Microchip-based ethanol/oxygen biofuel cell

Abstract

One of the limitations of lab-on-a-chip technology has been the lack of integrated power supplies for powering various devices on the chip. This research focused on design of a stackable, microchip-based biofuel cell. The biofuel cell is powered by the addition of ethanol through a flow channel to a bioanode. The bioanode contains a micromolded carbon ink anode that has been modified with two layers. The first layer is poly(methylene green), which is an electrocatalyst for NADH oxidation; the second layer is a membrane that contains an immobilized enzyme, alcohol dehydrogenase. Each layer was characterized electrochemically. It was found that the poly(methylene green) layer is kinetically-limited, but when the complete bioanode is formed, the bioanode is diffusion-limited due to slow mass transport of NADH within the modified Nafion membrane. When used relative to an external platinum cathode, the biofuel cell showed maximum open circuit potentials of 0.34 V and maximum current densities of 53.0 ± 9.1 µA cm⁻². This research demonstrates the feasibility of a microfabricated biofuel cell device.

---

Introduction

As proliferation of ubiquitous and increasingly complex electronic devices continues, forms of power that are alternative to existing battery technology must be developed. Fuel cells are an attractive option due to energy densities that are significantly greater than batteries, making them potentially capable of powering devices for extended periods. Conventional fuel cells rely on expensive non-renewable metal catalysts and are limited to hydrogen and methanol fuels and higher temperature applications. These conventional fuel cells have suffered from poor efficiencies due to passivation of the catalyst, crossover of the fuel through the polymer electrolyte membrane (PEM), and difficulties in miniaturizing and stacking the technology.¹

Biofuel cells are an attractive alternative to conventional fuel cells, because they eliminate the dependence on precious metal catalysts by replacing them with biological catalysts. Biofuel cells reported in the literature since the 1960's fall within two distinct categories: biofuel cells that utilize the chemical pathways of living cells (microbial fuel cells) and those that employ isolated enzymes.² Microbial fuel cells can achieve high efficiency in terms of conversion of chemical energy into electrical energy; however, problems associated with this approach include low volumetric catalytic activity of the whole organism and low power densities due to slow mass transport of the fuel across the cell wall.

Isolated enzymes are attractive catalysts for fuel cells due to their high catalytic activity and selectivity. The theoretical electrode current for an enzymatic catalyst with an activity of 10³ U mg⁻¹ is 1.6 A, a catalytic rate greater than platinum.² Such currents have yet to be observed as a result of losses of catalytic activity during enzyme immobilization at the electrode surface, transport limitations, and energy losses of the overall system. Additionally, enzymatic biofuel cells can operate on a wide variety of available fuels such as alcohols, sugars, fatty acids, or even waste materials. These advantages present opportunities for biofuel cells ranging from small power drain applications (e.g. sensors) to more demanding devices (e.g. cellular phones).

Biofuel cells have traditionally suffered from low power densities and short lifetimes due to the fragility of the enzyme catalyst. Previous research in the Minteer group has demonstrated a novel approach to immobilizing enzymes at the electrode surface using a quaternary ammonium salt-treated Nafion membrane.³ This approach has shown increases in biofuel cell power and lifetime compared to recent reports in literature. Ethanol fuel cells with immobilized enzymes have yielded higher power densities than current state-of-the-art alcohol biofuel cells. Open circuit potentials ranging from 0.61 to 0.82 V and power densities of 1.00 mW cm⁻² to 2.04 mW cm⁻² have been produced, corresponding to 1.82 mW h cc⁻¹.⁴ This represents a 16 to 32-fold increase in power density compared to the previous state-of-the-art biofuel cells reported in literature. Enzyme lifetimes in excess of 45 days have also been achieved as a result of this immobilization technique.³

Although previous research with alcohol dehydrogenase-based biofuel cells has led to the largest power densities reported for biofuel cells, the anode and cathode were housed in large diffusion chambers (50–75 mL total volume) and separated by large (1-in id) Nafion membranes.⁴ The need for small, portable power supplies for cell phones and PDAs as well as powering processes occurring on lab-on-a chip devices dictates that biofuel cells be miniaturized. Furthermore, to increase the power output from a fuel cell, it is typical to stack multiple cells in series.⁵ While this is feasible with the diffusion chambers that are typically used, in terms of size, stacking more than three cells together is unrealistic for powering electronic devices. An ideal situation is to make a biofuel cell that is small and amenable to stacking. The ability to reproducibly fabricate microchannels and microelectrodes by traditional lithography^6–9 can possibly afford small devices that fit these desired criteria.

