Microuid Nanouid (2008) 5:1321 DOI 10.

1007/s10404-007-0201-4

RESEARCH PAPER

A numerical study of an electrothermal vortex enhanced micromixer

J. Cao P. Cheng F. J. Hong

Received: 4 June 2007 / Accepted: 17 July 2007 / Published online: 1 September 2007 Springer-Verlag 2007

Abstract Temperature gradients aroused from the Joule heating in a non-uniform electrical eld can induce inhomogeneities of electric conductivity and permittivity of the electrolyte, thus causing an electrothermal force that generates ow motion. A 2D numerical investigation of a micromixer, utilizing electrothermal effect to enhance its mixing efciency, is proposed in this paper. Results for temperature and velocity distributions, as well as sample concentration distribution are obtained for an electrolyte solution in a microchannel with different pairs of electrodes under AC potentials with various frequencies. Numerical solutions were rst carried out for one pair of electrodes, with a length of 10 lm separated by a gap of 10 lm, on one side wall of a microchannel having a length of 200 lm and a height of 50 lm. It is found that the electrothermal ow effect, in the frequency range for which Coulomb force is predominant, induces vortex motion near the electrodes, thus stirring the ow streams and enhancing its mixing efciency. If more than one pair of electrodes is located on the opposite walls of the microchannel, the mixing efciency depends on the AC potential applied pattern and the electrodes arrangement pattern. The distance between two pairs of electrodes on two opposite walls is then optimized numerically. Sample mixing efciencies, using KCl solutions as the working uid in microchannels with different number of electrodes pairs at optimal electrodes arrangement pattern, are also investigated. If root mean squared voltages of 10 V in an AC frequency range of 0.110 MHz are imposed on 16 pairs of electrodes separated at an optimal distance, the numerical

results show that a mixing efciency of 98% can be achieved at the end of the microchannel having a length of 700 lm and a height of 50 lm at Re = 0.01 PeC = 100, and PeT = 0.07. However, the mixing efciency decreases sharply at a frequency higher than 10 MHz owing to the drastically decrease in the Coulomb force. Keywords Electrothermal ow Micromixer Numerical simulation AC frequency List of symbols a1 distance between channel entrance and the left side of the rst electrode on M side wall a2 distance between channel entrance and the left side of the rst electrode on N side wall Da distance difference C sample concentration C0 sample concentration prole with completely unmixed state C? sample concentration prole with completely mixed state C1in sample concentration at the upper part of the inlet C2in sample concentration at the lower part of the inlet Cpm heat capacity of the uid d1 length of the electrodes d2 gap between the two electrodes in one pair dt channel length used by two pairs of electrodes at the two side walls D sample diffusion coefcient fc cross over frequency k thermal conductivity of the uid L channel length p pressure patm atmospheric pressure ! u velocity

J. Cao P. Cheng (&) F. J. Hong School of Mechanical and Power Engineering, Shanghai Jiao Tong University, Shanghai 200240, China e-mail: pingcheng@sjtu.edu.cn

123

14

Microuid Nanouid (2008) 5:1321

u0 e V Vrms VI VR W

inlet velocity AC potential root mean squared value of the AC potential imaginary part of the AC potential real part of the AC potential channel height

Greek symbols r electric conductivity of the uid e the permittivity of the uid er relative permittivity of the uid e0 vacuum permittivity g dynamic viscosity of the uid c mixing efciency parameter qm density of the uid

