Introduction

Recently, carbon capture and utilization (CCU) methods have received significant levels of attention as technologies for effectively removing and utilizing atmospheric CO21,2. These technologies are attractive approaches as they treat CO2 as an unused resource and convert it into value-added chemicals and fuels. Among them, electroreduction is a promising technique, and C1, C2, and C2+ products have reportedly been obtained using various electrocatalysts through different CO2 reduction mechanisms3,4,5,6,7,8,9,10,11,12,13. In particular, CO production at Au and Ag, and hydrocarbon production at Cu, have been successfully elucidated to follow multistep proton-coupled electron-transfer pathways14,15. These reactions exhibit relatively high faradaic efficiencies; however, they require large overpotentials, which is disadvantageous. Consequently, their energy-conversion efficiencies are low, despite their high faradaic efficiencies.

Developing methods that ensure that the CO2-electroreduction reaction occurs with a small overpotential and a high energy-conversion efficiency is important. In this regard, platinum group metals have the potential to realize overpotential-free CO2 reductions. There are many reports in which CO is adsorbed on a metal (COads) at a positive potential rather than its theoretical potential16,17. However, the further reduction of COads is difficult because CO is strongly adsorbed to the metal through a donation–back-donation mechanism (Blyholder mechanism)18. To the best of our knowledge, the main product is H2 when a Pt electrocatalyst is used, even when it is negatively polarized19,20,21. We previously obtained the C1 compound by reducing CO2 near the theoretical potential in a proton-exchange-type membrane electrode assembly (MEA) with a carbon-supported platinum (Pt/C) electrocatalyst22. The use of a proton-exchange membrane and an ionomer was suggested to facilitate CO2 reduction; however, the C1 yield was quite low. We recently demonstrated that CH4 can be produced by the reduction of CO2 in the absence of an overpotential and with a faradaic efficiency of 6.8% using the MEA23. This CH4-generation reaction proceeds by a Langmuir–Hinshelwood (L–H) mechanism associated with COads and H adsorbed on the metal (Hads):

$${\text{CO}}_{{{\text{ads}}}} + {\text{ 6 H}}_{{{\text{ads}}}} \to {\text{CH}}_{{4}} + {\text{ H}}_{{2}} {\text{O }} + { 7}^*,$$
(1)

where * represents an active site on the metal catalyst. It follows that placing COads and Hads in the appropriate ratio on the metal surface is important for CH4 production, and this is realized by controlling the CO2-feed concentration as well as the electrode potential in the case of a Pt catalyst24. Based on these techniques, we reported CO2 reduction at a Pt catalyst to generate CH4 with a faradaic efficiency of 12.3% at 0.16 V vs. RHE using 4 vol% CO2 diluted with Ar (Ref.25).

Here, the most noteworthy point is that the CH4-synthesis potential is almost the same as the theoretical potential, which is more positive than that for the hydrogen oxidation reaction (HOR). Therefore, power can be generated by an H2-CO2 fuel cell26 by combining the HOR and the CO2 reduction reaction (to generate CH4) as the anodic and cathodic reactions, respectively25. The H2-CO2 fuel cell is a promising CCU technology that utilizes CO2 as a resource to generate electricity while producing a valuable compound (CH4). However, the CH4 yield as well as the amount of power generated need to be increased to further develop this technology, and the efficient use of COads, which requires weakening the CO-metal bond, would represent a potential breakthrough toward this goal.

Pt-Ru alloy catalysts are known to impact the ligand effect, in which Ru affects the electronic state of COads and weakens the CO-metal bond27,28,29,30. In our previous study, we investigated the reduction of CO2 using MEAs incorporated with Pt-Ru electrocatalysts and revealed that the CH4-generation efficiencies at the theoretical potential follow the order: Pt/C < Pt0.5Ru0.5/C < Pt0.8Ru0.2/C when 100 vol% CO2 was supplied31. For this reason, we expected that CH4 would be more-efficiently produced by the reduction of CO2 without an overpotential by combining both techniques, namely diluting the CO2 concentration and using a MEA with the Pt0.8Ru0.2/C electrocatalyst. In this work, we designed a polymer electrolyte fuel cell that incorporated a MEA with a Pt0.8Ru0.2/C cathode and a Pt/C anode. This paper reports the reduction of CO2 for the simultaneous production of CH4 and power by supplying CO2 and H2 to the cathode and anode, respectively.

