High-Power Abiotic Direct Glucose Fuel Cell Using a Gold-Platinum Bimetallic Anode Catalyst.

We developed a high-power abiotic direct glucose fuel cell system using a Au-Pt bimetallic anode catalyst. The high power generation (95.7 mW cm-2) was attained by optimizing operating conditions such as the composition of a bimetallic anode catalyst, loading amount of the metal catalyst on a carbon support, ionomer/carbon weight ratio when the catalyst was applied to the anode, glucose and KOH concentrations in the fuel solution, and operating temperature and flow rate of the fuel solution. It was found that poly(N-vinyl-2-pyrrolidone)-stabilized Au80Pt20 nanoparticles (mean diameter 1.5 nm) on a carbon (Ketjen Black 600) support function as a highly active anode catalyst for the glucose electrooxidation. The ionomer/carbon weight ratio also greatly affects the cell properties, which was found to be optimal at 0.2. As for the glucose concentration, a maximum cell power was derived at 0.4-0.6 mol dm-3. A high KOH concentration (4.0 mol dm-3) was preferable for deriving the maximum power. The cell power increased with the increasing flow rate of the glucose solution up to 50 cm3 min-1 and leveled off thereafter. At the optimal condition, the maximum power density and corresponding cell voltage of 58.2 mW cm-2 (0.36 V) and 95.7 mW cm-2 (0.34 V) were recorded at 298 and 328 K, respectively.

Introduction

Currently, there is an urgent demand on the circumvention from global warming.[1] For this purpose, a rapid change of energy sources from fossil fuels to clean and sustainable ones is indispensable. Glucose is one of the most abundantly available clean fuels, derivable by degradation of polysaccharides, for example, starch and cellulose.[2] Moreover, glucose is a safe (nonhazardous, nonvolatile, and biocompatible) energy source bearing a large inherent chemical energy, which can be released with the oxidation (2.87 MJ mol–1 or 15.9 kJ g–1 for the complete oxidation to carbon dioxide and water).[3]

Direct glucose fuel cell (DGFC) operates with the oxidation of glucose at the anode and the reduction of oxygen at the cathode.[4]

The DGFC can be classified into three categories:[5] enzymatic,[6−12] microbial,[13−15] and abiotic systems.[16−23] Each system has advantages. For instance, enzymatic and microbial systems have a high specificity for the substrate, which enables to construct separator-free fuel cell systems. Abiotic systems can work at harsh pH and temperature conditions. However, there is a severe issue in common with all these systems, that is, low current density (<10 mW cm–2 in most cases), which impedes wide applications of DGFC.[24,25] As such, their current application targets are limited to a power supply for implantable medical devices[3,26−28] and mobile electronic devices.[29] The low current density is partly due to a partial oxidation of glucose, that is, at conventional operating conditions only two electrons can be derived with the oxidation of glucose to gluconic acid in place of 24 electrons for the complete oxidation to CO2 and H2O,[3] but presumably it is not the all.

When confined to abiotic DGFC systems using metal catalysts, the cell performance can be affected by the following factors: (i) nature of the supported metal catalyst, (ii) concentration and pH of the glucose fuel solution, and (iii) operating condition of the fuel cell. The nature of the supported metal catalyst involves the size, shape, and composition of the metal nanoparticles, nature of the support, nature of the stabilizer if used, and dispersity of the metal catalyst on the support surface. Besides, in many cases, conductive polymers called ionomers are employed for the electric contact between the supported catalyst and the electrode (in particular, the anode), which can also influence the cell properties. The operating condition of the fuel cell involves the flow rate and operation temperature. For generation of high power density, all these parameters should be optimized.

To improve the power of the DGFC system, we previously reported that the concentration of a stabilizer for protecting metal nanoparticles from aggregation is crucial for their electrocatalytic properties.[30] There is an optimal stabilizer concentration, below which the current density decreased because of aggregation of the catalyst, whereas above which the current density also decreased because of that the stabilizer in excess coats the catalyst surface and impedes the approach of fuel molecules.

