Anal. Chem. 2006, 78, 5369-5374
Using Channel Depth To Isolate and Control Flow in a Micro Free-Flow Electrophoresis Device Bryan R. Fonslow,† Victor H. Barocas,‡ and Michael T. Bowser*,†
Department of Chemistry, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455, and Department of Biomedical Engineering, University of Minnesota, 312 Church Street SE, Minneapolis, Minnesota 55455
A multiple-depth micro free-flow electrophoresis chip (µFFE) has been fabricated with a 20-µm-deep separation channel and 78-µm-deep electrode channels. Due to the difference in channel heights, the linear velocity of buffer in the electrode channels is ∼15 times that of the buffer in the separation channel. Previous µ-FFE devices have been limited by electrolysis product formation at the electrodes. These electrolysis products, manifested as bubbles, decreased the electric field and disrupted the buffer flow profile, limiting performance and preventing continuous operation. Using channel depth to control buffer flow over the electrodes and in the separation channel effectively removes electrolysis products, allowing continuous operation. The linear velocities in the channels were confirmed using particle velocimetry and compared well with values predicted using lubrication theory. A separation potential of 645 V could be applied before significant Joule heating was observed. This corresponded to an electric field of 586 V/cm in the separation channel, a 4-fold increase over our previous design. A separation of fluorescent standards was demonstrated using the new µ-FFE device. Resolution increased by a factor of 1.3 over our previous design, even when operated under similar conditions, suggesting that effective removal of electrolysis products is more important than originally thought. Free-flow electrophoresis (FFE) is a preparative technique used to separate a continuously flowing stream of charged analytes.1 A thin sample stream is introduced into a planar separation channel with buffer running in parallel. An electric field is applied perpendicularly across the separation channel, and charged analytes are deflected laterally based on their electrophoretic mobility. FFE has proven useful in separating a range of analytes, including cells,2,3 cellular components,4-8 and proteins.9-12 * To whom correspondence should be addressed. E-mail: bowser@ chem.umn.edu. Phone: (612)624-0873. Fax: (612)626-7541. † Department of Chemistry. ‡ Department of Biomedical Engineering. (1) Roman, M. C.; Brown, P. R. Anal. Chem. 1994, 66, 86A-94A. (2) Graham, J. M.; Wilson, R. B. J.; Patel, K. Methodol. Surv. Biochem. Anal. 1987, 17, 143-152. (3) Heidrich, H. G.; Hannig, K. Methods Enzymol. 1989, 171, 513-531. (4) Hoffstetter-Kuhn, S.; Kuhn, R.; Wagner, H. Electrophoresis 1990, 11, 304309. (5) Hoffstetter-Kuhn, S.; Wagner, H. Electrophoresis 1990, 11, 451-456. (6) Hoffstetter-Kuhn, S.; Wagner, H. Electrophoresis 1990, 11, 457-462. (7) Kessler, R.; Manz, H.-J. Electrophoresis 1990, 11, 979-980. 10.1021/ac060290n CCC: $33.50 Published on Web 07/04/2006
© 2006 American Chemical Society
Recently, several researchers have demonstrated FFE on microfluidic chips (µ-FFE). As with other electrophoretic techniques, it should be possible to make gains through miniaturization as Joule heating and convective flow are reduced. Raymond et al. were the first to demonstrate a µ-FFE device.13 The device was fabricated in silicon, which limited the maximum potential that could be applied. More recent designs have been demonstrated in PDMS14 and glass15 substrates. Unfortunately, all of the µ-FFE devices reported to date have been limited in the electric field that can be applied across the separation channel. Generation of electrolysis products has proven problematic even at modest electric fields. For example, in our own glass µ-FFE device, we were only able to generate an electric field of 283 V/cm in the separation channel for short periods of time before bubble formation at the electrodes disrupted the separation.15 It would be advantageous to be able to remove electrolysis products from the electrode channels before they negatively affect the separation. In conventional FFE systems, the electrodes are physically isolated from the separation channel using ion exchange, nylon, or cellulose membranes.1 The membranes allow ions and, thus, current to flow through them. Electrolysis products such as bubbles are confined to the electrode bed and are flushed away by the