Porphyrin-Sensitized Nanoparticulate TiO2 as the Photoanode of a

Aug 6, 2004 - Cyclic voltammetry was done with a PAR 173 potentiostat and a 175 Universal Programmer. Acquisition of current ..... A fuel cell consist...
0 downloads 7 Views 105KB Size
Download PDF
8366

Langmuir 2004, 20, 8366-8371

Porphyrin-Sensitized Nanoparticulate TiO2 as the Photoanode of a Hybrid Photoelectrochemical Biofuel Cell Alicia Brune,† Goojin Jeong,† Paul A. Liddell,† Tadashi Sotomura,‡ Thomas A. Moore,*,† Ana L. Moore,*,† and Devens Gust*,† Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona 85287-1604, and Advanced Technology Research Laboratories/Nano Technology Research Laboratory, Matsushita Electric Ind. Co., Ltd., 3 Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-0237, Japan Received April 22, 2004. In Final Form: June 15, 2004 Porphyrin-sensitized nanoparticulate TiO2 on conducting glass has been investigated as a photoanode material for a new cell that converts light energy into electricity. The cell is a hybrid of a dye-sensitized nanoparticulate semiconductor photoelectrochemical solar cell, and a biofuel cell that oxidizes glucose. Porphyrin molecules excited by light inject electrons into the photoanode, from where they enter the external circuit. The resulting porphyrin radical cations are reduced by NADH in aqueous buffer, ultimately regenerating the photoanode and producing NAD+. Glucose dehydrogenase oxidizes glucose, and in the process recycles NAD+ back to NADH. The photoanode is coupled with a suitable cathode to make a functioning cell (Hg/Hg2SO4 was employed for evaluation purposes). The cell produces 1.1 V at open circuit and has a fill factor of 0.61. These values are both significantly higher than those for a previously reported cell of a similar type based on an SnO2 electrode.

Introduction A promising photosynthesis-based approach to the conversion of sunlight into electricity is the dye-sensitized nanoparticulate photoelectrochemical solar cell.1-3 Such cells employ a photoanode consisting of conductive glass bearing a coating of a nanoparticulate wide band gap semiconductor. The surface of the particles is covered with a monolayer of dye molecules that absorb light and inject electrons into the semiconductor. The oxidized dye is regenerated through reduction by a redox couple in the electrolyte, whose oxidized form is reduced in turn at the cathode. Another approach to solar conversion employs biofuel cells that produce electricity by oxidation of biological materials that are produced by photosynthesis. These often use enzymes4-8 or even entire organisms6,9-11 to catalyze oxidation of carbohydrates and related materials, and couple the oxidation to electrical current flow between an anode and a cathode. Typical fuels that have been investigated include glucose12-15 and methanol.16 * Corresponding author. E-mail: [email protected]. † Arizona State University. ‡ Matsushita Electric Industrial Co., Ltd. (1) O’Regan, B.; Gra¨tzel, M. Nature 1991, 353, 737-740. (2) Hagfeldt, A.; Gra¨tzel, M. Chem. Rev. 1995, 95, 49-68. (3) Sauve, G.; Cass, M. E.; Coia, G.; Doig, S. J.; Lauermann, I.; Pomykal, K. E.; Lewis, N. S. J. Phys. Chem. B 2000, 104, 6821-6836. (4) Yahiro, A. T.; Lee, S. M.; Kimble, D. O. Biochim. Biophys. Acta 1964, 88, 375-383. (5) Palmore, G. T. R.; Bertschy, H.; Bergens, S. H.; Whitesides, G. M. J. Electroanal. Chem. 1998, 443, 155-161. (6) Palmore, G. T. R.; Whitesides, G. M. Enzymatic Conversion of Biomass for Fuels Production; ACS Symposium Series 566; American Chemical Society: Washington, DC, 1994; pp 271-290. (7) Katz, E.; Willner, I.; Kotlyar, A. B. J. Electroanal. Chem. 1999, 479, 64-68. (8) Takeuchi, N. J. J. Electrochem. Soc. 1989, 136, 96-101. (9) Karube, I.; Suzuki, S. Methods Enzymol. 1988, 137, 668-674. (10) Tender, L. M.; Reimers, C. E.; Stecher, H. A., III; Holmes, D. E.; Bond, D. R.; Lowy, D. A.; Pilobello, K.; Fertig, S. J.; Lovley, D. R. Nat. Biotechnol. 2002, 20, 821-825. (11) Park, D. H.; Zeikus, J. G. Appl. Microbiol. Biotechnol. 2002, 59, 58-61. (12) Gongotri, K. M.; Regar, O. P.; Lal, C.; Kalla, P.; Genwa, K. R.; Meena, R. Int. J. Energy Res. 1996, 20, 581-585.

