Creating, transporting, cutting, and merging liquid droplets ... - CiteSeerX

70

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 12, NO. 1, FEBRUARY 2003

Creating, Transporting, Cutting, and Merging Liquid Droplets by Electrowetting-Based Actuation for Digital Microfluidic Circuits Sung Kwon Cho, Hyejin Moon, and Chang-Jin Kim, Member, IEEE, Member, ASME

Abstract—This paper reports the completion of four fundamental fluidic operations considered essential to build digital microfluidic circuits, which can be used for lab-on-a-chip or micro total analysis system ( TAS): 1) creating, 2) transporting, 3) cutting, and 4) merging liquid droplets, all by electrowetting, i.e., controlling the wetting property of the surface through electric potential. The surface used in this report is, more specifically, an electrode covered with dielectrics, hence, called electrowetting-on-dielectric (EWOD). All the fluidic movement is confined between two plates, which we call parallel-plate channel, rather than through closed channels or on open surfaces. While transporting and merging droplets are easily verified, we discover that there exists a design criterion for a given set of materials beyond which the droplet simply cannot be cut by EWOD mechanism. The condition for successful cutting is theoretically analyzed by examining the channel gap, the droplet size and the degree of contact angle change by electrowetting on dielectric (EWOD). A series of experiments is run and verifies the criterion. A smaller channel gap, a larger droplet size and a larger change in the contact angle enhance the necking of the droplet, helping the completion of the cutting process. Creating droplets from a pool of liquid is highly related to cutting, but much more challenging. Although droplets may be created by simply pulling liquid out of a reservoir, the location of cutting is sensitive to initial conditions and turns out unpredictable. This problem of an inconsistent cutting location is overcome by introducing side electrodes, which pull the liquid perpendicularly to the main fluid path before activating the cutting. All four operations are carried out in air environment at 25 Vdc applied voltage. [862] Index Terms—Contact angle, electrowetting, electrowetting on dielectric (EWOD), lab-on-a-chip, microfluidics, micro total analysis system ( TAS), surface tension.

I. INTRODUCTION

S

INCE the advent of MEMS, many microfluidic devices have been studied and developed to handle fluids on the microscale. In these devices, numerous actuation methods have been introduced to pump or regulate fluids: piezoelectric, electrostatic, thermopneumatic, electromagnetic, electrophoretic, electroosmotic, bimetallic, shape memory alloy, and so on [1]–[3]. Given the limitation and cost of micromachining and Manuscript received May 7, 2002; revised October 15, 2002. This work was supported by the National Science Foundation (NSF) CAREER Award, NSF Engineering Microsystems: “XYZ on a chip” Program, and Defense Advanced Research Projects Agency (DARPA) BioFlips Program. Subject Editor T. Kenny. The authors are with the Mechanical and Aerospace Engineering Department, University of California, Los Angeles (UCLA), Los Angeles, CA 90095-1597 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/JMEMS.2002.807467

Fig. 1. Envisioned digital microfluidic circuit and the four fundamental droplet operations necessary.

the importance of device reliability, the actuation mechanism that imposes the least mechanical complication in the device is highly needed for microfluidics. As MEMS technologies advanced in recent years, much attention has been drawn to using surface tension, which becomes a dominant force on the microscale, for the microfluidic actuation [4]. Micropumping has been reported using thermal control of surface tension, i.e., thermocapillary [5], [6]. Compared to thermocapillary, however, electrical control of surface tension, i.e., electrocapillary, is much more energy efficient and advantageous for most microfluid actuations [7]–[23]. Among the known configurations of electrowetting, electrowetting on dielectric (EWOD), which allows for control of the wettability of liquids on a dielectric solid surface using electric potential [10], [12]–[19], is considered most promising, thanks to the electrochemical inertness of the surface. Pollack et al. [10] showed that an aqueous liquid droplet could be transported with EWOD. Lee et al. [13], [14] demonstrated micropumping by both conventional electrowetting (i.e., liquid on metal) and EWOD (i.e., liquid on dielectrics) and further reported the concept of addressable micro liquid handling technique, envisioning an eventual digital microfluidic circuits such as the one in Fig. 1. This digital fluidic system can be used for biotechnology applications such as lab-on-a-chip or micro total analysis system ( TAS). In such digital microfluidic circuit devices, most fluidic operations can be carried out on a chip using discrete droplets rather