electrode buffer. Integrating similar membranes into a µ-FFE device is not practical at this time. Instead, membranes have been replaced by “membrane” channels, which run perpendicular to the electrode and separation channels.13-16 These membrane channels isolate buffer flowing through the separation channel from that flowing over the electrodes. Higher flow rates can therefore be applied in the electrode channels, which helps clear bubbles and other electrolysis products. A consequence of the physical isolation of the electrodes from the separation channel is a similar electrical isolation. The (8) Poggel, M.; Melin, T. Electrophoresis 2001, 22, 1008-1015. (9) Moritz, R. L.; Clippingdale, A. B.; Kapp, E. A.; Eddes, J. S.; Ji, H.; Gilbert, S.; Connolly, L. M.; Simpson, R. J. Proteomics 2005, 5, 3402-3413. (10) Wang, Y.; Hancock, W. S.; Weber, G.; Eckerskorn, C.; Palmer-Toy, D. J. Chromatogr., A 2004, 1053, 269-278. (11) Zischka, H.; Weber, G.; Weber, P. J. A.; Posch, A.; Braun, R. J.; Buehringer, D.; Schneider, U.; Nissum, M.; Meitinger, T.; Ueffing, M.; Eckerskorn, C. Proteomics 2003, 3, 906-916. (12) Zuo, X.; Lee, K.; Speicher, D. W. Proteome Anal. 2004, 93-118. (13) Raymond, D. E.; Manz, A.; Widmer, H. M. Anal. Chem. 1994, 66, 28582865. (14) Zhang, C.-X.; Manz, A. Anal. Chem. 2003, 75, 5759-5766. (15) Fonslow, B. R.; Bowser, M. T. Anal. Chem. 2005, 77, 5706-5710. (16) Raymond, D. E.; Manz, A.; Widmer, H. M. Anal. Chem. 1996, 68, 25152522.
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006 5369
electrical resistance of the “membrane” channels scales inversely with cross-sectional area and proportionally with length. The membrane channels must be narrow and long to effectively provide flow resistance, which results in a significant electrical resistance. For example, in our own previous µ-FFE design, we estimated that the resistance in the membrane channels was high enough that ∼50% of the potential applied between the two electrodes actually occurred in the membrane channels, not the separation channel.15 In other devices with longer membrane channels, only 4% of the applied potential occurred in the separation channel.14 This is problematic since it means that a significant amount of the applied potential is wasted in the membrane channels where no separation takes place. It may seem straightforward to increase the electric field in the separation channel by simply applying a higher potential between the electrodes. Only minimal gains can be found this way though since the increased potential increases current and subsequent generation of electrolysis products. These products can be washed away by increasing the flow rate over the electrodes but only if the lengths of the membrane channels are increased to maintain flow isolation from the separation channel. Increasing the length of the membrane channels increases their electrical resistance as well, resulting in more of the applied potential occurring in the membrane channels, not the separation channel. Consequently, there appears to be no optimum membrane channel geometry. It would be highly desirable to find an alternative method for decoupling the electrode and separation channel flows that would allow increased flow rates over the electrodes for electrolysis product removal without sacrificing the electric field in the separation channel. Lubrication theory describes flow in planar channels such as those used in µ-FFE, where the height is much less than the width, with the additional condition that17
FvH2/ηw , 1
(1)
where F is fluid density, v is linear velocity, H is channel height, η is viscosity, and w is the width of the channel. Similar to the more familiar Hagen-Poisuelle equation,18 which predicts that flow rate is proportional to the diameter to the fourth power,
q ) ∆Pπd4/128ηL
(2)
lubrication theory describes the flow rate in a planar microchannel:17,19
q ) ∆PH3w/12ηL
(3)
where q is volumetric flow rate, ∆P is pressure difference, d is capillary diameter, H is channel height, w is the channel width, η is viscosity, and L is the channel or capillary length. It is important to note that in a planar flow channel the flow rate is proportional to the height of the channel cubed. For example, the flow rate (17) Reynolds, O. Philos. Trans. R. Soc, London 1886, 177, 157. (18) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, 2nd ed.; John Wiley and Sons: New York, 2002. (19) Stay, M. S.; Barocas, V. H. J. Appl. Phys. 2004, 95, 6432-6443.