Microchip-based devices have mainly been used for analytical applications.^7–9 As described below, their application towards fuel cells have been limited but offer many advantages. Sample streams can be moved throughout the chip by either electrophoretic or hydrodynamic pumping.^6,10,11 It is possible to fabricate microelectrodes with many of the same photolithographic procedures that are used to construct the microchannels.^12,13 In addition, the electrode can be fabricated directly on the chip, leading to a fully integrated system. The sensitivity of electrochemistry is is increased, due to the increased flux towards the microelectrode surface¹⁴ and the reduced background current of microelectrodes¹⁵ leading to an increased coulometric efficiency. While the fluidic portion of these devices can be made in a variety of substrates, devices fabricated by soft lithography in poly(dimethylsiloxane) (PDMS) hold several unique advantages.^16,17 Rapid prototyping procedures can be utilized to make the initial master, greatly reducing the costs associated with making new designs. This same master can be used to make many devices and the resulting PDMS slab can be reversibly or irreversibly sealed to itself or other substrates such as glass.^16,17 Finally, PDMS is permeable to gases, which is essential for biofuel cells that contain enzymes that require oxygen as a cofactor.

As mentioned above, the application of microfabrication and microfluidics to fuel cell research has been limited,^18–20 with only one research group working with biofuel cells.^21,22 Of these works, several deserve further description. Work by Shah et al. involved development of miniature hydrogen-air proton exchange membrane (PEM) fuel cells on PDMS substrates using traditional photolithography. The electrodes were fabricated by physical vapor deposition, which formed smooth Pt microelectrodes with a small active catalyst area that led to current densities that were lower than desired.¹⁹ Work from Whitesides' lab utilized a membraneless approach to develop a vanadium redox fuel cell.²⁰ This work utilized the low Reynolds numbers that occur in microchannels to eliminate convective mixing of separate analyte streams. Here, two separate streams flow parallel to each other and vanadium redox couples are used to generate 35 mA cm⁻² at 1.1 V. This membraneless approach simplifies the chip fabrication and eliminates problems due to fouling or damage of the electrode; however, since analyte streams in microchannels do eventually fully mix by diffusion, the active electrode area was limited to 17 mm in length. Finally, the only descriptions of microfabricated biofuel cells came from Heller's group.^21,22 Their biofuel cells employed two 7 µm diameter, 2 cm long carbon fibers that were manually placed into a polycarbonate support. An electrocatalytic film that includes glucose oxidase was placed onto the anode to catalyze the oxidation of glucose while the film on the cathode consisted of laccase to catalyze the reduction of O₂. Due to its handmade nature, this was not truly a microfabricated fuel cell but the miniaturized format exhibited a power density as high as 244 µW cm⁻² at 0.36 V. This increase in power density was derived from the cylindrical mass transport at the carbon fibers and the use of redox polymers that are tailored for electron transport between the fibers and enzymes.²¹

In this paper, we describe the incorporation of a biofuel cell into a PDMS-based microchip device. Soft lithography is used to pattern microchannels through which fuel for an alcohol dehydrogenase-based biofuel cell is hydrodynamically introduced. A recently described method, micromolding of carbon inks, is utilized to pattern electrodes that run the entire length of the microchannel. Methylene green, an electrocatalyst necessary for the oxidation of NADH, was successfully immobilized on the carbon ink electrodes. In addition, a procedure to immobilize a Nafion/enzyme mixture on the carbon electrodes was developed. Electrochemical characterization of the device was performed as a function of the fuel flow rate through the microchannels. Finally, the fully assembled device was evaluated for its use as a functional biofuel cell by generation of a power density curve relative to an external platinum cathode. The work described here lays the foundation for creation of complete, miniature biofuel cells that are amenable to serial stacking.

Experimental

Materials and reagents

The following chemicals and materials were used as received: Nano SU-8 developer, SU-8 50 and SU-8 10 photoresist (Microchem Corp, Newton, MA); 30% hydrogen peroxide, Nafion (5% solution), nicotinamide adenine dinucleotide, alcohol dehydrogenase (ADH, E.C.1.1.1.1, initial activity of 300–500 U mg⁻¹), methylene green (Sigma, St. Louis, MO); tetrabutylammonium bromide (Eastman); colloidal silver (Ted Pella Inc., Redding, CA); JB weld epoxy (JB Weld, Sulfur Springs, TX); 100 mm silicon wafers (Silicon, Inc., Boise, ID); Sylgard 184 (Ellsworth Adhesives, Germantown, WI); isopropanol, sulfuric acid, sodium sulfate; 20 gauge copper wire, sodium borate, sodium nitrate (Fisher); high pressure fitting (F-120), leur adapter (P-659), 1/16″ od (250 µm id) teflon tubing (Upchurch Scientific, Oak Harbor, WA); Ercon E-978 (I) carbon ink, Ercon G-451 (I) graphite ink, N-160 solvent thinner (Ercon, Wareham, MA); Acheson Electrodag PF-407C, Acheson Electrodag 440B(49AB90) (Acheson, Brookfield, OH); a 1 mm diamond drill bit (Jules Borel and Co, Kansas City, MO), and soda lime glass (A-Affordable Glass, St. Louis, MO).