1057-7157/03$17.00 © 2003 IEEE

CHO et al.: CREATING, TRANSPORTING, CUTTING, AND MERGING LIQUID DROPLETS BY ELECTROWETTING-BASED ACTUATION

than the usual continuous flow. Fig. 1, for example, shows that m n different samples or reagents are introduced to digital microfluidic circuits and after several droplet-based microfluidic operations, various kinds of products can be produced on a single chip. In fact, typical fluidic operations such as pumping and mixing can be performed on the same chip by programming electric signals rather than adding up physical structures for a certain fluidic function. The proposed concept is very promising because fabrication process is much simpler with no need to build moving micromechanical parts in the device. In order to make the digital microfluidic circuit that we proposed fully functional as a lab-on-a-chip, various fluidic operations need to be established. Among them, droplet manipulations such as creating, transporting, cutting, merging of droplets are fundamental for developing more advanced fluidic operations such as mixing, separating, and concentration control of biological samples and reagents. Transportation of water droplets by EWOD was previously demonstrated by Pollack et al. [10] and Lee et al. [13], [14]. More recently, Fair et al. [16] and Pollack et al. [17] reported merging and generating water droplets. However, the experiment was done in silicone-oil environment, where droplet driving is much easier than in air environment. Manipulation of droplets by EWOD in air is more challenging due to the lower contact angle and higher contact-angle hysteresis than in an immiscible liquid and thus will be more general in terms of droplet driving. In this paper, we report the development of cutting, merging and creating droplets as well as transporting, completing all four fundamental operations in an air-filled environment, as illustrated in Fig. 1. Droplet creation has been performed on chip without any external actuation. First, we analyze the droplet cutting process, propose a criterion for the cutting to occur, and verify the proposed criterion by testing fabricated devices. Second, we report the operation of droplet merging, which is applicable to mixing enhancement. Third, we also report the creation of droplet from a reservoir and discuss the added challenges compared to the cutting process. In fact, the creation of droplet is critical for the success of eventual digital microfluidic circuits we envisioned as it is analogous to analog-to-digital (A/D) converter in electronic circuits. Finally, transporting droplets is verified. All four fundamental operations were successfully performed in an air-filled environment with 25 V , much lower than the EWOD actuation voltages in air previously reported [10], [12]–[14], [16], [17]. In this configuration, the threshold voltages at which the droplet movement is initiated are 12 V in silicone oil and 18 V in air, respectively, which are lower than those in Fair et al. [16] and Pollack et al. [17] (17 V in silicone oil and 48 V in air). We chose driving in air (versus immiscible liquid such as silicone oil) to present a more challenging scenario and be conservative in evaluating the success. However, we limited the tests with deionized (DI) water in this study. II. THEORY

71

Fig. 2. Principle of electrowetting on dielectric (EWOD). (a) Schematic configuration. (b) Pictures of basic EWOD demonstration (volume 5l).

the capillary forces at an interface. Introduced to the MEMS community first by Matsumoto and Colgate [7], the notion of electrical control of surface tension, i.e., electrocapillary or electrowetting, is quite attractive for microdevices because of its inherent effectiveness in microscale and simplicity in implementation. Recently, this principle was also reported valid in the configuration where the electrode is covered with a thin insulating film [21], [22], as illustrated in Fig. 2(a). When an electric voltage is applied, the electric charge changes free energy on the dielectric surface, inducing a change in wettabilty on the surface and contact angle of the droplet [Fig. 2(b)]. This phenomenon, which we name electrowetting on dielectric (EWOD), has an excellent reversibility with many kinds of dielectric layers compared to the conventional electrowetting, where liquids contact the conductive surface directly. See Lee et al. [13], [14] or Moon et al. [23]. According to Lippmann’s equation, the solid–liquid interfacan be controlled by the electric potential across cial tension the interface, (1) F m is the specific capacitance of the dielectric where layer. For the case of EWOD, where the electric potential at no external voltage is considered zero, is equivalent to an applied voltage. At the three-phase contact line, the relation among contact angle and interfacial tensions is given by Young’s equation (2)

A. Electrowetting on Dielectric (EWOD) Lippmann [20] recognized over a hundred years ago that an externally added electrostatic charge may significantly modify

is the solid–gas interfacial tension and the where liquid–gas interfacial tension. By substituting (1) into (2), the

72


Fig. 3. Instability of three liquid columns. (a) Liquid column in air, e.g., jet flow (unstable). (b) Liquid column on solid surface (unstable). (c) Liquid column in parallel-plate channel (stable). (d) Schematics of liquid column for perturbation method to check instability.

change of contact angle (cosine of contact angle, to be exact) by the electric potential can be described: (3) denotes the equilibrium contact angle at where V F m the permittivity of vacuum, the dielectric constant of the dielectric layer, and its thickness. Note in (3) that the contact angle change is not related to the polarity of the applied potential . B. Moving a Droplet A discrete liquid (droplet or liquid column) in a channel can be moved (i.e., pumped) by asymmetrically changing the interfacial tension. Several methods for changing the interfacial tension (e.g., thermocapillary [5], [6] or electrocapillary/electrowetting [7]–[19]) have been introduced to move a discrete liquid. No matter which method is used to change the interfacial tension, the asymmetric change in the interfacial tension induces an asymmetric deformation of the liquid meniscus, which establishes a pressure difference inside liquid, giving a rise to bulk fluid movement [see Fig. 8(b)].

C. Droplet Formation: Consideration of Hydrodynamic Instability An essential requirement for digital microfluidics we proposed is that discrete droplets be created from a continuous (i.e., bulk) state in the configuration of the present parallelplate channel. We first analyze if droplets can be formed spontaneously by hydrodynamic instability. It is relatively well understood how a long liquid column in air (e.g., a liquid jet issued to the air) behaves, as shown in Fig. 3(a). Being hydrodynamically unstable, a long liquid column does not keep its shape but readily breaks into beads. Jones et al. [24] demonstrated that a long liquid column on a solid surface pulled out of a large droplet or reservoir by dieletrophoretic force spontaneously breaks into beads (droplets) as soon as the dielectric force is turned off [Fig. 3(b)]. In the present study, however, liquid is placed and squeezed in a parallel-plate channel. The channel is defined by the top and bottom plates, but there are no side walls [see Fig. 3(c)]. In order to see whether droplets can be spontaneously created from the liquid column in the present configuration of the parallel-plate channel, we need to analyze hydrodynamic instability of the long liquid column. This in-