Regarding the anode catalyst, we focused our attention to the Au–Pt system among many bimetallic systems.[31−37] This is due to excellent electrocatalytic properties of monometallic Au and Pt in terms of high power density and durability. Several studies on the Au–Pt bimetallic system have been reported. Comotti et al. studied the catalytic activity of nonsupported and carbon-supported Au–Pt bimetallic nanoparticles and found the highest catalytic activity at Au/Pt = 2:1 (d = 2.8 nm from X-ray diffraction) for both systems.[31] Zhang and Toshima synthesized poly(vinylpyrrolidone) (PVP)-stabilized Au–Pt bimetallic nanoparticles with different compositions by alcohol or NaBH4 reduction.[32] They investigated the catalytic properties of the unsupported bimetallic nanoparticles and found the highest activity for glucose oxidation per total metal at Au80Pt20 (d = 1.6 nm) and that the bimetallic nanoparticles have a Pt-rich shell structure. They also reported that the Au–Pt bimetallic nanoparticles with a crown-jewel structure exhibit an extraordinary high catalytic activity for glucose oxidation.[33] Regarding the fuel cell properties utilizing glucose oxidation, Basu and Basu studied Pt–Au, Pt–Ru, and Pt–Bi bimetallic systems and concluded that the Pt–Au system provides the highest power and stability.[34] They reported 0.9 V and 0.72 mW cm–2 for the open-circuit voltage (OCV) and the maximum power density, respectively, as the fuel performance using Au–Pt supported on Vulcan XC-72 carbon as the anode catalyst and charcoal as the cathode working with 0.2 M glucose in 1 M KOH solution.[35] Habrioux et al. synthesized Au–Pt bimetallic nanoparticles in microemulsion systems and supported also on Vulcan XC-72 carbon to investigate fuel cell properties.[36] They found the highest electrocatalytic activity for Au80Pt20 (d = 5.02 ± 0.56 nm) working with a biocathode (ABTS + laccase) in phosphate buffer (pH 5.0) at 37 °C. Their maximum peak density of the fuel cell reported is 16 μW cm–2 at 0.16 V and 100 μA cm–2.[37,38]

Herein, we first revisit the optimal bimetallic composition of the Au–Pt system by cyclic voltammetry (CV), followed by the optimization of other conditions, that is, materials’ conditions (metal loading, ionomer, glucose, and base concentrations) and operating conditions (flow rate of the fuel solution and operating temperature). Consequently, we are successful to produce a high current density (∼100 mW cm–2) which has been rarely attained for the DGFC system in previous studies.

Results and Discussion

Charge Flow in the Present DGFC

In this DGFC system, a pair of graphite plates were used as an anode and a cathode, and a monovalent cation permeable membrane (CPM) was employed as a separator between the two electrodes. The flow of electronic and ionic charges is illustrated in Figure . On the anode, glucose molecules are oxidized to gluconolactone, and it further undergoes hydrolysis in solution to gluconic acid. During this process, two electrons are liberated and migrate to the cathode through the external circuit. At the same time, two protons are generated, but they rapidly dissipate by reacting with hydroxyl ions to yield water molecules because the fuel solution contains a high concentration of KOH. Accordingly, few protons migrate through the CPM, whereas potassium cations and electrically neutral water molecules can migrate though the CPM toward the cathode because a concentrated KOH solution was used as an electrolyte. On the other hand, on the cathode, atmospheric oxygen molecules are reduced by the electrons and combining with water to yield hydroxyl anions.

Figure 1

Flow of electronic and ionic charges in this DGFC system.