Results

Effect of CO2 concentration on CH4 yield

We first performed cyclic voltammetry (CV) to assess the cathodic reaction of the prepared cell (Fig. 1a) under various CO2 concentrations in Ar at a cell temperature of 40 °C. As shown in Fig. 1b, the oxidation current decreases with increasing CO2 concentration in the 0.08–0.43 V (vs. RHE) potential range, whereas it increases between 0.43 and 0.70 V (vs. RHE). Considering that the former and the latter are the oxidation currents that originate from H desorption and CO desorption, respectively32, this result reveals that the amounts of COads and Hads increase and decrease, respectively, with increasing CO2 concentration. The faradaic charges for Hads (QH) and COads (QCO) calculated from Fig. 1b are shown in Fig. 1c. In detail, QH and QCO were calculated as the integrated faradaic oxidation currents at 0.08–0.43 V (vs. RHE) and 0.43–0.70 V (vs. RHE), respectively, as depicted as the green filled area (for QH) and the red filled area (for QCO) in the inset in Fig. 1c. Based on Fig. 1c, a trade-off relationship between the QCO and QH are clearly observed. Therefore, the amounts of COads and Hads on the Pt0.8Ru0.2/C catalyst surface can be controlled by changing the concentration of CO2 supplied to the cathode. The onset potential for COads desorption was determined, as shown by the black arrow in Fig. 1b, which provided a value of 0.43 V (vs. RHE) for Pt0.8Ru0.2/C, which is more negative than that for Pt/C (0.45 V vs. RHE25), suggesting that the Pt0.8Ru0.2/C electrocatalyst exhibits a lower CO-adsorption energy.

Figure 1
figure 1

Schematic diagram and performance data. (a) Schematic diagram of the experimental setup used in this study. (b) Cathodic cyclic voltammograms at various concentrations of CO2 in Ar. (c) CO2-concentration-dependences of the faradaic charges of the oxidation-current peaks between 0.08 and 0.43 V (QH) and between 0.43 and 0.70 V (QCO) (vs. RHE) in the voltammograms shown in (b). The method used to calculate QH and QCO is shown in the inset in panel (c).

Full size image

CH4 generation from CO2 reduction at the Pt0.8Ru0.2/C electrocatalyst was next investigated. Figure 2a,b show cyclic voltammograms with in-line MS signals (m/z 2 and 15) at CO2 concentrations of 7 vol% and 100 vol%, respectively. The signal at m/z 15 was simultaneously detected with a reduction current at 7 vol% CO2 when the potential was below ~ 0.25 V (vs. RHE), whereas the m/z 15 signal only was weakly detected at 100 vol% CO2. The pattern in the mass spectrum of the cathodic output gas from the cell at 0.10 V (vs. RHE) in the 7 vol% CO2 atmosphere (Supplementary Fig. S1) is concordant with that of the CH4 standard gas; hence, the detected signal at m/z 15, which corresponds to CH3+ (not affected by H2O and CO2), is entirely derived from CH4 produced through the reduction of CO2. It should be noted that the signal at m/z 2 as hydrogen generation started to be detected at ~ 0.08 V (vs. RHE) at all CO2 concentrations. Figure 2c shows the dependence of the faradaic efficiency determined during CH4 generation on the CO2 concentration acquired during negative-potential-sweep CV between 0.20 and 0.10 V (vs. RHE). The faradaic efficiency was determined as a percentage of the methanogenic faradaic charge relative to the total faradaic charge. As a result, the highest faradaic efficiency of ~ 4.5% was calculated at 7 vol% CO2, which exceeds the efficiency for Pt/C (3.0% in a 5 vol% CO2 atmosphere24). On the other hand, the faradaic efficiency was only 0.61% in 100 vol% CO2, which corresponds to our previously reported efficiency31. Overall, we determined 7 vol% to be the preferred CO2 concentration for generating CH4 at the Pt0.8Ru0.2/C electrocatalyst.

Figure 2
figure 2

Cyclic voltammetry coupled with mass spectrometry. In-line m/z 2 (for H2) and 15 (for CH4) MS signals during CV at CO2 concentrations of: (a) 7 vol% and (b) 100 vol%. (c) Faradaic efficiency for the generation of CH4 as a function of CO2 concentration calculated by integrating the m/z 15-signals in the 0.20–0.10 V (vs. RHE) potential range during negative-scan CV.