73

stability can be determined by considering the pressure variations in the liquid column when it is perturbed by a small periodic disturbance of wavelength . As illustrated in Fig. 3(d), the liquid column will be deformed by infinitesimal perturbato tion from (4) . The where is the amplitude of the perturbation and system will be stable if the pressure at the region of greatest lateral extension, such as A in Fig. 3(d), is greater than that at the region of greatest lateral contraction, such as B. Then, the liquid will flow from the high pressure region A to the low pressure region B. As a result, the original straight shape will be restored. Thus, it is necessary to calculate this pressure difference at the two positions using Laplace equation. The pressure differences for each of the three cases (liquid jet flow, liquid column on solid surface and liquid column in parallel-plate channel) are obtained and shown in Fig. 3. For the detailed calculation, refer to Isenberg [25]. For the liquid column of jet flow and that on the solid surface, the small perturbations grow if their wavelengths are and , respectively. Then, longer than the liquid columns break up into droplets. In fact, these conditions are in agreement with Chandrasekhar [26] and Schiaffino and Sonin [27], respectively. On the other hand, the pressure diffor the long liquid column in the parallel-plate ference channel is positive definite, i.e., the liquid column is stable. Any small perturbation or disturbance decays away and the liquid column would keep its shape. Compared to the liquid column on the solid surface, the liquid in the parallel-plate channel has additional confining plate on top, which makes the liquid stable against small perturbations. This means that cutting and creating droplets in the present channel configuration require a large displacement of liquid, unlike on open surface [24], where droplets are formed spontaneously by instability. D. Droplet Formation: Cutting a Droplet At the end of the previous section, we concluded that droplets must be formed by actively inducing a large displacement of liquid. Consider a simple case of droplet formation: dividing a droplet into two. Fig. 4 schematically describes how a droplet can be cut by changing the wetting property locally (e.g., through EWOD). When cutting is in order, the droplet is elongated in the longitudinal direction by making the two ends wetting and keeping the middle nonwetting, thus pinching in the middle. Initially a droplet, shown by dotted lines, occupies the entire middle electrode as well as a portion of each control electrode at the ends. In the present cutting configuration, we consider the case where the diameter of the squeezed droplet in the top view [see Fig. 4(a)] is approximately equal to the diagonal length of the control electrode. Note that the liquid droplet contacts the top plate as well, enabling the electrical circuit to be closed. During stretching, the left and right electrodes are energized so the contact angles above them reduce according to (3), resulting in an increase of the radii of curvature in Fig. 4(c). In the mean time, the middle electrode is floated

Fig. 4. Droplet configuration for cutting. (a) Top view. (b) Cross-sectional view B-B’. (c) Cross-sectional view A-A’. Dotted lines show the initial shape of the droplet before electrodes are activated.

or grounded, keeping the middle section hydrophobic. As a result, the meniscus on the middle electrode starts to contract to keep the total volume of the droplet constant. That is, cutting is initiated by the elongation of the droplet in the longitudinal , shown in Fig. 4(a)] in the direction and necking [negative middle of the droplet. To cut a droplet efficiently and reliably, it is important to understand the physics of the cutting process. Let us first derive the relation among physical parameters. The radius of the meniscus curvature in a given region of the droplet is geometrically related to the local contact angles and the channel gap . (5) (6) indicates contact angle on the top plate, contact where angle on the bottom plate, the principal radii of curvature as shown Figs. 4(b) and 4(c), subscript 1 parameters in the middle region of the droplet and subscript 2 in the end regions of the droplet. These relations were derived based on the assumption that the meniscus shape in Fig. 4(b) and 4(c) is a circular arc because the effect of gravity force is negligible in the microscale we are interested in. Using Laplace equation, the pressures can be described by the principal radii of curvature at points 1 and 2: (7) (8) where is the principal radius of curvature as shown in Fig. 4(a) atmosphere pressure. In static equilibrium, the pressure and

74


should be equal inside the droplet, and and can be substituted with channel gap and contact angles using (5) and (6)

(9) Finally, the channel gap is related to contact angle difference and radii of curvature (10) Note that the contact angle on the top plate is cancelled out. Equation (10) provides an important criterion for cutting a droplet. The cutting is initiated by necking in the middle of becomes negative. This the droplet, which means the radius condition can be satisfied with a large contact angle change by EWOD), a (i.e., large positive value of at the end regions small channel gap , or a large radius of the droplet. As the ratio between the projection diameter of the initial droplet and electrode size is maintained as shown is proportional to the droplet size in Fig. 4(a), the radius and roughly determined by the size of the control electrode; the will be. Also larger the control electrodes, the larger radius note that in this analysis the effect of contact angle hysteresis (i.e., the difference between advancing and receding angles), far from being negligible in practice, is not taken into account for the sake of simplicity. The contact angle change by EWOD on the left or right electrode can be represented with the applied electric potential using (3): (11) where denotes the applied potential across the bottom dielectric layer. On the middle electrode, the contact angle does not change because no electric potential is applied there, that is, the is equal to . Thus, the contact angle contact angle difference in (10) can be described by the contact angle change by EWOD on the right or left electrode: (12) By substituting (12) into (10), we can obtain the direct relation and , the dielectric conamong , the radii of curvature stant and thickness of the dielectric layer, and the channel gap : (13) in the present channel However, it is difficult to estimate system, which may not be equal to the total voltage applied in the channel system. In addition, contact angle saturation [28], [29], where the contact angle ceases to decrease with increasing applied voltage, is not taken into account in (3). Therefore, in order to predict the channel gap required to cut a droplet under

Fig. 5. The measurement of the contact angle change by EWOD in a water sessile drop on a 1000 Å silicon dioxide dielectric layer coated with 200 Å Teflon.