Composition of the Bimetallic Anode Catalyst

In general, the catalytic properties of bimetallic nanoparticles are influenced by geometric factors (surface area[38] and crystal facet[39,40]) and electronic factors (surface metal composition).[41] Therefore, the size and composition of synthesized nanoparticles were analyzed with transmission electron microscopy (TEM) and energy-dispersive spectrometry (EDS). As shown in the TEM images of Figure , before depositing on the carbon support, most nanoparticles were individually dispersed without forming aggregates for all Au–Pt bimetallic compositions. Interestingly, we found a trend that the mean diameter decreases with the increasing Au fraction in the Au–Pt bimetallic systems from 2.7 to 1.4 nm (Figure S1), although the total metal concentration (0.2 mmol dm–3) and the reductant concentration (2.0 mmol dm–3) were identical in all systems. This can be attributed to a difference in the surface energy of the two metal elements,[42] which affects the aggregation between the nuclei of nanoparticles and the adsorption of stabilizer molecules. If we calculate the fraction of surface atoms being responsible for the catalytic reaction, it leads to a large variation from 64 to 92%, as listed in Table . This variation should be taken into account for their electrocatalytic properties. When the PVP-stabilized metal nanoparticles were deposited on the carbon support, the particle size was maintained. However, the deposition was not homogeneous, that is, there were some variations in the number density of deposited nanoparticles depending on the place of the carbon support (Figure S2). This is probably related to a strong hydrophobic attraction force between the metal nanoparticles and carbon support. Accordingly, in addition to the effect of surface area for different metal compositions, there may also be a substantial effect of heterogeneous deposition of the catalyst on the support. By contrast, no significant variation was found in the bimetallic composition depending on the particle size, as proved from EDS point analysis (Figure S3). Moreover, the EDS line analysis for individual nanoparticles showed no significant difference in the signal intensity at the central part and the peripheral part of a given nanoparticle, verifying homogeneous composition (Figure S4). Although there is a difference in the redox potential between Au3+/Au0 (E0 = 1.00 V) and Pt4+/Pt0 (0.71 V),[41] it does not necessarily lead to phase separation. Either the alloy formation or the phase separation depends on the power of a reducing agent.[43−49] Because we employed here sodium borohydride (NaBH4) bearing a strong reducing power, reduction of the two metal ions (Pt4+ and Au3+) to the metallic state was completed prior to the electron transfer from Pt(0) or Pt2+ to Au3+ or Au+, thus Au–Pt alloy nanoparticles were produced.

Figure 2

TEM images of Au–Pt bimetallic nanoparticles with different compositions. (a) Pt100, (b) Au10Pt90, (c) Au20Pt80, (d) Au30Pt70, (e) Au40Pt60, (f) Au50Pt50, (g) Au60Pt40, (h) Au70Pt30, (i) Au80Pt20, (j) Au90Pt10, and (k) Au100 (in molar ratio). Scale bar = 20 nm.

Table 1

Physical Characteristics of Au–Pt Bimetallic Nanoparticles

	composition	mean diameter (nm)	standard deviation (nm)	surface atom (%)
a	Pt100	2.72	0.80	64.0
b	Au10Pt90	2.55	1.04	67.0
c	Au20Pt80	2.77	0.77	63.5
d	Au30Pt70	2.00	0.74	78.1
e	Au40Pt60	2.11	0.76	75.9
f	Au50Pt50	2.13	0.80	75.7
g	Au60Pt40	1.72	0.51	84.9
h	Au70Pt30	1.42	0.42	91.9
i	Au80Pt20	1.50	0.54	90.2
j	Au90Pt10	1.57	0.57	88.8
k	Au100	1.59	0.62	88.4

Regarding the surface metal composition, UV–vis extinction spectra provide some information. Figure presents the UV–visible extinction spectra of the Au–Pt bimetallic nanoparticles dispersed in water. The pure Au nanoparticles showed a shoulder at 520 nm, which can be attributed to a localized surface plasmon resonance (LSPR) band. Whereas for the pure Pt nanoparticles, the extinction spectra monotonically increased toward shorter wavelengths without a distinct peak.[50] Note that the remarkable increase in the extinction below 240 nm is ascribed to PVP. In the Au–Pt bimetallic systems, the extinction spectra continuously evolved between those of the two monometallic systems. However, unlike the mixture of the nanoparticle dispersions of the two metal elements (not shown), the present samples showed a strong attenuation of the LSPR band of Au even at a small Pt content (e.g., Au90Pt10), which suggests the presence of Pt atoms at the surface of all nanoparticles.[43]

Figure 3

UV–visible extinction spectra of aqueous Au–Pt bimetallic nanoparticle dispersions. Total metal conc. = 0.2 mmol dm–3. PVP (0.2 mmol dm–3) is present as a stabilizer.