Full size image

CH4-generation dependence on the CO2 electroreduction potential

We next explored the optimum potential for CH4 production at the Pt0.8Ru0.2/C electrocatalyst with the supplied CO2 concentration fixed at 7 vol% under a potentio-static condition. The cathode potential was stepped 14 times in the negative direction in the 0.40–0.05 V (vs. RHE) range every 2 min. As shown in Fig. 3a, which was produced by analyzing Supplementary Fig. S2, the signal at m/z 15 for CH4 began to be detected at 0.30 V (vs. RHE), was most intense at around 0.20 V (vs. RHE), and began to decrease in intensity below this value. The cathode potential at which the maximum m/z 15 signal was observed was different under the potentio-dynamic (Fig. 2a) and potentio-static (Fig. 3a) conditions, probably be due to the slow reaction rate of CO2 → CH4. Meanwhile, the signal at m/z 2 for H2 was observed only at 0.08 V and 0.05 V (vs. RHE), where CH4 generation was suppressed. These results reveal that CH4 production occurs at a more positive potential than H2 evolution, with a maximum CH4 yield observed at 0.20 V (vs. RHE). The in-line MS signals (m/z 2 and 15) recorded during cyclic voltammetry at 0 vol% CO2 (100 vol% Ar) shown in Supplementary Fig. S3 reveal that the m/z 15 signal was hardly detected at all cathode potentials. Based on the standard CO2|CH4 electrode potential (0.169 V vs. SHE33), we successfully generated CH4 from CO2 in the absence of an overpotential in this study. Therefore, from the viewpoints of CH4 yield and product selectivity, 0.20 V vs. RHE was determined to be the preferred potential.

Figure 3
figure 3

Stationary-potential CO2 reduction. (a) Potential dependence of faradaic charge, and integrated m/z 2 and 15 MS signals when held at each potential for 2 min at a CO2 concentration of 7 vol%. (b) Potential program applied to the cathode (upper), current response (middle), and in-line m/z 2 and 15 MS signals (lower) at 7 vol% CO2.

Full size image

We subsequently directly stepped the potential from 0.40 to 0.20 V (vs. RHE). As shown in Fig. 3b, the intensity of the m/z 15 signal was observed to be almost constant when the potential was held at 0.20 V (vs. RHE). The faradaic efficiency of CH4 generation from CO2 reduction (CO2 + 8H+  + 8e− → CH4 + 2H2O) between 240 and 299 s in Fig. 3b was calculated by the following formula:

$${\text{Faradaic efficiency }}\left( {\text{\% }} \right) = \frac{nmF}{{It}} \times 100,$$
(2)

where n is the number of reaction electrons, m is the molar number of CH4, F is the Faraday constant, I is the mean current, and t is the time. Significantly, the faradaic efficiency was 18.2%, which exceeds the previously reported Pt/C efficiency (12.3%25). Supplementary Fig. S4 shows gas chromatograms of the cathodic output gas from the cell in which the cathode potential was held at 0.20 V (vs. RHE). Based on the results obtained using flame-ionization and thermal-conductivity detectors, only CH4 was produced during CO2 reduction at Pt0.8Ru0.2/C, and the faradaic efficiency was calculated to be 17.6%, as detailed in Supplementary Information S4. Overall, continuous CH4 generation with enhanced efficiency and zero overpotential was achieved at a CO2 concentration of 7 vol% and a holding potential of 0.20 V (vs. RHE).

Power generation as an H2-CO2 fuel cell

Figure 4 shows power generation characteristics as well as CH4-production rates determined from the data in Fig. 3a. The current densities shown in Fig. 4 are mean current densities (shown in Supplementary Fig. S2) for 60 s just before the next potential step. Figure 4 reveals that the power density (as an H2-CO2 fuel cell) and the CH4 yield rate exhibit similar trends, with maximum values of ~ 0.14 mW cm−2 and 86.3 μmol g−1 h−1, respectively, at a cell voltage of 0.20 V. Compared to a report on the formation of CH3OH (but not CH4) through CO2 reduction at Pt–Ru/C using a MEA34, the rate of CH4 production in this study is higher.

Figure 4
figure 4

Cell performance. CH4 yield, cell voltage, and power density as functions of current density at a CO2 concentration of 7 vol%.

Full size image

Finally, we compared the results obtained for Pt0.8Ru0.2/C with those for Pt/C. As listed in Table 1, the power density, CH4 yield rate, turnover frequency, and faradaic efficiency were ~ 10-, ~ 3.5-, ~ 5-, and ~ 1.5-times higher, respectively, when the Pt0.8Ru0.2/C electrocatalyst was used instead of the Pt/C electrocatalyst. Therefore, we demonstrated a H2-CO2 fuel cell that generates electric power while efficiently reducing CO2 to CH4.