Soft lithography

PDMS channel structures were produced based on previously published methods.^17,23 Masters for the production of PDMS-based flow channels were made by coating a 4-in silicon wafer with a SU-8 50 negative photoresist operating at 1750 rpm for 30 s. Micromolding channels were made in a similar manner with SU-8 10 negative photoresist and a spin program of 1800 rpm for 30 s. The photoresist was prebaked at 90 °C for 5 min prior to UV exposure with a near-UV flood source (Autoflood 1000, Optical Associates, Milpitas, CA) through a negative film containing the micromolding channel or flow channel design structures (Jostens, Topeka, KS). Following this exposure, the wafer was postbaked at 90 °C for 5 min and developed in Nano SU-8 developer. The thickness of the photoresist was measured with a profilometer (Alpha Step-200, Tencor Instruments, Mountain View, CA), which corresponded to the channel depth of the PDMS structures that followed. A 10∶1 mixture of Sylgard 184 elastomer and curing agent were then poured onto the silicon wafer and cured at 75 °C for approximately 2 h. The resulting PDMS flow channel was either 130 or 200 µm wide, 100 µm deep and 3.0 cm long. The electrode micromolding channels were 45 µm wide, 42 µm deep, and 2.5 cm long.

Electrode plate fabrication

The micromolded electrodes were patterned on soda-lime glass plates that were purchased from a local glass shop. The plates were 7 cm wide, 10 cm long and 2 mm thick. The glass plates were initially cleaned for 15 min in piranha solution (70% concentrated H₂SO₄/30% H₂O₂). Thin layer palladium electrodes were used to make a connection to the micromolded carbon electrodes (see Fig. 1). To produce these electrodes, the glass was first cleaned with piranha solution (7∶3 H₂SO₄∶H₂O₂) for approximately 15 min to remove organic impurities. The glass was rinsed thoroughly with NANOpure H₂O and dried with N₂. The glass was placed into a deposition system (Thin Film Deposition System, Kurt J. Lesker Co., Clairton, PA) for subsequent deposits of titanium (adhesion layer) and palladium (electrode layer). The thicknesses of the metals were monitored using a quartz crystal deposition monitor (Inficon XTM/2 Leybold Inficon, Syracuse, NY). Titanium was deposited from a titanium target at a rate of approximately 2.3 Å s⁻¹ to a depth of 200 Å. Palladium was deposited from a palladium target at a rate of approximately 2.0 Å s⁻¹ to a depth of 2000 Å. Positive resist (AZ 1518) was dynamically dispensed onto the metal covered plate at a rate of 100 rpm for 20 s. The spin coater was then spun at 2000 rpm for 23 s. The photoresist was prebaked at 100 °C for 1 min prior to UV exposure through a positive transparency (Jostens), which contained designs for palladium connector (Fig. 1). The electrode plates were developed in AZ 300 MIF developer. A post-exposure bake at 100 °C for 1 min followed the developing step. Aqua regia (8∶7∶1 H₂O∶HCl∶HNO₃) was used to etch unprotected palladium while titanium etchant (Transene) was used to remove titanium. Copper wire was epoxied (J.B. Weld) onto the glass plates and silver colloidal paste (Ted Pella) was used to make contact to the palladium electrode.

Fig. 1 Procedure for micromolding carbon ink electrodes using PDMS channels. (A) PDMS microchannel that is of the desired electrode dimensions is sealed to a glass plate that contains a fabricated palladium connector. This plate also contains a fluid access hole. (B) The microchannels are first primed with thinner and then filled with a 0.2% (v/w) solution of the carbon ink. (C) After heating for 1 h at 75 °C the PDMS is peeled up. Ink is removed from the reservoirs and the ink is activated by heating at 120 °C for 1 h. (D) A PDMS flow channel is reversibly sealed over the inlet and carbon electrode. The outlet reservoir is made by punching a hole in the PDMS.

A flow access hole was drilled through each glass plate, while immersed under water, with a 1 mm diamond drill bit and a Dremel rotary tool (Dremel, Racine, WI). To allow for connection of teflon tubing, the syringe connector portion of a leur adapter was removed with the Dremel rotary tool and accompanying cutting disc. After polishing with a sanding disc, the leur adapter was affixed to the glass plate with J.B. Weld. The epoxy was cured in an oven (75 °C) for 2 h before first use.

To fabricate carbon ink microelectrodes, the PDMS micromolding channel was reversibly sealed to the glass plate so that it was in between the fluid access hole and palladium connector but still overlaying the palladium connector (Fig. 1). The PDMS micromolding channels were first primed with solvent thinner (N-160). The thinner was removed by applying a vacuum to one of the reservoirs. As soon as the thinner had been removed, a mixture of commercially available carbon ink and solvent thinner was added to the channels and pulled through the channel by applying vacuum (via water aspirator) to the opposite end. The ink/thinner mixture was made so that the volume of added thinner was 0.2% (v/w) of the initial ink weight. After filling channels with carbon ink, the reservoir where vacuum had been applied was filled with the ink/thinner solution and the entire chip placed in an oven at 75 °C for one hour. After this period of time, the PDMS could be removed from the glass, leaving the carbon microelectrode attached to the glass surface. A final curing/conditioning step was achieved by placing the chip in a separate oven at 120 °C for one hour. The steps involved with micromolding of carbon inks are shown in Fig. 1. The height of the carbon microelectrode was measured with a profilometer and the width was measured via microscopy. Micrographs of a carbon ink microelectrode are represented in Fig. 2.