given parameters, we need to measure how much contact angle change is made by EWOD in the given channel configuration through experiments and use (10). III. CONTACT ANGLE DATA FOR THE GIVEN CHANNEL CONFIGURATION A. Contact Angle Measurement Measurements of the contact angle change by EWOD are made in a sessile drop using an optical contact angle measurement system developed for small droplets (First Ten Angstroms, FTA 4000A). The sessile drop is placed on the same dielectric layers as used in the droplet cutting. The platinum ground electrode is penetrated into the droplet from the top, and the electrical potential is applied between the liquid and the electrode underneath the dielectric layers as shown in Fig. 2. The data are in reasonably good agreement with (3), except for the saturation region, as shown in Fig. 5. The contact angle parabolically decreases as the applied potential increases until the contact angle is saturated at about 80 . The reason for the saturation is not clearly understood as of today. Verheijen and Prins [28] proposed the charge trapping in the dielectric layer could induce the saturation, and Peykov et al. [29] showed that the contact angle saturated when the solid-liquid surface energy became zero. In any case, there exists a limitation in the contact angle change by EWOD, beyond which higher potential does not decrease the contact angle any further. Basically, this limitation gives rise to the criterion in the cutting process with EWOD. In the channel system, as mentioned in the previous section, in (11)–(13) is not the total voltage applied to the channel system, but the voltage across the dielectric layers on the bottom plate. In principle, the total voltage drop in the channel system consists of the voltage drop across the dielectric layers on the bottom surface, plus that across the dielectric layers on the top surface, if the electric resistance of the water droplet is negligible compared to that of the dielectric layers. We have considered two channel systems: Channel I [see Fig. 6(a)] and Channel II [Fig. 6(b)]. Channel I has a 200 Å Teflon layer on the top surface and a 200 Å Teflon layer over an 1000 Å silicon dioxide layer on the bottom surface for the dielectric layers,


75

Fig. 7. Graphical expression of (10) with numerical values based on the contact angle change from 117 to 90 by EWOD. The lower solid curve represents necking in the middle of droplet (R < 0). The droplet is cut when the two menisci at the necking region meet each other (R = R ). For the given contact angle change, cutting occurs if d=R < 0:22, which corresponds, for example, to d < 154 m for 1.4 mm 1.4 mm electrodes.

0

2

Fig. 6. Contact angle measurements of a droplet squeezed in parallel-plate channels. Electrodes cover the entire area on the top and bottom surfaces: (a) Channel I (asymmetric). The top surface is coated with 200 Å Teflon and the bottom surface with 200 Å Teflon over 1000 Å silicon dioxide. When the voltage is applied, the contact angle on the bottom surface is changed by EWOD while the contact angle on the top surface remains virtually unchanged Droplet volume: 1.8 l. (b) Channel II (symmetric): Both the top surface and the bottom surface are coated with 200 Å Teflon over 1000 Å silicon dioxide. When the voltage is applied, the contact angles on the top surface and the bottom surface are changed in an equal amount by EWOD. Droplet volume: 1.6 l.

while Channel II has a 200 Å Teflon layer over an 1000 Å silicon dioxide layer for both the top and bottom surfaces. That is, Channel I is asymmetric and Channel II is symmetric in the way dielectric layers are fabricated on the top and bottom surfaces. In Channel I, it turns out that the voltage drop across the dielectric layer on the top surface is negligible because the Teflon layer is very thin. Fig. 6(a) shows the contact angle change in Channel I by EWOD with the total applied voltage of 25 V. The contact angle on the bottom surface changes from 117 to 90 , while the contact angle change on the top surface is less than 3 . Considering that the uncertainty of the measurement system is 3 , we can assume that the contact angle change on the top surface is negligible, indicating the voltage drop across the top dielectric layer is also negligible. Thus, in Channel I, the voltage drop across the dielectric layers on the bottom surface is almost equal to the total voltage drop in the channel system. On the other hand, in Channel II where the top surface has the same dielectric layers, the voltage drop across dielectric layers on the top surface is equal to that across the dielectric layers on the bottom surface. Fig. 6(b) shows the contact angle changes with the total applied voltage of 100 V . Note that negative curvature can be achieved with high voltage. After the wetting occurs, the contact angles on the top wall and the bottom wall are the same, indicating that voltage drops are also equal across the top and bottom layers. This experiment also confirms the argument in the earlier section that, in EWOD, the polarity of the applied potential does not affect the contact angle change.

B. Droplet Cutting The channel we used for droplet cutting is Channel I, i.e., the top surface is coated only with a very thin Teflon layer. Based on the contact angle measurements, the operating point for cutting is chosen at the total applied voltage of 25 V , where saturation of EWOD is about starting. In this case, the contact anand as shown gles in (9)–(12) will be in Fig. 6(a). Based on this contact angle change, the relation is calbetween the channel gap and radius of curvature is higher culated using (10), as illustrated in Fig. 7. If is positive and asympthan 0.45 (over dotted line in Fig. 7), , which means the droplet maintains a totically approaches is roughly circular shape without necking. However, if becomes negative, resulting in the necking lower than 0.45, in the middle of the droplet. That is, the smaller channel gap and the larger droplet (larger electrode) are favorable to cutting. Let us estimate the required channel gap needed for successful cutting under a given droplet size (electrode size) and a contact angle change. For square control electrodes we reason that for the two necking menisci meet and pinch off the droplet, should be roughly a half length of the control electrode side and should also be about the same length but of the opposite is 1, and we can see that is 0.22 sign. Then, should be from Fig. 7. For successful cutting, therefore, lower than 0.22. For example, if a control electrode is 1.4 mm 1.4 mm, the channel gap should be smaller than m mm for successful cutting. Although this estimation is made under many approximations, it clearly indicates the directions for how the device should be designed. The criterion is found to be in good agreement with the experiments, as will be shown in the next section. IV. FABRIACATION AND DEVICE TESTS A. Fabrication of Testing Devices A low-voltage operation is desirable for driving liquids with conventional electric circuits. The fabrication of testing