Table 1 Comparing Pt0.8Ru0.2/C and Pt/C cathode electrocatalyst data for the H2-CO2 fuel cell.
Full size table

Discussion

In the present study, we designed and demonstrated an H2-CO2 polymer-electrolyte fuel cell that generates CH4 from CO2 with enhanced efficiency, which was achieved by the strategic use of a Pt0.8Ru0.2/C cathodic catalyst. As mentioned in the introduction, the generation of CH4 through the reduction of CO2 follows the L–H mechanism involving COads and Hads (Ref.23). According to Eq. (1), this reaction theoretically proceeds at a COads-to-Hads molar ratio of 1:6; however, this reaction proceeded when the ratio was 1:11 or higher for a Pt/C electrocatalyst24, with the best ratio reported to be 1:18 (Ref.24). On the other hand, this ratio at the Pt0.8Ru0.2/C electrocatalyst was determined to be 1:8 at a CO2 concentration of 7 vol% according to the following equation:

$${\text{CO}}_{{{\text{ads}}}} :{\text{H}}_{{{\text{ads}}}} = \left( {\frac{{Q_{CO} }}{2}} \right):Q_{H} .$$
(3)

Hence, COads can be used to efficiently generate CH4 at Pt0.8Ru0.2/C because the ratio is close to the theoretical value of 1:6, which leads to a higher faradaic CH4-production efficiency, and was achieved by the lower CO-adsorption energy associated with the Pt0.8Ru0.2/C electrocatalyst.

The sustained generation of CH4 at 0.20 V (vs. RHE) (Fig. 3b) is also related to the COads-to-Hads ratio. Based on Supplementary Fig. S5, this ratio was determined to be 1:7 after holding at a potential of 0.20 V (vs. RHE) for 5 min at 7 vol% CO2. Therefore, CH4 is continuously generated at Pt0.8Ru0.2/C because the COads-to-Hads ratio was slightly different when the potential was held at 0.20 V (vs. RHE). It should be noted that the CH4-production reaction proceeds at a more-positive potential than its theoretical potential. Although the reason is unclear at present, there is one possible explanation for this observation. In our system, the CO2|CH4 equilibrium potential is influenced by the proton activity of the Nafion membrane electrolyte. As shown in Supplementary Fig. S3, the H2-evolution onset potential is ~ 0.08 V (vs. RHE); this value does not correspond to the theoretical potential (0.00 V vs. SHE33). In addition, the proton activity of the Nafion membrane has been reported to be different to that observed under SHE conditions35,36,37. Importantly, the CH4-generation reaction is likely to occur through a sequential CO2 → COads → CH4 reduction process, rather than through a one-step process (CO2 → CH4). Supplementary Fig. S6 shows that the onset potential for the formation of COads from CO2 through reduction at Pt0.8Ru0.2/C in 7 vol% CO2 is 0.375 V (vs. RHE). Hence, an electrode potential that progresses the COads → CH4 process can drive the overall reaction.

H2-CO2 fuel cells function when platinum group metals are employed as cathodic catalysts. In other words, a H2-CO2 fuel cell does not function using “active” electrocatalysts composed of only Cu, Au, and Ag, as well as their alloys, because CO2 reduction proceeds at a more negative potential than that for HOR (the overpotential is large)38. It should be noted that the highest faradaic CH4-yield efficiency was only 18.2% in this study, which is insufficient for practical applications. The reason for this low efficiency has not yet been clarified; one possible process that contributes to the rest efficiency involves the formation of COads and Hads, which are not associated with CH4 generation. Hence, further increasing the efficiency through catalyst design, including optimizing the Ru ratio in the Pt-Ru catalyst, will be important. One significant advantage of this technology is that the CO2-conversion reaction that produces CH4 occurs at a lower temperature (40 °C) than that used in chemical methanation technology39 and CO2-utilization technologies that rely on solid oxide electrolytes40 and molten solts41, which require temperature of several hundreds of degrees.

In conclusion, this work provides a novel approach to H2-CO2 fuel cells as a CO2-utilization technology. CH4 is produced continuously by the reduction of CO2 in the absence of an overpotential (at 0.20 V vs. RHE) in a cell that uses a MEA with a Pt0.8Ru0.2/C cathode at a CO2 concentration of 7 vol% and a cell temperature of 40 °C, and electrical energy is generated by combining the CO2-reduction and H2-oxidation (at a Pt/C anode) reactions. These results facilitate the carbon utilization strategy, although further investigations are necessary before it can be considered for practical applications.