Figure 1. Schematic diagram of the hybrid photoelectrochemical biofuel cell.

Recently, we reported a new strategy that combines these two approaches: a hybrid photoelectrochemical biofuel cell.17 The cell is illustrated schematically in Figure 1. The photoanode consists of an indium-tin oxide coated glass electrode to which SnO2 nanoparticles have been sintered. The particles are covered with a self-assembled layer of porphyrin sensitizer S (Chart 1). This anode is immersed in aqueous buffer. For evaluation purposes, it is electrically connected through an external circuit to a Hg/Hg2SO4 cathode bathed in a saturated K2SO4 solution. The cathode compartment is separated from the anode compartment by a Nafion ion-permeable membrane, which allows hydrogen ions to diffuse between the electrode (13) Chen, T.; Calabrese Barton, S.; Binyamin, G.; Gao, Z.; Zhang, Y.; Kim, H. H.; Heller, A. J. Am. Chem. Soc. 2001, 123, 8630-8631. (14) Katz, E.; Filanovsky, B.; Willner, I. New J. Chem. 1999, 481487. (15) Mano, N.; Mao, F.; Heller, A. J. Am. Chem. Soc. 2002, 124, 1296212963. (16) Daubmann, T.; Aivasidis, A.; Wandrey, D. Water Sci. Technol. 1997, 36, 175-182. (17) de la Garza, L.; Jeong, G.; Liddell, P. A.; Sotomura, T.; Moore, T. A.; Moore, A. L.; Gust, D. J. Phys. Chem. B 2003, 107, 10252-10260.

10.1021/la048974i CCC: $27.50 © 2004 American Chemical Society Published on Web 08/06/2004

Porphyrin-Sensitized Nanoparticulate TiO2 Chart 1

compartments. Light absorption by S generates the porphyrin excited singlet state, which donates an electron into the photoanode, leaving a porphyrin radical cation and initiating current flow through the external circuit to the cathode. The porphyrin radical cation is reduced by nicotinamide adenine dinucleotide (NADH) that is dissolved in the buffered electrolyte. This regenerates the photosensitizer S and, in a sequential two-electron process, produces NAD+. A suitable enzyme, such as glucose dehydrogenase, and its substrate (glucose) are present in the electrolyte solution. The glucose dehydrogenase oxidizes glucose to gluconolactone, consuming the coenzyme NAD+ and producing NADH and H+. Thus, the photoanode reactions, coupled with the cathodic reactions, produce electricity by consuming photons and oxidizing a carbohydrate fuel. Practical applications of such a cell would usually require a different cathodic reaction, such as oxygen reduction. The hybrid cell was evaluated with several fuels (and the appropriate enzymes), such as glucose, glucose-6phosphate, methanol, and ethanol. Under short-circuit conditions with 520-nm irradiation (1.0 mW/cm2), the cell produced a short-circuit current of 30 µA/cm2, and an opencircuit voltage of 0.75 V. The maximum power was 9.5 µW/cm2, obtained at 0.45 V. The fill factor was 0.42, and the incident photon to current efficiency was 12%.17 The hybrid approach, using two energy sources, has several potential advantages over either the nanoparticulate photoelectrochemical cell or the biofuel cell,17 including the production of higher voltages and higher power under some conditions. Nanoparticulate photoelectrochemical solar cells typically employ TiO2 rather than SnO2 as the wide band gap semiconductor. One reason is that TiO2 has a more negative conduction band potential, which could lead to a cell with a higher open-circuit voltage. For TiO2 nanoparticles on fluorine-doped tin oxide (FTO) conductive glass, the band potential has been determined18 and has been shown to be 0.5 V more negative than that of SnO2.19 This suggests that for the photoelectrochemical biofuel cell, an open circuit voltage enhancement of up to 0.5 V might be possible with TiO2. This would represent a significant increase in cell performance. We have therefore investigated the characteristics of a glucose-based photoelectrochemical biofuel cell similar to that diagrammed in Figure 1, but with a TiO2-based photoanode, to determine how cell performance is affected by the modification, and whether any enhancement is achieved. (18) Rothenberger, G.; Fitzmaurice, D.; Gra¨tzel, M. J. Phys. Chem. 1992, 96, 5983-5986. (19) Nasr, C.; Kamat, P. V.; Hotchandani, S. J. Phys. Chem. B 1998, 102, 10047-10056.

