Maximizing Flow Velocities in Redox-Magnetohydrodynamic

Oct 12, 2012 - surface, Magnet Sales Company) was used for the redox-MHD studies. The array chip was placed directly on the magnet. The. Figure 1. (a)...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/ac

Maximizing Flow Velocities in Redox-Magnetohydrodynamic Microfluidics Using the Transient Faradaic Current Melissa C. Weston, Christena K. Nash, Jerry J. Homesley, and Ingrid Fritsch* Department of Chemistry and Biochemistry, University of Arkansas, Fayetteville, Arkansas 72701, United States S Supporting Information *

ABSTRACT: There is a need for a microfluidic pumping technique that is simple to fabricate, yet robust, compatible with a variety of solvents, and which has easily controlled fluid flow. Redox-magnetohydrodynamics (MHD) offers these advantages. However, the presence of high concentrations of redox species, important for inducing sufficient convection at low magnetic fields for hand-held devices, can limit the use of redox-MHD pumping for analytical applications. A new method for redox-MHD pumping is investigated that takes advantage of the large amplitude of the transient portion of the faradaic current response that occurs upon stepping the potential sufficiently past the standard electrode potential, E°, of the pumping redox species at an electrode. This approach increases the velocity of the fluid for a given redox concentration. An electronic switch was implemented between the potentiostat and electrochemical cell to alternately turn on and off different electrodes along the length of the flow path to maximize this transient electronic current and, as a result, the flow speed. Velocities were determined by tracking microbeads in a solution containing electroactive potassium ferrocyanide and potassium ferricyanide, and supporting electrolyte, potassium chloride, in the presence of a magnetic field. Fluid velocities with slight pulsation were obtained with the switch that were 70% faster than the smooth velocities without the switch. This indicates that redox species concentrations can be lowered by a similar amount to achieve a given speed, thereby diminishing interference of the redox species with detection of the analyte in applications of redox-MHD microfluidics for chemical analysis.

A

Adding electroactive species, which undergo oxidation and reduction at the electrodes, can result in faradaic processes that yield high ionic currents with low voltages and can alleviate those problems. This approach, termed redox-MHD (or RMHD), is the focus of this work. The application of redoxMHD in microfluidics13−21 usually makes use of relatively high concentrations of redox species to induce convection in the presence of the low magnetic flux densities of small permanent and microfabricated electromagnets of interest for small devices. However, high concentrations of redox species used for pumping can cause interferences in detection methods either by “drowning out” the signal of the analyte of interest, as might be the case with electroanalytical methods, or through chemical or electron transfer reactions with the analyte, changing the analyte’s concentration. This would be a problem where both analyte detection and redox-MHD pumping are performed in the same chemical analysis system. For example, in redox-MHD enhancements of anodic stripping voltammetry,14,17,20 it was necessary to perform a manual dilution or rinsing step to significantly diminish or remove the pumping species prior to the detection step, complicating the procedure. In another

well-known phenomenon associated with an electrochemical reaction is the transient faradaic current that occurs when the voltage at the electrode is stepped to potentials sufficiently beyond the standard electrode potential to convert oxidized species to the reduced form, or vice versa. This results in a high, initial electronic current due to conversion of the electroactive species at the electrode surface, which are initially at the bulk concentration before they are depleted by electron transfer events. The work reported herein involves investigations into using this transient faradaic current to increase fluid velocities driven by redox-magnetohydrodynamics (MHD) in order to ultimately minimize the concentration of redox species and therefore improve the suitability of redox-MHD microfluidics for a wider variety of lab-on-a-chip applications. The use of MHD in applications of small-scale chemical analysis, and particularly for microfluidics, has not been widely investigated.1−3 MHD is based on the interaction of a localized ionic current density, j, perpendicular to a magnetic field, B, there to generate a magnetic force, FB, to induce fluid convection (FB = j × B, where FB can be proportional to velocity).4 Most of the MHD pumping procedures in the literature require high voltages to induce solution convection.5−12 These high voltages result in bubble formation, due to the electrolysis of water, which can interfere with the pumping. Electrode degradation is also a problem at these high potentials, thereby limiting the lifetime of these devices. © 2012 American Chemical Society

Received: August 1, 2012 Accepted: October 12, 2012 Published: October 12, 2012 9402