76


devices is focused on the selection of dielectric layers to reduce the driving voltage. For a given liquid, (3) implies that the electric potential required to generate a certain contact angle change can be reduced by using a thin dielectric layer or a high dielectric-constant layer. Following the above indication, we make the dielectric layer as thin as possible: 200 Å Teflon on the top of 1000 Å silicon dioxide. However, this combination of two layers sometimes causes high resistance against droplet movement. This phenomenon is not clearly understood yet and under investigation with different Teflon coating methods. We speculate that contact angle hysteresis is highly dependent on the condition of the coated Teflon layer. In order to circumvent the inconsistent coating effect, we first identified good dice that show good repeatability after preliminary EWOD testing. Devices made with known-good-dice give us good repeatable EWOD actuations at as low as 25 V , which is much lower than the previously reported EWOD actuation voltages in air (over 50 V) [10], [12]–[14], [16], [17]. After several tens of repeatable actuations the testing device still showed a good performance. Furthermore, there was no significant degradation on the testing device even after months of storage or usage. All four fundamental fluidic operations are accomplished at 25 V under this double layer condition. Thinner layers, however, are susceptible to electrolysis. When a very high dielectric-constant material such as Barium Strontium Titanate (BST, dielectric constant: 200–300) is deposited by a commercial MOCVD (metal organic chemical vapor deposition), we are able to reduce the operation voltage down to 15 V . For more details, refer to Moon et al. [23]. Fig. 8 shows schematic figures of the fabricated testing device. For the control electrode, 100 Å chromium and 700 Å platinum are evaporated on a glass substrate and patterned by wet etching. Many sets of testing devices with three different sizes of the control electrode (1.4 mm 1.4 mm, 1.0 mm 1.0 mm, and 0.7 mm 0.7 mm) are fabricated. However, most fluidic operations are tested on 1.4 1.4 mm electrodes. The inter-digitated fingers [10] are placed between adjacent electrodes to help facilitate continuous movement of droplets, although this was later found unnecessary for most cases. As a first dielectric layer, a 1000 Å LTO (low temperature oxide) layer is deposited by LPCVD (low pressure chemical vapor deposition). This deposition technique gives a better step coverage compared to PECVD (plasma enhanced chemical vapor deposition) so that the current leakage through the dielectric layers and the problem of electrolysis can be minimized. Then, a 200 Å Teflon AF layer is spin-coated on the oxide layer. This Teflon AF layer acts as the dielectric layer as well. The total capacitance is the serial combination of two dielectric layers. The entire area of the cover glass is coated with a transparent and conductive ITO (indium square) as a ground electrode. Then, a 200 Å tin oxide, Teflon AF layer is also spin-coated on the ITO covered top cover glass to make the surface hydrophobic. A spacer, for which we use a thick photoresist (NR9-8000) or double-stick tapes, is only to define the gap between the top and bottom plates in the parallel-plate channel; there is no side wall defining the channel. The channels 70, 150, and 300 m in gap size are tested. Liquid M -cm) is placed and squeezed in the (deionized water, gap space between the substrate and the glass cover. Liquid vol-

Fig. 8. Schematic of testing device (not to scale): (a) Artist’s view. Top plate is transparent. (b) Cross-section view. Dotted line indicates the shape of meniscus before actuation.

umes of droplets for each testing case are measured by multiplying the channel gap by the projection area of the droplet to the plate. The applied potential is adjusted by a power supply and the activation signals are controlled by a personal computer. B. Droplet Cutting Fig. 9 shows a series of experiments performed to verify the theory of droplet cutting. First, the channel gap effect is investigated with 1.4 mm 1.4 mm control electrodes and 25 V applied potential. In the case of the 300 m channel gap, cutting of a droplet is not realized. The droplet shape is shown in Fig. 9(a) after the droplet reaches the static equilibrium. The necking did not occur even at 32 V, over which electrolysis and electric breakdown occurred. When the gap is reduced to 150 m, we can see a necking in the middle of the droplet, where the two concave menisci are very close [see Fig. 9(b)]. When the channel gap further goes down to 70 m, the necks are pinched off and cutting can be achieved as shown in Fig. 9(c). Sequentially captured pictures for complete cutting process are shown in Fig. 10. The neck shape is initially formed and then pinched [see Fig. 10(b)], and finally the droplet is cut into two droplets [see Fig. 10(c)]. After the electric potential is turned off, the


77

Fig. 9. Droplet cutting and verification of the criteria at 25 V . (a) – (c) The effect of channel gap size. Electrode size is fixed at 1.4 mm 1.4 mm. The channel gaps are 300 m, 150 m, and 70 m and volumes are 0.9, 0.5, and 0.2 l, respectively. (c) – (e) The effect of electrode size. Channel gap d is fixed at 70 m. Electrode sizes are 1.4 mm 1.4 mm, 1.0 mm 1.0 mm, and 0.7 mm 0.7 mm volumes are 0.2, 0.1, and 0.05 l, respectively. Cutting is completed when channel gap d is 70 m and electrode is 1.4 mm 1.4 mm. Images are captured after reaching static equilibrium condition.