1 Introduction During the past decade, microuidic devices or Lab on a chip have attracted considerable attention for use in chemical and biological analyses. Compared with traditional laboratory techniques, microuidic devices offer many advantages, including reduced reagent consumption, a more rapid analysis and reduction of biological and chemical wastes. Sample mixing is one of the important processes in these devices for complete biochemical reactions (Reyes et al. 2002; Vilkner et al. 2004; Jahn et al. 2004). However, as the ow in these devices is usually at low Reynolds numbers, laminar ow rather than turbulent ow often occurs in microuidic chips. Since diffusion is the primary mechanism for mixing in a low Reynolds number laminar ow, it would take a long time and require a long microchannel to accomplish the mixing process. For this reason, the search for more efcient mixers for microuidic systems has been given a great deal of attention. In recent years, a number of enhanced microuidic mixing devices have been developed (Ramos et al. 1998; Liu et al. 2000; Erickson and Li 2002a, b; Wang et al. 2003; Chang and Yang 2004; Fu et al. 2005; Shin et al. 2005; Coleman et al. 2006). These devices can be categorized into passive mixers and active mixers. Passive mixers typically use particular channel geometry congurations to increase the interfacial area between liquids to be mixed. For example, Wang et al. (2003) presented a numerical investigation of mixing in microchannels with patterned grooves. They found that microchannels with patterned grooves could cause rotation in uid streams, thus enhancing the mixing efciency. A 3D serpentine channel that can achieve mixing through chaotic advection was proposed by Liu et al. (2000), who found that this technique was efcient when the Reynolds number of the ow was increased to 70. However, this device has a

complicated 3D structure that is not easy for micro-fabrication. The introduction of oppositely charged surface heterogeneities on microchannel walls can result in regions of localized ow circulation within the bulk ow. Erickson and Li (2002a, b) studied these circulation regions numerically, and found that heterogeneous surfaces could enhance mixing efciency in T-shape microchannels. Chang et al. (2004) presented a numerical investigation on electrokinetically driven ow mixing in microchannels with various numbers of incorporated rectangular blocks. They found that the introduction of rectangular blocks within the mixing channel enhanced sample mixing efciency only slightly, but patterning heterogeneous upper surfaces on the rectangular blocks was an effective method of enhancing the sample mixing efciency. On the other hand, active mixers utilize additional forces to generate disturbances in the ow eld to improve mixing efciency. For example, Fu et al. (2005) proposed the use of electroosmotic forces generated by the electrodes imposed with direct electric current in a double-T-form microuidic mixer. They found that the mixing efciency could be as high as 95% within a mixing length of 1,000 lm downstream from the secondary T-junction when a 100 V/cm driving electrical eld strength and a 2 Hz periodic switching frequency were applied. Shin et al. (2005) investigated experimentally a micromixer using time periodic electric eld to excite the instability in a cross channel, and found that the combined use of a static eld and an alternating eld could improve mixing effect. Coleman et al. (2006) demonstrated experimentally a eld switch microuidic mixing strategy, which utilized symmetric sequential injection geometry with an expansion chamber to achieve high mixing efciencies. Recently, AC electrokinetics generated by AC currents on the electrodes to manipulate the uid and particles has been given more and more attention. Compared with electrokinetics generated by direct currents, this technique has the advantages of operating at lower voltages in the range of 120 V (Ramos et al. 1998; Green et al. 2001; Hansen 2004; Perch-Nielsen et al. 2004; Brown et al. 2006; Urbanski et al. 2006; Chen and Du 2006; Gonzalez et al. 1954, 2006). When an AC electrical potential is imposed on the microelectrodes on the wall of the microchannels, the following two forces will be induced in the electrolyte solution depending on the frequency of the AC currents (Ramos et al. 1998): (1) AC electroosmosis is caused by the forces of the tangential component of the non-uniform AC electric elds on induced charges in the diffuse double layer of the electrolyte solution (Brown et al. 2006; Urbanski et al. 2006). Hansen (2004) showed that the effect of AC electroosmosis occurred mainly in

123

Microuid Nanouid (2008) 5:1321

15

(2)

the low frequency range of 100100 kHz, and diminished at other frequencies. The electrothermal ow arises when temperature gradients are generated in the electrolyte solution because of Joule heating, which induce the changes of the conductivity and permittivity of the solution (Ramos et al. 1998; Brown et al. 2006; Urbanski et al. 2006; Chen and Du 2006; Gonzalez et al. 2006). This gives rise to a time-averaged electrothermal force per unit volume given by Ramos et al. (1998]: 1a