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

40) was used to insulate the electrode leads and was purchased from Dow Corning Company. Redox-MHD and Tracking Beads. The redox-MHD setup (Figure 1a) and monitoring fluid flow by tracking microbeads

example, redox-MHD at pumping electrodes was demonstrated while simultaneously detecting a plug of solution by electroanalysis at a separate, much smaller electrode, where a low enough concentration (5 mM) of pumping redox species was used so that a rinse step was unnecessary and quantification of the analyte (enzymatically generated p-aminophenol) at the detecting electrode was still possible.22 Yet, the detection limits for the analyte were about five times higher with the presence of the pumping species than without. Recent studies have begun to diminish the concerns. Electrochemically quasireversible pumping species (e.g., 1,4-benzoquinone) can be chosen to eliminate the rinsing step in anodic stripping voltammetry.23 Also, in an investigation exploring the incorporation of redox-MHD microfluidics to fine-tune fluid flow around a tissue on a chip, it was shown that 5 mM of the pumping species [Ru(NH3)6]3+ does not seem to affect the viability of heart tissue.24 However, until now, the previous efforts did not involve modifying the methodology itself to improve fluid velocities so that a further decrease in concentration of the pumping redox species could be achieved to lower detection limits and lessen reactive interferences. Here, a new approach for optimizing the ionic current produced at pumping electrodes is explored that enhances the redox-MHD fluid flow for a given redox concentration. It takes advantage of the high current in the transient portion of the faradaic response achieved at the beginning of an applied potential step. Fluid velocities, as estimated by tracking microbeads over microelectrode arrays in a small cell containing an electrolyte solution of potassium chloride with the electroactive pumping species of potassium ferricyanide and potassium ferrocyanide, were studied as a function of time, in the presence of a small permanent Nd−Fe−B magnet, to determine factors such as potential waveform, electrode gap, and switching intervals that will lead to enhanced fluid flow. Velocities were improved by incorporating a “switch” between the potentiostat and electrochemical cell to alternate the turning of electrodes on, to keep fluid flowing, and off, to allow depleted redox species to recover at selected times to increase the frequency of faradaic transients in the system.

Figure 1. (a) Photograph of experimental setup showing method for tracking microbeads in solution. (b) Schematic of reinforcing flow between electrodes of opposite bias in a doubly-paired configuration showing a large gap assignment and dimensions of features in the array region.

■

EXPERIMENTAL SECTION Chemicals and Materials. All chemicals were of analytical grade and used as received. Aqueous solutions were prepared with reagent grade, 18 MΩ deionized water from Ricca Chemical Company (Arlington, TX). Potassium ferricyanide [K3Fe(CN)6] was obtained from EM Science (Gibbstown, NJ) and potassium ferrocyanide trihydrate [K4Fe(CN)6·3H2O] was from J.T. Baker (Phillipsburg, NJ). Potassium chloride was purchased from Aldrich Chemical Co. (St. Louis, MO). Polystyrene latex microspheres (10 μm diameter), functionalized with surface sulfate groups (2.5 wt % dispersion in water), were obtained from Alfa Aesar (Ward Hill, MA). Polydimethylsiloxane (PDMS) supplies (Sylgard 184 silicone elastomer base, Sylgard 184 silicone elastomer curing agent, and OS-30 solvent) were obtained from Dow Corning Corp. Silicon wafers (125 mm diameter and 600−650 μm thickness) with 2 μm of thermally grown SiO2 were purchased from Silicon Quest International (Santa Clara, California) and were used as the substrate materials for electrode arrays. A gold coin (Canadian Maple Leaf, 99.99%) and a chromium-plated tungsten rod (Kurt J. Lesker Company, Clairton, PA) were used for metal deposition onto the silicon wafer for the electrode arrays. Benzocyclobutene (BCB) (Cyclotene 4024−

(see the Supporting Information) are based on prior studies.15 The movement of the 10 μm beads is expected to follow fluid flow accurately, because their response time25−27 of 6.1 μs (for a particle density of 1.05 g/cm3 and dynamic viscosity of 9.55 × 10−4 kg/ms) is well below the shortest switching interval used in the experiments. A microelectrode array chip was placed on top of a Nd−Fe−B permanent magnet. A PDMS film (760 μm thick) with a rectangular opening (14.0 × 6.0 mm) was placed on the microelectrode array chip and filled with redox pumping solution containing microbeads and served to define the sidewalls and height of the cell. A glass microscope slide on top of the PDMS gasket confined the solution and formed the top wall. The entire apparatus was placed under a Nikon Eclipse ME600P microscope, interfaced to a Sony Handycam digital camera (model no. HDR-XR500 V, 30 frames per second with 1920 × 1080 pixels per frame), to record movies for visualizing the fluid flow. Magnetic Field. A 1 × 1 × 0.5 in. permanent, Nd−Fe−B sintered magnet (1.23 T residual induction, 0.55 T on the surface, Magnet Sales Company) was used for the redox-MHD studies. The array chip was placed directly on the magnet. The 9403

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

times. A high transition of the clock turns on a latch, which transfers the applied voltage from the potentiostat to the cell, for the duration of the clock. A low transition of the clock keeps the latch in the off state so the electrodes are left at open circuit.

