AC electrothermal manipulation of conductive fluids and particles for lab-chip applications

AC electrothermal manipulation of conductive fluidsand particles for lab-chip applications

M. Lian, N. Islam and J. Wu

Abstract: AC electrokinetics has shown great potential for microfluidic functions such aspumping, mixing and concentrating particles. So far, electrokinetics are typically applied onfluids that are not too conductive (,0.02 S/m), which excludes most biofluidic applications. Tosolve this problem, this paper seeks to apply AC electrothermal (ACET) effect to manipulateconductive fluids and particles within. ACET generates temperature gradients in the fluids, andconsequently induces space charges that move in electric fields and produce microflows. Thispaper reports two new ACET devices, a parallel plate particle trap and an asymmetric electrodemicropump. Preliminary experiments were performed on fluids with conductivity at 0.224 S/m.Particle trapping and micropumping were demonstrated at low voltages, reaching �100 mm/sfor no more than 8 Vrms at 200 kHz. The fluid velocity was found to depend on the appliedvoltage as V4, and the maxima were observed to be �20 mm above the electrodes.

1 Introduction

Microfluidic electrokinetics (EK) is gaining popularity as anactuation mechanism for lab-on-a-chip, owing to its simpleimplementation and reliability from no moving parts. EKcan be applied with DC or AC electric sources. Directcurrent electrokinetics (DC EK) has a long history of devel-opment, being investigated and applied extensively [1].However, DC EK suffers from high voltage operation(several kilovolts) and consequently excessive electroche-mical reactions and electrolysis at the electrodes. In thelast few years, alternating current electrokinetics (AC EK)receives increasing research interest as it has demonstratedgreat potential for microfluidic actuation. Compared withDC EK, AC EK has the following advantages and features:(1) low operating voltage (,10 Vrms against several kilo-volts), (2) alternating electric fields, minimising electrolysisand chemical reactions, and (3) non-uniform streamlines,which can be used to convect and mix fluids.

An AC electric field can interact with polarisable par-ticles and fluids to set them into motion, which is knownas AC EK [2, 3]. AC EK mainly includes dielectrophoresis(DEP), AC electro-osmosis (ACEO) and AC electrothermal(ACET) effect. DEP refers to the interaction between adipole moment on a particle and a non-uniform field [4].This technique has been studied in great detail for controlledmanipulation of particles, binary separation and characteris-ation of particles [5–8]. A particle experiencing DEP willexhibit a velocity as uDEP ’ 0.03(a21/h)(V2/r3), where eand h are the fluid permittivity and viscosity, respectively,a is the particle diameter, r is the distance between the par-ticle and electrode, and V is the applied voltage. It can beseen that DEP velocity is size dependent, and decreasesrapidly with the distance to the electrode.

# The Institution of Engineering and Technology 2007

doi:10.1049/iet-nbt:20060022

Paper first received 15th December 2006 and in revised form 16th March 2007

The authors are with the Department of Electrical and Computer Engineering,The University of Tennessee, Knoxville, TN 37996, USA

E-mail: [email protected]

36

Both ACEO and ACET apply a non-uniform electric fieldto produce fluid flow [2, 3]. In some cases, ACEO andACET produce very similar flow patterns, but they are ofdifferent origin. ACEO [9, 10] arises from the movementof ions in the electric double layer at the electrode/electro-lyte interface, producing microflows because of the fluidviscosity. Pioneer work and comprehensive review onACEO can be found in [2, 3, 9, 10]. The fluid velocity isapproximately given by uACEO ¼ 2(1/h) . Dj . Et, where1 and h are the permittivity and viscosity of the medium,Et is the electric fields parallel to the solid surface, and Djis the voltage drop over the charge layer, which is pro-portional to induced charge density. So ACEO requiresboth normal and tangential components of electric field atthe electrode surface, which leads to the frequency depen-dency of ACEO effect. At low frequencies, most of theapplied voltage drops across the double layer, ACEO isimportant. At high frequencies, electrode charging is negli-gible and ACEO becomes insignificant. Data in [9] alsoshow that ACEO is pronounced for frequencies lowerthan 100 kHz, beyond which its effect is minimal and canbe neglected. The optimal frequency for ACEO operationcan be estimated as V ’ (s/2p1)(lD/l ), where lD is theDebye length, typically less than 10 nm, and l is the charac-teristic length of the system, for example the electrodespacing. lD/l represents the process of charging thedouble layer through the resistive fluid bulk.ACEO velocity uEO � V2/r, and the resulting fluid flow