Fig. 2 (A) Carbon microelectrodes printed onto glass by micromolding technique. Dimensions: 55 µm wide, 85 µm high, and 2.5 cm long. (B) Carbon microelectrode sealed in a PDMS microchannel. Dimensions of flow channel: 200 µm wide, 100 µm in depth, and 3.0 cm long.

Assembly of final device

When hydrodynamic flow was desired through the device, a PDMS flow channel was reversibly sealed so that it overlaid the access hole and palladium connector while the carbon microelectrode was encased in the flow channel. Alignment of the flow channel over the microelectrode was aided by a stereomicroscope (SZ60, Olympus America, Melville, NY). Flow was directed to the chip through 250 µm id Teflon tubing (Upchurch Scientific, Oak Harbor, WA) via a syringe pump (Harvard Apparatus, Brookfield, OH).

Modified nafion preparation

Nafion^® membranes incorporated with tetrabutylammonium bromide were formed in a two-step process. The first step was to co-cast the tetrabutylammonium bromide salt with 5 wt% Nafion^® 1100 suspension into a weigh boat. All mixture-casting solutions were prepared so the concentration of tetrabutylammonium bromide is in a three-fold excess of the concentration of sulfonic acid sites in the Nafion^® suspension. Previous studies have shown that all of the bromide ions that were introduced into a membrane were ejected from the membrane upon soaking in water;²⁴ therefore, 18 MΩ water was added to the weigh boats and allowed to soak overnight. After this time, the water was removed, the films thoroughly rinsed with 18 MΩ water, and then resuspended in ethanol. This casting solution was used for the formation of the bioanode.

Creation of bioanode

Once formed, the carbon microelectrodes were modified to create the chip-based bioanode. For immobilization of methylene green, and for experiments that involved static conditions, a reservoir was created by removing a small section (1 cm × 3 cm) from a larger piece of PDMS (2 cm × 4 cm); this piece of PDMS was then sealed over the carbon electrode so that the entire length of the electrode was exposed to solution. To this reservoir was placed a conventional Ag/AgCl reference electrode (CH Instruments, Austin, TX) and a platinum auxiliary electrode. For immobilization of the electrocatlyst methylene green, a solution of 0.4 mM methylene green and 0.1 M sodium nitrate in 10 mM sodium borate (pH = 9.2) was pipetted into the PDMS reservoir. Polymerization of methylene green was performed via cyclic voltammetry using a CH Instruments 650 potentiostat. The potential was scanned from −0.3 V to 1.3 V for 14 scan segments at a scan rate of 50 mV s⁻¹. Following this, the methylene green solution was pipetted out of the reservoir and the PDMS removed. The poly(methylene green) modified carbon ink microelectrodes were then rinsed with 18 MΩ (Nanopure) water and allowed to dry overnight.

The alcohol dehydrogenase/Nafion^® mixture was immobilized onto the carbon microelectrode using hydrodynamic flow. To accomplish this, a PDMS microchannel (130 µm wide, 100 µm deep and 3 cm long) was sealed over the carbon electrode (40 µm wide, 27.5 µm in height, and 2.5 cm long, n = 4), so that the entire electrode was sealed inside the microchannel. The size of the coating channel is such that alignment over the microelectrode is possible but is not much wider than the electrode. A 2∶1 ratio of alcohol dehydrogenase (ADH) and tetrabutylammonium bromide modified Nafion^® mixture with 1 mg of NAD⁺ for each 20 µL of tetrabutylammonium bromide modified Nafion^® was prepared and vortexed until sufficiently mixed. The mixture was introduced to the coating channel through the attached leur fitting by use of a syringe pump at 1.0 µL min⁻¹. Once the mixture had traveled the entire length of the coating channel (monitored visually), the solvent was allowed to evaporate at room temperature. This process is possible since PDMS is permeable to gases. After evaporation was complete, the PDMS was removed, leaving a coated bioanode.

Electrochemical measurements

All modified electrodes were equilibrated in pH 7.15 phosphate buffer before electrochemical measurements were performed. The working electrodes were carbon ink microelectrodes modified as a bioanode. The reference electrode was a Ag/AgCl electrode and a platinum wire acted as the auxiliary or counter electrode. The bioanodes were studied by cyclic voltammetry from −0.5 V to 1.3 V in a 1.0 mM ethanol and 1.0 mM NAD⁺ solution in phosphate buffer (pH 7.15). Peak currents were recorded for each electrode in both a static system (defined by a reservoir in PDMS) and in a flow system (using 200 µm wide microchannels). All electrochemical experiments were performed in triplicate and uncertainties correspond to one standard deviation.