north pole pointed downward, perpendicular to the plane of the chip. The magnetic field at the arrays was measured to be 0.38 T by a DC magnetometer (AlfaLab Inc.). Microelectrode Array Chips. A 1 × 1 in. chip was used for all studies, containing two parallel linear arrays of individually addressable electrodes. The arrays are separated by a distance of 243 μm and each has nine 215 μm wide × 97 μm long rectangular pumping electrodes, separated by 29 μm gaps, to total 18 electrodes all together. A photograph of the overall chip showing contact pads and leads is shown in Figure S-1 of the Supporting Information. The reasons explaining the choice of electrode dimensions and a description of the microfabrication procedures are also included in the Supporting Information. A schematic of the electrode arrays showing dimensions is pictured in Figure 1b. Electrochemical Control. The chronoamperometry (CA) technique and only one working-electrode channel of a CHI 1030A multipotentiostat were used. The working electrode lead was attached to one connector of the switch (see below and in the Supporting Information). The auxiliary and reference electrode leads were shorted together and attached to a different connector on the switch. Two configurations of “doubly paired” electrodes on the microelectrode array chip were hooked up to the switch via the wires on the edge connector. In one configuration of doubly paired electrodes, two nonadjacent electrodes shorted together in one array served as the anode (when connected by the switch to the working lead of the potentiostat), and two electrodes in the other array, directly across the 243 μm gap between arrays, served as the cathode (when connected by the switch to the shorted auxiliary and reference leads of the potentiostat). In a second doubly paired configuration, electrodes were also connected to the switch and could be activated by the potentiostat through the switch, but not at the same time as the first configuration. The electrodes in the second configuration were a fixed distance away from the corresponding electrodes of the first configuration, with two anodes in one array and two cathodes in the other. Switching experiments involved alternately activating (+0.3 V) and inactivating (open circuit) the two configurations of doubly paired electrodes. This assignment of anodes and cathodes produces MHD forces that act in a reinforcing way, resulting in solution movement in a path between the two electrode arrays. A schematic representation of the ionic current vectors, magnetic field, and resulting MHD force (and flow) is shown for a single configuration of doubly paired electrodes in Figure 1b. The starting solution contained a 1:1 mol ratio of 0.1 M K3Fe(CN)6 and 0.1 M K4Fe(CN)6 in a 0.1 M KCl supporting electrolyte. The microbeads were pipetted into the redox solution to achieve a 30× dilution of the beads (from 2.5 to 0.083 wt % dispersion). Switch Design. An electronic circuit was designed to automatically switch the two sets of doubly paired electrodes between active and inactive modes. Details about the switch are provided with a schematic of the electronic circuit in Figure S-2 of the Supporting Information. Basically, the circuit consists of a clock pulse generator for timing and an analog switch to connect the leads from the potentiostat to the electrodes (cell). The clock pulse generator has two momentarily stable states between which it continuously alternates, remaining in each form for a period of time and controlled by the adjustable circuit parameters. The clock is applied to the latches which switch the output of the potentiostat to the cell at designated

■

RESULTS AND DISCUSSION The Transient Portion of the Faradaic Current Response during a Potential Step. A feature of the electrochemical signal obtained when a potential step (like that in CA) is made to a value well past E° for a redox couple is an initial spike in the electronic current due to charging of the double layer at the electrode−solution interface (if different than the potential of zero charge) and electrochemical conversion of redox species immediately adjacent to the electrode. Once the charging current has diminished, having an exponential falloff with time, the faradaic current dominates and has a t−1/2 decay due to a linear diffusion-limited arrival of redox species at the electrode. For a microelectrode at long time scales in static solution, the electronic current eventually reaches a steady- or pseudosteady state due to the radial component of diffusion of the redox species. A plot of electronic current as a function of time for a simple potential step procedure (using a single configuration of doubly paired electrodes as designated in Figure 1b) is shown in Figure 2a and follows the expected trend. A plot of the corresponding

Figure 2. (a) Overlay of electronic current and averaged instantaneous bead velocity (for four different beads) as a function of time resulting from a single potential step experiment (control study) at one paired set of 215 μm wide × 97 μm long rectangle electrodes for a large gap electrode configuration. The combined auxiliary and quasi-reference cathodes (two, 215 μm wide × 97 μm long Au rectangles) were 243 μm away from the paired working anodes. (b) Microscope image (focused at 300 μm above the chip) on which is superimposed black and white markers that show the locations of two different beads (each color represents one bead) at 0.2 s intervals that were tracked in solution over a 2.6 s period for the experiment in (a). The working (anode, +) and combined auxiliary and quasi-reference (cathode, −) electrodes are indicated on the figure. 9404