exerts a drag force on particles. Therefore ACEO can beused to transport particles as well as fluids, and there isno size dependency. Particle manipulation and fluid flowcontrol using ACEO have been reported in various forms,such as biased ACEO [11, 12], and 3D ACEO pump [13,14], travelling wave ACEO pump [15], asymmetric elec-trode ACEO pump [16] and particle traps [17–20]. ACEOhas been demonstrated to generate surface velocity of 50–100 mm/s by applying AC signals in the order of 1 Vrmswith an electrode spacing of 5–25 mm. However, complexityin fluid dynamics takes place at higher voltages, forexample, flow reversal. It was tentatively attributed toFaradaic charging of the double layer (i.e. generation of

IET Nanobiotechnol., 2007, 1, (3), pp. 36–43

co-ions by electrochemical reaction), as ACEO originatesfrom the electric stress at the electrode/electrolyteinterface.ACEO is typically limited to fluids with low ionic

strength, for example de-ionised water. High conductivitycompresses the thickness of the double layer, renderingelectro-osmosis ineffective. For the same applied voltage,more conductive fluids have a lower peak velocity at V. Ithas been reported that ACEO is not observable for fluidswith conductivity above 140 mS/m [16].However, labs-on-a-chip frequently involve samples with

conductivity higher than 0.1 S/m. Biological applicationsregularly use saline solutions (1–2 S/m). So it is very desir-able to develop an electrokinetic technique suitable forconductive fluids, and ACET effect has shown promise inthis aspect [21–25].ACET arises from non-uniform electric fields and temp-

erature gradients in the fluids, which produces spacecharges that move under the influence of electric fieldsand consequently induce microflows. In the case of aplanar electrode pair, ACET velocity can be approximatelyexpressed as uET ¼ 5 � 1024(@s/s@T )(1s/kh)(V4/r),where s and k are the electric and thermal conductivitiesof the fluid, respectively. Higher conductivity leads tohigher fluid velocity because of increased heat generationand temperature gradients.Electrothermal effect was first utilised in macro appli-

cations. Development of miniaturised ACET devices washindered by the fact that ACET could become suppressedat microscale. ACET exerts volume force to the fluid bulkand obeys no-slip boundary condition, unlike ACEOwhich works by surface stress. Too small a hydraulic diam-eter will inhibit the formation of ACET flow [2, 3]. One ofthe earliest microscale prototypes was reported by Fuhret al. in early 1990s, which applied a travelling wave offour-phase signals to drive fluids [21]. Using an electrodearray of 10 mm width and 10 mm gap, a peak velocity of�180 mm/s was achieved at 35 Vp-p, 100 kHz. The conduc-tivity of fluid they used was up to 9.3 mS/m, which wasmuch higher than 1026 S/m reported for electrohydrody-namic micropumps at that time. Recently, ACET startedto attract research interest again because of its increasingimportance in shrinking electrokinetic chips [22]. Furtherminiaturisation leads to higher energy density being dissi-pated in the fluid, and the heating source could be external(e.g. illumination) or internal Joule heating [23]. A veryrecent paper [24] by Gonzalez et al. has studied in greatdetail the ACET effect in the presence of a constant externaltemperature gradient (strong illumination), and in this caseuET ’ 3 � 1023(1V2/h)j@ T/@yk(1/s) (@s/@T ), wherej@T/@yj is the external thermal gradient. The experimentsverified many theoretic predictions for ACET, forexample fluid velocity changes with illumination intensityand exhibits quadratic relationship with voltage. With a sol-ution of s ¼ 2.5 mS/m, a peak velocity of 80–110 mm/s at10 Vrms was observed. The above suggests that ACET flowcan be controlled and is strong enough to be utilised.This paper reports two new types of ACET devices, a par-

ticle trap using two electrode plates that face each other inparallel at 500 microns separation, and an ACET micro-pump based on asymmetric interdigitated electrode arrays.No external temperature gradient is provided, as the fluor-escent microscope was at its lowest setting in our exper-iments. So a lower fluid velocity is expected of ourexperiments, and uET / V4. Also, much higher fluidconductivities were used, since the goal is to developACEK devices for conductive fluids.