Creation of power curve

Power curves were generated by measuring current and potential at various electrical loads. In this case, an external platinum cathode was used. This was made affixing (epoxy) a Nafion^® 117 membrane to the end of a piece of glass tubing (4 mm id). The glass tube was filled with phosphate buffer (pH 7.15) and a piece of platinum wire (22 gauge) was inserted. The bioanode remained within a flow channel and 1.0 mM ethanol and NAD⁺ was pumped through the system at 1.0 µL min⁻¹. The cathode was placed in the waste reservoir at the end of the flow channel.

Results and discussion

Enzyme system

The enzyme system used in these studies is alcohol dehydrogenase. Ethanol diffuses into the enzyme immobilization membrane and is oxidized by alcohol dehydrogenase in the presence of NAD⁺ to form acetaldehyde and NADH. The NADH is then oxidized at the poly(methylene green) electrocatalyst to regenerate NAD⁺. The chemistry involved in each of the layers is depicted in Fig. 3.

Fig. 3 Schematic of the microscale bioanode.

Microchip fabrication and assembly

The microchips used in these studies consisted of an electrode plate, where carbon microelectrodes were affixed, and PDMS-based flow channels (Fig. 1). Palladium connectors were first patterned onto the electrode plate by traditional sputtering and lithographic methods.²⁵ The connector served two purposes, to make an electrical connection to the carbon microelectrode and ensure that the PDMS flow channel sealed over the thin (0.2 µm) palladium connector. The latter ensures that the PDMS-based flow channel can be reversibly sealed on the electrode plate and that the carbon electrode completely resides in the microchannel, ensuring maximum exposed surface area. Palladium was utilized since similar platinum targets are much more expensive, but electrode conductivity could be improved by employing platinum, gold, or copper connectors. It is expected that the increased conductivity would result in an increased open circuit potential and increased power densities for the biofuel cell. After drilling fluid access holes and attaching fittings to the glass, the carbon microelectrode was placed on the plate so that the electrode was confined to the area between the access hole but over the palladium connector (Fig. 1). The carbon microelectrode was fabricated as recently described,²⁶ except in this case larger electrodes were made. Commercially available carbon inks are introduced into microchannels (45 µm wide, 42 µm deep, and 2.5 cm long) via vacuum, and after two heating steps, the PDMS micromolding channel is removed, leaving the carbon electrode on the glass surface. The final electrode that results is smaller than the original micromolding channel (40 µm wide, 27.5 µm in height, and 2.5 cm long), with shrinkage due to evaporation of thinner and other lightweight components of the ink.

In terms of operational aspects of the microchip-based fuel cell, the use of reversibly sealed PDMS flow channels simplifies the electrode coating/immobilization procedures that are required. Once the carbon microelectrode was coated with methylene green, the Nafion/enzyme layer was coated onto the electrode through a 130 wide, 100 µm deep and 3.0 cm long PDMS flow channel. After the channel was filled, the solvent was allowed to evaporate overnight, and the PDMS removed from the glass surface. When the bioanode was needed, a wider flow channel (200 µm) was reversibly sealed over the electrode and fuel was hydrodynamically introduced to the chip through the attached fittings. Both the ability to reversibly seal PDMS and the gas permeable nature of PDMS makes this procedure possible.

Electrode characterization

The carbon microelectrodes were characterized under both static and hydrodynamic conditions with cyclic voltammetry. A solution of 1 mM Ru(bpy)₃⁺² with 0.1 M sodium sulfate as the electrolyte was used to characterize the response of the electrode using cyclic voltammetry. A current density of 3.38 ± 0.34 × 10⁻⁴ A cm⁻² was obtained for a carbon ink electrode in a static solution. This compares to 2.06 × 10⁻⁴ A cm⁻² for a conventional glassy carbon macroelectrode. This increase in current density is expected because of radial transport to microelectrodes and an increased roughness for the micromolded electrodes compared to the polished glassy carbon electrode.²⁷ After static experiments were performed, a carbon microelectrode that was sealed within a 200 µm wide channel was studied with the 1.0 mM Ru(bpy)₃⁺² solution, which was pumped at various flow rates. Fig. 4 compares the current densities obtained at the various flow rates. As can be seen, as the flow rate was increased, the current density increased, which is expected due to increased transport of the analyte to the electrode surface as the flow rate increases. An electrochemical pretreatment was utilized to clean the electrode by applying 1.5 V for 3 min in a 50 mM phosphate buffer (pH 7.4).²⁷ However, the pretreatment showed little effect on the cyclic voltammograms when compared to non-treated electrodes and therefore was not utilized in further studies.