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

bead velocity for this electronic current signal is also shown in Figure 2a, and an image of two beads tracked in solution (black and white markers) over 2.7 s is shown in Figure 2b. The similar trends of electronic current and velocity with time, as determined by slopes that are within error of each other when the data are plotted as a log−log function and fit to a line [the slope for log velocity (μm s−1) vs log time (s) was −0.298 ± 0.0097, while the slope for log electronic current (μA) vs log time (s) is −0.301 ± 0.0042], suggest that the electronic current passing through the electrodes is proportional to the velocity and is consistent with previously reported results for redox-MHD in small volumes.4 Thus, it would be desirable to use just the large amplitude of the initial, transient portion of the faradaic current repeatedly to achieve the highest fluid velocities in redox-MHD. The challenge that arises in using a simple potential step to harness the faradaic transient current to enhance fluid propulsion is that recovery of the depleted redox species cannot take place. This causes the ionic current, and therefore the fluid velocity, to decrease over time. Repeated sweep-step potential waveforms at a single configuration of doubly paired electrodes were investigated to maximize the contribution to MHD, where an initial, high amplitude, transient portion of the faradaic current was caused by a potential step, and then a linear sweep segment was intended to replenish the diffusion layer. However, this approach did not provide a net advantage over a simple potential step waveform. It produced a noncontinuous fluid flow that occurred during the sweep segment and was unable to repeat equally high transient faradaic currents because recovery of depleted redox species was insufficient after several sweepstep cycles.28 Thus, a switching procedure was investigated instead (as described below), to better retain the high, initial transient faradaic current and sustain an enhanced fluid flow over long periods of time. Switching Experiments. It has been demonstrated that redox-MHD is capable of affecting fluid flow at locations quite distant to the microelectrodes (e.g., >1 cm) that generate the ionic current.29 Thus, it should be possible to allow almost complete recovery of depleted redox species while sustaining fluid flow enhanced by faradaic transients, by alternately switching on and off different configurations of electrodes that are located outside of the other's diffusion layers and with the right timing. To accomplish this, a device was fabricated with a design that divides up the pumping electrodes into multiple, individually addressable pairs, separated length-wise by gaps. A switch was employed between the potentiostat and microelectrode chip to alternately turn on and off every other pair to allow pumping to continue in the same direction (by the activated electrodes) while allowing enough time for recovery of redox species by diffusion and convection (at the inactivated electrodes). The switching procedure uses different sets of paired electrodes and is illustrated in Figure 3a for the small gap and Figure 3b for the large gap configurations. Two electrodes in one array (set A) originally serve as the working anodes, while two other electrodes (set C) serve as the combined auxiliary and quasi-reference cathodes (ΔE = 0.3 V) for some amount of time based on the frequency of the switch, starting at time 1 (t1) in Figure 3. At time 2 (t2), this first configuration of doubly paired electrodes (sets A and C) is then switched to open circuit as two other working anodes (set B) and two other cathodes (set D, as combined auxiliary and quasi-reference) are activated with ΔE = 0.3 V, for the same amount of time. While

Figure 3. Diagrams of switching procedures for (a) small and (b) large gap configurations. Set A initially serves as the working electrodes (anodes, +), while Set C serves as the combined auxiliary and quasireference electrodes (cathodes, −) at time 1 (t1). The potential is held constant until time 2 (t2), when the electrodes are switched. Set A and Set C go to open circuit, while Set B is activated as the working anodes, and Set D serves as the combined auxiliary and quasi-reference cathodes. Again, the potential is held constant. This cycle is then repeated. The gap between edges of similarly biased pumping electrodes in the small gap configuration, at any given time, is 155 μm. The gap in the large gap configuration is 407 μm. (The electrodes used for the control procedure, when there was no switching, consisted of activating only Set A and Set C, while Set B and Set D were left at open circuit for the duration of the experiment.) The microscope image over which the diagrams are superimposed are of the two parallel linear arrays of electrodes, with the focus at the chip surface. A small stretch of the edge of the BCB insulator is highlighted with a dashed line for clarity, shown in (a).

sets A and C are active, sets B and D are inactive (i.e., at open circuit) and vice versa. This switching is repeated and was investigated here with three different times between the switching events (119 ms, 220 ms, and 440 ms) for a total experiment time of 90 s. For the control procedure (no switching), A is the set of working anodes, and C is the set of combined auxiliary and quasi-reference cathodes; meanwhile, sets B and D are left at open circuit for the duration of the 90 s. The 90 s time frame is a practical period for evaluation, because it is sufficient for some analytical applications. For example, 9405

