erickson_expfluids.pdf

8/10/2019 erickson_expfluids.pdf

1/10

Microbubble lensing-induced photobleaching (l-BLIP)with application to microflow visualization

D. Sinton, D. Erickson, D. Li

Abstract The curvature of gasliquid interfaces and thestep change in properties across these interfaces in mi-crochannels are shown here to create a powerful lens/mirror effect. In a hydrophilic system, light incident on thebubble is focused into the surrounding liquid, resulting ina locally increased total light exposure. The optical phe-nomena leading to this are discussed, and the effect isdemonstrated experimentally by imaging the increasedphotobleaching rate of fluorophores in the near-bubbleregion. Numerical simulations of the system are performed

to investigate the electrical potential and flow fieldsresulting from the application of an axial electric field.Microbubble lensing-induced photobleaching (l-BLIP) isthen applied as a method to inject a negative scalar flowmarker for flow visualization in microchannels. Onceformed, the electrokinetic transport of this marker isanalyzed to determine the cross-channel velocity profile ofthe liquid phase and the liquid velocity in the film.Experimental data is verified by comparison with numer-ical predictions and previous experimental studies. Thiscontribution represents both a new application of micro-scale gasliquid interfacial phenomena, and a new tech-nique for microfluidic flow visualization, particularly

applicable (though not limited) to the study of multiphasemicrochannel flows.

1IntroductionMicrosized chemical synthesis and analysis systems canoffer many advantages over their traditional, macrosizedcounterparts. Typical advantages include reduced reagent/sample use, increased sensitivity, increased speed of

analysis, and potential for mass-manufactured, portable,disposable systems. The desire to miniaturize conventionalmultiphase systems has fostered recent interest in phasechange and multiphase transport phenomena in micro-channels (Chang 2002; Takhistov et al. 2002; Zhao et al.2001). Potential applications include microsized evapora-tors, condensers, distillation units, gasliquid reactors,multiphase extraction and separation units, and heatexchangers. Latent heat effects, central to the performanceof many conventional heat exchangers, could be particu-

larly useful in the thermal control of temperature-sensitivemicrofluidic chip applications. Bubbles may also be usedin microfluidic chips to separate discrete liquid samples(Zhao et al. 2001), or employed as pumping mechanisms(Jun and Kim 1998). Knowledge of multiphase phenomenain microsystems is also required to avoid unwanted boil-ing of electrolyte in microchannels due to a lack of dissi-pation of Joule heating. This will be of increasingimportance as microfluidic chip technologies mature intomass-produced polymer-based chips (Duffy et al. 1998;Ross et al. 2001), which are less efficient at dissipating heatthan conventional glass chips, due to their low thermalconductivity.

A theoretical analysis of pressure-driven motion of asemi-infinite bubble in a circular capillary was presentedby Bretherton (1961). Those results have since been ex-tended to include discrete bubbles (Ratulowski and Chang1989), the effect of surfactants (Ratulowski and Chang1990), and other aspects. The effect of surfactant on thepressure required to advance a semi-infinite bubble is ofparticular relevance to pulmonary airway reopening con-ditions (Ghadiali and Gaver 2000). In microfluidic chipapplications, however, pressure-driven multiphase trans-port is often not practical, as it can require orders ofmagnitude higher pressure gradients than single-phaseliquid transport (Chang 2002). For this reason, electroki-

netic bubble transport, described in detail by Takhistovet al. (2002), is an attractive option.

To study liquid flow in microchannels, various micro-flow visualization methods have evolved. Microparticleimage velocimetry (microPIV) is a method that wasadapted from well-developed PIV techniques for flows inmacrosized systems (Taylor and Yeung 1993; Santiago etal. 1998; Meinhart et al. 1999). In that technique, the fluidmotion is inferred from the motion of submicron tracerparticles. To eliminate the effect of Brownian motion,temporal or spatial averaging must be employed. Particleaffinities for other particles, channel walls, and free sur-faces must also be considered. In electrokinetic flows, the

Experiments in Fluids 35 (2003) 178187

DOI 10.1007/s00348-003-0645-6

8

Received: 21 January 2003 / Accepted: 24 April 2003

Published online: 27 June 2003 Springer-Verlag 2003

D. Sinton, D. Erickson, D. Li (&)Department of Mechanical and Industrial Engineering,University of Toronto, 5 Kings College Road,Toronto, Ontario, M5S 3G8, CanadaE-mail: [email protected]: +1-416-9787753

Financial support of this work by the Natural Sciences andEngineering Research Council (NSERC) of Canada, through post-graduate scholarships to D.S. and D.E. and a research grant toD.L., is gratefully acknowledged. Financial support from GlynnWilliams, through a post-graduate scholarship to D.S. is alsogratefully acknowledged.