1 ea b 1 rT EE eajEj2 rT h Fe i 2 1 2pf e=r2 4 where a oe=oT=e; b or=oT=r

1b

respectively with a height of 50 lm. Computations were carried out for the electrodes having an imposed voltage of 10 and 10 V (root mean square value) at a frequency range from 0.1 to 40 MHz. This frequency range is higher than 100 kHz for which the AC electroosmosis effect can be neglected (Perch-Nielsen et al. 2004), and is below the crossover frequency of 74.6 MHz for which the Coulomb force is predominant. The temperature distribution and the ow streamlines as well as the mixing efciency of this micromixer are presented. It is found that if the microchannel has 16 pairs of such electrodes, a mixing efciency of 98% can be achieved at the end of the microchannels at a frequency range of 0.110 MHz. However, the mixing efciency decreases drastically at a frequency higher than 10 MHz because of the decrease of the Coulomb force.

with r and e being the electric conductivity and the permittivity of the uid, both are functions of temperature. As shown from Eq. (1a), the electrothermal force becomes zero when the change of permittivity and electric conductivity with respect to temperature are zero (i.e., a = 0 and b = 0). Note that the rst term on the right hand side of Eq. (1a) denotes the Coulomb force which depends on the frequency and the second term denotes the dielectric force (Chen and Du 2006). Letting the rst term equal to the second term in Eq. (1a) gives the cross over frequency fc: fc r p 1 2b=a 2pe 2

2 Formulations 2.1 Governing equations A 2D model of the proposed micro-mixer is schematically shown in Fig. 1. Two uids with different concentrations of samples from two inlets at upstream (not shown) are converged into a single microchannel having a length L and a height of W. Electrodes pairs connected with the function generator are fabricated on opposite side walls of the microchannel, and the distances between the channel entrance and the left side of the rst electrode on two side walls M and N are a1and a2, respectively, with the distance difference Da = a2 a1. The lengths of all the electrodes are d1, the gap between the two electrodes in one pair is d2, and the distance between two pairs of electrodes at two opposite side walls is dt. The thickness of the electrodes is assumed to be innitely thin. A nonuniform electric eld is generated in the electrolyte solution when AC voltages with the phase of 0 and 180

which shows that the cross over frequency depends only on physical properties of the electrolytic solution. For example, if KCL (with r = 100 mS m1, a = 0.4% K1, b = 2.0% K1, e = 80e0) is used as a working uid, the cross over frequency given by Eq. (2) is fc = 74.6 MHz. Chen and Du (2006) showed that the electrothermal force has its maximum effects in a frequency range below the cross over frequency where the Coulomb force dominates. In this paper, we propose a microuidic mixer using an array of electrodes fabricated on the walls of a microchannel, which is imposed with an AC current operating in the frequency range in which the Coulomb force is predominant. Under this condition, the electrothermal force gives rise to a swirling ow pattern in the microchannel, which enhances the mixing of uids having different concentrations of sample. Computations were carried out for two streams of uids having sample concentrations of C1in and C2in entering microchannels at a uniform velocity u0, driven by a pressure gradient. Pairs of electrodes, having a length of 10 lm with a separate distance of 10 lm, are located on opposite walls of a microchannel having a length of 200 and 700 lm,

y
L

a2
side N C=C1in W C=C2in

0
a1

d1 d2
dt

side M

Fig. 1 Schematic of the mixing channel with electrodes at the two opposite side walls

123

16

Microuid Nanouid (2008) 5:1321

are applied to the adjacent electrodes. The non-uniform electric eld gives rise to an inhomogeneous temperature eld, leading to variations of the temperature-dependent electric conductivity and permittivity of the uid. This results in an electrothermal force consisting of the Coulomb force and the dielectric force that generates uid motion. If the imposed frequency is other than 100 100 kHz, the electrothermal effect rather than electroosmosis effect dominates (Ramos et al. 1998). Under these assumptions, the mathematical model given by Chen and Du (2006) can be used for numerical investigation of the problem under consideration. The Laplaces equation is used to determine the electric eld, which can be expressed as e r2 V 0 3

solution components. The convectiondiffusion equation for the sample under steady ow condition is ! rC Dr2 C 0 u 8

where C is the concentration of sample and D s the sample mass diffusivities. The degree of mixing efciency at any cross section in the mixing channel is given by the parameter c (Erickson and Li 2002) RW cx 1
0 RW 0