In the following, the electrocatalytic properties of the Au–Pt nanoparticles for glucose oxidation were investigated from CV. Figure depicts CV profiles associated with glucose electrooxidation at Au–Pt catalyst systems of different compositions. For the pure Pt system (a), in the reference (0.2 mol dm–3 KOH without glucose), a single peak was observed each for anodic and cathodic scans. The oxidation peak at −0.2 V (vs Ag|AgCl) and the reduction peak at −0.3 V correspond to the oxidation of Pt to Pt–OH and the reduction of Pt–OH to Pt, respectively.[51,52] In the presence of 0.1 mol dm–3 glucose, two oxidation peaks were observed with the anodic scan, a small hump at −0.2 V and a more distinct peak at 0.15 V. These peaks are associated with the oxidation of glucose on the oxidized Pt surface.[51,52] Although it is not apparent here, another oxidation peak could have been observed at a lower potential (∼−0.7 V) because of dehydrogenative adsorption of glucose on the Pt surface[51,52]

Figure 4

Cyclic voltammograms for glucose electrooxidation with (a) Pt, (b) Au, (c) Au50Pt50, (d) Au70Pt30, (e) Au80Pt20, and (f) Au90Pt10 bimetallic catalysts. [KOH] = 0.2 mol dm–3, scan rate = 20 mV s–1.

Because this reaction takes place at low potential where the Pt surface is not oxidized, the hydrogen atom dissociated from the glucose molecule is adsorbed on the Pt surface. However, the hydrogen atom is immediately converted to a proton with the release of an electron and combines with OH– in the solution[51,52]

On the other hand, the dehydrogenated glucose is oxidized to yield gluconate, which adsorbs to the Pt or Pt–O surface, by forming a monodendate or a bidentate intermediate, depending on the potential[52]

These intermediates readily desorb as δ-gluconolactone and undergo hydrolysis in the solution to yield gluconate.

At high potential, the adsorbed dehydrogenated glucose is directly oxidized to δ-gluconolactone[42]

The δ-gluconolactone readily desorbs from the Pt surface and undergoes hydrolysis to yield gluconate

During the cathodic scan, a weak cathodic peak was observed at −0.20 V, which can be assigned to the reduction of δ-gluconolactone to glucose.[52]

In the pure Au system (b), in the absence of glucose, an oxidation peak of Au and a reduction of Au–OH were observed at 0.2 and 0.1 V for anodic and cathodic scans, respectively.[53−56] In the presence of glucose, a remarkable increase of the current with glucose adsorption was set in at −0.4 V on the anodic scan, which continued up to −0.2 V where a small shoulder was observed. After that, the current further increased, and after passing a peak at 0.2 V, it decreased. The two peaks can be ascribed to the dehydrogenative adsorption of glucose on the Au–OH surface and oxidation to gluconolactone, respectively.[53−56] While here in the course of the cathodic scan, a prominent oxidation peak was observed at 0.2 V. This potential agrees with that of the oxidation peak in the anodic scan in the absence of glucose. Accordingly, this peak can be assigned to the oxidation of glucose, which simultaneously proceeds with the reduction of Au=O to Au–OH.[53−56] Another interpretation for the anodic peak during the cathodic scan is the oxidation of glucose by a reactive oxygen species (O2–) generated with the reduction of the Au–OH surface.[53]