5370
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
would be expected to increase 64-fold if the depth of the channel is increased 4-fold. Similarly, the linear velocity (v) in the channel scales with the square of the channel height:
v ) ∆PH2/12ηL
(4)
The linear velocity would increase 16-fold for a 4-fold increase in channel depth. These equations demonstrate the power of channel depth as a variable in controlling flow rate. In the current paper, we describe a µ-FFE device that uses channel depth to isolate and control fluid flow in the separation and electrode channels. EXPERIMENTAL SECTION Reagents and Chemicals. Unless otherwise noted, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Deionized water (18.3 MΩ, Barnstead, Dubuque, IA) was used for all preparations unless otherwise noted. Separation buffer consisted of 25 mM HEPES, adjusted to pH 7.0 using 1 M NaOH (Mallinckrodt, Paris, KY) and filtered through a 0.2-µm membrane filter (Fisher Scientific, Fairlawn, NJ). Stock solutions of fluorescent standards were prepared in ethanol (Fisher Scientific, Fairlawn, NJ), and dilutions were made in separation buffer. Particle velocimetry buffer consisted of 6 × 10-6% by mass, 1-µm carboxylate-modified yellow/green fluorescent FluoSpheres (Molecular Probes, Eugene Oregon) in 25 mM Borate (pH 9.0). Piranha solutions (4:1 H2SO4/H2O2, Ashland Chemical, Dublin, OH) were used to clean glass wafers and etch deposited Ti. GE-6 (Acton Technologies, Inc., Pittston, PA) was used to etch or remove Au. Concentrated HF (Ashland Chemical) was used to etch the glass wafers. Silver conductive epoxy (MG Chemicals, Surrey, BC, Canada) was used to make electrical connections to the chip. Chip Fabrication. A 1.1-mm-thick Borofloat wafer (Precision Glass & Optics, Santa Ana, CA) was cleaned for 5 min in Piranha solution. The ∼500-nm amorphous silicon (a-Si) was deposited on both sides of a single wafer using plasma enhanced chemical vapor deposition (PECVD; Plasma-Therm, St. Petersburg, FL) and then annealed at 400 °C for 2 h. Standard photolithography was performed to define electrode beds. Following photolithography, the glass was unmasked using reactive ion etching (Surface Technology Systems 320PC, Newport, UK) for 3 min. Electrode channels were etched into the unmasked regions for 5 min, resulting in a depth of 60 µm as measured using a Fowler dial gauge (Painesville, OH). Photolithography and unmasking were repeated to define the remaining channels. The electrode channels and remaining channels were etched for 2 min to a final depth of 78 and 20 µm, respectively. The a-Si mask was removed by removing photoresist and then repeating reactive ion etching for 3 min. A Temescal electron beam evaporation system was used to deposit 100- and 150-nm layers of Ti and Au, respectively. Photolithography was repeated with a mask for patterning electrodes into the side channels and unwanted metal was removed. Access holes (all 1-mm diameter except for the sample inlet, which was 635-µm diameter) were drilled by the UMN Electrical Engineering/Computer Science (EE/CS) Machine Shop in an unprocessed wafer with an ultrasonic diamond drill oscillating at 25 kHz at a feed of 1/8
Figure 2. Image of fluorescent beads pumped through the µ-FFE channels with an exposure time of 5 ms. The dashed vertical line through the center of the image was added to illustrate the boundary between the 78-µm-deep electrode channel (left) and the 20-µm separation channel (right). Arrows identify the streaks formed as the beads traveled through the channel during the exposure time of the CCD. The linear velocity of a bead can be determined by dividing the length of the streak by the exposure time.
Figure 1. (A) Schematic of the pumping and detection system for the µ-FFE chip. The cross is used to ensure equal pressure is applied at all buffer inlets. (B) A top view of the µ-FFE mask with the following features: (1) separation buffer inlet, (2) sample inlet, (3) electrode buffer inlets, (4) Au electrodes in electrode channels, (5) separation channel, (6) electrode buffer outlets, and (7) separation buffer outlet. (C) A side view of the electrode and separation channels etched into the bottom glass wafer and bonded to the top glass wafer. Note that this drawing is not to scale.