jC C1 jdy

! 100% 9

jC0 C1 jdy

e where V VR iVI indicates the potential phasor. For the two phase electrode array in the model, the imaginary part VI is zero, so only the real part VR needs to be solved by the Laplaces equation (Ramos et al. 1998; Chen and Du 2006]. For the 2D problem under consideration, Eq. (3) can be simplied to r2 V 0 4

where C denotes the concentration prole of the sample across the width of the mixing channel, C0 and C? denote the sample concentration proles with completely unmixed and completely mixed states, respectively. In order to obtain dimensionless forms of the above equations, the following dimensionless variables are introduced x y V u x ; y ; V ; u ; d d Vref u0 v p patm T C v ; p ; T ; C ; u0 Tref Cref qref u2 0 10 where patm is the atmospheric pressure,Vref, qref, Tref, and Cref denote the reference electrical potential, uid density, temperature and sample concentrations, respectively. Therefore, Eqs. (3)(7) in dimensionless form are expressed as r2 V 0
PeT ! r T r2 T Ajr V j2 u

where V = VR enotes the voltage applied to the electrodes. The temperature of the uid can be calculated by the following energy equation with a source term because of the electrical heat generation (Ramos et al. 1998) qm Cpm ! rT kr2 T rjrVj2 u 5

where qm, Cpm, k and r are the density, heat capacity, thermal conductivity and the electric conductivity of the uid, respectively. The inhomogeneities of medium properties induce a body force that affect the velocity eld of the uid. For an incompressible uid at low Reynolds numbers, the steady NavierStokes equation with the electrothermal body force and with inertia force neglected can then be expressed as r!0 u rp gr u hFe i 0
2!

11 12 13

r! 0 u

1 2 ! r u r p Re h i B Rr T r V r V jr V j2 r T
u Pec ! r C r2 C

14 15

6 7

where p is the pressure; g is the dynamic viscosity of the uid; ! is the uid velocity, and the body force hFei is u given by Eq. (1a). The mass transfer of the sample particles in the mixing channel is accomplished by two mechanisms including convection and diffusion. In this analysis, it is assumed that no adsorption of sample particles onto the mixing channel wall and no interaction between sample and the electrolyte

where Re qf u0 d=g; Pec u0 d=D; PeT qm Cpm u0 d=k (with d being the length of the electrode) are the Reynolds number, the Peclet number for the energy equation and for the concentration equation, respectively. The other three dimensionless parameters relating to the electric eld are 2 2 2 A rVref =kTref ; B eaTref Vref = 4d2 qref Uref ; and 2 R 21 b=a=1 2pf e=r ; with the last dimensionless parameter denoting the relative importance of the Coulomb force to the dielectric force.

123

Microuid Nanouid (2008) 5:1321

17

2.2 Boundary conditions Figure 1 shows a 2D microchannel with two opposite walls M and N having a length L, separating by a distance W. The electric potentials on each pair of the electrodes are set as the root mean squared values Vrms and Vrms, respectively, and the other boundaries are set as electrical insulation condition. Two streams with sample concentrations C* = 0 and C* = 1, driven by a pressure difference, 1in 2in enter the microchannels at x* = 0 at uniform inlet velocity u0. The outlet boundary condition for concentration is qC*/ qx* = 0, and all other boundaries are assumed to be insulated. The mixed uid ows to zero pressure at the outlet, and no-slip boundary conditions are imposed on two opposite side walls. The temperatures of the electrodes, as well as the inlet temperature of the uid are all assumed to be 298 K, and the boundary condition qT*/qx* = 0 is imposed at the outlet. Thermal insulation boundary conditions are imposed on other microchannel walls.

(corresponding to 0.1 < f < 40 MHz), respectively. The FEM method was used to solve the controlling equations. All the numerical results presented in the following have been checked for grid-independence.