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

The concept of particle trapping is to induce microflowsto convey particles from the bulk of the fluid onto theelectrode surface, where the fluid slows down and depositthe particles. Although DEP has been extensively used forparticle manipulation, the technique will be much moreeffective if it works in conjunction with ACET or ACEO[19]. The effectiveness of DEP, or DEP force, scales withthe particle volume, which makes DEP unfavourablewhen handling submicron particles, whereas ACET andACEO use fluid flows to convect particles to certainlocations, hence has no dependence on particle sizes.Moreover, DEP decreases much faster than ACET/ACEOwith respect to the distance from the electrode (approxi-mately 1/r3 for DEP and 1/r for ACET).

There are prior reports on convecting and trapping par-ticles with electrohydrodynamic methods. Most of themuse a side-by-side electrode configuration [25], and arebased on ACEO [18–20], which limits their application tolow ionic strength solutions and the trapping range intothe fluid is in the order of the system characteristic length.Our ACET particle trap consists of a pair of electrodesthat face each other. The electrodes have a spacing of500 mm, exerting force on the micro-litre fluid betweenthem for particle collection, which is desirable for manytypes of lab-chip sample processing.

Our work demonstrates ACET microfluidic devicescapable of manipulating fluids and particles at a fluid con-ductivity of interest to biochemical analysis and environ-mental monitoring. Preliminary data on the ACET particletrap and pump will be presented for working fluids withs ¼ 0.224 S/m. ACET experiments have been successfullyperformed on fluids with conductivity as high as 1.58 S/m(phosphate buffered saline). However, the flow motionwas not recorded well enough to be analysed, as flow rateincreases linearly with fluid conductivity and becomes toohigh to be measured.

2 ACET effect

ACET effect refers to fluid motion induced by temperaturegradients in the fluid in the presence of AC electric fields.When an electric field E is applied over the fluid with elec-trical conductivity s, Joule heating of the fluid will takeplace according to the energy balance equation

kr2T þ1

2ksE2l ¼ 0 (1)

where T is temperature and k is the thermal conductivity.For microsystems, heat convection is small compared toheat diffusion [2, 3]. So here the temperature equationassumes the simplified form with Joule heating as theenergy source. If the field strength E is non-uniform, therewill be spatial variation in heat generation, which leads totemperature gradients rT in the fluid. The temperature gra-dient rT produces spatial gradient in local conductivity andpermittivity by r1 ¼ (@1/@T )rT and rs ¼ (@s/@T )rT.Further, rs and r1 generate mobile space charges, r, inthe fluid bulk, by r ¼ r . (1E) ¼ r1 . Eþ 1r . E and @r/@tþ r.(sE) ¼ 0 with @/@t ¼ iv in AC fields. The freespace charges experience a volume force fE in the electricfield E, fE ¼ rE2 1/2jEj2r1, exerting force on fluidthrough viscosity and leading to fluid transport. The timeaverage of electric force on fluid is given as [10]

kFetl ¼ �1

2

�rs

s�r1

1

��

1jEj2

1þ (vt)2�1

4r1jEj

2 (2)

37

where s and 1 are the electrical conductivity and permittiv-ity of the medium, t ¼ 1/s is its charge relaxation time, andv ¼ 2pf is radian frequency. For aqueous media at 293K,we have [22]

1

1

@1

@T¼ �0:004¼)

r1

1¼

1

1

@1

@TrT ¼ �0:004rT (3)

1

s

@s

@T¼ 0:02¼)

rs

s¼

1

s

@s

@TrT ¼ �0:02rT (4)

giving

kFetl ¼ �0:012 � rT �1jEj

2

1þ (vt)2þ 0:001 � rT � 1jEj

2 (5)

The two terms at right hand side of (2) and (5) represent theCoulomb force and dielectric force, respectively. At lowfrequency, 1/1þ (w t) 2 ’ 1, Coulomb force is much stron-ger (about 11 times) than dielectric force. Coulomb forcedecreases with frequency, and becomes less than dielectricforce at high frequency. Consequently, the flow will reverseits direction. The crossover frequency can be expressed asfc ¼

ffiffiffiffiffi11

p/2pt. The solution used in our experiments has

a conductivity of s ¼ 0.224 S m21, so its crossover fre-quency is estimated to be �1.7 � 108 Hz. With ACsignals at 200 kHz, the dielectric force is much smallerthan the Coulomb force, so the flow direction is dictatedby the Coulomb force. In the meanwhile, the applied ACsignals have high enough frequency, so that doubly layercharging can be ignored and ACEO is negligible. The char-ging time of the double layer is estimated as (1/s) . (l/lD).In this work, the electrode separation is 500 mm (parallelplate) and 20 mm (pump). At an ionic strength of 17.8mM NaCl (s ¼ 0.224 S m21), the Debye length is calcu-lated to be less than 2.3 nm, and the double layer has a char-ging frequency much lower than 200 kHz.