CV profiles of Au–Pt bimetallic systems were compared in (c–f). In the absence of glucose, a very weak peak or shoulder was observed in the range of −0.1 to 0.0 V, an intermediate potential between the oxidation potentials of pure Pt and pure Au. While regarding the reduction peak, two very broad and weak peaks are discernible at around 0.3 and −0.4 V. On the other hand, in the presence of glucose, in all bimetallic compositions except Au80Pt20, a single peak was observed at 0.2 V. Meanwhile, in Au80Pt20, three peaks were observed at −0.25, 0.20, and 0.35 V. The apparent difference in the shape of CV profiles is reminiscent of the effect of the crystal surface of Au electrodes.[40] Namely, the CV profile for Au80Pt20 and slightly Au100 resembles well that for the glucose electrooxidation on the Au{200} surface and the rest is similar to the CV profile on the Au{111} surface.[40] In addition, both the peak potential and the relative values of current density are in good agreement with the literature.[40] Accordingly, one of the reasons for a large glucose oxidation current on Au80Pt20 might be exposure of highly active {200} surfaces. On the other hand, during the cathodic scan, a sharp oxidation peak was observed at 0.20 V in all bimetallic compositions, similar to the pure Au system. The increasing peak intensity is consistent with the Au content in the bimetallic systems.

Figure depicts the net current density by glucose oxidation in Au–Pt bimetallic systems, where the net current density was calculated by subtracting the current density in the absence of glucose from that in the presence of glucose. The onset of the current increase is accompanied with the dehydrogenative adsorption of glucose molecules on the metal surface. As shown, the onset potential is remarkably dependent on the bimetallic composition. In particular, the onset potentials at Au80Pt20 are lower than those at other compositions. The lower onset potential implies the smaller overpotential required for the glucose oxidation, thus the greater cell voltage can be expected in the fuel cell. Besides, the Au80Pt20 system provides a significantly large oxidation peak current compared with other bimetallic compositions. Accordingly, the Au80Pt20 system is the optimal composition for serving as the anode of the DGFC system, thus we used this bimetallic catalyst hereafter.

Figure 5

Net current density by glucose oxidation on Au–Pt bimetallic catalysts with different compositions. [Glucose] = 0.1 mol dm–3, [KOH] = 0.2 mol dm–3, scan rate = 20 mV s–1.

Ionomer/Carbon Weight Ratio

In liquid fuel cell systems, conductive polymers called ionomers are often used for making an electric contact between the catalyst support and the electrode.[57] In this case, it is readily inferred that the cell properties are dependent on the concentration of the ionomer. Figure shows the CV profiles for different ionomer (Nafion) concentrations. Virtually, the cell power is strongly dependent on the ionomer concentration. The maximum power density was recorded at the ionomer/carbon weight ratio (referred to as I/C ratio) being 0.2. A too low concentration of the ionomer makes it difficult to form a sufficient electric contact between the catalyst support and the electrode, whereas a too high concentration of the ionomer might cover the entire surface of the metal catalyst on the support and impedes the approach of the reaction substrate to the catalyst. Therefore, the I/C ratio was fixed hereafter at 0.2.

Figure 6

Effect of the ionomer/carbon ratio on the I–W characteristics of DGFC. [Glucose] = 0.4 mol dm–3, [KOH] = 4.0 mol dm–3, flow rate of glucose solution = 10 cm3 min–1.

Loading Amount of the Anode Catalyst

Figure shows the variation of cell power with a loading amount of the anode catalyst. As can be seen, the cell power increased with an increase in the loading amount of the catalyst up to 2.0 mg cm–2, but it decreased at higher loadings. Visually, at higher loadings, the catalyst ink stacked on the surface of the carbon cloth anode and the catalyst surface was less exposed to the contact with fuel molecules. Accordingly, all experiments were performed hereafter at 2.0 mg cm–2 of the anode catalyst loaded on the anode.

Figure 7

Variation of the power of DGFC with a loading amount of the anode catalyst. [Glucose] = 0.4 mol dm–3, [KOH] = 4.0 mol dm–3, flow rate of glucose solution = 10 cm3 min–1.