3 Results and discussions 3.1 The mixing effect of a channel with one pair of electrodes The microchannel with one pair of electrodes located at side wall M is used to study the mixing efciency of the electrothermal ow. Results of the computations for L* = 20, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105, and R = 12 (0.1 < f < 1 MHz) are presented in Figs. 2, 3, 4. Figure 2 shows the temperature distribution and thermal contour of the computational domain. From simulation results, the highest temperature appears at the gap between the electrodes, and the largest temperature increase in the electrolyte is about 7.8 K, which has little inuence on the sample and is therefore acceptable. The streamline and the sample concentration distributions are presented in Fig. 3. It is shown that two asymmetric vortices induced by the electrothermal effect are formed at the top region of the electrodes, and the electrolyte at the upper part of the channel with zero concentration is drawn to the lower region, which stirs the ow and promotes the mixing process, thus enhancing the mixing efciency. The sample concentration proles at the channel inlet and outlet with and without electrodes are presented in Fig. 4. If the AC electric eld is not applied, it is noted that sample mixing effect in the microchannel is weak and mixing is primarily from the diffusion mechanism. Under this situation, the mixing efciency at the outlet of the channel is about 17.4% according to Eq. (9). When one pair of electrodes is added at the side wall, the mixing efciency is enhanced to 33.3% because of the vortex motion induced by the electrothermal effect. However, the mixing efciency is still very low when one pair of electrodes is added at one side of the channel wall. It is expected that higher mixing efciencies can be achieved if more electrodes are added on the walls of the microchannel.

2.3 Computation parameters Computations were carried out for all electrodes and the gap between each pair of electrodes in the mixer have the length of d = d1 = d2 = 10 lm (see Fig. 1). The width of the microchannel is W = 50 lm, and the length of the channel depends on the number of electrodes pairs. a1 = 50 lm, and the value of a2 as well as the distance difference Da are varied to study their effects on the mixing efciency. The voltages applied at the adjacent electrodes are 10 and 10 V, respectively, while the frequency is in the range of 0.140 MHz. The inlet velocity of the uid is assumed to be u0 = 1 mm s1, and sample concentrations at the upper and lower half part of the inlet are C1in = 0 and C2in = 1 mol m3, respectively. The diffusion coefcient of the sample (Rhodamine B) is D = 1010 m2 s1 (Fu et al. 2005). The KCl electrolyte solution is used as the working uid, and the physical properties of the solution are given by: Cpm = 4,180 J kg1 K1, qm = 1,000 kg m3, r = 100 ms m1, g = 0.001 Pa s, k 0.61 W m1 K1, a = 0.4% K1 and b = 2.0% K1 (Chen and Du 2006). If the temperature variation in the electrolytic solution is small, e (as well as r) can be assumed to be constant and can be calculated approximately from e = e0 er (Chen and Du 2006), where er and e0 denote the relative permittivity of the liquid and the vacuum permittivity (i.e., er = 80, e0 = 8.85 1012 C V1 m1), respectively. The reference quantities are chosen to be Tref = 298 K Vref = 10 V, qref = 1,000 kg m3, Cref = 1 mol m3. Thus, the dimensionless parameters used in the present study are L* = 20, 70, W* = 5, d* = d* = 1, A = 0.055, B = 2.11 1 2 105, Re = 0.01, Pec = 100, PeT = 0.07, and 2.88 < R < 12

299.6K 301.9K 301.1K 300.4K 302.7K 303.5K 302.7K

298.8K

Fig. 2 Temperature eld in the mixing channel with one pair of electrodes for L* = 20, W* = 5,Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

123

18

Microuid Nanouid (2008) 5:1321

Fig. 3 Streamline and concentration distributions in the mixing channel with one pair of electrodes for L* = 20, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

3.2 Effects of AC potential and electrodes arrangement patterns In order to get higher mixing efciency, another pair of electrodes is added at the opposite wall N of the microchannel. The voltage applied pattern, as well as the position of the second pair of electrodes a2 relative to the rst one located at a1 (i.e., Da = a2 a1) are investigated. Two AC potential applied patterns are considered in this section. In Case 1, as shown in Fig. 5a, the AC potentials with the same phase are applied at the rst electrodes at the side M and the side N, so both have the values of +Vrms. In Case 2, the AC potential with phasor difference 180 is applied at the electrodes at the sides M and N, so the value is +Vrms on side M and is Vrms on side N as shown in Fig. 5b. The results of numerical calculations for L* = 20, R = 12 (f = 1 MHz), Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 are presented in Figs. 6 and 7.