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

transient portion of the faradaic response each time the electrode sets are switched and is seen in the electrochemical signal (Figure 4a) in which a spike and subsequent falloff in the electronic current is observed every 220 ms, minimizing the pseudosteady-state portion of the curve. For the simple potential step procedure, the faradaic current is at a maximum only once, at the beginning of the experiment and then falls off to reach a low pseudosteady-state current over the remaining time. Likewise, the bead velocities from the switching procedure (220 ms interval) were also consistently higher than those from the control studies as shown in Figure 4b for the large gap. The plotted bead velocities in Figure 4b are the average velocities of four different beads over a 1.67 s period at six different times during the experiment (1, 5, 10, 30, 60, and 90 s into the experiment). A velocity was determined by measuring the net displacement of each bead over a 1.67 s period and dividing by that time. Velocity transients are not observed in Figure 4b, and the speeds appear almost continuous as represented by the small error bars in that figure. One would expect, however, the fluid flow to pulse, following the electronic current transients, moving faster at the beginning and slowing at the end of each transient. The periodicity is easily evident when watching the video for the 440 ms switching interval at the small gap configuration (see video ac302063a_si_002.mpg of the Supporting Information). A quantitative assessment of the fluid flow at the time resolution required to follow it during a switching event is described in the Supporting Information and shown in Figure S-4a over a 1.8 s duration at 60 s into the experiment. The 440 ms periodicity in velocity is present, but difficult to see amidst the large measurement error. For comparison, the control data for the same expanded period (Figure S-4b of the Supporting Information) also exhibit large error values for the same reasons, but do not display the periodicity. The percent enhancement of velocities obtained with switching over the control procedure, at each of the six sampling times, are indicated in Figure 4b, where % enhancement = 100% [(velocityswitching − velocitycontrol)/ (velocitycontrol)]. Unfortunately, the high velocities achieved at the beginning of the experiment are not maintained throughout. For both control and switching operations, the ionic current and resulting velocities initially start high and then decrease to a pseudosteady state. A plot of the average electronic current over 1.67 s periods (i.e., 417 points, each point sampled every 4 ms) at six different times over the entire experiment of 90 s for both procedures is shown in Figure 5a. (Note that error bars were added to the plot for a sense of completion, but caution should be taken so as not to overinterpret them. They are very large for the switching procedure because each includes 7.59 sets of faradaic transients with 220 ms intervals over the period represented by each marker. For the control procedure, the largest change in current occurs at the beginning of that experiment, and thus the error bars are largest there. The noise introduced by the switch itself is best reflected in the error at long times for the control procedure.) An approximate 32% decrease in electronic current over time was observed for the control studies; a 36% decrease was observed for the switching procedures. Thus, even by allowing time for the diffusion layer to become replenished via the electrode switching in addition to the presence of convection from the MHD force which should introduce fresh species, it is still not possible to sustain the high current

commercially available point-of-care systems have analysis times in the tens of seconds and some separations on a chip can last a few minutes.30 Two different gap dimensions between each set of similarly biased electrodes of the doubly paired electrode configurations were studied. The “small gap” had a distance of 155 μm between the edges of the anodes and cathodes (Figure 3a). The “large gap” distance was 407 μm (Figure 3b). Velocity measurements were made approximately centered between the four biased electrodes, where the maximum flow rates were observed. There was no visible difference in the overall flow trajectory between the switching (not shown) and the control procedures (Figure 2b). It is interesting to note that the flow is linear between oppositely biased electrodes down the length of the two arrays, rather than circulating around individual pumping electrodes, even for a large gap size of 407 μm. This phenomenon demonstrates an advantage of redox-MHD for fluid flow: pumping electrodes positioned far from one another can still direct flow without the need for channel sidewalls.22 Figure S-3 of the Supporting Information illustrates the overall fluid flow and the general shift of flow that occurs upon switching. The raw electronic current responses obtained from the “no switching” (gray) and switching (black, 220 ms interval) procedures for the large gap configuration in Figure 3b and over a short time of 3 s are given in Figure 4a. The electronic currents achieved from the switching procedure were consistently higher than those from control studies. This is a result of repetitively generating the large current of the

Figure 4. Comparison of (a) raw electronic current responses and (b) corresponding bead velocities over time for the control (gray curve and black square, sets A and C only in Figure 3b,) and switching (black curve and ●, alternating sets A and C with sets B and D in Figure 3b at 220 ms switching intervals) procedures for the large gap configuration. Each velocity marker in (b) corresponds to an average of four different beads whose velocities were determined by measuring the displacement of each bead over a 1.67 s period and then dividing by 1.67 s. Velocities were sampled at six different times throughout the experiment, and the markers were placed at the starting time for the time period used to determine the velocity. 9406

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

over the times investigated here, due to an inability for the solution adjacent to the electrodes to fully recover from the depletion of redox species that initially occurs. However, by using the switching approach, the velocities have significantly improved by an average 63% over control experiments for results shown in Figure 4b (large gap, 220 ms). We do not know at this point how much longer than 90 s the switching procedure can maintain this advantage over the control procedure. However, an extrapolation of the velocities in Figure 4b beyond 90 s suggests that the enhancement should be sustainable over several minutes at least. (Note that we do not expect a “thin layer effect” to occur within 90 s. Even if the electrodes are “on” continuously for 90 s and there is no supply of redox species from MHD convection, the diffusion length, 357 μm for a diffusion coefficient of 7.09 × 10−6 cm2/s, should not exceed the cell height, 760 μm. In addition, because alternating electrodes are switched on and off, the composition change is less than if all electrodes were activated for the entire 90 s.) The gap size and frequency of the switch were varied in attempts to further increase velocities. A smaller gap size decreases the distance between similarly biased sets of electrodes, so that they are more likely to affect solution composition at adjacent electrodes, and is therefore expected to allow for less recovery of the redox species between the sets of active electrodes. Faster switching times (shorter switching intervals) should, (1) decrease the length of the diffusion layer, allowing for improved redox species recovery at adjacent electrodes and (2) result in higher average electronic currents because of the shorter time frame (less time for the current to decrease from depletion of the redox species for each interval). Electronic currents, fluid velocities, and percent enhancements between switching and control procedures obtained for these studies are presented in Table 1 (for the small gap configuration) and Table 2 (for the large gap configuration). Although all three switching intervals produced fluid flow that was significantly greater than the control, there was only a very slight impact on electronic currents and velocities when

