Anal. Chem. 2005, 77, 6374-6380
Characterization of a Membrane-Based, Electrochemically Driven Pumping System Using Aqueous Electrolyte Solutions Mya A. Norman,† Christine E. Evans,† Anthony R. Fuoco,† Richard D. Noble,‡ and Carl A. Koval*,†
Department of Chemistry and Biochemistry (215 UCB) and Department of Chemical and Biological Engineering (424 UCB), University of Colorado at Boulder, Boulder, Colorado 80309
Electrokinetic flow provides a mechanism for a variety of fluid pumping schemes. The design and characterization of an electrochemically driven pump that utilizes porous carbon electrodes, iodide/triiodide redox electrolytes, and Nafion membranes is described. Fluid pumping by the cell is reversible and controlled by the cell current. Chronopotentiometry experiments indicate that the total available fluid that can be pumped in a single electrolysis without gas evolution is determined solely by the initial concentration of electrolyte and the applied current. The magnitude of the fluid flow at a given current is determined by the nature of the cation in the electrolyte and by the water absorption properties of the Nafion membrane. For 1 M aqueous electrolytes, pumping rates ranging from 1 to 14 µL/min were obtained for current densities of 10-30 mA/cm2 of membrane area. Molar volume changes for the I3-/I- redox couple and for the alkali cation migration contribute little to the observed volumetric flow rates; the magnitude of the flow is dominated by the migration-induced flow of water. When the compartments of electrochemical cells are separated by a porous medium having fixed charges on its walls, the migration current that accompanies electrolysis often induces the flow of solvent. For porous media in which the pore structures are well defined, that is, cylindrical tubes or pores, and for which the Debye length associated with ions attracted to the walls is small compared to the pore radius, this migration-induced solvent flow is referred to as electroosmotic (EO) flow.1 In EO flow, movement of solvent in the center of the pores is induced by migration of charge, which predominantly occurs at the pore walls. Even though exact structures are frequently unknown for ionexchange membrane separators, the flow of ions and solvent through these materials is also often referred to as EO flow. The flow of solvent that accompanies electrochemically driven solvent flow in commercially important ion-exchange membranes such as Nafion2 is critical to the performance of these materials in devices such as fuel cells, electro-dialysis cells and water electrolysis cells. Such electrochemically driven flow (EDF) can also act as a mechanism for the nonmechanical pumping of fluids. †
Department of Chemistry and Biochemistry. Department of Chemical and Biological Engineering. (1) Adamson, A. W.; Gast, A. P. Physical Chemistry of Surfaces, 6th ed.; Wiley: New York, 1997. ‡
6374 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
EO flow is essential for separation methods such as capillary electrophoresis and is often used for fluid manipulation in microchannels for lab-on-a-chip applications. The characteristics of EO flow-based microfluidic pumps have been reviewed recently by Laser and Santiago.3 In these devices, microporous media, for example, glass frits or silica-packed capillaries, are used to generate EO flow. Because these media are relatively resistive, potentials in the kV range are needed to obtain significant flow rates, especially if the EO flow is to be maintained against backpressure greater than a few psi. On the basis of the concepts initially described by Tan et al.,4 Hayakawa et al.5 developed an EO pumping system for gel electrophoresis. This type of EO pump utilized dialysis membranes with various pore sizes. For most of these membranes, potentials of 100 V and currents of 30 mA generated flow rates between 10 and 100 µL/min. Primarily because of their high conductivity and relatively low hydraulic permeability, we have been exploring the use of permselective ion-exchange membranes (IEMs) for use in electrolysis-based fluid pumping devices.6 We prefer to describe this type of pumping as electrochemically driven (EDF) flow because it seems unlikely that the conventional description of electroosmosis is consistent with the morphology of solvent-swollen, ionexchange membranes for which effective pore diameters range from 1 to 5 nm. The goal of this research is to develop a scaleable design for a membrane-based electrochemical fluid pump that can produce sustained and controlled flow rates ranging from nanoliters to milliliters per minute. Recently, we reported large EDF rates across Nafion membranes in cells containing tetraalkylammonium iodide/dimethylformamide electrolytes.6 High concentrations of the I3-/I- redox couple were used to allow operation at significant currents without gas generation. The overall volume increase in the cathode compartment of these cells approached 700 mL/mole of electrons, which translated into pumping rates of 50 µL/min at a current density of ∼20 mA/cm2 of membrane area. Due to the relatively low conductivity of nonaqueous (2) Nafion is a registered trademark of E.I. du Pont de Nemours. (3) Laser, D. J.; Santiago, J. G. J. Micromech. Microeng. 2004, 14, R35-R64. (4) Tan, H.-v.; Kitzis, A.; Berthollet, T.; Hamard, G.; Beldjord, C.; Benarous, R. Nucliec Acids Res. 1988, 16, 1921-1930. (5) Hayakawa, M.; Hosogi, Y.; Takiguchi, H.; Saito, S.; Shiroza, T.; Shibata, Y.; Hiratsuka, K.; Kiyama-Kishikawa, M.; Abiko, Y. Anal. Biochem. 2001, 288, 168-175. (6) Norman, M.; Noble, R. D.; Koval, C. A. J. Electrochem. Soc. 2004, 151, E364-E369. 10.1021/ac0508705 CCC: $30.25
© 2005 American Chemical Society Published on Web 09/03/2005
Figure 1. Cell diagram. A and D, solution loading port; B, glassy carbon; C, acrylic cell body; E, syringe barrel; F, glass frit; G, porous carbon electrode disks; H, membrane holder; I, membrane.
