp. 963–966[17] L.W. Ho, J. Marchetti, C. Gallegos, in: Proc. of the 1999 Modeling a

Simulation of Microsystems Conference, 1999, pp. 530–533.[18] A. Hibara, M. Tokeshi, K. Uchiyama, H. Hisamoto, T. Kitamori, Ana

nSci. 17 (2001) 89–93.

[19] R. Dreyfus, P. Tabeling, H. Williams, Phys. Rev. Lett. 90 (2003)144505.

d and

ra,

o,

i,

ev.

V. Reddy, J.D. Zahn / Journal of Colloi

[20] M. Suremeian, M.N. Slyyadnev, H. Hisamoto, A. HibaK. Uchiyama, T. Kitamori, Anal. Chem. 74 (2002) 2014–2020.

[21] M. Tokeshi, T. Minagawa, K. Uchiyama, A. Hibara, K. Sat

H. Hisamoto, T. Kitamori, Anal. Chem. 74 (2002) 1565–1571.

Interface Science 286 (2005) 158–165 165

[22] A. Hibara, M. Nonaka, H. Hisamoto, K. Uchiyama, Y. KikutanM. Tokeshi, T. Kitamori, Anal. Chem. 74 (2002) 1724–1728.

[23] J.B. Knight, A. Vishwanath, J.P. Brody, R.H. Austin, Phys. R

Lett. 80 (17) (1998) 3863–3866.