Langmuir, Vol. 20, No. 19, 2004 8367

Experimental Section Reagents. The buffer trishydroxylaminomethane (Tris), β-nicotinamide adenine dinucleotide reduced form disodium salt (β-NADH), β-nicotinamide adenine dinucleotide (β-NAD+), and β-D-glucose were purchased from Sigma-Aldrich. The sensitizer (S), 5-(4-carboxyphenyl)-10,15,20-tris(4-methylphenyl)porphyrin, was prepared using procedures described in the literature.20 The enzyme glucose dehydrogenase (GDH) from Bacillus megaterium was obtained from USB Corp. The enzyme activity was assayed following a protocol provided by the manufacturer. Electrodes. Glass plates coated with a conductive layer of fluorine-doped tin oxide (FTO), which in turn bears a coating of nanoparticulate TiO2, were purchased from two manufacturers: transparent FTO/TiO2 layers (20 nm particle size, film thickness ∼10 µm) were obtained from INAP, Germany, and transparent and relatively opaque FTO/TiO2 layers (16 and 35 nm particle size, respectively, film thickness ∼11 µm) were purchased from ECN, The Netherlands. To coat the photoanode with the sensitizer S, the porphyrin was dissolved in toluene (∼5.0 × 10-4 M). This solution was dropped on the TiO2 surface using a Pasteur pipet and was left in contact with it for different periods of time (5 min to 24 h), to achieve different loads of sensitizer on the semiconductor. Afterward, the plates were rinsed with toluene-hexane (1:1) to remove unbound material and were dried under a stream of nitrogen. The FTO/TiO2 plates used as controls were exposed only to toluene or to water. The cathode in the photobiofuel cell consisted of the reference system Hg/Hg2SO4/saturated K2SO4. It was separated from the photoanode compartment by a Nafion ion-permeable membrane.17 Photoelectrochemical Cell. The two-compartment cell was constructed as previously described.17 The anodic compartment was filled with 13 mL of a solution of 0.25 M Tris at pH 8 (adjusted with HCl) that contained 0.1 M KCl as a supporting electrolyte. The glucose fuel, glucose dehydrogenase, NADH, and NAD+ were added as required. The cathodic compartment was filled with a saturated K2SO4 solution. Provision was made for purging the anodic compartment with argon gas to exclude oxygen. Instrumentation and Measurements. A xenon lamp and a Jobin-Yvon monochromator system provided light in the 300800 nm wavelength range. The back (nonconductive, not TiO2 coated) side of the photoanode faced the excitation beam. The illuminated area was 2 cm2. The power density of the light at the photoanode was measured with a silicon photodiode. Experiments were carried out at 520 nm and 1 mW/cm2, unless stated otherwise. A model 617 Keithley electrometer, interfaced to a computer, was used in current versus time measurements. Cyclic voltammetry was done with a PAR 173 potentiostat and a 175 Universal Programmer. Acquisition of current versus time and cyclic voltammetric data was done with LabView software. A 236 Keithley source and a Keithley 2400 source meter were used to measure current versus voltage characteristics of the cell. Absorbance was measured with a UV-visible spectrophotometer (Shimadzu UV 2100 UV). X-ray diffraction patterns were obtained with a Rigaku D/max-IIB diffractometer. Optical microscopy observations were carried out with a Mititoyo Ultraplan FS-110 microscope. All measurements were done at room temperature.