Fig. 4 Current densities for a carbon ink microelectrode as a function of flow rate using 1 mM tris(2,2'-bipyridyl)dichlororuthenium(II) hexahydrate and 0.1 M sodium sulfate as electrolyte.

Bioanode characterization

Ethanol bioanodes were fabricated in a layered approach (Fig. 3). The first layer was poly(methylene green), a known electrocatalyst for NADH. The electrocatalyst decreases the overpotential for NADH oxidation at a carbon electrode, which directly increases the open circuit potential of the cell. The second layer was the enzyme immobilization layer, which consists of alcohol dehydrogenase immobilized within a tetrabutylammonium bromide-treated Nafion membrane. The membrane acts to three-dimensionally constrain the enzyme within the enlarged micellar pore structure of the modified Nafion, while providing a buffered pH environment for the enzyme.²⁸

The effectiveness of using carbon ink microelectrodes as bioanodes was first investigated by characterizing the poly(methylene green) layer. Methylene green was successfully immobilized onto the carbon microelectrodes using 14 scan segments from −0.3 V to 1.3 V, the same procedure employed for macro-scale carbon electrodes.³ The polymerization voltammograms resembled those obtained with macro-sized carbon electrodes, except for a 200 mV shift in the adsorption peak due to a decrease in conductivity of the electrode.^27,29 NADH was used to measure the electrocatalytic properties of the poly(methylene green)-coated carbon ink electrode. Under static conditions, a current density of 5.09 × 10⁻⁴ A cm⁻² was obtained. Further studies used hydrodynamic flow conditions at various flow rates to pump the analyte solution to the electrode surface through PDMS flow channels. Commercially available microfittings were used to deliver the solution to the reversibly sealed microchannels. NADH was pumped through the PDMS flow channels at various flow rates of 0.5 µL min⁻¹ to 15.0 µL min⁻¹. Current densities for these conditions are presented in Table 1. In addition, current densities obtained under similar conditions using a flow cell with a planar disc (3 mm) glassy carbon electrode are shown in Table 1. The results of these experiments show that the current densities of poly(methylene green) modified electrodes are independent of flow rate. In this hydrodynamic system, the limiting current is a function of the mass transport rate of NADH within the poly(methylene green) film (i_s), the electron transfer rate for the reaction between NADH and poly(methylene green) (i_k), and the rate of mass transfer of NADH in solution(i_A). This is shown in eqn. (1).

Table 1 Current densities for 1 mM NADH in phosphate buffer (pH 7.15) at poly(methylene green) modified carbon electrodes. A current density of 5.09 × 10⁻⁴ A cm⁻² was measured for a static microelectrode

Flow rate/µL min^−1	Current density/A cm^−2
	Planar disc electrode (flowcell)	Carbon microelectrodes
0.5	5.92 × 10^−5	5.47 × 10^−5
1	5.89 × 10^−5	5.32 × 10^−5
5	5.72 × 10^−5	5.28 × 10^−5
10	5.63 × 10^−5	5.63 × 10^−5
15	5.53 × 10^−5	5.50 × 10^−5

---

For sufficiently thin coatings (i_s ≫ i_k), with the assumption of rapid desorption of NAD⁺ from the methylene green, it is possible to predict i_k from the known rate constant for the oxidation of the adsorbed NADH (k_cat) and the equilibrium constant for the NADH adsorption (K_m), since the limiting current becomes a function of only the rate of cross-exchange between NADH and poly(methylene green) and the mass transfer rate of the NADH in solution (eqns. (2) and (3)).

where Γ is the electrode coverage, b is the channel height, v is the flow rate, and D is the diffusion coefficient of NADH in solution.

Previous experimental results at pH 7.15 have shown that the rate of NADH oxidation at the poly(methylene green)-modified electrodes (k_cat) is 0.558 ± 0.095 s⁻¹ and K_m is equal to 1.22 ± 0.21 × 10⁻⁶ M. When using this catalytic information to determine i_k, theory predicts a kinetic current (i_k) of 6.1 ± 1.5 × 10⁻⁵ A cm⁻². When comparing the theoretically predicted i_k with the experimental limiting current (Table 1), they are not statistically different. This can only be explained by a kinetically-limited system. The system could be kinetically-limited due to a limited concentration of active sites on the electrode surface or due to slow Michaelis–Menton-type kinetics for the reaction between NADH and the electrocatalyst methylene green. Future work will examine thicker poly(methylene green) membranes to increase the number of active sites on the electrode surface, along with an examination of other NADH electrocatalysts that may have higher catalytic rates than methylene green.