2/10

electrophoretic motion of the particles (relative to the bulkflow) is an additional consideration. Dye-based microflowvisualization methods have also evolved from their mac-rosized counterparts. However, traditional mechanical dyeinjection techniques are difficult to apply on the micro-scale. Specialized caged fluorescent dyes have been em-ployed to facilitate this injection using selective lightexposure. The photoinjection is accomplished by exposing

an initially nonfluorescent solution seeded with cagedfluorescent dye to ultraviolet light. As a result of theultraviolet exposure, caging groups are broken and fluo-rescent dye is released. Lempert et al. (1995) first employedcaged fluorescent dyes for flow tagging velocimetry inmacrosized flows. This technique has since been used tostudy a variety of liquid flow phenomena in microstruc-tures (Paul et al. 1998; Herr et al. 2000; Johnson et al. 2001;Molho et al. 2001; Sinton et al. 2002a, 2002b). The disad-vantages of this technique are that it requires specializedcaged dye (which is expensive), extensive infrastructure tofacilitate the photo injection, and channel materials thattransmit ultraviolet light. As an alternative to photoinjec-tion of fluorescent dye, a flow marker can be created in auniform solution of fluorescent dye by local photoble-aching (Molho et al. 2001; Mosier et al. 2002). When a dyemolecule is photobleached (or photodestroyed) it nolonger fluoresces. Although the literature on this process issparse, it is well established that the rate of photobleachingcorrelates directly with excitation light exposure (Becker1996). None of these flow velocimetry methods have beenapplied to direct measurements of transport in multiphasemicrochannel applications. Instead, liquid velocities havebeen inferred from observed interface motion or bymeasuring reservoir levels by means of a capillary(Takhistov et al. 2002).

In this paper, the ability to intensify light exposure inthe near-bubble liquid using the optical characteristics ofthe gasliquid interface is demonstrated. The opticalphenomena leading to this focusing are discussed, and theeffect is shown experimentally by imaging the increasedphotobleaching rate of fluorophores in the near-bubbleregion. Numerical simulations are performed to investi-gate the electrical potential field and flow field in thissystem, resulting from the application of an axial electricfield. Microbubble lensing-induced photobleaching is thenapplied as a method to inject a flow marker for flowvisualization. The bubble film flow rates and the cross-channel velocity profiles determined are verified by com-parison with numerical predictions and previous experi-

mental studies. This contribution represents both a newapplication of microscale gasliquid interfacial phenom-ena, and a new technique for microfluidic flow visualiza-tion, particularly applicable to (though not limited to) thestudy of multiphase microchannel flows.

2ExperimentsIn the experiment, a length of capillary was filled with anaqueous fluorescein/buffer solution such that it containeda single bubble at the midpoint of the capillary. Each endof the capillary was connected to a reservoir, and thebubble region was viewed under an oil-immersion

epi-illumination fluorescent microscope. The fluorescentemission of the liquid was imaged with a progressive scanCCD camera, digitized, and stored on the computer. Fol-lowing the experiment, image processing was performed,and the images were analyzed. In cases where fluid flowwas involved, scalar image velocimetry was performed todetermine liquid velocities.

2.1ImagingImaging was performed with a previously developedfluorescent microscope system (Sinton et al. 2002a,2002b). A continuous uniform flood of excitation lightwas provided by a single-line 200-mW, 488-nm argonlaser (American Laser Corp.), through the 25, NA=0.75oil immersion microscope objective (Leica). The index ofrefraction of the oil (noil=1.48) was matched to that of thecapillary (nfused_silica=1.46), to avoid lensing caused bythe curvature of the capillary. The camera was run invideo mode at 15 Hz with individual exposure times of1/60 s. Because of the geometry of the CCD chip, thecamera captured only a square central portion of the fieldof view illuminated by the microscope optics. Theacquired images had a resolution of 640484 pixels. Thiscorresponded to a 543-lm visible length of capillary, witheach pixel representing a 0.85-lm square in the objectplane. The camera orientation was carefully adjusted suchthat the pixel grid was aligned with the radial and axialdirections.

2.2Image processingTo remove any nonuniformities present in the imagingsystem, dark-field image subtraction and bright-field im-age normalization were performed with each image (Russ1999). The bright-field image was obtained by imaging thechannel filled with a uniform concentration of fluorescentdye without a bubble present. The pixel intensity valueswere then scaled linearly by a single factor such that thedegree of photobleaching that occurred spanned thegrayscale range. Finally, the images were smoothed with adistance based 1010 pixel kernel. Pixel intensities ofpostprocessed images were directly interpreted as dyeconcentration based on a previously determined linearcamera response characteristic.