1.0 0.8 0.6

inlet outlet without electrodes outlet with 1 pair of electrodes

Figure 6 shows the mixing efciency at the outlet of the microchannel for Cases 1 and 2 with Da* = 06. It is apparent that the mixing efciencies are lowest when the two pairs electrodes are set at the same position at the two opposite side walls (i.e., Da* = 0) for both Cases 1 and 2. For Case 1, the mixing efciency increases with a larger Da until it reaches the maximum value at Da* = 2, and then it decreases with further increase in Da*. However, for Case 2, the mixing efciency increases with Da*, but the increase is nearly vanished when Da* = 6. Since the maximal mixing efciency of Case 1 with Da* = 2 is higher than Case 2 with any value of Da*, Case 1 phasor arrangement is chosen for further study in the following paragraphs. As seen from this gure, the maximum mixing efciency is less than 50% when two pairs of electrodes are added on the opposite walls of the microchannel. Figure 7 presents streamlines and concentration distributions in the mixing microchannels with Case 1 AC potential pattern and with various Da*. It can be seen that streamlines and concentration distributions vary with Da*, since the velocity elds induced by the electrothermal ow are different. When the distance Da* is small, vortices induced by the electrodes overlapped with each other, thus the electrothermal effect of two pairs of electrodes is not apparent. However, when the distance is too large, the induced vortices are too far apart to inuence each other.

0.4 0.2 0.0 0 1 2 3 4 5

3.3 The effect of the electrodes pairs number In order to get higher mixing efciency, it is necessary to increase the number of electrodes pairs. For more than two pairs of electrodes, the distance between adjacent two pairs of electrodes dt (see Fig. 1) will also be an important factor on the mixing efciency. From the consideration of the minimum microchannel length and better mixing
(b)

y*
Fig. 4 Concentration proles at the inlet and outlet of the mixing channel with one pair of electrodes for L* = 20, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

(a)

+V -V

side N

-V

+V

side N

+V -V

side M

+V -V

side M

Fig. 5 a The AC potentials applied at one pair of electrodes at opposite walls having the same phasor; b the AC potentials applied at one pair of electrodes at opposite walls having opposite phasor

123

Microuid Nanouid (2008) 5:1321

19

0.50 0.45 0.40

Case 1 Case 2

0.35 0.30 0.25 0.20 0 1 2 3 4 5 6

a*
Fig. 6 Mixing efciency in a mixing channel with two pairs of electrodes at different arrangements with Case 1 for electrode pairs having the same phasor and with Case 2 for electrode pairs having opposite phasor. The parameters are L* = 20, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

efciency, the potential applied pattern of Case 1 with Da* = 2, 3 and d* = 68 are chosen for further study. t Simulation results for L* = 70 and W* = 5 for up to 16 electrodes pairs (i.e., eight pairs at each side) are presented in Fig. 8 for three different electrodes arrangement patterns: (a) Da *= 3 and d* = 6; (b) Da* = 2 and d* = 6; (c) t t Da* = 2 and d* = 8. It can be seen that the mixing eft ciency becomes higher as more electrodes pairs are added. Comparing with patterns (a) and (b), which have the same distance between two adjacent pairs (d* = 6) of electrodes t but different values of Da*, the mixing efciency of pattern (a) with a larger value of Da* is higher. With the same electrodes pairs, the highest mixing efciency can be obtained for pattern (c) with Da* = 2 and d* = 8, for which t the mixing efciency reaches 98% for 16 pairs of electrodes.