Results Characterization of the Photoanode. The absorption spectrum of porphyrin sensitizer S in toluene solution includes a Soret band at 414 nm and Q-bands at 516, 551, 594, and 650 nm (Figure 2). The spectrum of nanoparticulate TiO2-coated FTO glass shows light-scattering effects that increase toward the blue, but little or no absorption. Soaking the electrodes in toluene solutions of S, as described in the Experimental Section, results in adsorption of porphyrin onto the electrode surface. The absorbance of the S-coated electrodes in the Soret region is too large to be measured accurately; essentially all of the light is absorbed. The absorption spectrum of the electrodes in the Q-band region is little changed from that (20) Anton, J. A.; Kwong, J.; Loach, P. A. J. Heterocycl. Chem. 1976, 13, 717-725.

8368

Langmuir, Vol. 20, No. 19, 2004

Brune et al.

Figure 2. Absorption spectra of porphyrin S in toluene (- - -), an FTO/TiO2 photoanode in air (- ‚ ‚ -), and FTO/TiO2 photoanodes that have been treated with toluene solutions of S for 5 min (‚ ‚ ‚ ‚), 30 min (- ‚ - ‚ -), and 24 h (-).

Figure 4. Oxidation of NADH in the photoelectrochemical cell during the generation of photocurrent in the external circuit. (a) NADH concentration in the electrolyte solution, measured spectrophotometrically, as a function of time with irradiation at 520 nm (1.0 mW/cm2) (O), or in the dark (3). (b) Moles of NADH consumed in an irradiation experiment like that described in (a) as a function of the number of moles of electrons flowing through the external circuit. The dotted line is a linear least-squares fit to the data.

Figure 3. X-ray diffraction patterns obtained from an FTO electrode coated with nanoparticulate TiO2 (a), and a similar electrode that had been soaked in a solution of porphyrin S in toluene for 24 h, and then rinsed, as described in the Experimental Section (b). At the bottom are shown the patterns expected for SnO2 and TiO2 (anatase).

in toluene. On the electrodes there is a slight shift of the Q-band maxima to longer wavelengths that increases somewhat with longer soaking times (and consequently denser coverage of the electrode surface by porphyrin). This suggests that the red shift may be due to interchromophore interactions and/or adsorption at different types of surface sites. The spectra indicate that adsorption of the porphyrin to the TiO2 surface does not significantly perturb the porphyrin electronic structure. The spectrum of a porphyrin-coated electrode does not change significantly after it has been employed as the photoanode in the photoelectrochemical biofuel cell for at least 6 h, whether exposed to air or kept under an argon atmosphere. X-ray powder diffraction patterns were obtained from the electrodes to further characterize their structure. An FTO electrode coated with TiO2 nanoparticles (INAP) gives a sharp diffraction pattern, as shown in Figure 3a. The peaks in the diffractogram are assigned to TiO2 (anatase phase) and SnO2. After an electrode was soaked in a toluene solution of S for 24 h, an additional peak is seen at low diffraction angles (Figure 3b). This may be due to supramolecular structures consisting of stacks of porphyrins that might form at high loading of the substrate with S. Indeed, optical microscopy of the TiO2 surface

with high porphyrin loading revealed the presence of microcrystals. Similar X-ray studies of the electrodes from ECN gave similar results, with the exception that some rutile phase of the TiO2 was also detected in the electrodes having a 16-nm particle size. Cell Performance - Faradaic Efficiency of NADH Oxidation. The first step in the evaluation of cell performance with the TiO2 photoanode was characterization of the operation of the cell with NADH. The NADH serves as the electron donor to the porphyrin radical cation that results from electron transfer to the TiO2 by the porphyrin excited singlet state. After two sequential electron donations, the NADH has been converted to NAD+. The cell was set up as shown in Figure 1 and described in the Experimental Section. The photoanode compartment contained a solution of 0.25 M Tris buffer and 0.10 M KCl at pH 8 that was stirred and purged with argon to minimize dissolved oxygen. NADH was added to a concentration of 1.0 mM, the photoanode was illuminated with 520 nm light (1.0 mW/cm2), and the photocurrent was measured. Current in the dark was negligible. The concentration of NADH was monitored by periodically withdrawing aliquots of solution and measuring the disappearance of the characteristic absorbance at 340 nm.21 Typical results are shown in Figure 4a. The NADH concentration decreases with illumination time, reaching approximately 70% of its initial value after 7 h. A similar cell, kept in the dark, shows no decrease in the concentration of NADH from its initial value of 0.93 mM over this time period, within experimental error. Thus, NADH is consumed by the cell as photocurrent is produced. Figure (21) Rover, L.; Fernandes, J. C. B.; Olivera, N. G.; Kubota, L. T.; Katekawa, E.; Serrano, S. H. P. Anal. Biochem. 1998, 260, 50-55.