2.3MicrochannelsThe microchannels were 100-lm i.d., 15-cm long, circularcross-section fused silica capillaries (Polymicro Tech.). Atthe capillary midpoint, the exterior polyimide coating wasoxidized and removed to create a viewing window. Newcapillaries were prepared by flushing with pure waterfollowed by buffer, and then buffer with fluorescent dye.To inject the bubble, air was drawn into the capillary bysuction provided by a low-volume high-pressure syringe(10 lL, Hamilton Gastight). The air was then slowlypumped out until only the bubble length desired remainedin the capillary. At that point, the capillary end was rein-serted into the solution and the bubble was drawn to thecapillary midpoint. Each end of the capillary was then


3/10

connected to a small reservoir with embedded platinumelectrodes.

2.4ChemicalsThe fluorescent dye employed here was fluorescein,332 MW (Molecular Probes). The dye was dissolved to aconcentration of co=50 lM in sodium (bi)carbonate buffer

at pH=9.0, I=0.050 M. Rhodamine B was also used inpreliminary tests. Although microbubble lensing-inducedphotobleaching (l-BLIP) was clearly observed with rho-damine B, its fluorescence intensity is strongly tempera-ture dependant. To isolate photobleaching from thermaleffects, fluorescein (which is relatively insensitive to tem-perature) was used. Immediately before use, all solutionswere filtered using 0.2-lm pore size syringe filters.

3Results and discussion

3.1Microbubble lensing-induced photobleaching (l-BLIP)When light is incident on an interface, it refracts accordingto Snells Law (Hecht 2001)

n1sin h1 n2sin h2; 1

wheren and h are the index of refraction and transmissionangle (relative to the local interface normal), respectively,and subscripts 1 and 2 denote the medium. Thus, when thelight is traveling from a higher-index medium (n1) to alower-index medium (n2), the light bends away from thenormal. Light incident at the critical angle h1=hc isrefracted along the interface (h2=90) where

hc sin1 n1=n2 ; n1> n2 2

Any light incident at an angle greater than the criticalangle is totally reflected back into the higher-index med-ium. The effect of this phenomenon, termed total internalreflection (TIR), is illustrated in the backlit image shownin Fig. 1a. The edges of the air bubble are visible in thetransparent aqueous solution because the index ofrefraction of air (nair=1.0) is less than that of water(nwater=1.3). The edges of the bubble appear dark because

the light (from below) is reflected in this region, where theangle of incidence relative to the local interface normal isgreater than the critical angle ofhc=50. This darkeningdoes not occur in the center of the bubble where the anglebetween the local interfacial normal and the incident lightis less than the critical angle (i.e. the interface is moreperpendicular to the incident light). The same phenomena,though to a lesser extent, darkens the channelliquid

interface where nchannel=1.46510 nm) is passed to the camera. In a similarmanner, the blue excitation light from the microscope isreflected at the bubble interface as illustrated in Fig. 2.Figure 2a illustrates that uniform excitation light exposurein a liquid-only capillary would cause a spatially uniformrate of fluorescence, and in turn, a spatially uniform rate ofphotobleaching. The addition of an air bubble alters the

Fig. 1a,b. Images of a bubble: a With backlighting. b Withfluorescent emission from the liquid phase

Fig. 2a,b.Illustrative schematics of optical phenomena:a Uniformphotobleaching in a liquid-filled channel.b Increased photoble-aching in the near-bubble liquid because of the presence of abubble

0


4/10

path of the excitation light depending on its angle ofincidence with respect to the local interface normal. Thus,excitation light may be refracted or reflected at the inter-face as illustrated in Fig. 2b. Light incident at anglesgreater than the critical angle is reflected back into theliquid as shown on the right in Fig. 2b. Light incident atangles less than the critical angle is refracted at bothinterfaces and transmitted back into the liquid as shown

on the left in Fig. 2b. In both ways the intensity of theexcitation light is increased in the liquid near the bubble.This results in higher total excitation light exposure, ini-tially higher fluorescence intensity, and a higher rate ofphotobleaching, which in time results in lower fluores-cence intensity. This local photodestruction is the result ofmicrobubble lensing-induced photobleaching (l-BLIP).