in./min. Drilled wafers were cleaned with Piranha solution, and ∼90-nm a-Si was deposited using PECVD. A diced rectangular wafer was placed on the separation region to prevent deposition of a-Si in the separation channel. The drilled, a-Si deposited wafer was aligned with the etched, electrode deposited wafer on the wafer chuck of a SB-6 bonder (Karl Suss, Munich, Germany), and 900 V was applied for 2 h at 450 °C and 5 µbar. The voltage was applied to the a-Si deposited wafer with the other wafer grounded. If bonding was not complete, the wafer was rotated and bonding was repeated. Chips were diced into rectangles with a 1.5-cm border around access holes. Nanoports (Upchurch Scientific, Oak Harbor, WA) were attached to the access holes using manufacturer procedures. Electrodes were connected to wires using silver conductive epoxy. The chip was perfused with 0.1 M NaOH until the channels were clear (150 min) to remove unwanted a-Si. Instrumentation and Data Collection: A SMZ 1500 stereomicroscope (Nikon Corp., Tokyo, Japan) mounted with a 100-W X-Cite fiber-optic metal halide lamp (Nikon Corp.) and a Cascade 512B CCD camera (Photometrics, Tucson, AZ) were used for fluorescence imaging. Figure 1A gives a schematic of the microscope setup. The microscope was equipped with an Endow GFP band-pass emission filter cube (Nikon Corp.) containing two bandpass filters (450-490 and 500-550 nm) and a dichroic mirror (495nm cutoff). A 1.6× objective was used for collection with a 0.7× CCD camera lens. MetaVue software (Downington, PA) was used for image collection and processing. Analysis of line scans across
the analyte streams (analogous to electropherograms) was performed using Cutter 5.0.20 Constant Pressure Pumping. As shown in Figure 1A, a single syringe and syringe pump (Harvard Apparatus, Holliston, MA) were used to introduce buffer into the microchip. A mixing cross (Upchurch Scientific) was used to split the buffer between the three inlets. The tubes (0.75-mm i.d.) connecting the mixing cross to the nanoports were all the same length and fed the separation channel and two electrode channels. Particle Velocimetry. Buffer containing FluoSpheres was pumped into the three buffer inlets at 250 µL/min, 750 µL/min, 1 mL/min, and 3 mL/min. The microscope was focused on the interface between the separation and electrode channels to allow the velocity of particles in both regions to be compared simultaneously. At the magnification and exposure times used, fluorescent beads were imaged as streaks. The 11.25× zoom used allowed a fluorescent bead to travel 640 µm vertically through the square imaging region. Exposure times on the CCD camera were chosen to allow full streaks to be contained in the imaging region. Many images were acquired, but only those with a full streak within the image were analyzed. Twenty images were analyzed for each linear velocity. µ-FFE Separations. Separation buffer was pumped into the three buffer inlets at 1.484 mL/min (0.50 cm/s in separation channel). Fluorescent standards were pumped into the sample channel at 0.21 µL/min (0.50 cm/s in sample channel). Separation voltages ranging from 0 to 290 V were applied at the right electrode with the left electrode grounded. Analyte detection was performed using the 0.75× zoom to image the 1-cm-wide separation channel 2.5 cm downstream from the sample inlet. Ω Plot Generation. Separation buffer was pumped into the three inlets at 298 and 742 µL/min, and 1.484, 2.224, 2.968, 3.71, 4.46, 5.20, and 5.94 mL/min. The corresponding linear velocities in the separation channel are shown in the Figure 3 caption. (20) Shackman, J. G.; Watson, C. J.; Kennedy, R. T. J. Chromatogr., A 2004, 1040, 273-282.