Porphyrin-Sensitized Nanoparticulate TiO2

Figure 5. Cycling of NADH by the photoelectrochemical biofuel cell. The cell, prepared as described in the Experimental Section and containing 1.0 mM NADH in the electrolyte, was illuminated for 7 h, after which time ∼30% of the NADH had been oxidized by the porphyrin radical cation. Next, glucose (to 0.10 M) and GDH (to 2.4 units/mL) were added. The NADH concentration returned to near its original value, due to reduction of NAD+ by the enzyme system.

4b shows a plot of the moles of NADH consumed against the moles of electrons passing through the external current. The dotted line is the best straight line fit to the data from a regression analysis. The inverse of the slope of this line represents the number of moles of electrons passing through the circuit per mole of NADH consumed. Data from experiments with four different photoanodes yielded an average of 1.8 ( 0.2 electrons obtained per molecule of NADH oxidized. As the oxidation of NADH to NAD+ is an overall two-electron process, the results indicate that under these cell conditions, the photocurrent derives from NADH oxidation, and that there is little loss of NADH to side reactions. Cell Performance - Effect of Glucose and Glucose Dehydrogenase. If the TiO2 photoanode performs in the same general manner as the SnO2 photoanode investigated earlier, the product of NADH oxidation by the porphyrin radical cation is NAD+ in an enzymatically useable form. To investigate this possibility, glucose to a concentration of 0.10 M and glucose dehydrogenase to an activity of 2.4 units/mL were added to the cell described in the last section after it had been irradiated at 520 nm for 7 h, at which time the concentration of NADH had been reduced by about 30% (Figure 5). The additions resulted in regeneration of NADH to a concentration close to, but somewhat below, the initial concentration. This result confirms that the major product of NADH oxidation during operation of the photoelectrochemical biofuel cell is NAD+ and that the product can be reduced back to NADH by GDH during oxidation of glucose. The performance of the cell over time, with glucose and GDH present, was also investigated. The cell was set up as described in the Experimental Section, with 4.0 mM NADH, 0.10 M glucose, and 0.5 units/mL GDH present. The cell was irradiated, and the NADH concentration was monitored with time (Figure 6). The initial current in the external circuit was 92 µA, although this decreased with time (vide infra). The figure indicates that there is no change in NADH concentration, within experimental error, over a 24 h period. During this time, a total of 55 × 10-6 mol of electrons passed through the circuit, oxidizing 27.5 × 10-6 mol of NADH to NAD+, which was then reduced enzymatically. During the 24 h period, half of the NADH present was cycled through the redox loop, but there was no measurable net loss of NADH. Cell Performance - Characteristics of the Photoelectrochemical Biofuel Cell. The current-voltage

Langmuir, Vol. 20, No. 19, 2004 8369

Figure 6. NADH regeneration as a function of illumination time. A cell was prepared as described in Figure 5, and contained 4.0 mM NADH, 0.10 M glucose, and 0.5 units/mL GDH. It was illuminated, with current flowing through the external circuit, for a period of 24 h. The concentration of NADH (0) does not change appreciably with time, due to enzymatic regeneration of NADH from the NAD+ formed at the photoanode.