The induced fluid motion is described by Navier-Stokes(N-S) equation

r@u

@tþ r(r � u)u� hr2uþ rP ¼ Fet (6)

where r is the fluid density, h is the dynamic viscosity of thefluid, P is the external pressure and u is the velocity of thefluid. Since we are considering fluid motion in a microsys-tem, that is, low Reynold number, the time-averaged fluidvelocity can be found by the simplified N-S equation as

� hr2uþ rP ¼ kFetl (7)

Together with r . u ¼ 0 for incompressible fluid, (7) is laterused in numerical simulation to obtain the fluid flow fieldsin our ACET devices.

This paper presents two new ACET devices, a long rangeparticle trap and a micropump for conductive fluids. Bothnumeric simulation and experiments have been performedto design and prototype the new microfluidic devices, andfunctional devices have been demonstrated with prelimi-nary experiments.

3 Parallel plate particle trap

Our ACET particle trap consists of a pair of parallel plateelectrodes at a spacing of 500 mm, as shown in Fig. 1a.The bottom electrode is shorter than the top one in orderto create a non-uniform electric field, which in turninduces non-uniform temperature field and, further, fluidicflow field. Microflows convey particles in the bulk solutionto the shorter electrode and deposit them in its centre,increasing particle count at that location.

38

3.1 Numerical simulation

Parallel plate ACET trap was simulated with Femlab/Comsol Multiphysics package (www.comsol.com;Stockholm, Sweden). The model used in the simulation isshown in Fig. 1b, including one bottom electrode with alength of 600 mm. Periodic boundary conditions for allthe variables (electric fields, temperature and velocity)were applied to the sides.There are several steps involved in the numerical simu-

lation. The first one is to use Laplace’s equation r2V ¼ 0

to derive the electric field distribution in the fluidicchamber. The resulting electric field distribution is used tocalculate the temperature field according to the energyequation (2). Then the fluid volume force in the chamberis calculated using the temperature gradient and the electricfield distribution from the first two steps. Lastly, the fluidflow field is obtained by N-S equation (7).In electric field simulation, the top and bottom electrodes

are set at certain electric potentials, whereas other bound-aries are left at neutral and insulative. Then the heat transfermodule is invoked, using the resulting E to calculate thetemperature field. The top and bottom boundaries alongwith electrodes are set at ambient temperature. Treatingthe electrodes as isothermal is appropriate consideringtheir sub-micron thickness [25]. In the fluid dynamicmodule, an AC frequency of 200 kHz is used to calculatethe fluid volume force, which corresponds to our exper-imental conditions. The top and bottom boundaries aretreated as non-slip, that is, v ¼ 0, and the left and rightwalls are defined to be periodic.The electric field simulation (Fig. 2a) indicate the exist-

ence of electric field strength (E ¼ 2rV ) maxima aroundthe edge of the bottom electrode, with both tangential andnormal components. Since there is no flow into the electro-des, it is the tangential component of the electric field thatinduces the microflow. For a potential difference of 20.3Vp2p, the highest electric field strength is calculated as0.522 V/mm (Fig. 2a). Accordingly, the temperature gradi-ent takes the highest value of 0.609 K/mm at the samelocations, shown by the arrows in Fig. 2b. The peak tempera-ture rise is calculated to be 2.44 K, comparable to 2.3 Kby the analytical predictions DT ¼ sV2

rms/8k in [2].

Fig. 1 Schematic of parallel plate particle trap and its simu-lation model

a Schematic of microfluidic chamberb Simulation model

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

Fig. 2 Results from electrical field and thermal simulation

a Electrical field profile shows the maxima of field strength at the edges of the bottom electrode. The direction of ACET microflow follows those ofelectric fieldb Temperature distribution. The maximum temperature rise is about 2.4 K at 20 Vp2p. The temperature gradient (in arrows) has its maxima close tothe electrodes

The result of fluidic simulation is shown in Fig. 3, wherethe arrow length scales with the fluid velocity. Two counter-rotating vortices are produced above the bottom electrodes.The flow direction is upwards in the middle and downwardsat the sides. Global velocity maxima are seen at the edge ofbottom electrode, where the electrical field strength is at thehighest. Flow velocity decreases along the electrodeinwards until it becomes zero at the middle. Fluid motionsin the middle of bottom boundary are cancelled out byflows in opposite directions and consequently producingstagnation. It is expected that micro/nano particles aredeposited in that area.