Glucose Concentration in the Fuel Solution

Subsequently, the effect of glucose concentration on the cell power was investigated. Figure a shows that the cell power increased with the increasing glucose concentration up to 0.4–0.6 mol dm–3, whereas at higher fuel concentrations, the cell power decreased significantly. On the other hand, the I–V profiles (Figure b) indicate that the cell voltage hardly changed with the glucose concentration up to 1.0 mol dm–3 and decreased at higher concentrations. The reason for the decrease would be related to the decrease in the diffusion rate of glucose fuel because of increased viscosity of the solution. On the other hand, Figure c presents the evolution of OCV and IR stack. The OCV was constant or very slightly increased with the increasing glucose concentration up to 0.6 mol dm–3, and then it decreased at higher glucose concentrations. Meanwhile, the IR stack was almost constant up to 0.6 mol dm–3 and then increased with the increasing glucose concentration. These results suggest that up to 0.6 mol dm–3 of glucose, adsorption of the fuel molecules and desorption of the product from the catalyst surface efficiently occurred, but above this concentration, the number of glucose molecules in excess increases progressively within the electric double layer of the anode, which impedes a rapid catalytic reaction cycle.

Figure 8

Effect of glucose concentration on (a) I–W characteristics, (b) I–V characteristics, and (c) OCV and IR stack of DGFC at 298 K. [KOH] = 4.0 mol dm–3, flow rate of glucose solution = 10 cm3 min–1.

Base Concentration in the Fuel Solution

Subsequently, the effect of base (KOH) concentration on the cell performance was studied. Here, the role of KOH is threefold: it is used as the supporting electrolyte, to oxidize the metal surface, and to induce the tautomeric change of glucose from a ring form to the chain form, which is more vulnerable to oxidation. Importantly, the evolutions of OCV and IR stack with the KOH concentration showed an opposite trend to those with the glucose concentration. As shown in Figure a, the power density and the cell voltage increased with the increasing KOH concentration up to 4.0 mol dm–3 and then leveled off. At 8.0 mol dm–3 KOH, a decrease in the potential at the peak current density was observed. Also, in the I–V characteristics (Figure b), the decrease in the cell voltage was more effectively suppressed with the increasing KOH concentration up to 4.0 mol dm–3. The OCV increased up to 1.05 V when the KOH concentration was increased to 4.0 mol dm–3 and then leveled off at higher concentrations (Figure c). By contrast, the IR stack steeply decreased with the increasing KOH concentration up to 4 mol dm–3 and leveled off thereafter. With the increasing KOH concentration, the surface of the anode metal catalyst becomes more vulnerable to the oxidation and dehydrogenative adsorption of glucose molecules is activated, thus the power density and OCV increased. However, the increase of OH– ions promotes combination with H+, which accounts for the increase in OCV and decrease in the resistance in IR stack. The increase in OCV with the increasing KOH concentration is the opposite trend to the previously reported results using an anion exchange membrane as the separator.[17]

Figure 9

Effect of KOH concentration on (a) I–W characteristics, (b) I–V characteristics, and (c) OCV and IR stack of the DGFC at 298 K. [Glucose] = 0.4 mol dm–3, feed rate of glucose solution = 10 cm3 min–1.

Temperature Dependence

Figure depicts the temperature dependence on the I–W and I–V characteristics. The cell power increased when the temperature was elevated from 298 to 318 K, but the power did not increase further at higher temperature, rather it slightly decreased at 348 K. This can be reasonably explained by spontaneous oxidation of glucose molecules in the strongly basic solution prior to reaching the catalyst surface, which becomes more remarkable at high temperature. On the other hand, the drop of cell voltage with the increasing current density was less remarkable at lower temperatures but not detrimental even at 328 K.

Figure 10

Temperature effect on (a) I–W and (b) I–V characteristics of the DGFC. [Glucose] = 0.4 mol dm–3, [KOH] = 4.0 mol dm–3, flow rate = 10 cm3 min–1.