Figure 5. Comparisons of (a) electronic current and (b) fluid velocity as a function of time for control (squares) and switching (circles) procedures, both performed with the small gap (□ and ○) and large gap (■ and ●) configurations. The switching interval was 220 ms. The electronic current for switching experiments was sampled every 4 ms over 1.67 s, beginning at each time indicated on the plot, and is reported as an average of 417 points. Error bars of electronic current represent plus or minus one standard deviation of the 417 points. These are large for the switching procedure because they include 7.59 sets of faradaic transients and smaller for the control procedure because there is only one faradaic transient whose largest change in current is at the beginning of the experiment. Markers in (b) correspond to an average of speed determined for four beads, based on measuring the net displacement of each bead over a 1.67 s period and dividing by that time. Error bars represent plus or minus one standard deviation over the four measurements.

Table 1. Electronic Currents and Velocities for the Small Gap Configuration (Figure 3a)a time during experiment switching interval/ms ∞ 440

220

119

velocity (μm s−1) current (μA) velocity (μm s−1) % enhancement current (μA) % enhancement velocity (μm s−1) % enhancement current (μA) % enhancement velocity (μm s−1) % enhancement current (μA) % enhancement

1s

5s

10s

30 s

60s

90s

107 ± 3.6 9.14 171 ± 11 60 13.9 ± 5.4 52 163 ± 3.5 53 13.8 ± 4.0 51 158 ± 1.5 48 13.9 ± 3.2 52

77.0 ± 3.2 7.44 121 ± 2.9 57 10.0 ± 4.0 34 112 ± 2.3 32 9.90 ± 2.9 33 117 ± 5.8 52 10.4 ± 2.5 40

72.7 ± 2.1 6.79 104 ± 6.7 43 9.13 ± 3.6 34 101 ± 6.1 39 9.07 ± 2.7 34 104 ± 2.7 43 9.07 ± 2.2 34

67.5 ± 0.9 6.34 93 ± 4.2 38 8.36 ± 2.9 32 92.4 ± 3.0 37 8.39 ± 2.6 32 94.3 ± 3.5 40 8.35 ± 2.3 32

66.0 ± 2.4 6.18 92 ± 3.2 39 8.05 ± 3.4 30 85.3 ± 5.2 29 8.12 ± 2.6 31 93.7 ± 5.5 42 8.11 ± 2.2 31

67.0 ± 2.9 6.10 89 ± 3.7 32 7.64 ± 3.0 25 86.2 ± 5.5 29 7.91 ± 2.6 30 86.9 ± 5.4 30 7.99 ± 2.2 31

a These measurements were taken over periods of 1.67 s, at different times throughout each experiment, for control conditions (indicated as ∞ switching interval) and different switching intervals. Reported values for velocities are the average and standard deviation of four replicate bead displacement measurements. Reported values for electronic current are the average and standard deviation of current sampled every 4 ms for a total of 417 points and include 3.80, 7.59, and 14.0 sets of faradaic transients for switching intervals of 440 ms, 220 ms, and 119 ms, respectively. Percent enhancements of electronic currents and velocities during switching over those of control experiments are also indicated.

9407

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

Table 2. Electronic Currents and Velocities for the Large Gap Configuration (Figure 3b)a time during experiment switching interval/ms ∞ 440

220

119

−1

velocity (μm s ) current (μA) velocity (μm s−1) % enhancement current (μA) % enhancement velocity (μm s−1) % enhancement current (μA) % enhancement velocity (μm s−1) % enhancement current (μA) % enhancement

1s

5s

10 s

89.4 ± 3.5 9.44 144 ± 5.2 62 14.6 ± 4.9 55 152 ± 4.0 61 15.3 ± 4.8 62 149 ± 5.1 67 15.0 ± 3.5 59

71.4 ± 3.5 7.73 111 ± 4.9 56 11.7 ± 3.7 51 122 ± 6.5 71 12.6 ± 4.1 63 120 ± 2.6 69 12.2 ± 2.7 58

62.8 ± 2.8 7.11 99.4 ± 2.9 57 10.5 ± 3.2 48 107 ± 4.6 70 11.4 ± 3.7 60 103 ± 1.4 64 11.2 ± 2.6 58