electrolytes, Joule heating became a significant complication at higher operating currents in these experiments. This paper describes low-pressure EDF/IEM-based fluid pumping using aqueous iodide electrolytes and Nafion membranes with different equivalent weights and water absorption properties. The measurements use a cylindrical cell that contains porous carbon electrodes to allow constant current operation for easily predicted time periods. This cell allows for accurate fluid flow rate determination using either volume or mass measurements. Flow rates were found to depend on the nature of the alkali cation in the electrolyte and on the equivalent weight and water sorption properties of the membrane. A number of researchers have examined the transference of ions and solvent across Nafion membranes.7-14 Analysis of flow rates obtained in this study yield values of solvent molecules transferred per ion that are slightly less than previously reported values obtained using electrokinetic methods. Flow rates obtained using an electrolyte containing a large yet highly water-soluble cation appear to confirm the idea originally described by Okada et al.14,15 that a large cation effectively pushes solvent molecules through the channels in Nafion membranes. EXPERIMENTAL SECTION Electrochemical Pumping Experiments. An acrylic electrochemical cell was constructed from two identical cylindrical tubes that contain a membrane holder between the two compartments (Figure 1). The membrane holder is designed with 32 0.32cm-diameter holes that give a total exposed membrane surface area of 2.5 cm2. The membrane area that is exposed to solution provides ∼5 × 1019 ion exchange sites (for Nafion 117), as calculated from the equivalent weight (1100 g/mol ion-exchange (7) Okada, T.; Moller-Holst, S.; Gorseth, O.; Kjelstrup, S. J. Electroanal. Chem. 1998, 442, 137-145. (8) Okada, T.; Satou, H.; Okuno, M.; Yuasa, M. J. Phys. Chem. B 2002, 106, 1267-1273. (9) Okada, T.; Xie, G.; Gorseth, O.; Kjelstrup, S.; Nakamura, N.; Arimura, T. Electrochim. Acta 1998, 43, 3741-3747. (10) Pintauro, P. N.; Tandon, R.; Chao, L.; Xu, W.; Evilia, R. J. Phys. Chem. 1995, 99, 12915-12924. (11) Lehmani, A.; Turq, P.; Perie, M.; Perie, J.; Simonin, J.-P. J. Electroanal. Chem. 1997, 428, 81-89. (12) Vishnyakov, A.; Neimark, A. V. J. Phys. Chem. B 2001, 105, 9586-9594. (13) Huang, K.-L.; Holsen, T. M.; Selman, J. R. Ind. Eng. Chem. Res. 2003, 42, 3620-3625. (14) Xie, G.; Okada, T. Electrochim. Acta 1996, 41, 1569-1571. (15) Xie, G.; Okada, T. J. Chem. Soc., Faraday Trans. 1996, 92, 663-669.
