Rechargeable PEM Fuel-Cell Batteries Using Quinones as Hydrogen Carriers

The proton conductor used was Sn_0.95Al_0.05P₂O₇ because this material possesses a proton conductivity of ca. 0.05 S cm⁻¹ in the temperature range from room temperature to 100°C, as reviewed elsewhere.^16–18 Sn_0.95Al_0.05P₂O₇ powder was synthesized according to a previously reported procedure.^19,20 Polytetrafluoroethylene (PTFE) powder (0.04 g) was added to 1.00 g of Sn_0.95Al_0.05P₂O₇ powder, kneaded using a mortar and pestle, and then cold-rolled to a thickness of 250 μm using a laboratory rolling mill.

The AQ/C fuel electrode was prepared as follows. 0.067 g of AQ (Aldrich) was dissolved in 30 mL of benzene at 70°C, followed by the addition of 0.133 g of carbon powder (Kansai Coke and Chemicals). The suspension was stirred until the benzene was completely evaporated. After drying at 120°C for 2 h, the solid product was dispersed with 0.5 g of 105% H₃PO₄ ionomer (Aldrich) using a mixer (Thinky AR-100) for 15 min. The obtained slurry was deposited on the surface of a carbon fiber paper (Toray TGP-H-090). The AQ loading was adjusted according to the electrode thickness; a maximum AQ loading of 3.68 mg was obtained at an electrode thickness of 0.4 mm. Two different air electrodes were examined: a commercially available Pt/C (40 wt% Pt, ElectroChem) and a physical mixture of 50 wt% RuO₂ (Aldrich) and carbon. After the addition of a small amount of 105% H₃PO₄ ionomer, the electrode slurries were coated on carbon fiber paper in a similar manner to that for the AQ/C electrode.

Transmission electron microscopy (TEM) and energy dispersive X-ray (EDX) images were recorded using a JEOL JEM2100HK instrument at an accelerating voltage of 200 kV with a beam current of 92 μA. X-ray diffraction (XRD) patterns were obtained using a Rigaku Miniflex II diffractometer operated at 45 kV and 20 mA with Cu Kα radiation (λ = 1.5432 Å). Nitrogen adsorption-desorption isotherms were measured using a Nihon Bell Belsorp-28SA gas analyzer after samples were degassed at 200°C. The micro- and mesopore volume distributions were determined using the standard micropore method (MP) and the Barrett-Joyner-Halenda (BJH) method, respectively. Brunauer-Emmett-Teller (BET) specific surface areas were calculated from adsorption data in the relative pressure range of 0.03 to 0.23. The CO₂ concentration in the outlet gas from the electrode was analyzed using online gas chromatography (GC; Varian CP-2002).

Cyclic voltammograms (CV) for the fuel electrode were obtained using a Hokuto Denko HZ-5000 galvanostat/potentiostat in three-electrode mode, where the fuel electrode, an Ag/AgCl reference electrode, and a Pt counter electrode were set in a 0.5 M H₂SO₄ solution. Charge-discharge measurements were performed between room temperature and 75°C as follows. An electrolyte membrane (13 mm diameter) was sandwiched between the fuel and air electrodes (0.5 cm²). The fuel electrode was attached to a stainless steel current collector and then sealed using thermal- and chemical-resistant PTFE tape (Nitto Denko). The air electrode was supplied with atmospheric air at a relative humidity of ca. 50% and a flow rate of 30 mL min⁻¹. The applied potential range between the two electrodes was 0–1.0 V, which was galvanostatically controlled using a Solartron SI 1260 impedance analyzer and Solartron 1287 electrochemical interface. The polarization resistance of each electrode was analyzed by measuring the impedance spectrum of a symmetrical cell consisting of AQ/C|Sn_0.95Al_0.05P₂O₇-PTFE|AQ/C, Pt/C|Sn_0.95Al_0.05P₂O₇-PTFE|Pt/C, or RuO₂/C|Sn_0.95Al_0.05P₂O₇-PTFE|RuO₂/C. The frequency range for the measurements was 0.1–10⁶ Hz, and the AC amplitude was 10 mV.