It is important to note that Fig. 2 and the aboveexplanation are simplifications of the real case. The two-dimensional model of the bubble cap is reasonable con-sidering the hemispherical geometry of the interface.However, the excitation light beam is, in reality, conical,due to the high numerical aperture objective used. Thisadds a layer of complexity to the optical analysis, but thephysics remains the same as that illustrated in Fig. 2. Inaddition, lensing and increased photobleaching in theliquid film between the bubble and the wall (similar towaveguide or fiber-optic light transmission) is alsoexpected. Because of the small thickness of the film,however, the volume of liquid photobleached in this regiondoes not contribute significantly compared to thatphotobleached at the bubble caps. For the same reason, theextent of photobleaching is essentially independent ofbubble length.

The image sequence in Fig. 3 demonstrates the l-BLIPprocess. Figure 3a is an image of the uniform fluorescentemission of the dye-filled channel without a bubble. Theimage in Fig. 3b was taken shortly after the bubble wasmoved into position. Although both caps contributeequally to the photobleaching, the cameras field of view isfocused on the right cap, the point of which has a bright-spot reflection as discussed previously. In Fig. 3b thefluorescent emission appears fairly uniform throughoutthe liquid at the level of the original dye concentrationwithout a bubble (Fig. 3a). The images in Fig. 3bf weretaken in sequence at 20/15-s (1.33-s) intervals, and pro-cessed identically. Significant darkening in the near-bub-ble liquid is apparent. The radial orientation of the dark/bright ray pattern (believed to be caused by interference)is further evidence of the bubble lensing phenomena. After

5.3 s (Fig. 3f), a significant photobleached region isformed around the bubble, where as the dye at the right-hand side of the camera view has retained the originalintensity level. This is more apparent in the axial con-centration profiles shown in Fig. 4, where the five profilescorrespond to the image sequence in Fig. 3bf. As shownin Fig. 4, the bulk of the photobleaching occurs during thefirst 2 s. The broadening of the dark region beyond that ismostly due to diffusion of the photobleached dye.Increasing the excitation light energy and correspondinglydecreasing the exposure time could reduce this axialbroadening. Allowing time for radial diffusion, however,creates a more uniform, axially symmetric photobleached

region. For the most part, experimental parameters such asthese (solutions, light intensities, exposure durations) werechosen out of convenience, and it is likely that perfor-mance could be improved with optimization. One optionis to use high molecular weight dyes such as fluorescein-dextran conjugates (Molecular Probes), which are lesssusceptible to diffusion than standard dyes.

The image in Fig. 3g was taken after an axial electricfield was applied to the channel. The resulting electro-osmotic liquid flow transformed the dye photobleached atboth bubble caps into a dark cross-stream liquid flow

Fig. 3. a Image of fluorescent emission from a liquid-filledcapillary.bf Image sequence demonstrating the l-BLIP process.gImage of photobleached dye advected with electro-osmotic flow

Fig. 4. Axial fluorescence intensity profiles corresponding to theimage sequence in Fig. 3bf. The bulk of the photobleachingoccurs close to the interface and in the first few seconds of

exposure


5/10

marker. The electrokinetic transport of the bulk liquid inthis system is calculated numerically. These simulationresults, particularly the shape of the electric field, establishthat it is reasonable to assume that the electrophoreticvelocity of the marker is purely axial at distances greaterthan 1 diameter from the bubble cap. Scalar image ve-locimetry is then applied to a l-BLIP-generated flowmarker for experimental determination of the cross-

stream liquid velocity profile in the microchannel. Accu-rate cross-stream velocity measurements were obtainedwith this technique purely by experimental means.

3.2Numerical analysisIn order to develop the underlying physics of the l-BLIPprocess, a series of numerical simulations were conductedusing the BLOCS (Bio-lab-on-a-chip simulation) finiteelement code (Sinton et al. 2002a; Erickson and Li 2002a;Erickson and Li 2002b). Here we provide some brief detailson the numerical method, concentrating on the imple-mentation of the code to this application. For further de-

tails, including comments on general verification of thecode and computational expense, the reader is referred toErickson and Li (2002a). In these simulations we consid-ered a long axially symmetric capillary, with a diameterequivalent to those used in the aforementioned experi-ments (100 lm), with a 200-lm long stationary bubblelocated at the midpoint. The liquid domain geometrywas then discretized using nine-noded biquadratic

elements, which were significantly refined in the regionnear the bubble and coarsened near the capillary entranceand exit.