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
5371
Voltages were applied across the electrodes using a PS310/1250V25W High Voltage Power Supply (Stanford Research Systems, Inc., Sunnyvale, CA) and currents were recorded manually from the display. Safety Considerations: Caution: Piranha solution self-heats to ∼70 °C and is extremely caustic. RESULTS AND DISCUSSION A schematic of the µ-FFE device is shown in Figure 1. Multiple etching steps were used to generate electrode channels ∼4-fold deeper than the separation channel. Actual depths were measured to be 20 and 78 µm giving a ratio of 3.9. Clearly the depths of these channels are much less than their widths (1 and 0.2 cm for the separation and electrode channels, respectively). Even under the most extreme conditions tested here eq 1 is less than 1, suggesting that it is appropriate to use lubrication theory to predict flow rates in this device (FvH2/ηw ) 0.49 if F ) 1 g/cm3, v ) 16 cm/s, H ) 0.0078 cm, η ) 0.01 dyn cm-1 s-1, w ) 0.2 cm). According to eq 3, increasing the depth of a channel 3.9-fold should result in a 59-fold increase in flow rate if a constant pressure is applied. Similarly, the linear velocity of the buffer would be expected to increase by a factor of 15.2. Care was taken to ensure that the pressure applied to all inlets was the same. All buffer entering the µ-FFE device was driven from a single syringe connected to the inlets with tubing of the same inner diameter and length to ensure equal pressure. The result is a much higher flow rate over the electrodes than in the separation channel. This is the ideal situation since the increased flow over the electrodes effectively removes electrolysis products. The lower flow rate in the separation channel increases analyte residence time in the electric field, increasing resolution. Particle velocimetry was used to confirm that the multipledepth design generated higher linear velocities in the electrode channels than in the separation channel. Buffer containing fluorescently loaded beads was pumped through the µ-FFE device at various flow rates. Images were collected of a 640 µm × 640 µm region centered on the boundary between the 78-µm-deep electrode channel and the 20-µm-deep separation channel. Figure 2 shows an example of an image acquired during a particle velocimetry experiment. The streaks observed in the image are the paths the particles followed through the channels during a single exposure of the CCD camera. The linear velocity of a particle was determined by dividing the length of a streak by the exposure time. Table 1 shows the average linear velocities observed in each region for flow rates ranging from 0.25 to 3.00 mL/min. Clearly, much higher velocities were observed in the deeper electrode channel than in the separation channel. The ratios of the observed linear velocities matched well with those predicted from lubrication theory, especially at higher flow rates. It is important to note the trajectory of the particles near the electrode channel/separation channel interface. All of the trajectories were nearly parallel with the channels, suggesting that there is little crossover between the high-velocity flow in the electrode channel and the lower velocity flow in the separation channel. It should be noted that lubrication theory cannot be used to predict flow at the boundary. Negligible boundary effects are observed though due to the nearly straight streamlines and very high aspect ratio of the channels. This is important since it allows a high5372 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
Table 1. Comparison of Linear Velocities Measured Using Particle Velocimetry in the Electrode and Separation Channels to Those Predicted Using Lubrication Theory (Eq 4)a
flow rate 250 µL/min 750 µL/min 1 mL/min 3 mL/min
a
separation channel electrode channel ratio separation channel electrode channel ratio separation channel electrode channel ratio separation channel electrode channel ratio
measured linear velocity (cm/s)
predicted linear velocity (cm/s)
0.10 ( 0.01 1.0 ( 0.3 11 ( 3 0.20 ( 0.04 3.4 ( 0.8 17 ( 5 0.28 ( 0.05 5(1 16 ( 6 1.1 ( 0.1 16 ( 3 15 ( 3
0.0843 1.28 15.2 0.253 3.84 15.2 0.337 5.13 15.2 1.01 15.4 15.2
Errors are the 95% confidence intervals of the means (n ) 20).
velocity flow to be applied to the electrode channel without disrupting the flow profile in the separation channel. The increased linear velocity in the electrode channels removes electrolysis products more efficiently, allowing higher separation voltages to be applied. In our own earlier µFFE device, which employed membrane channels to isolate buffer flow in the electrode channels, current did not increase linearly when the separation voltage was increased above 300 V.15 A negative deviation from linearity was observed, which we attributed to the generation of electrolysis products that in turn increased resistance in the electrode and membrane channels. At higher potentials, the separation became unstable due to bubble formation and the voltage could only be applied for short periods of time. It should be noted that due to the high resistance of the membrane channels only 50% of the applied potential difference actually occurred in the separation channel, giving rise to a maximum electric field in the separation channel of 150 V/cm. Figure 3 shows an Ohm plot collected using the multiple-depth µ-FFE device. Generation of electrolysis products no longer limits the maximum separation voltage that can be applied. Continuous operation at potentials as high as 1100 V was possible with no observable bubble formation. The positive deviation observed at higher potentials follows the trend expected for Joule heating. Joule heating became significant above 645 V, over double the maximum voltage possible with our earlier design.15 The increase in electric field in the separation channel is actually higher. Increasing the depth of the electrode channel means that the resistance is actually lower in this region than in the separation channel. Therefore, most of the applied potential difference occurs in the separation channel, not the electrode channels. In the current design, we estimate that 91% of the applied potential difference occurs in the separation channel, giving rise to an electric field of 586 V/cm, a 4-fold increase over our earlier design. Higher electric fields are accessible if some Joule heating is acceptable. Interestingly, there does not appear to be a linear velocity dependence on the Joule heating. As shown in Figure 3, the relationship between current and voltage is independent of
Figure 3. An Ω plot of applied voltage versus current in the µ-FFE device. Linear velocity of the buffer in the separation channel: 0.10 (×), 0.25 (+), 0.50 (*), 0.75 (b), 1.0 (2), 1.25 (0), 1.5 (4), 1.75 (3), and 2 cm/sec (]). A positive deviation is observed above 645 V (1.25 W, 586 V/cm in the separation channel), indicating Joule heating. The data fits the equation I ) (2.5 × 10-6)V2 + (1.9 × 10-3)V + 8.5 × 10-2, where I is current in mA and V is the potential applied to the electrodes in volts with R2 ) 0.9987.