sites), the dry membrane thickness (175 µm), and the exposed membrane area. All fittings and valves in this cell design are nonmetallic and composed of either PEEK or nylon. The electrolyte solutions used were one of four iodide salts (LiI, NaI, KI, TrisHI) at a concentration of 1.0 M. The solutions also contain iodine (0.1 M), which reacts with excess iodide ion to form triiodide ion. Potassium iodide, lithium iodide, sodium iodide, and iodine were obtained from Aldrich (99% or higher purity) and used without further purification. TrisH iodide was synthesized from tris(hydroxymethyl)aminomethane and hydroiodic acid and recrystalized from a 55:45 methanol/ethanol mix. Ultrahigh-purity water (18 MΩ) was used throughout this study (Millipore, MilliQ Plus). Each compartment of the cell contained ∼20 mL of solution. These electrolyte solutions undergo electrolysis, which interconverts the iodide and triiodide in the compartments of the cell and drives cations across the Nafion membrane to maintain electroneutrality. Electrolysis occurred at porous reticulated vitreous carbon electrodes (Electrolytica, Inc., 30 pores/in.; 97% void space) in each compartment; these electrodes filled ∼90% of the interior volume of the cell. Glassy carbon rods inserted in the side of the cell provided the electrode connection through pressure contact. A Princeton Applied Research model 173 galvanostat was used to control current. To monitor volume changes in the compartments, a syringe barrel was connected to the top of each compartment of the cell through a Luer-loc fitting. The barrels used were 50, 100, or 500 µL, depending on the electrolyte being investigated and the total charge passed during electrolysis. Air bubbles were removed from the cell and membrane holder before starting an electrolysis experiment by pulling a mild vacuum through the side port with bubbles being flushed out the top port. Liquid level equilibration in the cell was considered complete when the volume in either syringe did not change for 15 min. Most electrolysis experiments were conducted for a total charge of 100 C in each direction at constant currents ranging between 25 and 75 mA. This corresponded to current densities based on exposed membrane area of 10-30 mA/cm2. The liquid volume level in each syringe and the corresponding amount of charge passed was recorded periodically. The measured flow rates were reproducible to 10% relative standard deviation or less. Electrodes were cleaned with acetone and rinsed with water between runs, and a new or reconditioned piece of Nafion (E.I. du Pont de Nemours & Co.) was used each time the cell was assembled. Nafion pieces were reconditioned by boiling in 1 M nitric acid for ∼45 min and then soaking in a 1 M hydroxide solution consisting of the cation of interest for 18-24 h. All new Nafion membranes were boiled in water for ∼30 min then soaked overnight in a 1 M aqueous hydroxide solution of the cation of interest prior to use. Membrane Swelling and Water Uptake. The initial dimensions and mass for vacuum-desiccated Nafion membranes in the hydrogen form were recorded periodically until constant values were obtained. The membranes were then boiled in water for 1 h and subsequently immersed in aqueous solutions of the appropriate hydroxide (1 M) for at least 24 h to ensure exchange of the desired cation for protons. After removal, the membranes were blotted dry, and their mass and dimensions were remeasured. Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
6375
The exchanged membranes were again dried by vacuum desiccation until constant mass was achieved. The masses of the wet exchanged and dried membranes were used to calculate the number of water molecules associated with each cation in the exchanged membranes. On the basis of the various errors associated with these measurements, the determinations of water uptake for different samples of the same membrane were reproducible to within 5% relative standard deviation. A series of Nafion membranes were obtained from the DuPont Company. These membranes were characterized previously by Manley et al.16 For this paper, N002 (EW 1300, dry thickness 125 µm) and N004 (EW 1500, dry thickness 125 µm) were examined for solvent uptake as described above and compared with Nafion 117 membranes in electrolysis experiments. Determination of Transference Number. Although Nafion is generally assumed to have a cation transference number of one; this assumption was verified for the type of electrolyte systems used in these studies. An electrolysis experiment similar to those described in the Results and Discussion section was performed with 1.0 M NaI/0.1 M I2 in the cathode compartment and 1.0 M NaCl in the anode compartment. After 200 C of charge was passed through the cell, the side containing sodium chloride was sampled and examined by square wave voltammetry and compared to standard samples containing the iodide ion in 1.0 M NaCl. No iodide ion was observed at levels as low as the limit of detection, 1 × 10-6 M, indicating that the transference number for I- was less than 10-5. RESULTS AND DISCUSSION Design and Characterization of the EDF/IEM Cell. Figure 1 illustrates a detailed drawing of the fixed-volume, nonstirred EDF/IEM cell. The two identical acrylic compartments of the cell were fitted with filling ports, a graphite rod used as an electrical connection, and a Luer-lock port for volume change determinations. Each compartment of the cell contained porous glassy carbon disks which were held in place with a porous glass frit. The membrane holder and IEM separated the compartments of the cell. To prevent deformation of the membranes during electrolysis, membrane holders incorporated an array of 0.32-cm holes. Initial characterization of the cell’s ability to serve as a fluid pump for predictable time periods without gas evolution was evaluated using constant current electrolyses (CCE) in a threeelectrode configuration.17 The potential of the porous GC electrode in the cathode compartment was recorded vs time versus a platinum pseudoreference electrode. The pseudoreference was inserted into the compartment through a filling port and positioned as far as possible from the porous electrode so that its potential remained close to the initial solution potential. Since the electrolytes used in these studies were composed of 1.0 M iodide ion (16) Manley, D. S.; Williamson, D. L.; Noble, R. D.; Koval, C. A. Chem. Mater. 1996, 8, 2595-2600. (17) For the electrolytes and applied currents used, the overall resistance of the cell was 35 ( 5 Ω prior to the transition time. Therefore, the potentials required to maintain currents of 25-100 mA were in the range of 1-4 V, and the power required for pumping was