30 s 60.5 6.69 91.1 50 9.74 46 95.4 58 10.3 54 93.5 54 10.1 51

± 2.9 ± 3.8 ± 3.4 ± 4.3 ± 3.4 ± 3.2 ± 2.5

60 s 58.2 6.53 87.7 51 9.34 43 93.0 60 9.97 53 91.3 57 9.71 49

± 2.4 ± 1.6 ± 3.2 ± 2.1 ± 3.5 ± 1.1 ± 2.5

90 s 56.2 6.45 90.9 62 9.41 46 90.3 61 9.81 52 88.1 57 9.56 48

± 3.2 ± 0.43 ± 3.4 ± 4.4 ± 3.45 ± 1.6 ± 2.6

The measurements are taken over periods of 1.67 s, at different times throughout each experiment, for control conditions (indicated as ∞ switching interval) and different switching intervals. Reported values for velocities are the average and standard deviation of four replicate bead displacement measurements. Reported values for electronic current are the average and standard deviation of current sampled every 4 ms for a total of 417 points and include 3.80, 7.59, and 14.0 sets of faradaic transients for switching intervals of 440 ms, 220 ms, and 119 ms, respectively. Percent enhancements of electronic currents and velocities during switching over those of control experiments are also indicated. a

consider. One is that there is a greater recovery of redox species that occurs at increased distances between active electrodes, therefore providing a consistently greater current for the large gap configuration. And secondly, depletion is much more significant for smaller gaps between electrodes, because of the shielding effect,31 thus lowering the electronic current. The velocities, however, for both large and small gap configurations are similar throughout (Figure 5b) and can be explained with the second phenomenon, which is the same argument as was used to explain the velocities in the control experiments. That is, the smaller cross section for the ionic current to pass in the smaller gap configuration produces an ionic current density j that balances the lower electronic current, producing an MHD force and velocities that are more comparable to those obtained for the large gap. However, marginally greater velocities were still obtained with the large gap, giving it a slight advantage.

comparing the different switching intervals (119, 220, and 440 ms) with either gap size. One might expect that as the switching interval increased, the results should approach those of the control experiment (infinite interval). A careful ANOVA evaluation of the large gap data revealed that there was indeed a significant increase in bead speeds at the 95% confidence level between the 440 and 220 ms switching intervals for the 1 s, 5 s, 10 s, and 60 s times and between the 440 and 119 ms switching intervals for the 5 and 60 s times. The ANOVA evaluation of the small gap data showed only a significant difference in bead speeds at the 95% confidence level between the 440 and 119 ms switching intervals for the 1s time and between the 220 and 119 ms switching intervals for the 60 s time. (Unlike other data in the tables, the speed for the 1 s time at the small gap had the opposite trend than expected: it significantly decreased with a decrease in the switching interval from 440 to 119 ms but might be explained by the difficulty in making an accurate velocity measurement in that rapidly changing part of the transient.) A comparison of results between the two gap configurations exhibits clear differences. Figure 5 shows data over the entire experiment for controls and the 220 ms switching interval for both the small gap and large gap configurations. For the control studies, the electronic current for the small gap configuration was similar or only slightly smaller than that for the large gap (see Figure 5a), presumably because of a slightly greater shielding effect of adjacent electrodes. In contrast, the velocities shown in Figure 5b were consistently higher for the small gap than for the large gap configuration. This can be explained by the smaller cross-sectional area in the small gap configuration from similarly biased electrodes that are closer together that causes a higher ionic current density (greater |j|) that balances the slightly lower net current and produces a greater MHD force comparable to the large gap configuration. For the switching procedure, and in contrast to the control experiments, gap size affected the electronic current (Figure 5a) much more than the velocity magnitude (Figure 5b). More extensive data for all switching intervals are listed in the tables. In the presence of switching, there are two main phenomena to

■

CONCLUSIONS The high, faradaic transient current that occurs when the potential is stepped at an electrode results in fast, initial flow velocities in redox-MHD. A novel method using a switch to alternately turn on and off electrodes, in order to maximize contributions from this transient faradaic current while maintaining a fairly continuous flow, was implemented and exhibited promising results. The switching procedure enhanced velocities over those of a simple potential step function by a minimum of 30% and as large as 70%. Thus, redox species concentrations could be further decreased to achieve sufficient pumping speeds. This will have significant implications for the future of redox-MHD as a pumping technique for lab-on-a-chip applications, where microfluidics and analyte detection coexist in the same device. Lower concentrations of the pumping redox species would diminish the extent of interference with the analyte, and thereby improve the detection limit. (The quantitative impact on the detection limit will depend on the nature of the interference and the detection method and thus should be determined experimentally on a case-by-case basis. If detection is to be performed via electroanalysis, it is recommended that it take place at a different set of 9408