Fig. 7 Sample streamlines (left) and concentration (right) distribution in the mixing channel with two pairs of electrodes at different relative positions for Case 1 arrangement. The parameters are L* = 20, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

123

20

Microuid Nanouid (2008) 5:1321

1.0

0.9

0.8
pattern (a): a*=3, dt*=6 pattern (b): a*=2, dt*=6 pattern (c): a*=2, dt*=8

0.7

0.6 2 4 6 8 10 12 14 16

Number of electrodes pairs

Fig. 8 The mixing efciency for a mixing channel with different number of electrodes pairs at three electrodes arrangement patterns for Case 1 with electrodes pair having the same phasor. The parameters are L* = 70, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12
1.0 0.8 0.6
no electrodes 4 pairs of electrodes 8 pairs of electrodes 16 pairs of electrodes

increase number of electrodes pairs, more uniform sample concentration distribution at the outlet of the mixing microchannel can be obtained. When 16 pairs of electrodes are added on two opposite walls of the microchannel, the sample concentration is nearly uniform at the outlet. Figure 10a and b present the concentration and temperature distribution in the microchannels with 16 pairs of electrodes on the walls, respectively. As can be seen from Fig. 10a, the velocity eld of the uid is signicantly disrupted in such a short distance by the electrothermal effect, and a nearly homogeneous sample concentration is obtained at the outlet. It is also important to note that the maximum temperature increase in the uid is less than 10 K from Fig. 10b, and the highest temperature appears at the electrode gaps in the channel. Therefore, temperature rise in the solution is not a limiting factor for concern when more electrodes on the opposite walls are added.

3.4 The effect of the AC frequency Figure 11 shows the mixing efciency c with the variation of AC frequency from 0.1 to 40 MHz (12 < R < 2.88) in a microchannel having a length of L* = 70 and height W* = 5 with 2 and 16 pairs of electrodes, respectively. The arrangement pattern (c) is used for the 16 pairs of electrodes. It can be seen that with the increase of frequencies, the mixing efciencies c remains constant in the range of 0.1 MHz < f < 1 MHz (where R is nearly constant at 12). However, c decreases gradually when f increases from 1 to 10 MHz (i.e., a decrease of R from 12 to 10), and a sharp decrease are shown when f > 10 MHz (R < 10). When the frequency is increased to 40 MHz (R = 2.88), it can also be seen that the mixing efciency decreases almost to the same value for channels with 2 and 16 pairs of electrodes. This is because the Coulomb force decreases with the increase of AC frequencies, and becomes weaker especially when the frequencies are larger than 10 MHz (R < 10). When the frequency is larger than 40 MHz

C
0.4 0.2 0.0 0 1 2

Fig. 9 The concentration proles at the outlet of mixing channel with different electrodes pairs for Case 1 arrangement. The parameters are L* = 70, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055, B = 2.11 105 and R = 12

Figure 9 shows concentration proles at the outlet of the mixing channel when there are 0, 4, 8 and 16 pairs of electrodes on the opposite side walls. The channel length is L* = 70, Da* = 2 and d* = 8. It is clear that with the t
Fig. 10 a Sample concentration distribution in the mixing channel with 16 pairs of electrodes for Case 1 arrangement. b Temperature distribution in the mixing channel with 16 pairs of electrodes for Case 1 arrangement (part of the channel including the area where the maximum temperature appears, i.e., from x = 280350 lm)

123

Microuid Nanouid (2008) 5:1321

1.0 0.9 0.8 0.7
16 pairs of electrodes 2 pairs of electrodes

21 No. 50536010, and by Science and Technology Committee of Shanghai City Government through Key Fundamental Project No. 05JC14025.