Figure 4. µ-FFE separation of fluorescein, rhodamine 110 impurity, rhodamine 110, and rhodamine 123 (left to right). (A) was acquired using an earlier µ-FFE design15 (linear velocity 0.54 cm/s, electric field in the separation channel 200 V/cm). (B) was recorded on the multiple depth µ-FFE device described here (linear velocity 0.50 cm/ s, electric field in the separation channel 160 V/cm).
the buffer flow rate through the µFFE device, suggesting that fluid flow does not actively cool the chip even at relatively high flow rates. A separation of three fluorescent standards was performed to make a direct comparison with separations performed on our earlier µ-FFE device.15 On the earlier device we achieved a maximum resolution of 2.1 between rhodamine 110 and rhodamine 123 at 200 V/cm with a linear velocity of 0.54 cm/s in the separation channel. Figure 4 compares the separation achieved with our earlier device to the current multiple-level µ-FFE device under similar conditions (linear velocity 0.50 cm/s, electric field 160 V/cm). The resolution improved significantly even though a lower separation voltage was used. The resolution between rhodamine 110 and rhodamine 123 in the newly designed µ-FFE device was 2.7, a 1.3-fold increase over the earlier design. Note that this improvement was observed under similar separation conditions. If anything, the conditions chosen should favor the earlier design since the electric field used on the multiple-depth
Figure 5. Effect of increasing electric field on the µ-FFE separation of fluorescein, rhodamine 110 impurity, rhodamine 110, and rhodamine 123 (left to right). The electric field in the separation channel is listed for each separation. The buffer linear velocity was 0.50 cm/s for all separations.
device was lower. We hypothesize that even low separation potentials electrolysis product formation in the earlier design increased resistance in the electrode and membrane channels, reducing the electric field in the separation channel from that predicted by considering channel geometry alone. This would explain why improved resolution was observed in the new µ-FFE design even under similar separation conditions. It should be noted that the new design is no longer limited by generation of electrolysis products and the resolution can therefore be further improved by increasing the separation voltage. Figure 5 demonstrates the improved resolutions that can be achieved by increasing the electric field in the separation channel. As shown in Figure 4, increased band broadening was observed in the new µ-FFE device. We are currently performing a detailed study of the sources of band broadening in µ-FFE. Electric field, buffer linear velocity, and electroosmotic flow are all important contributors to band broadening in µ-FFE. Optimization of these parameters will be discussed in detail in a later paper. Regardless of the cause, the increased separation of the analyte streams overcame the increased broadening, resulting in a significant improvement in resolution. CONCLUSIONS We have demonstrated the application of a µ-FFE device that uses channel depth as a mechanism for controlling and isolating flow in the electrode and separation channels. The flow through the deeper electrode channels was significantly faster than that in the shallower separation channel, as predicted by lubrication theory. The increased buffer flow through the electrode channels effectively removed electrolysis products without disrupting the flow pattern in the separation channel. Electrolysis product formation is no longer the factor limiting the maximum separation voltage that can be applied. Removing the membrane channels Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
5373
allows 91% of the applied potential drop to occur in the separation channel where it actually impacts the separation. The multipledepth µ-FFE device described here can be operated continuously at electric fields (in the separation channel) as high as 589 V/cm, a 4-fold improvement over previous designs. Joule heating, not electrolysis product formation, has become the limiting factor for higher separation potentials. Fortunately, this problem can be avoided since a combination of low flow rates and low electric fields can be used to achieve similar resolution. Further work is under way that focuses on exploiting the linear velocity and
5374
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
electric field dependence of resolution to find conditions where resolution is maximized under non-Joule heating conditions. ACKNOWLEDGMENT Funding for this research was provided by the National Institutes of Health (GM 063533 and NS 043304). Received for review February 15, 2006. Accepted May 26, 2006. AC060290N
