Abstract
In this study, a carbon-neutralized direct methanol fuel cell (DMFC) using two bifunctional electrodes, Pd-Ag and Pt-Zn, has been designed. This system has two modes, which are fuel-cell mode and spontaneous CO2 reduction mode. In the operation of fuel-cell mode, the methanol has been oxidized into CO2 on the Pd-Ag electrode and generates electricity. In the next step of the operation, CO2, which is the product of fuel-cell mode, has been spontaneously reduced to CO on Pd-Ag, and electricity has been obtained. In contrast, the Pt in the Pt-Zn electrode catalyzes the oxygen reduction reaction in the fuel-cell mode, and the Zn in Pt-Zn is oxidized sacrificially in the CO2 reduction mode. During operation in the fuel-cell mode, a power density of 12.11 mW/cm2 has been obtained with the production of CO2. On the other hand, the power density out of the CO2 reduction mode has been 11.76 mW/cm2. In each mode, the faradaic efficiencies of CO2 and CO have been 98.81% and 89.11%, respectively.
1. Introduction
Recently, research on renewable energy as an alternative to fossil fuels has been extensively conducted to solve the problems associated with environmental pollution and power shortages [1, 2]. Among them, fuel cells that can generate energy through the redox reactions of the fuels are seen to be promising prospects [3, 4]. Several fuel cells differ in terms of fuel type, cost, efficiency, and application. Polymer electrolyte membrane fuel cells (PEMFC) are the most widely used fuel-cell type. PEMFC have high energy conversion efficiency and power density. However, there are some disadvantages caused by using hydrogen, such as difficulty in transport and storage requirements due to hydrogen explosions [5]. In an effort to replace hydrogen, some other fuel cells that use liquid as fuel have attracted attention. Among them, direct methanol fuel cells (DMFCs) using the oxidation of methanol (CH3OH+H2O→CO2+6H++6e−) and the oxygen reduction reaction (ORR, 3/2O2+6H++6e− ⟶3H2O) are in the spotlight. Methanol is easy to transport and store because it is in the liquid state. Additionally, it is safe to use and possesses a high energy density [6–8]. Despite these advantages, DMFCs generate carbon dioxide (CO2), which causes global warming, when methanol is oxidized.
The CO2 is one of the greenhouse gases that contribute to global warming. Since the beginning of the industrial revolution, the concentration of CO2 in the atmosphere has increased by over 40% and reached 421 ppm in 2022, which is the highest record until now [9]. Accordingly, many countries are establishing regulations on CO2 emissions, and at the same time, many researchers are attracting attention to CO2 capture and utilization [10, 11]. There are various methods for capturing and utilizing CO2, such as thermochemical [12], chemical [13], photochemical [14–16], and electrochemical methods [17, 18]. Among them, the electrochemical CO2 reduction process is currently being studied extensively to convert CO2 to high-value-added materials such as hydrocarbons (HCs), carbon monoxide (CO), formic acid, and alcohol, and its products depend on the catalyst [19]. For example, Au and Ag catalysts are widely known to produce CO during the electrochemical CO2 reduction and have high selectivity of CO. On the other hand, the Cu catalysts are only used to produce significant amounts of hydrocarbons [17–19]. Generally, an electrochemical CO2 reduction system consists of two electrodes, a cathode and an anode, in which the CO2 reduction reaction and oxygen evolution reaction (OER, H2O⟶1/2O2+2H++2e−) occur, respectively. Like a typical electrochemical cell, there is a membrane that separates both electrodes to prevent a short circuit, and this membrane provides the role of a channel through which ions can move quickly. However, conventional electrochemical CO2 reduction has some limitations. The expensive noble metals such as Pt, Ru, and Ir have been used as anode materials in conventional electrochemical CO2 reduction systems because OER should occur [17, 18]. The standard reduction potential of CO2 is more negative than OER potential (1.229 V vs. SHE), as shown in Table 1 [20–22]. This means that the has a negative value and has positive value by calculation equations: where is the potential, is the Gibbs free energy, is the number of moles of electrons transferred, and is the Faraday constant (96485 C/mol). Therefore, conventional electrochemical CO2 reduction is a nonspontaneous reaction that requires external energy to occur CO2 reduction. In addition, the sluggish reaction kinetics of the OER, which charges the provision of electrons for electrochemical CO2 reduction, results in a large overpotential and high energy input [23–26].