3.2 Experiments

Parallel particle trap presented uses a silicon wafer with pat-terned Au layer as the bottom electrode and an indium-tinoxide glass slide as the top electrode, as shown schemati-cally in Fig. 1a. A 500mm thick spacer (PC8R-0.5, GraceBio-Labs, Inc.) is placed between the electrodes to formthe microfluidic chamber. NaCl solution (s ¼ 0.224S/m)seeded with 500 nm fluorescent particles (molecular

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

probes) is used as the working fluid. Particle motion isrecorded through a Nikon eclipse LV100 microscope anda CCD camera in real time. The images are then processedby Image Pro 3DS (www.mediacy.com, Cybernetics Inc.) toextract particle velocity information. A signal generator(Agilent 33220A function generator, Agilent Technology,CO, USA) and an amplifier (Model 354-1-50, HeicoCompany) are used to provide a desired voltage level forelectrothermal experiments. Voltages of 11.9222.7 Vp2p

at 200 kHz were used in the experiments.Particles started to make fast circular movements after an

AC signal is applied (e.g. 12 Vp2p, 200 kHz). At the wafersurface, the particles moved from the edge of the conductivearea towards its centre. When the focal plane of the micro-scope was elevated to be above the wafer, the flow direc-tions were reversed. So it is certain that the fluid wasmoving in vortices.

Figs. 4a and b compare the wafer surface before and 5min after applying an AC signal. Within a few minutes, alarge proportion of the particles became immobilised andaccumulated at the stagnation points/lines. As shown inFig. 4b, particles deposited along the field minima,

39

Fig. 3 Simulated flow field profile in a parallel plate particle trap

Above the bottom electrode, two counter-rotating vortices are formed that move from the electrode edges inwards

forming the cell-like structures. For the same reason,particles also form smaller clusters at the centres of non-conductive regions. It can also be seen that some particlesaccumulate at the edges of electrodes (bright lines), indicat-ing the existence of positive DEP effect.

The trapping effect was also quantified. A rectangle areaof 360 � 225 mm2, with its location indicated in Fig. 4b,was used to measure the particle deposition. Fig. 5 showsthe number of trapped particles as a function of time. A par-ticle concentration of 6 � 105particles/mm3 was used in theexperiment. At the beginning of experiment (t ¼ 1 s), aninitial reading of 30 particles was obtained in the specifiedrectangle. The number of trapped particles shows a linearrelationship with time at a positive slope with a trappingrate of about 0.5 particles/s. The particle count tripled att ¼ 120s.

4 ACET micropump

The above experiments also demonstrate that ACET pro-vides an effective method to manipulate fluids. With anapplied voltage of 20 Vp2p, the velocity of ACET flow isin the order of 100 mm/s at the edge of the electrode. Soit is natural to explore ACET effect for micropumping.

Our research indicates that the net pumping motion canbe achieved by asymmetric co-planar electrode arrays.Prior research has shown that reflective-symmetric, counter-rotating vortices can be generated over a pair of identical

Fig. 4 Wafer surface before and after AC signals being appliedto generate convection and to trap particles

Particles were directed towards null points of electric fields, andbecame trapped (Bright areas indicate high density of particles.)a Before applying AC voltageb 5 min after applying AC voltage

40

electrodes [22]. Unequal width of electrodes in a pairbreaks the symmetry of electric and thermal field distri-bution, resulting in net velocity vector and therefore thenet flow.The electrode array used in our simulation and exper-

iments has the dimensions of 100 mm narrow electrode/20 mm gap/180 mm wide electrode/100 mm in-pair gap.The height of the chamber is 500 mm. One period of theACET pump was simulated, with periodic boundary con-ditions applied to the sides (inlet and outlet). For otherboundaries, boundary conditions similar to the paralleltrap were applied.Fig. 9 shows the fluid field profile from 0 to 500 mm

above the electrodes. The initial fluid velocity is set tozero. The fluid motions are generated by applied electricsignals, and the net flow is directed from the narrower elec-trode towards the wider one.The simulation results have been verified by experiments.