2

2

2

2

2

separated droplets are relaxed to circular shapes. Total cutting process takes only 8/30 s. This small channel gap requirement for successful cutting also helps slowing evaporation of liquid droplet. For sealed devices, which is more typical in real application, the inner volume is expected to be saturated with humidity by extremely small volume of liquid evaporation (e.g., 10–100 pl) [14]. In addition to the test of the channel gap variation above, the effect of the droplet size (electrode size) is also investigated. As m), shown in Fig. 9(d) and (e), for a given channel gap ( the cutting becomes more difficult with smaller electrodes. All the conditions except electrode size are maintained the same as those in Fig. 9(c). Moreover, we can see the extent of the necking decreasing as the electrode size becomes smaller from 1 mm [see Fig. 9(d)] to 700 m [see Fig. 9(e)]. This trend agrees with (10). C. Droplet Merging Merging droplets is essential for mixing in many microfluidic devices [16], [19]. Fig. 10 shows that merging is achieved by moving two droplets toward each other. The merging process, although opposite to cutting, does not follow the reverse procedure of cutting [see Fig. 10(c)–(e)]. Note that there is no state of necking involved during the merging process. Also note in Fig. 10(d)–(e) that the droplet shape does not relax to a circle even after all electrodes are turned off. This phenomenon is more likely to occur in a smaller channel gap and larger droplet, where the effect of contact angle hysteresis becomes more apparent. We speculate that contact angle hysteresis hinders the movement of the three-phase contact line and prevents the formation of a complete circle. This effect, heavily dependent on surface condition, is often significant in surface-tension-driven flows.

Fig. 10. Sequential images of successful cutting and merging of droplets at 25 V (gap size d 70 m, electrode is 1.4 mm 1.4 mm, volume is 0.2 l).

=

2

D. Droplet Creating From a Reservoir To generate droplets from an on-chip reservoir, liquid needs to be first pulled out of the reservoir and separated from it. This process is highly related with cutting process, and a smaller channel gap makes the droplet creation easier. However, the creation process becomes more difficult when a large reservoir is placed between the parallel channel plates. There exists only a weak force holding the liquid back, i.e., a weak reaction force against the force pulling the liquid out especially if electrowetting is not effective on the surface inside the reservoir. When the front meniscus is pulled out of the reservoir by EWOD, the liquid may continue to follow the leading meniscus to some extent, forming a long liquid column. The cutting location depends on the initial shape of the meniscus in the reservoir, on the surface condition, and on how the control electrodes are activated. Fig. 11 shows droplets can be created from a large reservoir droplet (2–3 l in volume) by pulling liquid out along the electrode path. The reservoir droplet is squeezed between channel plates without any confined sidewalls, as drawn in Fig. 11(a). The sizes of tested reservoir droplets are kept comparable for all the creation tests in this paper. A pair of electrodes is sequentially activated from the left all the way to the right with one electrode overlapping with the previous pair. The two electrodes marked “side electrode” are not used in this experiment. The switching interval is 0.7 second. However, it is rather unpredictable how far to the right the liquid needs to be pulled out before a droplet is created. Usually, 5 to 7 control electrodes are used to complete the droplet creation as shown in Fig. 11(a) and (b). This seems to be greatly affected by the initial condition of meniscus. For example, in case the first electrode at the reservoir entrance remains activated throughout the pullout, the

78


Fig. 11. The captured images right before droplet separation. Side electrodes are not used. The minimum number of electrodes needed to create a droplet depends on the initial condition and is not predictable. After pulling liquid out along the straight electrode path, all electrodes except those marked “on” are turned off. Reservoir droplet has been drawn for clarification. In all the cases, 3 l): (a) five electrodes the sizes of reservoir droplets are comparable (2 required for droplet creation; (b) seven electrodes required; and (c) the liquid is stretched even longer when the first electrode remains activated.

liquid column is stretched even longer and cannot be separated even when all 7 electrodes (excluding the first electrode at the reservoir entrance) are used in the process, as shown Fig. 11(c). The above problem of unpredictable separating location is overcome by introducing two side electrodes beside the main fluid path (see Fig. 12). By actively pulling the liquid sideways from the main fluid path, the liquid can be virtually pulled back, enhancing the necking [second and third pictures from top in Fig. 12(a)]. The use of side electrodes method makes the creation of droplets more consistent. Fig. 12(a) shows that two droplets are consecutively generated at a fixed location with side electrodes, and they are almost equal in size (two electrode size, approximately 270 nl in volume). Note in the first picture of Fig. 12(a) that a long liquid column (five-electrode long), after being pulled out by activating five electrodes, does not break into droplets by instability as predicted in Theory section, even after all the electrodes are turned off. The liquid column keeps its shape unless any large external force is applied. Moreover, the liquid column released from actuation does not go back into the reservoir droplet. This is attributed to contact angle hysteresis. Since the reservoir droplet is squeezed in the microscale channel gap, the pressure difference between the leading front of the liquid column and in the reservoir droplet is not large enough to overcome the meniscus pinning by contact angle hysteresis. In fact, we observed that liquid columns spontaneously go back into a reservoir when liquid in the reservoir is exposed to air through a large hole (1.5 mm in diameter) drilled through the top

Fig. 12. Captured images for droplet creation with side electrodes. (a) Droplets can be generated more consistently with side electrodes. Note that two droplets were generated in equal size (volume 270 nl). (b) Generation and transportation of a three-electrode size droplet (volume 400 nl).

cover glass (i.e., large pressure difference). The size of the droplet is determined by how many electrodes are activated simultaneously. Fig. 12(b) shows a three-electrode size droplet is generated when three electrodes near the leading meniscus are activated at the same time.