Four commercially available carbon inks typically used in screen-printing of carbon electrodes were tested.¹⁴ The carbon ink microelectrodes were first polymerized with methylene green. A mixture of alcohol dehydrogenase and tetrabutylammonium bromide modified Nafion^® membrane was coated on the electrode through a 130 µm wide PDMS channel using hydrodynamic flow. The channel was removed after all the solvent had evaporated and replaced with a PDMS channel that was 200 µm wide. Cyclic voltammetry was employed and a 1.0 mM ethanol and 1.0 mM NAD⁺ fuel solution in pH 7.15 phosphate buffer was pumped through the channel at 1.0 µL min⁻¹. Peak currents were recorded and current densities calculated for each type of ink employed (Table 2). For all inks, the current densities are flow rate independent, which is a common phenomena for enzymes immobilized within polymer matrices due to a decrease in mass transport of the fuel and coenzyme within the polymer.³⁰ The Ercon E-978(I) carbon ink demonstrated the highest current densities by at least an order of magnitude and was used for further studies. Maximum current densities obtained for the microelectrode bioanodes were 3.26 mA cm⁻² which are comparable to macroscale bioanodes that typically shown maximum current densities between 2–5 mA cm⁻² depending on polymer loading.

Table 2 Current densities (A cm⁻²) for bioanodes fabricated with various commercially available screen printing carbon inks. All inks diluted with Ercon N160 solvent thinner to 0.2% (v/w) solution of the carbon ink

Flow rate/µL min^−1	Ercon E-978(I)	Ercon G-451(I)	Acheson Electrodag PF-407C	Acheson Electrodag 440B
0.5	1.77 × 10^−3 ± 9.06 × 10^−4	1.00 × 10^−4 ± 2.19 × 10^−5	1.03 × 10^−4 ± 1.82 × 10^−5	4.70 × 10^−5 ± 2.07 × 10^−6
1	1.92 × 10^−3 ± 9.70 × 10^−4	1.01 × 10^−4 ± 2.18 × 10^−5	1.01 × 10^−4 ± 1.96 × 10^−5	4.73 × 10^−5 ± 6.92 × 10^−6
5	1.90 × 10^−3 ± 8.58 × 10^−4	1.07 × 10^−4 ± 3.77 × 10^−5	1.06 × 10^−4 ± 1.27 × 10^−5	4.27 × 10^−5 ± 6.59 × 10^−6
10	1.91 × 10^−3 ± 8.41 × 10^−4	9.69 × 10^−4 ± 8.16 × 10^−6	1.05 × 10^−4 ± 1.63 × 10^−5	4.07 × 10^−5 ± 4.44 × 10^−6
15	1.96 × 10^−3 ± 7.79 × 10^−4	9.80 × 10^−4 ± 2.44 × 10^−5	1.13 × 10^−4 ± 2.97 × 10^−5	3.83 × 10^−5 ± 4.16 × 10^−6

---

Use of bioanode with platinum cathode

An external platinum cathode was paired with the microchip-based bioanode. The cathode consisted of a platinum wire that was placed in a piece of glass tubing; the two solutions were separated by a Nafion^® 117 membrane. The microchip-based bioanode remained within a flow channel and a solution containing 1.0 mM ethanol and 1.0 mM NAD⁺ was pumped through the system at 1.0 µL min⁻¹. The cathode was placed in the reservoir at the end of the flow channel. A representative power curve obtained for this biofuel cell is presented in Fig. 5. Maximum open circuit potentials of 0.34 V have been obtained with maximum current densities of 53.0 ± 9.1 µA cm⁻² for a microfluidic system employing alcohol dehydrogenase. These values are lower than the macroscale biofuel cell in part due to the thick membrane film on the electrode surface. In the macroscale system,⁴ control of the film thickness is obtained by pipetting a smaller or larger volume on the electrode. In this work, the microelectrodes are coated by hydrodynamically flowing the casting solution through the channels. The thickness of the membrane is dependent on the size of the microchannels, the size of the electrodes, and the percent of the membrane that is present in the solution. Future work will investigate use of smaller channels for coating the channels. Another possibility is use of larger electrodes. Diluting the Nafion solution is not an option, as when Nafion is diluted it forms unstable films.^31,32

Fig. 5 A representative power curve for an ethanol/oxygen microfabricated biofuel cell containing a 1.0 mM fuel solution.

Conclusions

In the case of portable electronics, small, lightweight, and long-lasting energy sources are among the most urgently needed technologies. Totally self-sufficient lab-on-a-chip devices require even smaller power sources, as weight and volume of on-board energy sources are disproportionately large compared to the systems that they power. Power sources such as fuel cells, microcombustors, thin film batteries, and beamed-in RF or optical energy are being explored. By miniaturization, the consumption of energy and materials is decreased and integration with electronics is possible, leading to a simplified system where cost is decreased without affecting performance. Through the use of traditional photolithography and micromolding of carbon inks, components to a biofuel cell were integrated into a microchip. With the use of an external platinum cathode and an integrated bioanode on a chip, open circuit potentials of 0.34 V and current densities of 53.0 ± 9.1 µA cm⁻² were produced. While these results show great promise, improvements in the conductivity of micromolded electrodes and thickness of the modifying layers are needed. Further research to implement both the bioanode and the biocathode on a microchip will be explored as will attempts at serially stacking multiple biofuel cells together.