The distribution state of AQ on the carbon surface is an important contributing factor for acceleration of the AQ/AH₂Q redox reaction. The AQ particle size in the fuel electrode was analyzed and compared with that of a physical mixture of AQ and carbon with the same AQ content (33 wt%) as that of the fuel electrode [Figs. 1a and 1b]. While ca. 2 μm AQ particles were in contact with some carbon particles in the physical mixture, only aggregates of carbon-like particles was observed in the fuel electrode sample. The AQ particles were almost invisible, even in TEM images of the fuel electrode sample obtained at higher magnification (data not shown). The crystallinity of the two samples was investigated by XRD analysis. Figure 1c shows that the diffraction peak intensities of AQ in the fuel electrode sample were considerably lower than those of the physical mixture, which indicates that the AQ in the fuel electrode sample had a less crystalline structure or a nanocrystalline structure, in contrast to the AQ structure in the physical mixture. The porosity of the two samples was characterized using N₂ adsorption-desorption isotherm measurements. Both samples produced type I isotherms, which is characteristic of mainly microporous materials.²¹ However, Fig. 1d shows a large difference in the pore size distribution between the two samples. The micropore volume calculated by the MP method was 0.73 cm³ g⁻¹ for the fuel electrode sample, but 1.07 cm³ g⁻¹ for the physical mixture. In addition, the mesopore volume calculated using the BJH method was 0.45 cm³ g⁻¹ for the fuel electrode, but 0.58 cm³ g⁻¹ for the physical mixture. The BET specific surface areas were 1451 m² g⁻¹ for the fuel electrode sample and 1731 m² g⁻¹ for the physical mixture. Based on these results, it is likely that AQ molecules were deposited on the surface of micro- and mesopores in the carbon support of the fuel electrode.

Zoom In Zoom Out Reset image size

AQ/C|Sn_0.95Al_0.05P₂O₇-PTFE|Pt/C

The voltage window of RFCB operation was determined first from CV measurements. Figure 2a shows CVs collected at a scan rate of 50 mV s⁻¹ for the carbon-supported AQ (AQ/C) electrode and a carbon electrode without AQ. The carbon electrode exhibited pure capacitive behavior in the absence of AQ, as evident from the rectangular shape of the CV curve. In contrast, a large redox couple at E_1/2 = −0.14 V (vs. Ag/AgCl) was observed in the presence of AQ, where both the peak intensity and potential remained almost unchanged after 100 scans. This E_1/2 value is sensitive to the pH of the electrolyte solution used because AQ is in equilibrium with various intermediates (radical anions, dianions, and anions) or fully reduced AH₂Q, depending on the proton and AQ concentrations.²² At low pH (such as those of H₂SO₄ solutions), AQ coexists with AH₂Q according to the following two-electron and two-proton reaction:

where the oxidation reaction of water molecules and the reduction reaction of oxygen molecules are speculated to occur as the corresponding counter reactions at the Pt electrode. Reaction 1 generates an E_1/2 value of ca. −0.18 V (vs. Ag/AgCl) at pH = 0,²² which is comparable to the value observed in this study. This suggests that AQ is directly converted to AH₂Q without the use of hydrogen as a reductant, which is supported by the E_1/2 value, which was more positive than the potential for hydrogen reduction (−0.2 V or more negative), as shown in Fig. 2a. Thus, a charge voltage of 1.0 V was confirmed to be sufficiently large to charge AQ to AH₂Q.

Zoom In Zoom Out Reset image size

To determine whether the expected charge reaction occurs in the battery with the Sn_0.95Al_0.05P₂O₇-PTFE electrolyte, the cell was charged to 1.0 V and the circuit was then opened at room temperature. Transient changes in the cell voltage are presented in Fig. 2b. The open-circuit voltage (OCV) of the cell before charge was ca. 0.3 V; however, this value was not reproducible. (Most OCVs were between 0.2 and 0.4 V.) As a result of the charge, a plateau appeared at ca. 0.8 V, which is near the OCV value expected from the E_1/2 value shown in Fig. 2a. The OCV obtained may be attributable to the potential difference between the anodic (AH₂Q oxidation) and cathodic (O₂ reduction) reactions. On the other hand, preliminary experiments with the battery charged to 1.5 V showed a rapid decrease of the voltage from 1.5 V to an OCV of ca. 0.9 V, which is comparable to the OCV when charged to 1.0 V. This rapid decrease indicates that the hydrogen production reaction proceeds during charge at over 1.0 V; the produced hydrogen escapes from the fuel electrode through an aperture of the sealing tape.