Effect of the Flow Rate

The effect of the flow rate of glucose fuel solution on the cell properties was investigated at 298 and 328 K. As shown in Figure , the larger current density was observed at 328 K at a given flow rate, implying the larger reaction rates for both glucose oxidation. Moreover, at each temperature, the cell voltage increased with the increasing flow rate up to 50 cm3 min–1, whereas it leveled off at higher flow rates. This implies a transition from the diffusion-limited current to the reaction-limited one. At the low flow rate regime (≤50 cm3 min–1), the current is determined by the diffusion of fuel solution to the catalyst surface, whereas at the high-rate regime (>50 cm3 min–1), the current is governed by the reaction rate. The maximum power density and corresponding cell voltage were 58.2 mW cm–2 (0.36 V) and 95.7 mW cm–2 (0.34 V) at 298 and 328 K, respectively.

Figure 11

Effect of the flow rate of glucose fuel solution on (a,b) I–W and (c,d) I–V characteristics of DGFC at (a,c) 298 K and (b,d) 328 K. [Glucose] = 0.4 mol dm–3, [KOH] = 4.0 mol dm–3.

Stability of the Power

Finally, stability of the power density of this DGFC was investigated at the optimal condition determined above. As shown in Figure a, under the constant voltage of 0.37 V, the initial power density at 328 K was 89.7 mW cm–2. After 28 min, the power density dropped by 8.0% to 82.5 mW cm–2. At 298 K (Figure b), the initial power density was 64.2 mW cm–2, which dropped by 27.4% to 46.6 mW cm–2 after 180 min. A plausible reason for the deactivation is adsorption of reaction intermediates and byproducts on the catalyst surface.

Figure 12

Stability of the power density at the constant voltage of 0.37 V at (a) 328 K and (b) 298 K. [Glucose] = 0.4 mol dm–3, [KOH] = 4.0 mol dm–3, flow rate = 50 cm3 min–1.

Conclusions

We demonstrated here that a high power density (maximum 95.7 mW cm–2) can be generated with DGFC by optimizing the following seven parameters: composition of the Au–Pt bimetallic anode catalyst, ionomer/carbon weight ratio, loading amount of the catalyst on the anode, glucose and KOH concentrations in the fuel solution, and flow rate and operation temperature. It was found that among different bimetallic compositions, the Au80Pt20 system shows the highest catalytic activity for the glucose oxidation. Optimal conditions were found at I/C = 0.2 (in weight ratio), 0.4 mol dm–3 glucose, 4.0 mol dm–3 KOH, and 50 cm3 min–1 flow rate of fuel solution. Regarding the I/C ratio, too low concentration of the ionomer is ineffective for making the electric contact between the metal catalyst and the anode, whereas a too high concentration of the ionomer covers the catalyst surface and impedes the approach of the fuel molecules to the catalyst surface. A too high concentration of glucose remarkably increases the solution viscosity, thereby reducing the diffusion of the fuel to the catalyst surface. Although the increase in the base concentration promotes the transformation of glucose molecules from the ring to chain conformation, too high a base concentration induces the spontaneous oxidation of glucose in the solution, which impedes the electrooxidation on the catalyst surface. The cell power is also affected by the flow rate of the glucose fuel solution. It increased with an increase in the flow rate up to 50 cm3 min–1 but stagnated at higher flow rates, implying the transition from the diffusion-determining current to reaction-determining one. At the condition where all these parameters were optimized, the maximum power densities and corresponding cell voltage of 58.2 mW cm–2 (0.36 V) and 95.7 mW cm–2 (0.34 V) were recorded at 298 and 328 K, respectively.

Experimental Section

Materials

HAuCl4·4H2O, H2PtCl6·6H2O, NaBH4, ethanol, 1-propanol, and β-d-glucose were of reagent grade from Wako and used without further purification. PVP (Mw = 50 000) and Nafion solution (5 wt % in lower aliphatic alcohol/water) were provided from Sigma-Aldrich. Ketjen Black 600 (KB600) was purchased from Lion. Water was deionized with a Milli-Q system (18.2 mΩ cm in resistivity).