Asymmetric electrode arrays were patterned by lift-off ontosilicon wafers. Polymer channels were used. Solution ofsodium chloride (s ¼ 0.224S/m) was again used as theworking fluid. To indicate the microflow, 1mm latex par-ticles were injected into one end of a pre-filled channel.The particles were carried along by the fluid and transportedfrom one end of the channel to the other (Fig. 6), when an

Fig. 5 Number of particles collected in designated area as afunction of time

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

AC signal of 18.8 Vp2p at 200 kHz was applied. Imagesequence in Fig. 10 illustrates the pumping action on a par-ticle cluster advancing through electrode pairs (Fig. 7),which was taken at �20 mm above the wafer surface. Theheight of the focal plane was controlled through a compu-terised stepper (Optic scan II, CS152Z, Prior ScientificInstrument LTD). The average fluid velocity was approxi-mately 117 mm/S.

Fig. 6 Simulated flow pattern over a pair of asymmetric electro-des by AC electrothermal effect

Net fluid transport is generated. Maximum velocity is 162 mm/s at15.6 Vp2p

Fig. 7 Image sequence showing a particle cluster advancingthrough the electrodes

The focal plane is �20 microns above the wafer. The image colourwas reversed to illustrate the particle more clearly (Dark areas are elec-trodes at 400 mm pitch.)a t ¼ 0 sb t ¼ 1 sc t ¼ 1.8 sd t ¼ 2.3 se t ¼ 2.9 sf t ¼ 3.4 s. V ¼ 18.8 Vp2p at 200 kHz

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

The micropump design presented here has not been opti-mised. Research on asymmetric ACEO micropumps indi-cates that a width ratio of �7 for the electrode pair seemsto produce a high flow rate for a given voltage [16]. Sinceboth types of micropumps are based on the geometry asym-metry, various ratios will be investigated in the future toimprove pumping efficiency.

The channel height is another important factor. Our studyshows that fluid motion becomes suppressed when thechannel height is reduced from 500 to 200 mm. This is inagreement with other groups’ observation [2]. This attributeis opposite to that of ACEO pumps, which produce highersurface velocity with decreased channel height.

5 Discussion

In theory, ACET velocity is expected to follow a quarticrelationship with respect to applied voltage. The voltagedependency of fluid velocity was experimentally studiedin this research by varying the applied voltage from11.9222.7 Vp2p at 200 kHz. These voltages are the actualpotential drop across two electrodes, measured by an oscil-loscope. For each voltage settings, four velocity readingswere taken and averaged to reduce the effect of Brownianmotion.

Our experiment has shown that the fastest particlemotion, that is, fluid velocity, occurs at �20 micronsabove the wafer surface, which is in agreement with thesimulation results. This is consistent with the characteristicsof ACET flows, since electrothermal effect induces volumeforce on the fluid, and fluid velocity at the boundaries is zeroaccording to no-slip condition.

Fig. 8 gives a comparison between velocities from theor-etical prediction (quadratic dependency to applied voltage0.000303 Vp2p

4 or 0.0194 Vrms4 ), simulation and experimental

measurements. The three sets of data exhibit a close agree-ment. Simulated velocities are slightly higher than exper-imental data. However, at 12 Vp2p, the experimentsyielded a velocity slightly higher than calculation and simu-lation. That is probably because of other forces such asACEO and DEP. ACEO can generate almost identicalfluid flow patterns. Even though ACEO is weak in fluidswith high ionic strength, it is possible that it contributes tothe slightly higher velocity. As the applied voltageincreases, ACEO fluid velocity goes up as V2, whereasuACET / V4, the contribution from ACEO becomes lessnoticeable.

Fig. 8 Particle velocity as a function of applied voltage

Three data sets are from simulation (triangle), curve-fitting (dot) andexperiments (square)

41

Our work also uses impedance analysis to assess the rela-tive importance of double layer charging (ACEO) at theelectrode surfaces and space charge force (ACET) in thefluid bulk. A pair of electrodes in a fluidic cell can be elec-trically represented by the equivalent circuit as shown inFig. 9. At the interface of electrolyte and electrodes, thereare double layer capacitances, Cdl, for charging at the inter-face (which do not behave ideally, but act like a constantphase element (CPE) as defined in [9, 14].) The fluid bulkis treated as a resistor Rsolu and electricity passing it gener-ates heat according to Ohm’s law. It is in series with theinterfacial impedances on both ends. Ccell represents direct