The reversibility of the AQ/C fuel electrode was evaluated in the battery at room temperature. Constant current charge-discharge curves for the cell are presented in Fig. 2c. A charge-discharge cycle with a coulombic efficiency of 100%, within experimental error, was established, which indicates no loss of the electrical charge stored in the fuel electrode. The Pt/C air electrode can also reversibly oxidize water vapor and reduce oxygen:

However, no voltage plateau was observed in Fig. 2c during charge and discharge, which is related to both the low storage capacity of the AQ used and the large internal resistance of the cell. The resulting power density of the cell decreased with time: 16.8, 9.9, 6.2, and 2.9 mW cm⁻² at 1, 5, 10, and 20 s, respectively.

Figure 2d presents the influence of the amount of AQ in the fuel electrode on the discharge process of the battery at room temperature. A current density of 14 mA cm⁻² could be drawn from the cell, even when no AQ was present in the fuel electrode. This is not surprising, because an electrical double layer (EDL) is formed by charging the carbon, which yields an electrical capacity.²³ The discharge time of the cell was extended by the presence of AQ in the fuel electrode; this effect was increased with the amount of AQ, although it was observed only at low cell voltages. The effect was attributed to the hydrogen storage capacity based on the AQ/AH₂Q redox reaction. However, the electrical capacity estimated from the discharge curve was 35 mAh g⁻¹, which is only 13% of the theoretical value. Therefore, the performance of this cell could not be significantly improved merely by increasing the amount of AQ in the fuel electrode.

Not only the fuel electrode, but also the air electrode, influences the performance of the RFCB because the poor acid-resistant properties of Pt are important in PEM fuel cells.²⁴ It should be noted that both the Sn_0.95Al_0.05P₂O₇-PTFE electrolyte and the H₃PO₄ ionomer show high acidity. Figure 3a shows that the diffraction peak intensities of Pt were considerably decreased by acid treatment of the Pt/C electrode sample at 75°C for 3 h. TEM observation showed that agglomeration of Pt particles was more significant after the acid treatment than before (Supplementary Information, Fig. S1). Dissolution and re-deposition of Pt particles can easily occur by respective charging and discharging of the cell; therefore, the air electrode may already be degraded during the 1^st cycle, which would reduce the apparent capacity of the battery. Transition metal oxides such as RuO₂,²⁵ Mn₂O₃,²⁶ and IrO₂²⁷ are promising catalysts for the air electrode. In particular, the good acid-resistant properties of RuO₂ shown in Fig. 3b make this catalyst compatible with the Sn_0.95Al_0.05P₂O₇-PTFE electrolyte. Figure 3c shows a TEM micrograph and EDX elemental maps for the RuO₂/C electrode sample. The RuO₂ particles have an almost spherical to hexagonal geometry with diameters of ca. 100 nm. These particles appear to be mixed with the carbon particles in the submicron range. (Note that the EDX element map for C also shows the signal from the carbon microgrid.) For comparison, TEM and EDX results for the Pt/C electrode are given in Fig. S2 (Supplementary Information).

Zoom In Zoom Out Reset image size

AQ/C|Sn_0.95Al_0.05P₂O₇-PTFE|RuO₂/C

The reaction kinetics of the Pt/C and RuO₂/C electrodes were compared using room temperature impedance measurements and the results are presented in Fig. 4a. The intercept of the impedance line and the real axis, which corresponds to the ohmic resistance, was somewhat lower for Pt/C than for RuO₂/C, which reflects their structural or morphological differences; Pt/C is a cluster of Pt-loaded carbon particles and RuO₂/C is a physical mixture of RuO₂ and carbon particles. The depressed arcs for both electrode samples in the middle frequency range (20 Hz–16 kHz), which are associated with the charge-transfer resistance, were almost comparable. More importantly, the impedance curves were skewed in the low frequency range (0.1–20 Hz), which is due to the slow diffusion or adsorption of oxygen and/or water vapor.²⁸ Both the skew angle and degree for RuO₂/C were much smaller than those for Pt/C. It is reasonable to consider that this difference could be attributed to their stability in the acid medium [Figs. 3a and 3b]. Thus, the adsorption site density of the Pt particles is most likely severely decreased by corrosion in the acid medium.