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409

Analytical Chemistry

Article

(16) Arumugam, P. U.; Clark, E. A.; Fritsch, I. Anal. Chem. 2005, 77, 1167−1171. (17) Clark, E. A.; Fritsch, I. Anal. Chem. 2004, 76, 2415−2418. (18) Grant, K. M.; Hemmert, J. W.; White, H. S. J. Am. Chem. Soc. 2002, 124, 462−467. (19) Leventis, N.; Gao, X. Anal. Chem. 2001, 73, 3981−3992. (20) Weston, M. C.; Anderson, E. C.; Arumugam, P. U.; Yoga Narasimhan, P.; Fritsch, I. Analyst 2006, 131, 1322−1331. (21) Wilkes, J. S.; Williams, M. L.; Musselman, R. L. Electrochemistry 2005, 73, 742−744. (22) Weston, M. C.; Nash, C. H.; Fritsch, I. Anal. Chem. 2010, 82, 7068−7072. (23) Ensafi, A. A.; Nazari, Z.; Fritsch, I. Analyst 2012, 137, 424−431. (24) Cheah, L. T.; Fritsch, I.; Haswell, S. J.; Greenman, J. Biotechnol. Bioeng. 2012, 109, 1827−1834. (25) Melling, A. Meas. Sci. Technol. 1997, 12, 1406−1416. (26) van Hout, R. Int. J. Multiphase Flow 2011, 37, 345−357. (27) Lindken, R.; Rossi, M.; Große, S.; Westerweel, J. Lab Chip 2009, 9, 2551−2567. (28) Weston, M. C. Redox-Magnetohydrodynamic (MHD) Microfluidics: Fundamentals, Optimization, and Applications to Analytical Chemistry. Ph.D. Dissertation, University of Arkansas, Fayetteville, AR, 2010. (29) Scrape, P. G.; Gerner, M. D.; Weston, M. C.; Fritsch, I. J. Electrochem. Soc. 2012, to be submitted. (30) Tudos, A. J.; Besselink, G. A. J.; Schasfoort, R. B. M. Lab Chip 2001, 1, 83−95. (31) Bard, A. J.; Crayston, J. A.; Kittlesen, G. P.; Shea, T. V.; Wrighton, M. S. Anal. Chem. 1986, 58, 2321−2331.

electrode(s), independent of the pumping electrodes, to avoid the noise introduced by the switching events.) It is anticipated that the proper design of electrode gap, geometry, and timing, as well as activating more electrodes (beyond a pair at a time) and multiple sets of electrodes (greater than two) over a larger region or path will provide fluid flow over longer distances with higher velocities and with more efficient recovery of redox species.

■

ASSOCIATED CONTENT

* Supporting Information S

Reasons behind the design and dimensions of the electrodes, details of microfabrication of electrode array chips, a figure showing the entire chip with contact pads, leads, and array region, a description of video processing and quantification of bead velocities, a schematic of the electronic circuit for the switch with a description of its components and construction, a schematic illustrating fluid flow circulation and shifting that occur with switching, a figure and discussion of quantifying the periodicity of bead speed for switching (440 ms interval) and for the control for the small gap configuration, and a video clip showing an example of bead movement during a switching experiment (small gap, 440 ms interval). This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Email: [email protected]. Tel: (479) 575-6499. Fax: (479) 575-4049. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank the National Science Foundation (CHE-0719097) for financial support. We acknowledge Henry S. White for initially suggesting sweep-step experiments for a nonredox solution, which ultimately led us to the switching studies presented here for the redox-containing system.

■

REFERENCES

(1) Pamme, N. Lab Chip 2006, 6, 24−38. (2) Qian, S. Z.; Bau, H. H. Mech. Res. Commun. 2009, 36, 10−21. (3) Weston, M. C.; Gerner, M. D.; Fritsch, I. Anal. Chem. 2010, 82, 3411−3418. (4) Weston, M. C.; Fritsch, I. Sens. Actuators, B 2012, 173, 935−944. (5) Bau, H. H.; Zhong, J. H.; Yi, M. Q. Sens. Actuators, B 2001, 79, 207−215. (6) Bau, H. H.; Zhu, J. Z.; Qian, S. Z.; Xiang, Y. Sens. Actuators, B 2003, 88, 205−216. (7) Eijkel, J. C. T.; Dalton, C.; Hayden, C. J.; Burt, J. P. H.; Manz, A. Sens. Actuators, B 2003, 92, 215−221. (8) Homsy, A.; Koster, S.; Eijkel, J. C. T.; van den Berg, A.; Lucklum, F.; Verpoorte, E.; de Rooij, N. F. Lab Chip 2005, 5, 466−471. (9) Homsy, A.; Linder, V.; Lucklum, F.; de Rooij, N. F. Sens. Actuators, B 2007, 123, 636−646. (10) Lemoff, A. V.; Lee, A. P. Sens. Actuators, B 2000, 63, 178−185. (11) Nguyen, B.; Kassegne, S. K. Microfluid. Nanofluid. 2008, 5, 383− 393. (12) Qian, S.; Bau, H. H. Sens. Actuators, B 2005, 106, 859−870. (13) Aguilar, Z. P.; Arumugam, P. U.; Fritsch, I. J. Electroanal. Chem. 2006, 591, 201−209. (14) Anderson, E. C.; Fritsch, I. Anal. Chem. 2006, 78, 3745−3751. (15) Anderson, E. C.; Weston, M. C.; Fritsch, I. Anal. Chem. 2010, 82, 2643−2651. 9409

dx.doi.org/10.1021/ac302063a | Anal. Chem. 2012, 84, 9402−9409