The side-electrodes are also effective for droplet creation from a large droplet liquid confined by the reservoir wall (not shown), which represents real applications better than previous case of the large droplet reservoirs squeezed in the channel gap without physical boundary. These side-electrodes present two main advantages in droplet creation: 1) controllable cutting location by changing the position of side electrodes and 2) no control electrodes required in the reservoir because it is not necessary to pull liquid back with EWOD actuation. It is interesting to note that if the entire inner surface of the reservoir is made hydrophobic, all liquid can be taken out by EWOD with no dead volume remaining in the reservoir. E. Droplet Transporting Transporting is verified by sequentially energizing the electrode beneath the leading meniscus of the droplet. Fig. 12(b) includes the images of a droplet being transported after being created from the reservoir. Pollack et al. [17] showed that the movement of droplet is initiated at around 48 V and thereafter the droplet is repeatedly transported. In the present study, we can initiate to move a droplet at a much lower applied potential, about 18 V. At 25 V, we can get repeatable transport action of droplets. The combination of the thin Teflon and oxide layers enables this low voltage operation. Furthermore, we found that when high-voltage ac potential ( 150 V) is applied, the transporting speed increases up to 250 mm/s [18], one order of magnitude faster than previously reported [10], depending on the frequency of the ac potential and the channel gap. V. CONCLUSION In this paper, we report the completion of all four fundamental microfluidic operations as building blocks for digital microfluidic circuits: 1) cutting, 2) merging, 3) creating, and 4) transporting liquid droplets, based on electrowetting-on-dielectric (EWOD) actuation. First, cutting a single droplet in a parallel-plate microchannel is achieved. Through theoretical analysis, we derive the relation among the contact angle change by EWOD, the channel gap and the droplet size. This relation reveals that a smaller channel gap, a larger droplet and a larger contact angle change make droplet cutting easier. Experimental verifications for these criteria are made by fabricating and testing devices with different channel gaps (70, 150, and 300 m) and control electrodes (1.4 mm 1.4 mm, 1.0 mm 1.0 mm and 0.7 mm 0.7 mm) at 25 V applied voltage. For electrodes of 1.4 mm 1.4 mm, a droplet is successfully cut into two when the channel gap is reduced to 70 m. It has also been verified that for a given channel gap cutting becomes easier as the electrode size increases. Second, merging, a reverse process of cutting and essential for microfluid mixing, is realized by moving two droplets toward each other. Third, the challenge associated with droplet creation by pulling a liquid column out of reservoir is discussed. This is improved by pulling the liquid side ways from the main fluid path using side electrodes. Finally, droplet transporting is demonstrated. All four fluidic operations (creating, transporting, cutting and merging droplets) are achieved in air environment at the applied voltage of 25 V , much lower than the EWOD actuation voltages previously reported.

79

ACKNOWLEDGMENT The authors would like to thank J. Fowler, S.-K. Fan, and P. P. de Guzman in the UCLA Micromanufacturing Lab for their valuable discussions and F. Saeki, J.-Y. Yoon, and Prof. R. L. Garrell in the Department of Chemistry and Biochemistry in UCLA for the assistance in using the contact angle measurement system. REFERENCES [1] M. Madou, Fundamentals of Microfabrication. Boca Raton, FL: CRC, 1997, ch. 9. [2] G. T. A. Kovacs, Micromachined Transducers Sourcebook. New York: McGraw-Hill, 1998, ch. 9. [3] C.-M Ho, “Fluidics-the link between micro and nano sciences and technologies-,” in Proc. IEEE Int. Conf. MEMS, Interlaken, Switzerland, Jan. 2001, pp. 375–384. [4] C.-J. Kim, “Micromachines driven by surface tension,” in AIAA 99-3800, 30th AIAA Fluid Dynamics Conference, Norfolk, VA, June–July 1999, pp. 1–6. [5] T. K. Jun and C.-J. Kim, “Valveless pumping using traversing vapor bubbles in microchannels,” J. Appl. Phys., vol. 83, no. 11, pp. 5658–5664, 1998. [6] T. A. Sammarco and M. A. Burns, “Thermocapillary pumping of discrete drops in microfabricated analysis devices,” AIChE J., vol. 45, no. 2, pp. 350–366, 1999. [7] H. Matsumoto and J. E. Colgate, “Preliminary investigation of micropumping based on electrical control of interfacial tension,” in Proc. IEEE MEMS Workshop, Napa Valley, CA, Feb. 1990, pp. 105–110. [8] J. Lee and C.-J. Kim, “Liquid micromotor driven by continuous electrowetting,” in Proc. IEEE Int. Conf. MEMS, Heidelberg, Germany, Jan. 1998, pp. 538–543. , “Surface-tension-driven microactuation based on continuous elec[9] trowetting,” J. Microelectromech. Syst., vol. 9, no. 2, pp. 171–180, 2000. [10] M. G. Pollack, R. B. Fair, and A. D. Shenderov, “Electrowetting-based actuation of liquid droplets for microfluidic applications,” Appl. Phys. Lett., vol. 77, no. 11, pp. 1725–1726, 2000. [11] K.-S. Yun, I.-J Cho, J.-U. Bu, G.-H. Kim, Y.-S. Jeon, C.-J. Kim, and E. Yoon, “A micropump driven by continuous electrowetting actuation for low voltage and low power operations,” in Proc. IEEE Int. Conf. MEMS, Interlaken, Switzerland, Jan. 2001, pp. 487–490. [12] M. W. J. Prins, W. J. J. Welters, and J. W. Weekamp, “Fluid control in multichannel structures by electrocapillary pressure,” Science, vol. 291, pp. 277–280, 2001. [13] J. Lee, H. Moon, J. Fowler, C.-J. Kim, and T. Schoellhammer, “Addressable micro liquid handling by electric control of surface tension,” in Proc. IEEE Int. Conf. MEMS, Interlaken, Switzerland, Jan. 2001, pp. 499–502. [14] J. Lee, H. Moon, J. Fowler, T. Schoellhammer, and C.-J. Kim, “Electrowetting and electrowetting-on-dielectric for microscale liquid handling,” Sens. Actuators, Phys. A, vol. 95, pp. 259–268, 2002. [15] S. K. Cho, H. Moon, J. Fowler, and C.-J. Kim, “Splitting a liquid droplet for electrowetting-based microfluidics,” in International Mechanical Engineering Congress and Exposition, New York, NY, Nov. 2001, IMECE2001/MEMS-23 831. [16] R. B. Fair, M. G. Pollack, R. Woo, V. K. Pamula, R. Hong, T. Zhang, and J. Venkatraman, “A micro-watt metal-insulator-solution-transport (MIST) devices for scalable digital bio-microfluidic systems,” in Electron Devices Meeting, IEDM Technical Digest International, Washington DC, Dec. 2001, pp. 16.4.1–16.4.4. [17] M. G. Pollack, A. D. Shenderov, and R. B. Fair, “Electrowetting-based actuation of droplets for integrated microfluidics,” Lab Chip, vol. 2, pp. 96–101, 2002. [18] S. K. Cho, S.-K Fan, H. Moon, and C.-J Kim, “Toward digital microfluidic circuits: creating, transporting, cutting and merging liquid droplets by electrowetting-based actuation,” in Proc. IEEE Int. Conf. MEMS, Las Vegas, NV, Jan. 2002, pp. 32–35. [19] J. Fowler, H. Moon, and C.-J. Kim, “Enhancement of mixing by droplet based microfluidics,” in Proc. IEEE Int. Conf. MEMS, Las Vegas, NV, Jan. 2002, pp. 97–100. [20] M. G. Lippmann, “Relations entre les phénomènes electriques et capillaires,” Ann. Chim. Phys., vol. 5, no. 11, pp. 494–549, 1875. [21] B. Berge, “Electrocapillarity and wetting of insulator films by water,” Comptes Rendus de l’Academie des Sciences Serie II, vol. 317, pp. 157–163, 1993.