Fig. 9 RC equivalent circuits of an electrode/fluid system

Fig. 10 Impedance spectra of the ACEO devices from 100 Hz to30 MHz

a Magnitudeb Phase

42

dielectric coupling between electrodes and its value is deter-mined by dielectric properties of the fluid.At low frequencies, the impedance from Cdl is much

larger than that from Rsolu. The system exhibits mostlycapacitive characteristics, and most of the voltage dropacross the interfacial double layer, which is desired forACEO techniques. As the frequency increases, the impe-dance from double layer capacitances and constant phaseimpedances decrease, and less voltage drops over the elec-trode/electrolyte interfaces. More voltage drops across thefluid bulk, resulting in higher current flowing through theresistive fluid, and consequently larger temperature gradi-ent. As a result, fluid volume force from ACET effectstarts to dominate. Therefore resistive characteristics at asufficiently high frequency indicate the dominance ofACET effects. As the frequency increases further, thedielectric coupling between the two electrodes willbecome dominant. The whole system exhibits capacitivecharacteristics again.Impedance analysis from 100 Hz to 30 MHz was applied

to both the parallel plate particle trap and the asymmetricmicropump (Fig. 10). For the plate trap, the magnitude ofimpedance decreases from 1472 V at 100 Hz to 79 V at50 kHz and remains constant to several MHz (Fig 10a)and the phase angle changes from 274.831 to 22.149 atcorresponding frequencies (Fig. 10b). The same character-istics are also observed for the pump. The impedancemeasurement shows a frequency range from 50 kHz to afew MHz to be suitable for ACET operation. The phaseangle at low frequency does not go to 908. It is because ofthe non-ideality of the double layer capacitance (CPE)and also the frequency is not low enough. The impedanceanalysis has also been performed on micropump presentedabove. The overall impedance gives 26 V /2128 at200 kHz. Therefore at the operating frequency of200 kHz, the system is electrically functioning as a resistorand electrothermal effect dominates.Electrothermal effect, as the name implies, will lead to

temperature rise in the device. The electrodes are depositedon a 400 silicon wafer, which is a good thermal conductor.During the experiments, the silicon wafer acting as a heatsink dissipates the heat generated in fluid. The temperaturerise was measured by an infrared thermometer (model52224, Mastercool, Inc.) to be �2.1 K for the set up inFig. 2b, close to the predicted value. This small temperaturerise is insufficient to produce observable buoyancyphenomenon.

6 Conclusion

This paper presents two new ACET devices to achievemicrofluidic functions of particle trapping and pumping ata fluid conductivity of interest to biochemical analysis.Preliminary experiments of ACET devices were successful.As theoretically predicted, fluid velocity scales to the fourthpower of applied voltage, and it increases with fluid conduc-tivity. Effective particle and fluid manipulation weredemonstrated. Improved performance of ACET devices isexpected with optimised design. The advance with ACETdevices will greatly expand the application scope of electro-kinetics in microfluidic chips.

7 Acknowledgment

The project has been supported by the US National ScienceFoundation under grant number ECS-0448896 andTennessee Science Alliance Award.

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

8 References

1 Li, D.: ‘Electrokinetics in microfluidics’ (Elsevier, 2004)2 Ramos, A., Morgan, H., Green, N.G., and Castellanos, A.: ‘AC

electrokinetics: a review of forces in microelectrode structures’,J. Phys. D: Appl. Phys., 31, pp. 2338–2353

3 Castellanos, A., Ramos, A., Gonzalez, A., Green, N.G., andMorgan, H.: ‘Electrohydrodynamics and dielectrophoresis inmicrosystems: scaling laws’, J. Phys. D: Appl. Phys., 2003, 36,pp. 2584–2597

4 Pohl, H.A.: ‘Dielectrophoresis’ (Cambridge University Press,Cambridge, 1995)

5 Pethig, R.: ‘Application of A.C. electrical fields to the manipulationand characterisation of cells’ in Karube, I. (Ed.): ‘Automation inbiotechnology’ (Elsevier, 1991), pp. 159–185

6 Gascoyne, P.R.C., and Vykoukal, J.V.: ‘Dielectrophoresis-basedsample handling in general-purpose programmable diagnosticinstruments’, Proc. IEEE, 2004, 92, pp. 22–42

7 Morgan, H., Hughes, M.P., and Green, N.G.: ‘Separation of submicronbioparticles by dielectrophoresis’, Biophys. J, 1999, 77, pp. 516–525