Figure 7. Plot of current against voltage for the photoelectrochemical biofuel cell operating with 4.0 mM NADH, 0.10 M glucose, and 0.5 units/mL GDH under an argon atmosphere. The curve was obtained after the cell had been operating under illumination for 1 h.

characteristics of the cell, prepared with an INAP photoanode, 4.0 mM NADH, 0.10 M glucose, and 0.5 units/ mL GDH, were determined after 1 h of illumination (Figure 7). From this and similar experiments, the average current at short circuit (Isc) and voltage at open circuit (Voc) were determined to be 110 µA and 1.10 V, respectively. The maximum power density (Pmax) was 74 µW per 2 cm2 and was obtained at 0.84 V. From these parameters, the fill factor (FF) was calculated as 0.61 from the equation FF ) Pmax/IscVoc. The incident photon to current efficiency (IPCE), the number of electrons injected into the external circuit of the cell per unit time divided by the number of incident photons per unit time, is a measure of the overall performance of a solar conversion cell. Figure 8a shows the IPCE as a function of irradiation wavelength for a cell containing 5.0 mM NADH, but no glucose or GDH. (The IPCE was determined soon after illumination began, when the concentration of NADH was not a limiting factor.) In the 400 nm (Soret band) region of the spectrum, the sample is optically thick (absorbance .1) such that essentially all of the light is absorbed. That is not true in the Q-band region at >500 nm. In this region, the IPCE tracks the peaks and valleys of the absorption spectrum (Figure 2) because the IPCE is a function of the fraction of incident light actually absorbed. The light-harvesting efficiency at any particular wavelength is given by 1-10-A, where A is the absorbance. The absorbance may be estimated from a spectrum such as

8370

Langmuir, Vol. 20, No. 19, 2004

Brune et al.

Figure 9. Cell current as a function of irradiation time. The cell was prepared as described in the Experimental Section, and was stirred under an argon atmosphere during illumination at 520 nm. It was initially charged with 4.0 mM NADH, 0.10 M glucose, and 0.5 units/mL GDH. The steep vertical deviations are the result of disrupting cell operation to remove electrolyte samples. Table 1. Characteristics of the Photoelectrochemical Biofuel Cella Using TiO2 or SnO2 Electrodes electrode

Isc (µA/cm2)

Voc (V)

Pmax (µW/cm2)

FF

TiO2 SnO2

55 30

1.10 0.75

37 9.5

0.61 0.42

a Cell parameters are as described in the text and ref 17. Illumination was at 520 nm using 1 mW/cm2.

Figure 8. (a) Incident photon to current efficiency (IPCE) as a function of irradiation wavelength for a photoelectrochemical biofuel cell set up as described in the Experimental Section, and containing 5.0 mM NADH in the electrolyte, but no glucose or GDH. The trace has been corrected for variations in excitation system output with wavelength. In the Q-band region, where not all of the excitation light is absorbed, the IPCE tracks the maxima and minima of the absorption spectrum (Figure 2). (b) Quantum yield of electron injection into the external circuit as a function of wavelength. The trace was obtained by dividing the IPCE by the light-harvesting efficiency, which was estimated from the absorption spectrum of the electrode. The actual quantum yield may be higher, due to artifacts in the absorption spectrum caused by light scattering by the photoanode.

that in Figure 2, although there is some error involved because the samples not only absorb, but also scatter light, leading to an overestimation of A. Dividing the IPCE by the light-harvesting efficiency yields the quantum yield of electron injection into the external circuit (Figure 8b). The quantum yield throughout the visible spectral region is on the order of 0.1 or greater, depending on the degree of overestimation of A. The quantum yield, as shown in Figure 8b, appears to be reduced at the Q-band absorption maxima, relative to regions where less light is absorbed. The quantum yield variation shown in the figure may be due at least in part to problems in the estimation of the amount of light absorbed that arise because of scattering. Aging Effects on Cell Performance. The cell current output was found to decrease with operation time. Figure 9 shows the short circuit current as a function of time from a cell initially containing 4.0 mM NADH, 0.10 M glucose, and 0.50 units GDH per mL. The cell was stirred, kept under an argon atmosphere, and irradiated at 520 nm during the entire experiment. After 24 h, the current had decreased to about half its initial value. The fill factor and the open-circuit voltage did not change significantly during this irradiation period. The decrease in current