Zoom In Zoom Out Reset image size

To evaluate the electrochemical properties of the RuO₂/C air electrode for the RFCB, a galvanostatic charge-discharge profile was measured with a current density of 14 mA cm⁻² at room temperature. Figure 4b shows that the charge voltage was much smaller with RuO₂/C than with Pt/C at each time, and vice versa for the discharge voltage, which enhanced the electrical capacity of the cell to 67 mAh g⁻¹ with the RuO₂/C electrode. However, this result does not mean an increase in the inherent capacity of the cell using the RuO₂/C electrode, but results from a lesser dependence of the current density on the capacity compared to that with the Pt/C electrode. This point will be discussed in the next paragraph.

The difference in the performance of the two air electrodes is further emphasized by the galvanostatic charge-discharge profiles measured at different current densities. The cell voltage with the Pt/C electrode quickly dropped to ca. 0.2 V and then slowly decreased with time, especially at low current densities [inset of Fig. 4c]. In contrast, the cell voltage with the RuO₂/C electrode decreased almost linearly with time at all the tested current densities [inset of Fig. 4d]. As a result, the electrical capacity of the cell with the RuO₂/C electrode was higher than that of the cell with the Pt/C electrode [Figs. 4c and 4d]. It should also be noted that this difference becomes larger as the current density increases. Therefore, the difference in capacity between the two cells is very small at a low current density of 3 mA cm⁻², where there is no significant current density dependence. This result can be explained by assuming that the inherent capacity of the two cells is determined by the hydrogen storage capacity of the AQ/C fuel electrode and that the current density dependence of the capacity is strongly influenced by the type of air electrode used. Subsequent experiments were conducted using the RuO₂/C air electrode.

AQ/C|Sn_0.95Al_0.05P₂O₇-PTFE|RuO₂/C

AQ was identified as close to the carbon surface in the fuel electrode; therefore, the storage capacity could be further enhanced by an increase in AQ utilization, i.e., the kinetics for the AQ/AH₂Q redox reaction. Operation of the RFCB at elevated temperatures increases the reactivity of AQ and AH₂Q molecules and promotes redox reactions. This also allows for low internal resistance of the fuel electrode and/or the air electrode, which improves cell performance. The battery was charged and then discharged at various temperatures. Figure 5a shows that the cell voltage increased gradually with the operating temperature at each time. However, the voltage drop within 10 s from the start of discharge was the most significant at 75°C, most likely due to a decrease in adhesion of the sealing tape at this temperature, which introduced a small quantity of air into the fuel electrode. Figure 5b shows that the electrical capacity also increased significantly with the operating temperature. A high capacity of 143 mAh g⁻¹, which corresponds to 56% of the theoretical value, was achieved at 75°C. The capacity then reached 224 mAh g⁻¹ at 100°C (data not shown); however, this should be noted as only additional information, because such a high temperature is not suited for RFCB applications.

Zoom In Zoom Out Reset image size

To elucidate the details of this process, AC impedance spectra were measured independently for the fuel and air electrodes between room temperature and 75°C, as shown in Figs. 5c and 5d, respectively. The results reveal three important features. First, the ohmic resistance is not largely dependent on the operating temperature, which reflects the small temperature dependence of proton conductivity for the Sn_0.95Al_0.05P₂O₇-PTFE electrolyte.²⁹ Second, the charge-transfer resistance was reduced by an increase in temperature, and was more pronounced for the fuel electrode than for the air electrode, which indicates that the improvement of cell performance shown in Figs. 5a and 5b was due mainly to the enhanced rate of the AQ/AH₂Q redox reaction. Third, the sharp rise in the low frequency range was not parallel to the imaginary axis, which indicates that these electrodes do not function as simple EDL electrodes.³⁰ More importantly, the skew degree was not significantly influenced by an increase of temperature, especially for AQ/C, which means that mass transport in the micropores of the carbon is sufficiently fast over the range of temperature tested. This can be explained by the unrestrictive diffusion of ionomer ions in the narrow micropores of the carbon; all of the micropore surfaces in this sample were wetted by the ionomer.