80


[22] M. Vallet, B. Berge, and L. Vovelle, “Electrowetting of water and aqueous solutions on poly (ethylene terephthalate) insulating films,” Polymer, vol. 37, no. 12, pp. 2465–2470, 1996. [23] H. Moon, S. K. Cho, R. L. Garrell, and C.-J. Kim, “Low voltage electrowetting-on-dielectric,” J. Appl. Phys., vol. 92, no. 7, pp. 4080–4087, 2002. [24] T. B. Jones, M. Gunji, M. Washizu, and M. J. Feldman, “Dielectrophoretic liquid actuation and nanodroplet formation,” J. Appl. Phys., vol. 89, no. 2, pp. 1441–1448, 2001. [25] C. Isenberg, The Science of Soap Films and Soap Bubbles. New York: Dover, 1992, pp. 131–136. [26] S. Chandrasekhar, Hydrodynamic and Hydromagnetic Stability. New York: Dover, 1961, ch. XII. [27] S. Schiaffino and A. A. Sonin, “Formation and stability of liquid and molten beads on a solid surface,” J. Fluid Mech., vol. 343, pp. 95–110, 1997. [28] H. J. I. Verheijen and M. W. J. Prins, “Reversible electrowetting and trapping of charge: model and experiments,” Langmuir, vol. 15, pp. 6616–6620, 1999. [29] V. Peykov, A. Quinn, and J. Ralston, “Electrowetting: a model for contact-angle saturation,” Colloid Polymer Sci., vol. 278, pp. 789–793, 2000.

Sung Kwon Cho received the B.S., M.S., and Ph.D. degrees in mechanical engineering from the Department of Mechanical Engineering, Seoul National University, Korea, in 1990, 1992, and 1998, respectively. In 1999, he joined in Mechanical and Aerospace Engineering Department at the University of California, Los Angeles (UCLA) as a Postdoctoral, where he has worked on the fabrication of MEMS shear stress sensors to study endothelial cell dynamics and on microdroplet manipulations by electrically controlling surface tension (electrowetting). He is currently interested in bio-MEMS and microfluidics.

Hyejin Moon received the B.S. and M.S. degrees in chemical engineering from Sogang University, Seoul, South Korea, in 1995 and 1997, respectively. She is currently a doctoral student with the University of California, Los Angeles (UCLA) MEMS program. Her major research interest is in microactuators using surface tension.

Chang-Jin “CJ” Kim (S’89–M’91) received the Ph.D. degree in mechanical engineering from the University of California at Berkeley in 1991. He received the B.S. degree from Seoul National University, Korea, and the M.S. degree from Iowa State University, Ames, along with the Graduate Research Excellence Award. Upon joining the faculty at University of California, Los Angeles (UCLA) in 1993, he has developed several MEMS courses and established a formal MEMS Ph.D. major field in Mechanical and Aerospace Engineering Department. His research is in MEMS and nanotechnology, including design and fabrication of micro/nanostructures, actuators, and systems, including the use of surface tension. Prof. Kim is the recipient of the 1995 TRW Outstanding Young Teacher Award, the 1997 NSF CAREER Award, and 2002 ALA Achievement Award. He served as Chairman of the Micromechanical Systems Panel of the ASME DSC Division and co-organized the MEMS Symposia between 1994 and 1996 for the ASME International Mechanical Engineering Congress & Exposition. He also organized the 1996 ASME Satellite Broadcast Program on MEMS and the 6th IEEE International Conference on Emerging Technologies and Factory Automation. He is serving in various Technical Program Committees, including the IEEE MEMS Conference, Transducers, and the SPIE Symposium on Micromachining and Microfabrication. He is also serving in the U.S. Army Science Board as Consultant, in the Executive Committee of ASME MEMS Subdivision, and as a Subject Editor for the IEEE/ASME JOURNAL OF MICROELECTROMECHANICAL SYSTEMS. He is a Member of the ASME.