8 Ramos, A., Morgan, H., Green, N.G., and Castellanos, A.: ‘The role ofelectrohydrodynamic forces in the dielectrophoretic manipulation andseparation of particles’, J. Electrostatics, 1999, 47, pp. 71–81

9 Green, N.G., Ramos, A., Gonzalez, A., Morgan, H., and Castellanos,A.: ‘Fluid flow induced by nonuniform ac electric fields in electrolyteson microelectrodes. I. Experimental measurements’, Phys. Rev. E,2000, 61, pp. 4011–4018

10 Green, N.G., Ramos, A., Gonzalez, A., Morgan, H., and Castellanos,A.: ‘Fluid flow induced by nonuniform ac electric fields in electrolyteson microelectrodes. III. Observation of streamlines and numericalsimulation’, Phys. Rev. E, 2002, 66, p. 026305

11 Wu, J.: ‘Electrokinetic microfluidics for on-chip bioparticleprocessing’, IEEE Trans. Nanotech., 2006, 5, (2), pp. 84–89

12 Lian, M., Islam, N., and Wu, J.: ‘Particle line assembly/patterning bymicrofluidic AC electroosmosis’, J. Phys.: Conf. Series, 2006, 34,pp. 589–594

13 Bazant, M.Z., and Ben, Y.: ‘Theoretical prediction of fast 3D ACelectro-osmotic pumps’, Lab Chip, 2006, 6, pp. 1455–1461

IET Nanobiotechnol., Vol. 1, No. 3, June 2007

14 Urbanski, J.P., Thorsen, T., Levitan, J.A., and Bazant, M.Z.: ‘Fast ACelectro-osmotic pumps with non-planar electrodes’, Appl. Phys. Lett.,2006, 89, p. 143508

15 Ramos, A., Morgan, H., Green, N.G., Gonzalez, A., and Castellanos,A.: ‘Pumping of liquids with traveling-wave electroosmosis’, J. Appl.Phys., 2005, 97, p. 084906

16 Studer, V., Pepin, A., Chen, Y., and Ajdari, A.: ‘An integrated ACelectrokinetic pump in a microfluidic loop for fast tunable flowcontrol’, Analyst, 2004, 129, pp. 944–949

17 Wu, J., Islam, N., and Lian, M.: ‘High sensitivity particle detection bybiased ac electro-osmotic trapping on cantilever’. 19th IEEE Int. Conf.Micro Electro Mechanical Systems (MEMS 2006), Istanbul, Turkey,January 2006, pp. 566–569

18 Wong, P.K., Chen, C.-Y., Wang, T.-H., and Ho, C.-M.: ‘An ACelectroosmotic processor for biomolecules’. TRANSDUCERS’ 03,June 2003, pp. 20–23

19 Hoettges, K.F., McDonnel, M.B., and Hughes, M.P.: ‘Use ofcombined dielectrophoretic/electrohydrodynamic forces forbiosensor enhancement’, J. Phys. D: Appl. Phys., 2003, 36, pp.L101–L104

20 Wu, J., Ben, Y., Battigelli, D., and Chang, H.-C.: ‘Long-range ACelectrokinetic trapping and detection of bioparticles’, Indust. Eng.Chem. Res., 2005, 44, (8), pp. 2815–2822

21 Fuhr, G., Schnelle, T., and Wagner, B.: ‘Traveling wave-drivenmicrofabricated electrohydrodynamic pumps for liquids’,J. Micromech. Microeng., 1994, 4, pp. 217–226

22 Green, N.G., Ramos, A., Gonzalez, A., Castellanos, A., and Morgan,H.: ‘Electrothermally induced fluid flow on microelectrodes’,J. Electrostat., 2001, 53, pp. 71–87

23 Green, N.G., Ramos, A., Gonzalez, A., Castellanos, A., and Morgan,H.: ‘Electric field induced fluid flow on microelectrodes: the effect ofillumination’, J. Phys. D: Appl. Phys., 2000, 33, pp. L13–L17

24 Gonzalez, A., Ramos, A., Morgan, H., Green, N., and Castellanos, A.:‘Electrothermal flows generated by alternating and rotating electricfields in Microsystems’, J. Fluid Mech., 2006, 564, pp. 415–433

25 Sigurdson, M., Wang, D.Z., and Meinhart, C.D.: ‘Electrothermalstirring for heterogeneous immunoassays’, Lab Chip, 2005, 5,pp. 1366–1373

43