was more rapid when air was admitted to the cell. In a separate experiment, it was found that no detectable bleaching of the porphyrin absorption of a similar electrode in a similar cell (5.0 mM NADH, but with no glucose or GDH) occurred after 6 h of irradiation under these conditions, whether under an argon atmosphere or open to the air (the Voc and Isc were both lower in the cell open to the air). During the first 6 h of operation, the cell current in Figure 9 dropped by about 20%. These results suggest that the loss of current is not due to destruction of the sensitizer S. Figure 6 shows the NADH concentration in the cell described in Figure 9 as a function of time. It indicates that NADH is continually regenerated by the enzyme system during the course of the experiment and that the NADH concentration after 24 h is essentially the same as it was at the beginning. Thus, the decrease in cell performance shown in Figure 9 is not due to degradation of NADH, GDH, or any other part of the NADH regeneration cycle. In this regard, it was determined from cyclic voltammetric measurements on a porphyrin-coated photoanode in aqueous solution containing Tris (pH ) 8.0) and KCl that NAD+ was not reduced at the photoanode at the voltages present during cell operation. A similar situation was found earlier for the cell using a SnO2 photoanode.17 Discussion Cell Performance. Table 1 shows the characteristics of the photoelectrochemical biofuel cell with a TiO2 electrode, as well as those for a similar cell employing a SnO2 electrode.17 The Isc and Pmax will, of course, depend on the light-harvesting efficiency of the cell, which in turn depends on the absorbance. This was not strictly controlled in the experiments giving rise to the data in the table. The Voc (which contributes to Pmax) and the FF were both significantly enhanced in the TiO2 cell. Thus, from these points of view, the TiO2 cell represents a significant

Porphyrin-Sensitized Nanoparticulate TiO2

improvement in performance over the analogous SnO2based cell. The higher open-circuit voltage, in particular, which is due to the more negative conduction band potential of TiO2, makes the TiO2-based photoanode suitable for driving a wider range of reductive processes when coupled to suitable cathodic half cells. The lower fill factor for the SnO2-based photoanode may be due in part to an increase in sheet resistance for the ITO substrate that results from sintering.22 The increase in maximum voltage for the TiO2 cell is in accord with what is known about the conduction band potentials in the two semiconductors.18,19,23-25 The Vmax is assumed to be a function of the potentials of the conduction band of the semiconductor and the cathode (Hg/Hg2SO4). The conduction band potential for TiO2 on FTO in an electrode of a type similar to those studied here reportedly equals (-0.400-0.06 pH) V vs SCE.18 It is also reported that the conduction band potential for SnO2 on similar electrodes is more positive than that of TiO2 by 0.5 V.19 Assuming that this difference is applicable at pH 8.0, we can estimate the potentials of the TiO2 and SnO2 electrodes at pH 8 as -0.88 and -0.38 V vs SCE, or -0.64 and -0.14 V vs NHE, respectively. The potential of the Hg/Hg2SO4 electrode is taken as +0.640 V vs NHE.26 Thus, the maximum theoretical potential of the TiO2 cell is estimated to be around 1.28 V, and that of the SnO2 cell is ca. 0.78 V. The measured Voc values in Table 1 are consistent with the theoretical maxima, although somewhat lower. No effort has been made to maximize the performance of either cell. The quantum yield obtained for the cell is similar to those found for porphyrins in dye-sensitized nanoparticulate wide band gap semiconductor photoelectrochemical cells. Time-resolved spectroscopic studies have shown that charge injection from an excited porphyrin into TiO2 is extremely rapid, suggesting a quantum yield for injection of near unity.27 If this is the case for all of the porphyrin molecules on the electrode used in this study, then quantum yields less than unity must be due to recombination processes of some kind. However, other work28 suggests that electron injection is not optimal in porphyrin-based cells and that this may be due to unfavorable electronic coupling between the dye and the nanoparticles. Aging of the Cell. The Isc of the cell decreases with irradiation time (Figure 9). This is not due mainly to destruction of the sensitizer, or to loss of NADH. A similar decrease in Isc with irradiation time was noted for the SnO2-based cell studied earlier.17 As the absorption spectra (22) Ngamsinlapasathian, S.; Sreethawong, T.; Suzuki, Y.; Yoshikawa, S. Effect of substrate on dye-sensitized solar cell performance using nanocrystalline titania. Abstracts of the 205th Meeting of the Electrochemical Society 2004; Abstract 605. (23) Dutoit, C.; Cardon, F.; Gomes, W. Ber. Bunsen.-Ges. Phys. Chem. 1976, 80, 475-481. (24) Mollers, F.; Memming, R. Ber. Bunsen.-Ges. Phys. Chem. 1972, 76, 469-580. (25) Watson, D. F.; Marton, A.; Stux, A. M.; Meyer, G. J. J. Phys. Chem. B 2003, 107, 10971-10973. (26) Pletcher, D.; Greef, R.; Peat, R.; Peter, L. M.; Robinson, J. Instrumental Methods in Chemistry, 1st ed.; Horwood: Chichester, 2001; p 362. (27) Tachibana, Y.; Haque, S. A.; Mercer, I. P.; Durrant, J. R.; Klug, D. R. J. Phys. Chem. B 2000, 104, 1198-1205. (28) Odobel, F.; Blart, R.; Lagre´e, M.; Villieras, M.; Boujtita, H.; El Murr, N.; Caramori, S.; Bignozzi, C. A. J. Mater. Chem. 2003, 13, 502510.