The first stage of the numerical analysis is the deter-mination of the applied potential field in the liquid system,which, for the case of a constant conductivity electrolyte,can be determined from the Laplace equation

r2/ 0; 3

where / is the applied potential field. Requiring that thesolution remain finite at the capillary axis and applyinginsulation conditions, //n wheren is the normal to thesurface, were used along the bubble and capillary walls.The electric field lines generated for the case equivalent to2600 V applied over the 15-cm capillary (consistent withthe experiments discussed in Sect. 3.3) are shown inFig. 5a. Since the liquid cross-sectional area is significantlyreduced, and thus the local channel resistance is greatlyincreased, the electric field lines are concentrated in thethin film that surrounds the bubble. This has the effect ofincreasing the gradient of/ within this region and

reducing it far away from the bubble. This is demonstratedin Fig. 5b, which shows the change in/ along the length ofthe capillary for the no-bubble case, a bubble with a 0.5-lm film thickness, and a bubble with a 0.1-lm filmthickness. The presence of the bubble reduces //x to86% and 96% of that for the no-bubble case for the 0.1-lmfilm thickness and 0.5-lm film thickness, respectively.Also of interest in Fig. 5a is the shape of the isopotential

Fig. 5. a Numerically determined isopo-tential lines in the presence of an insulatingbubble.b Influence of bubble film thick-ness on the global applied electric field

2


6/10

lines, which assume a slightly curved shape beyond thethin film on either side of the bubble. At a distance lessthat one capillary diameter from the edge of the bubble,however, they are nearly perpendicular to the channel wall.This limited influence of the bubble on the shape of theisopotential field lines facilitates the use ofl-BLIP as amicroflow visualization technique.

With the potential field solution developed, the flow

field can be determined using the low-Reynolds-numberStokes equations, Eq. 4a and the compressibility conditionEq. 4b

0 rplr2u; 4a

r u 0; 4b

whereu is the fluid velocity,p is the pressure and l is theviscosity. In principal, Eq. (4a) should contain an electricalbody force term resulting from the application of theelectric field to the net charge density in the double layer.In this analysis, however, we assume an electro-osmoticslip velocityueo=meoE, where meo is the electro-osmotic

mobility (taken as 6.2x10

)8

m

2

/Vs (Duffy et al. 1998)) andE is the gradient of/ tangential to the surface, appliedalong the capillary wall. This simplification eliminates thedouble-layer formulation and has proven accurate intransport cases, such as that examined here, where thedouble-layer thickness is very thin compared with thecapillary diameter (Sinton et al. 2002a; Erickson and Li2002a). In this case the condition is slightly more stringent

as the double-layer thickness must be thinner than the filmsurrounding the bubble. In the case examined the ionicstrength of the buffers is on the order of 10-2 M, whichgives us a double-layer thickness on the order of 3 nm. Asshown in the proceeding section, this is several orders ofmagnitude smaller than the film thickness encounteredhere. Applying this slip boundary condition along thecapillary wall, a bounded condition along the capillary

axis, and a zero tangential shear condition along thebubble (Erickson et al. 2002c) yields the near-bubble flowfield shown in Fig. 6a. The velocity profiles at 0.1, 0.4 and1.0 capillary diameters away from the edge of the bubbleare shown in more detail in Fig. 6b. As can be seen, verynear the bubble the velocity close to the capillary wall isslightly higher than that at the center; however, it veryrapidly evolves to the traditional plug-type velocity profileexpected for electro-osmotic flow. This result is used as averification of the microflow visualization techniquedeveloped in the following section.

3.3Microflow visualization with a l-BLIP-generatedflow markerMethods for inferring a bulk fluid velocity by analyzing asequence of images of dye transport fall under the broadcategory of scalar image velocimetry (Dahm et al. 1992;Tokumaru and Dimotakis 1995). In general, the goal ofthese methods is to extract the bulk fluid velocity vector

Fig. 6. a Numerically determined electro-osmotic flow field in the near bubbleregion. b X-direction velocity profiles atvarious distances from the bubble edge.L denotes distance from the edge of thebubble


7/10

u from the advective terms of the mass conservationequation satisfied by the imaged dye species

@c

@tu rc Dr2c; 5

wherecis the concentration of dye species, and D is thediffusion coefficient. The analysis may be greatly simpli-fied when applied to internal, fully-developed flows by

using a discrete dye sample. This is because the motion ofthe locus of dye concentration maxima is, for the mostpart, not affected by diffusion and reflects the bulk motionof the fluid. This maximum concentration trackingmethod has been applied in caged-dye-based microfluidicflow visualization studies (Herr et al. 2000; Sinton et al.2002b). In those cases, a photoinjection of fluorescent dyeprovided the only nonzero dye concentration cin anotherwise nonfluorescent solution. In the case ofl-BLIP,however, the solution contains a uniform dye concentra-tion,c=c0, and a portion of the dye is photochemicallydestroyed. Since the fluorescent dye concentration cis aconserved quantity, the concentration of photodestroyed

dyec is also conserved. Thus Eq. (5) is satisfied for c=c,where

c c0c : 6

Thus maximum concentration tracking methods can bedirectly applied to inverted and linearly scaled intensityimages of a photobleached sample. Equivalently one maythink of the conserved quantity as the photodestroyedfluorophores whose concentration is quantified by a lackof fluorescence.