Methods

Materials

Pt0.8Ru0.2/C (42.5 wt%; TECRuE43) and Pt/C (46.2 wt%; TEC10E50E) electrocatalyst powders were obtained from Tanaka Kikinzoku Kogyo Co., Ltd. Nafion-117 membranes (0.18-mm thick) were purchased from DuPont and boiled successively in Milli-Q water, 0.5 M H2O2, 0.5 M H2SO4, and Milli-Q water (1 h each) prior to use. All chemicals (H2O2, H2SO4, acetone, 2-propanol, methanol, and 5 wt% Nafion solution) were obtained from the Fujifilm Wako Pure Chemical Corporation. Water-repellent carbon paper (TGP-H-060H) was purchased from Toray Industries, Inc, and polymer electrolyte cell components (gasket, separator with parallel flow paths, and stainless steel plate) were purchased from Miclab.

Cell fabrication

A polymer electrolyte fuel cell (PEFC) was fabricated by essentially following the same procedure reported previously23,24,25, with the exception that Pt0.8Ru0.2/C was used as the cathode instead of Pt/C. Briefly, a 6 × 6 cm2 Nafion 117 membrane and 3 × 3 cm2 pieces of carbon paper pretreated with acetone were used as the proton-exchange membrane and gas diffusion layers, respectively. The electrocatalyst dispersion was prepared by mixing the Pt0.8Ru0.2/C catalyst with 5 wt% Nafion (1:1 v/v) and an aqueous solution containing 1:2:1 (w/w/w) 2-propanol, methanol, and Milli-Q water, followed by spraying onto one piece carbon paper to prepare the cathode. The anode was prepared by spraying a Pt/C electrocatalyst dispersion onto another piece of carbon paper. The amount of loaded metal and the apparent electrode surface area were 1.0 mg cm−2 and 9.0 cm2, respectively, on both electrodes. The MEA was prepared by bringing these electrodes into contact with each side of the Nafion-117 membrane, followed by hot-pressing at 140 °C with a 4.5 kN load for 10 min. It should be noted that the Pt/C electrocatalyst dispersion was dropped onto the Nafion 117 membrane on the anode side to provide a reference reversible hydrogen electrode (RHE). Finally, the MEA, gasket, separator, and stainless steel plate were assembled to complete the PEFC used in this study, as shown in Fig. 1a.

Electrochemical CO2 reduction and product analysis

A schematic diagram of the experimental setup used in this study is shown in Fig. 1a. Electrochemical experiments were conducted using a PEFC-operating apparatus (FCG-20S, ACE), a potentiostat/galvanostat (HA-310, Hokuto Denko), and a function generator (HB-104, Hokuto Denko). Fully humidified 100 vol% H2 and CO2 diluted with Ar (CO2 concentration: 0, 4, 7, 10, 20, 50, and 100 vol%) gas were fed to the anode and cathode at 50 cm3 min−1, respectively, in all experiments. Fully humidified 100 vol% H2 gas was supplied to the reference electrode at 10 cm3 min−1. The cell temperature was set to 40 °C because the cell humidification at least 40 °C is required to operate the MEA. The H2, CO2, and Ar gases were 99.999%, 99.995%, and 99.998% pure, respectively. Before electrochemical measurements, the cathodic potential sweep in the 0.05–0.70 V (vs. RHE) range at 50 mV s−1 was repeated until the current–potential curve of the cathode became stable. The initial cathode potential during introduction of CO2-containing gas was ~ 0.13 V (vs. RHE). The electrochemical surface area (ECSA) of the cathode electrocatalyst was obtained to be 0.294 m2 according to the hydrogen adsorption method30,42,43. The cathodic potential was scanned in the 0.08–0.70 V (vs. RHE) range at 10 mV s−1 during CV. It should be noted that that a 0.05–0.70 V (vs. RHE) potential range was used for CV with in-line product analysis. In the potential-step experiment, the cathodic potential was stepped through 14 levels in the 0.40–0.05 V (vs. RHE) range in the negative direction every 2 min at a CO2 concentration of 7 vol%. In addition, the cathodic potential was directly stepped from 0.40 to 0.20 V (vs. RHE) at 7 vol% CO2 and held there for 5 min, after which it was stepped to 0.05 V (vs. RHE). In-line mass spectrometry (MS) was carried out during the electrochemical experiments by introducing the cathode exhaust gas directly to a mass spectrometer (JMS-Q1050GC, JEOL). The ionization voltage was 23 eV. Note that the lag time for in-line MS product detection was adjusted by the H2 evolution response (7 s). A calibration curve, which was obtained using CH4 gas (purity: 99.999%) diluted with Ar, was used to calculate the amount of CH4 generated, for determining the faradaic efficiencies and CH4-yield rates. All the current densities were calculated using the apparent electrode surface area (9.0 cm2).