Acknowledgements

This authors wish to thank the Office of Naval Research for funding this project and the Clare Booth Luce Foundation for C.M.M.'s graduate fellowship. The authors would also like to thank the research group of Professor Susan M. Lunte (University of Kansas) for use of their microfabrication facility.

References

1. K. Kordesch and G. Simader, Fuel Cells and Their Applications, Wiley-VCH, Weinheim, 2001.
2. G. Tayhas, R. Palmore and G. M. Whitesides, Enzymatic Conversion of Biomass for Fuels Production, American Chemical Society, Washington, D.C., 1994.
3. C. M. Moore, N. L. Akers and S. D. Minteer, Biomacromolecules, 2004, 5, 1241–1247.
4. N. L. Akers and S. D. Minteer, Fuel Sci. Eng. Technol., 2004, 2, 497–501 .★This paper introduced high current density and long lifetime ethanol/oxygen biofuel cells.
5. C. M. A. Brett and A. M. O. Brett, Electrochemistry Principles, Methods, and Applications, Bookcraft (Bath) Ltd., Avon, 1993.
6. D. R. Reyes, D. Iossifidis, P. A. Auroux and A. Manz, Anal. Chem., 2002, 74, 2623–2636.
7. P. A. Auroux, D. Iossifidis, D. R. Reyes and A. Manz, Anal. Chem., 2002, 74, 2637–2652.
8. T. Vilkner, D. Janasek and A. Manz, Anal. Chem., 2004, 76, 3373–3386.
9. A. J. Gawron, R. S. Martin and S. M. Lunte, Eur. J. Pharm. Sci., 2001, 14, 1–12.
10. N. A. Polson and M. A. Hayes, Anal. Chem., 2001, 73, 312A–319A.
11. K. Seller, Z. H. Fan, K. Fluri and D. J. Harrision, Anal. Chem., 1994, 66, 3485–3491.
12. W. R. Vandaveer, IV, S. A. Pasas, R. S. Martin and S. M. Lunte, Electrophoresis, 2002, 23, 3667–3677.
13. M. A. Schwarz and P. C. Hauser, Lab Chip, 2001, 1, 1–6.
14. J. Wang, B. Tian, V. B. Nascimento and L. Angnes, Electrochim. Acta, 1998, 43, 3459–3465.
15. R. M. Wightman, Anal. Chem., 1981, 53, 1125A–1134A.
16. J. C. McDonald and G. M. Whitesides, Acc. Chem. Res., 2002, 35, 491–499.
17. J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. A. Schueller and G. M. Whitesides, Electrophoresis, 2000, 21, 27–40.
18. J. S. Wainright, R. F. Savinell, C. C. Liu and M. Litt, Electrochim. Acta, 2003, 48, 2869–287.
19. K. Shah, W. C. Shin and R. S. Besser, J. Power Sources, 2003, 123, 172–181.
20. R. Ferrigno, A. D. Strook, T. D. Clark, M. Mayer and G. M. Whitesides, J. Am. Chem. Soc., 2002, 124, 12930–12931 .★This is the first reported microfluidic fuel cell.
21. T. Chen, S. C. Barton, G. Binyamin, Z. Gao, Y. Zhang, H. H. Kim and A. Heller, J. Am. Chem. Soc., 2001, 123, 8630–8631.
22. N. Mano and A. Heller, J. Electrochem. Soc., 2003, 150, A1136–A1138.
23. R. S. Martin, A. J. Gawron, S. M. Lunte and C. S. Henry, Anal. Chem., 2000, 72, 3196–3202.
24. M. Schrenk, R. Villigram, N. Torrance, S. Brancato and S. D. Minteer, J. Membr. Sci., 2002, 205, 3–10.
25. N. A. Lacher, S. M. Lunte and R. S. Martin, Anal. Chem., 2004, 76, 2482–2491.
26. M. L. Kovarik, N. J. Torrence, D. M. Spence and R. S. Martin, Analyst, 2004, 129, 400–405.
27. K. Grennan, A. J. Killard and M. R. Smyth, Electroanalysis, 2001, 13, 745–750.
28. T. J. Thomas, K. E. Ponnusamy, N. M. Chang, K. Galmore and S. D. Minteer, J. Membr. Sci., 2003, 213, 55–66.
29. J. Wang, M. Pedrero, H. Sakslund, O. Hammerich and J. Pingarron, Analyst, 1996, 121, 345–350.
30. A. J. Bard and L. R. Faulkner, Electrochemical Methods Fundamental and Applications, John Wiley & Sons, New York, 2001.
31. G. Fortier, M. Vaillancourt and D. Belanger, Electroanalysis, 1992, 4, 275–283.
32. J. Rishpon, S. Gottesfeld, C. Campbell, J. Davey and T. A. J. Zawodzinski, Electroanalysis, 1994, 6, 17–21.

---