Once the near-bubble liquid was photobleached (asshown in Fig. 3), an axial electric field was applied bysetting the upstream (left) reservoir potential to 2600 V,

and connecting the downstream (right) reservoir toground. The camera recorded the resulting transport ofthe photobleached sample. A five-image sequence takenat 1/15 s intervals is given in Fig. 7a. The electro-os-motic liquid flow has combined the dye photobleachedat both bubble caps into a dark, cross-stream, band,which is advected downstream with the electro-osmoticflow. The bubble itself is shown to have a finite velocityin the direction opposite to the electro-osmotic liquidflow.

The concentration field of photobleached dye c wascalculated for each image using image-processing soft-ware developed in-house. Axial marker concentration

profiles of the images shown in Fig. 7a are given inFig. 7b. The vertical lines on the left in Fig. 7b indicatethe motion of the interface, and the profiles on the rightshow the marker transport. A relatively clear concentra-tion maximum is apparent in each profile. To determinethe bubble velocity the distance between the vertical linesmay be divided by the corresponding time step. Here abubble velocity ofububble=)190 lm/s was determined. Todetermine the velocity profile in the liquid region, thepoint of maximum concentration was located along eachaxial line of pixels, using a weighted average of thehighest concentration values in the liquid phase. The setof maximum concentration points from each image

formed the cross-stream concentration maxima profilesshown in Fig. 7c. In a given sequence, any pair of con-centration profiles could provide a velocity distributionby dividing the distance between them by the corre-sponding time-step. Here, all five profiles were used in anerror-weighted average to determine the velocity data.Once calculated, this velocity data represents theobserved velocity of the marker uob. Since the dyemolecules are charged, the observed velocity of the

Fig. 7. aAn image sequence of the l-BLIP-generated flow markerin the electro-osmotic flow. b Corresponding axialconcentration profiles of the flow marker.c Correspondingcross-stream profiles of marker concentration maxima

4


8/10

concentration maxima is the summation of the bulk fluidvelocity ueo and the electrophoretic velocity of the dyeuph as follows:

uob ueouph: 7

The electrophoretic velocityuphmust be calculated di-rectly from the electrophoretic mobilitymphand the ap-plied electrical field strength E as

uphmphE: 8

The electrophoretic mobility of fluorescein may be ta-ken asmph=)3.310

)8 m2/Vs (Paul et al. 1998; Sinton et al.2002b). The local axial applied electric field, however, re-quires special consideration here. Through the numericalsimulations it was shown that the isopotential lines be-came almost totally radial less than one diameter awayfrom each bubble cap (Fig. 5a). Thus it is reasonable toassume that the electrophoretic velocity of the marker ispurely axial (aligned with the potential gradient vector E

*

).In the case of a channel containing no bubble and filled

with a uniform electrolyte solution, the magnitude of theaxial electric field may be calculated simply by dividing theapplied voltage differential DVby the length of the channelL. As shown in the numerically determined axial potentialprofiles in Fig. 5b, the bubble adds resistance to thechannel and hence causes a field reduction in the liquid faraway from the bubble. To experimentally determine thisreduced field strength in the liquid, the current was mea-sured both with and without the bubble present, and thefollowing calculation performed:

EB ENBiB

iNB

; 9

wherei is the electrical current draw, subscript B indicatesthe presence of a bubble, and subscript NB indicates thatno bubble is present. The bracketed current ratio inEq. (9) is the factor by which the electric field is reducedfrom the overall value (DV/L) due to the presence of thebubble. A plot of the currents measured with and withoutthe bubble present versus the overall applied electricalfield strength is given in Fig. 8. Both curves trend slightlyupwards because of Joule heating-induced increases influid temperature (Swinney and Bornhop 2002; Sinton andLi 2003). The current ratio (in Eq. (9)), however, wasfound to be relatively constant at 0.850.025 over thisrange of applied field. Thus according to these current

measurements and Eq. (9), the applied voltage drop ofDV=2600 V over the L=0.15-m length of capillary gener-ated an electrical field strength ofE=14.7 kV/m in theliquid region. This is consistent with the decrease inE observed for the 0.1-lm film thickness case shown inFig. 5b, suggesting the film thickness is of this order.Using thisE value to calculate the electrophoretic markervelocity in Eq. (8), and substituting the result into Eq. (7),gave the bulk liquid electro-osmotic velocity profile shownin Fig. 9. The flat plug-like profile observed is character-istic of electro-osmotic flows and is in keeping with thecorresponding numerically predicted velocity profile inFig. 6. From the wall velocity, the electro-osmotic mobility