Preparation of the Anode Catalyst

The anode catalyst used here was Au–Pt bimetallic nanoparticles prepared by the chemical reduction of mixed HAuCl4/H2PtCl6 aqueous solutions in presence of a stabilizer (PVP). For instance, Au80Pt20 nanoparticles were prepared by mixing 8.0 cm3 of HAuCl4 (2 mmol dm–3), 2.0 cm3 of H2PtCl6 (2 mmol dm–3), 20 cm3 of PVP (10 mmol dm–3 in repeating unit concentration), and 68 cm3 of water in a 150 cm3 vial and stirred for 15 min. Subsequently, to this solution, 2 cm3 of NaBH4 (100 mmol dm–3) was quickly added under vigorous stirring to reduce metal ions. Separately, a calculated amount of carbon powder (KB600) was placed in a 200 cm3 vial. After adding 50 cm3 of water, it was sonicated for 1 h to obtain homogeneous dispersion. Then these two dispersions were mixed together and vigorously stirred for 12 h to attain an adsorption/desorption equilibrium. Finally, the carbon-supported metal catalysts were filtrated, washed with copious amount of water, and dried at 60 °C for 24 h in vacuum. The metal catalysts were observed with TEM (Hitachi H-7650 at 120 kV and JEOL JEM-2100F at 200 kV). EDS analysis was performed with a JEOL JED-2300T using Au Lα line (9.441 keV) and Pt Lα line (9.712 keV). Particle size distributions were determined from about 2000 particles using ImageJ program.

Cyclic Voltammetry

The electrocatalytic properties of Au–Pt bimetallic nanoparticles for glucose oxidation were evaluated from CV. The samples were prepared as follows. In a plastic centrifugation tube, 2 mg of the carbon–supported metal catalysts (5 wt % metal/carbon) was suspended in 2 cm3 of mixed ethanol/water (1/1; v/v), then it was centrifuged, and the supernatant was discarded. The precipitate was mixed with 8 mm3 of 5 wt % Nafion in 1-propanol/water (7/3; v/v) and sonicated for 20 min to obtain the catalyst ink. Then, a 1.5 mm3 aliquot of the ink was placed on a Teflon-lined glassy carbon disk electrode (Toyo Corporation G0229, 2 mm ϕ). After drying at 393 K for 15 min, it was used as a working electrode. A Pt wire (Toyo G0228) was used as a counter electrode, and an Ag|AgCl electrode with saturated KCl (Toyo K0265) was employed as a reference electrode. These three electrodes were immersed in 10 cm3 of 0.2 mol dm–3 KOH solution in presence or absence of 0.1 mol dm–3 glucose. The CV profile was recorded in the range between −0.8 and +0.8 V versus Ag|AgCl (saturated KCl) at the scan rate of 20 mV s–1 on a Hokuto Denko HZ5000 potentio-galvanostat.

Fuel Cell Setup and Operation

The DGFC used here was a CHEMIX DFC-010-01 comprising a cation exchange membrane, two electrodes, two silicon gaskets, two graphite separators, two current collectors made of Au-plated Cu, and two stainless plates, located from the center to both ends in this order. The anode was carbon cloth (Fuel Cell Earth CCP40, 3.2 × 3.2 cm2) on which the catalyst ink was deposited. Here, the metal concentration in the catalyst ink was 60 wt % with respect to the carbon support. The cathode used here was a commercial Pt/C paper (Toray EC-E20-10-07). These two electrodes were separated from each other with a monovalent CPM (Astom CIMS, 0.15 mm thick). On the graphite separators, a serpentine flow channel was graved to provide the glucose fuel solution or oxygen gas. Concentrations of glucose and KOH were 0.4 and 4.0 mol dm–3, respectively, unless otherwise noticed. The flow rates of glucose fuel solution and air were 10 cm3 min–1 and 2.0 dm3 min–1, respectively. The current–power (I–W) and the current–voltage (I–V) properties of the glucose fuel cell were conducted with a Toyo PEMTest8900DM system. The cell properties were measured at 298 K unless otherwise noticed.