Langmuir, Vol. 20, No. 19, 2004 8371

of the electrodes do not change with irradiation time, it is unlikely that the overall rate of electron injection from the dye into the semiconductor changes with time (barring major changes in the molecule-surface interface). If this is the case, the decrease in current is likely due to increased charge recombination at the photoanode. The reasons for this behavior are unknown and remain the subject of investigation. If an operating cell is exposed to air, both the Isc and the Voc values decrease significantly within minutes, and the decrease in Isc with time is much more pronounced. For the experiments described in this paper, the cell was purged with argon, but it is not known to what extent the observed decrease in photocurrent is a result of incomplete removal of oxygen. Conclusions A photoelectrochemical biofuel cell has been constructed with a nanoparticulate TiO2 electrode, a free base porphyrin sensitizer, NADH/NAD+ as the redox relay, glucose dehydrogenase as the redox enzyme, and glucose as the fuel. The cell combines the chemistry of a dye-sensitized photoelectrochemical solar cell with that of an enzymecatalyzed biofuel cell. The one-electron reactions at the photoanode are coupled to the two-electron reactions of the fuel cell by NADH/NAD+. The open-circuit voltage realized, 1.1 V, is significantly higher than that obtained from a similar cell with a SnO2 electrode, and other cell characteristics are also improved, although degradation in the cell current with illumination time remains a potential problem if such cells were to be used in actual devices. In such a device, an alternative to the Hg/Hg2SO4 cathode (used here simply to permit evaluation of the photoanode reactions) would be employed. If efficient oxygen reduction were possible at the cathode (0.82 V vs NHE at pH 7),29 the combination with the TiO2 photoanode would in principle generate 1.4 V at pH 7. As was pointed out earlier in connection with the SnO2based cell,17 a photoelectrochemical biofuel cell can in principle produce more power than either a photoelectrochemical cell or a biofuel cell working individually. The hybrid cell can operate at a higher voltage than a conventional fuel cell with the same fuel and cathodic half-cell, because the anode potential is established by the sensitizer/semiconductor combination chosen. In addition, it can operate at a higher voltage than a dyesensitized photoelectrochemical cell using the same semiconductor and sensitizer, due to the fact that it is a nonregenerative cell with the ability to use a separate cathodic half cell. The cell described here can potentially avoid some of the energy-wasting charge recombination reactions found in dye-sensitized photoelectrochemical cells and can generate a higher current under some conditions.17 As the fuel employed is specific to the enzymatic catalyst chosen, sensor applications of the system can also be envisioned. Acknowledgment. This work was supported by a grant from the Matsushita Electric Industrial Co., Ltd. This is publication 594 from the Arizona State University Center for the Study of Early Events in Photosynthesis. LA048974I (29) Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solutions, 2nd ed.; National Association of Corrosion Engineers: Houston, TX, 1974.