was calculated to be meo=6.810)8 m2/Vs, in reasonable

agreement with the value ofmeo=6.210)8 m2/Vs reported

by Duffy et al. (1998). The plug-like velocity profileachieved with the l-BLIP-generated flow marker extendsto within 4 lm from each wall. This degree of near-wallresolution is comparable to, and in many cases improvedover, that achieved with caged-dye based microflow visu-

alization techniques that involve increased infrastructureand specialized chemicals. In the general context ofmicroflow visualization techniques, a clear disadvantage ofl-BLIP is the requirement that a bubble be present tofacilitate the photoinjection of the flow marker. It wasdemonstrated here, however, that measurements of thecross-stream velocity profile in the microchannel can beobtained, free from bubble-induced two-dimensionalelectrical and hydrodynamic effects. Thus, this techniqueis a candidate for studying a variety of other aspects ofelectrokinetic channel flow for which caged-dye imaginghas been employed such as the effects of joule heating,band spreading in corners, and pressure disturbances. An

Fig. 8. A plot of the current values measured versus the overallelectrical fields applied to the channel with and without a bubblepresent

Fig. 9. Cross-stream electro-osmotic liquid velocity profileobtained by applying scalar image velocimetry to the transport ofthe l-BLIP-generated flow marker


9/10

advantage of this technique is the ability to concurrentlyimage the bubble geometry, bubble velocity, and the localfluid velocity in multiphase systems. In this case, theeffective film thickness was both measured (throughcurrent measurements) and calculated (through numericalsimulations shown in Fig. 5) to be less than 1 lm. Near thebubble caps, the interface is optically indistinguishablefrom the wall, indicating that the film in that area is less

than 1 lm. However, the images in Fig. 7a also show sig-nificant film thickening in the middle region of the bubble.This film thickening is a result of elongation of the bubble,which occurs upon application of the applied electric field.We expect that this elongation and the motion of thebubble are caused by the presence of mobile ionic surfacecharges inducing a differential net charge on the bubblecaps, potentially in combination with changes in surfacetension because of the presence of the high electric fieldstrength in the film (electrocapillary effects). Film thick-ening results in lower liquid velocities in that portion ofthe film. By integrating the cross-stream velocity profile(Fig. 9), and determining the bubble velocity from theinterface movement in Fig. 7b, the liquid film velocity wasdetermined to be 3.3 mm/s at the bubble midpoint and anestimated 30 mm/s near the bubble caps. This demon-strates the applicability of this technique to the study ofcoupled dynamic transport phenomena, characteristic ofmicroscale multiphase systems.

4ConclusionsIn this work, a method to intensify light exposure in thenear-bubble liquid using the optical characteristics of thegasliquid interface is proposed and demonstratedexperimentally. The method takes advantage of the cur-vature and the step change in properties across a gasliquid interface to create a lens/mirror optical arrangementin which light incident on the bubble is focused into thesurrounding liquid, resulting in a locally increased totallight exposure. The effect is demonstrated experimentallyby imaging the increased photobleaching rate of fluoro-phores in the near-bubble region.

Based on these phenomena, a microbubble lensing-in-duced photobleaching (l-BLIP) technique is then devel-oped and applied as a method to inject a marker for flowvisualization. A series of numerical simulations on themultiphase system demonstrated that both the isopotentiallines and the flow field are disturbed by the presence of thebubble, however, the effect is limited to the near-bubble

region (typically less than 1 capillary diameter from theedge of the bubble). Using the l-BLIP technique, theelectrokinetic transport of the photobleached marker isanalyzed to determine the cross-channel velocity profile ofthe liquid phase and the liquid velocity in the film. Theseresults are in good agreement with the numerical predic-tions and are consistent with velocity measurements fromprevious studies. The numerical simulations establish thatit is reasonable to assume that the electrophoretic velocityof the marker is purely axial (aligned with the potentialgradient vector E

*

) at distances greater than 1 diameterfrom the bubble cap. As demonstrated, accurate cross-stream velocity measurements were obtained with this

technique purely by experimental means. This contribu-tion represents both a new application of microscale gasliquid interfacial phenomena and a new technique formicrofluidic flow visualization, particularly applicable to(though not limited to) the study of multiphase micro-channel flows.