References
Reyes DR, Lossidis D, Auroux PA, Manz A (2002) Micro total analysis systems. 1. Introduction, theory, and technology. Anal Chem 74:26232636 Vilkner T, Janasek D, Manz A (2004) Micro total analysis systems, recent developments. Anal Chem 76:33733386 Jahn A, Vreeland WN, Gaitan M, Locascio LE (2004) Controlled vesicle self-assembly in microuidic channels with hydrodynamic focusing. J Am Chem Soc 126:26742675 Wang HZ, Lovenitti P, Harvey E, Masood S (2003) Numerical investigation of mixing in microchannels with patterned grooves. J Micromech Microeng 13:801808 Liu RH, Stremler MA, Sharp KV, Olsen MG, Santiago JG, Adrian RJ, Hassan A, Beebe DJ (2000) Passive mixing in a threedimensional serpentine microchannel. J Microelectromech Syst 9:190197 Erickson D, Li DQ (2002) Inuence of surface heterogeneity on electrokinetically driven microuidic mixing. Langmuir 18:18831892 Erickson D, Li DQ (2002) Microchannel ow with patchwise and periodic surface heterogeneity. Langmuir 18:89498959 Chang CC, Yang RJ (2004) Computational analysis of electrokinetically driven ow mixing in microchannels with patterned. Blocks J Micromech Microeng 14:550558 Fu LM, Yang RJ, Lin CH, Chien YS (2005) A novel microuidic mixer utilizing electrokinetic driving forces under low switching frequency. Electrophoresis 5:18141824 Shin SM, Kang LS, Cho YK (2005) Mixing enhancement by using electrokinetic instability under time-periodic electric eld. J Micromech Microeng 15:455462 Coleman JT, McKechnie J, Sinton D (2006) High-efciency electrokinetic micromixing through symmetric sequential injection and expansion. Lab Chip 6:10331039 Ramos A, Morgan H, Green NG, Castellanos A (1998) AC electrokinetics: a review of forces in microelectrode structures. J Phys D Appl Phys 3:23402353 Green NG, Ramos A, Gonzalez A, Castellanos A, Morgan H (2001) Electrothermally induced ow on microelectrodes. J Electrostatics 53:7178 Hansen TS (2004) Simulation and testing of AC electroosmotic micropumps, Master thesis, Technical University of Denmark Perch-Nielsen IR, Green NG, Wolff A (2004) Numerical simulation of travelling wave induced electrothermal uid ow. J Phys D Appl Phys 37:2323- 2330 Brown ABD, Smith CG, Rennie AR (2006) Pumping of water with AC electric eld applied to asymmetric pairs of microelectrodes. Phys Rev E 63 016305 Urbanski JP, Thorsen T, Levitan JA, Bazant MZ (2006) Fast AC electro-osmotic pumps with non-planar electrodes. Appl Phys Lett 89:143508 Chen DF, Du H (2006) Simulation studies on electrothermal uid ow induced in a dielectrophoretic microelectrode system. J Micromech Microeng 116:24112419 Gonzalez A, Ramos A, Morgan H, Green NG, Castellanos A (2006) Electrothermal ows generated by alternating and rotating electric elds in Microsystems. J Fluid Mech 564:415433

0.6 0.5 0.4 0.3 0.1 1 10 40

Frequency (MHz)
Fig. 11 Mixing efciencies for the mixing channels with different number of electrodes pairs at different AC frequencies for Case 1 arrangement. The parameters are L* = 70, W* = 5, Re = 0.01, Pec = 100, PeT = 0.07, A = 0.055 and B = 2.11 105

(R < 2.88), and the ow pattern is not be affected by the imposed AC voltage and therefore the mixing efciency is nearly the same in the channel with and without electrodes. 4 Concluding remarks This paper presents a 2D numerical simulation of a micromixer using vortices induced by the electrothermal effect to enhance the mixing efciency of sample in a KCI solution in a microchannel with a height of W = 50 lm at Re = 0.01, Pec = 100, and PeT = 0.07. The simulation results show that the mixing efciency of two uids with concentrations of 0 and 1 mol m3 increases from 17.4 to 33.3% at the end of the microchannel with a length of L = 200 lm after a voltage of 10 and 10 V are imposed at one pair of the electrodes (with d = d1 = d2 = 10 lm) at an AC frequency of 1 MHz (R = 12). The mixing efciency increases to 49.6% in the microchannel with two pairs of electrodes on opposite side walls at the optimal position of Da = 20 lm with the same AC potential phasor (Case 1). The mixing efciency can reach to 98% when 16 pairs of electrodes are added on the walls of the microchannel having a length of L = 700 lm when the electrodes arrangement pattern (c) is used. The mixing efciency decreases drastically when the imposed frequency is higher than 10 MHz (R < 10) because of the decrease of the Coulomb force at higher frequencies. The largest temperature increase in the electrolyte in all the simulations is less than 10 K, so the effect of temperature increase in the solution is not of serious concern.
Acknowledgments This research work was supported by the National Natural Science Foundation of China through key project

123