The present RFCB is a type of secondary battery. The energy density was thus estimated by integrating the area under the voltage-time curve during discharge, multiplying by the current, and then dividing by the masses of the AQ, AQ/C, and entire cell. The power density was obtained by dividing the energy density by the discharge time. Based on these data, a Ragone plot³¹ for the cell is presented in Fig. 6a. The energy densities (0.8–3.4 Wh kg⁻¹), which were normalized by the total cell weight, are lower than those for other batteries, although the corresponding power densities (9.5–258.9 W kg⁻¹) are comparable with those for lead-acid batteries, but lower than those for nickel-metal hydride batteries. The electrolyte occupies approximately 70–80% of the total cell weight; therefore, the synthesis of a more lightweight electrolyte membrane is a necessary requirement for practical applications. One promising approach to meet this requirement is to form a hybrid of Sn_0.95Al_0.05P₂O₇ and sulfonated polystyrene-b-poly(ethylene/butylene)-b-polystyrene (sSEBS) materials, the thickness of which could be controlled from 25 to 80 μm to provide a proton conductivity of approximately 0.01 S cm⁻¹ without humidification.³²

Zoom In Zoom Out Reset image size

The present RFCB is also regarded as a type of PEM fuel cell, although it does not require any fuel supply. For fuel cell applications, the specific power (W kg⁻¹) and power density (W cm⁻³) are important performance factors. Figure 6b shows the specific power of the cell as a function of power density, where the powers estimated above are normalized according to the total cell weight and volume. Both the specific power (9.5–258.9 W kg⁻¹) and power density (0.001–0.26 W cm⁻³) are within the wide range of PEM fuel cells and higher than those for direct methanol fuel cells. However, it should be noted that the slope of the current-voltage curves was strongly dependent on the current-sweep rate, as shown in Fig. 6c, due to the limited hydrogen storage capacity of the fuel electrode.

The cyclability of the battery between room temperature and 75°C was evaluated using galvanostatic charge-discharge measurements. In all tests, no significant overshoots beyond 1.0 V upon charge at 40 mA cm⁻² or under 0 V upon discharge at 40 mA cm⁻² were observed. Plots of electrical capacity retention and coulombic efficiency against cycle number are presented in Figs. 7a and 7b, respectively. Although coulombic efficiencies greater than 90% were obtained for the charge-discharge cycles at all the tested temperatures, the cyclability was sensitive to the operation temperature. At room temperature, the capacity was either unchanged or very slightly increased with cycle number. At 50°C, the capacity remained at 91% of the initial value after 300 cycles. In contrast, at 75°C, the capacity continuously decreased as the cycle number increased. Nevertheless, ca. 63% of the initial performance was retained after 300 cycles, which has not been achieved using other organic chemical hydrides as hydrogen storage media.

Zoom In Zoom Out Reset image size

Figure 7c shows that the electrochemical reaction region (100 Hz–4.0 kHz) and the diffusion region (0.1–100 Hz) were enlarged compared to the ohmic region (4.0 kHz–10³ kHz) after 300 cycles at 75°C. This was due to degradation of the air electrode rather than the fuel electrode, because the discharge time recovered to approximately 100% of the initial value by replacing the air electrode with a new one in the test cell, as shown in Fig. 7d. Electrode degradation could be attributed to corrosion of the catalytically active nanoparticles or corrosion of the support. However, the agglomeration of RuO₂ particles was not observed at 75°C [Fig. 3b]; therefore, this effect may be excluded in the present case. In contrast, carbon corrosion is a known phenomenon in PEM fuel cells and electrolyzer cells, caused by the oxidation of carbon to CO₂ by water vapor:^33,34

This was confirmed by analysis of the outlet gas from the anodically polarized air electrode. The CO₂ concentration increased with the anodic current density at 75°C [Fig. 7e], although the linearity is poor, most likely due to the adsorption of CO₂ on the carbon surface. Semiconductive metal oxides, such as SnO₂³⁵ and TiO₂,³⁶ have been investigated as alternative non-carbon supports for PEM fuel cells. In addition, mesoporous SnO₂^37,38 and TiO₂³⁹ support materials with high specific surface areas have been successfully synthesized. However, further research is required to evaluate the effectiveness of such alternative support materials.