ReferencesBecker PL (1996) Quantitative fluorescence measurements. In: WangXE, Herman B (ed) Fluorescence imaging spectroscopy andmicroscopy. Wiley, New York

Brethreton FP (1961) The motion of long bubbles in tubes. J FluidMech 10:166188

Chang HC (2002) Bubble/drop transport in microchannels. In: Gad-el-Hak M (ed) The MEMs handbook. CRC, Boca Raton

Dahm WJA, Su LK, Southerland KB (1992) A scalar imaging veloci-metry technique for fully resolved four-dimensional vectorvelocity fluid measurements in turbulent flows. Phys Fluids A4:21912206

Duffy DC, McDonald JC, Schueller OJA, Whitesides GM (1998) Rapidprototyping of microfluidic systems in poly(dimethylsiloxane).Anal Chem 70:49744984

Erickson D, Li D (2002a) Influence of surface heterogeneity onelectrokinetically driven microfluidic mixing. Langmuir 18:18831892

Erickson D, Li D (2002b) Microchannel flow with patch-wise andperiodic surface heterogeneity. Langmuir 18:89498959

Erickson D, Li D, Park C (2002c) Numerical simulations of capillarydriven flows in nonuniform cross sectional capillaries. J ColloidInterface Sci 250:422430

Ghadiali SN, Gaver DP (2000) An investigation of pulmonary sur-factant physicochemical behavior under airway reopening condi-tions. J Appl Physiol 88:493506

Hecht E (2001) Optics, 4th edn. Addison-Wesley, New YorkHerr AE, Molho JI, Santiago JG, Mungal MG, Kenny TW, Garguilo MG

(2000) Electro-osmotic capillary flow with nonuniform zeta po-tential. Anal Chem 72:10531057

Johnson TJ, Ross D, Gaitan M, Locascio LE (2001) Laser modificationof preformed polymer microchannels: application to reduce bandbroadening around turns subject to electrokinetic flow. AnalChem 73:36563661

Jun TK, Kim CJ (1998) Valveless pumping using traversing vaporbubbles in microchannels. J Appl Phys 83:56585664

Lempert WR, Magee K, Ronney P, Gee KR, Haugland RP (1995) Flowtagging velocimetry in incompressible flow using photoactivatednonintrusive tracking of molecular motion (PHANTOMM). ExpFluids 18:249257

Meinhart CD, Wereley ST, Santiago JG (1999) PIV measurements of amicrochannel flow. Exp Fluids 27:414419

Molho JI, Herr AE, Mosier BP, Santiago JG, Kenny TW (2001) Opti-mization of turn geometries for microchip electrophoresis. AnalChem 73:13501360

Mosier BP, Molho JI, Santiago JG (2002) Photobleached-fluorescenceimaging for microflows. Exp Fluids 33:545554

Paul PH, Garguilo MG, Rakestraw DJ (1998) Imaging of pressure- andelectrokinetically driven flows through open capillaries. Anal

Chem 70:24592467Ratulowski J, Chang HC (1989) Transport of gas bubbles in capil-

laries. Phys Fluids A 1:16421655Ratulowski J, Chang HC (1990) Marangoni effects of trace impurities

on the motion of long gas bubbles in capillaries. J Fluid Mech210:303328

Ross D, Johnson TJ, Locascio LE (2001) Imaging of electro-osmoticflow in plastic microchannels. Anal Chem 73:25092515

Russ JC (1999) The image processing handbook, 3rd edn. CRC, BocaRaton

Santiago JG, Wereley ST, Meinhart CD, Beebe DJ, Adrian RJ (1998) Aparticle image velocimetry system for microfluidics. Exp Fluids25:316319

Sinton D, Erickson D, Li D (2002a) Photoinjection based sample de-sign and electro-osmotic transport in microchannels. J MicromechMicroeng 12:898904

6


10/10

Sinton D, Escobedo-Canseco C, Ren L, Li D (2002b) Direct andindirect electro-osmotic flow velocity measurements in micro-channels. J Colloid Interface Sci 254:184189

Sinton D, Li D (2003) Electro-osmotic velocity profiles in micro-channels. Colloids Surf A, in press

Swinney K, Bornhop DJ (2002) Quantification and evaluation of Jouleheating in on-chip capillary electrophoresis. Electrophoresis23:613620

Takhistov P, Indeikina A, Chang HC (2002) Electrokinetic displace-ment of air bubbles in microchannels. Phys Fluids 14:114

Taylor JA, Yeung ES (1993) Imaging of hydrodynamic and electro-kinetic flow profiles in capillaries. Anal Chem 65:29282932

Tokumaru PT, Dimotakis PE (1995) Image correlation velocimetry.Exp Fluids 19:115

Zhao B, Moore JS, Beebe DJ (2001) Surface-directed liquid flow insidemicrochannels. Science 291:10231026

Documents

erickson_expfluids.pdf