To overcome this limitation, many researchers have used organic material oxidation instead of the OER [23–26]. Verma et al. investigated alternatives to OER and demonstrated that the use of glycerol oxidation reaction at the anode lowered the electrochemical CO2 reduction potential by ~0.85 V, resulting in a reduction in the power usage by up to 53% [23]. Na et al. reported a technoeconomic study of the coproduction of the organic oxidation reaction-electrochemical CO2 reduction reaction through conceptual process design and suggested feasible economic combinations [24]. Bi et al. used the anodic 5-hydroxymethylfurfural (HMF) oxidation reaction (HMFOR) to form a new paired electrolysis system. The CO2 reduction-HMFOR pairs required a 1.06 V onset cell potential for efficient conversion of both CO2 and HMF, compared with 1.77 V for a conventional CO2 reduction-OER system [25]. Li et al. used the methanol oxidation reaction (MOR) in an anode reaction. The CO2 reduction-MOR system produced formic acid in both CO2 reduction and MOR. In addition, the total energy consumption is reduced by over 40% compared with CO2 reduction coupled with the OER [26]. These studies lowered the total cell potential along with producing the high valuable materials instead of oxygen evolution.
Despite the efforts of many researchers, the conventional electrochemical CO2 reduction system still operated when external energy is applied. In our previous study, we developed a sacrificial Zn electrode to overcome these limitations and make the CO2 reduction spontaneously [27]. The Zn oxidation reaction in alkaline solution has a more negative value (-1.23 V vs. SHE) compared with the standard reduction potential of CO2. In this study, we combined a methanol fuel cell with the spontaneous electrochemical reduction of CO2 to generate electricity and convert CO2 to value-added products, which are produced by methanol oxidation, as shown in Figure 1. In our system, the electrode materials are important because the electrodes are used for methanol oxidation/CO2 reduction and ORR/Zn oxidation, respectively. In an electrode for methanol oxidation, the noble metals such as Pt, Rh, and Pd and nonnoble metals including Cu, Co, and Ni have been used [28–30]. Among the metal candidates, Pd catalysts show high performance and are suitable for use in the alcohol oxidation reaction (AOR) [30]. Besides, it is well known that porous structures such as 1-D nanowires, 2-D nanosheets, 3-D nanospheres, and nanodendrite offer a high mass transfer efficiency because of their large surface area [28]. On the other hand, metal-based catalysts such as Cu, Ag, Cd, Zn, and Sn are generally used for electrochemical CO2 reduction [31–35]. Among these metal electrodes, the Ag electrode has attracted many researchers because CO2 is reduced to CO on the Ag surface, its high electrical conductivity is S/m, and it is stable in air or solution [31]. In addition, when CO2 is reduced to CO, the number of electrons required to obtain 1 mol of CO is 2 mol, which is less than that of other products. In addition, it has been reported that Ag catalysts have better resistance to CO toxicity [36]. Therefore, we have designed the Pd-Ag bifunctional electrode for methanol oxidation/CO2 reduction. On the other hand, Pt/C is commercially used for ORR catalyst [37, 38], and Zn is used for the sacrificial electrode in alkaline solution to occur spontaneous CO2 reduction. Therefore, we have fabricated Pt-Zn bifunctional electrodes for ORR and Zn oxidation. In addition, a strong alkaline solution was used for Zn oxidation and to prevent damage to the other electrode.

2. Materials and Methods
2.1. Preparation of Pd-Ag Bifunctional Electrode
For the carbon-neutralized direct methanol fuel cell, the two-electrode system with the channels has been used in this work. One of the bifunctional electrodes was the Pd-Ag for methanol oxidation/CO2 reduction. A dendritic Pd-Ag bifunctional electrode was fabricated through electrodeposition with a hydrogen evolution reaction (HER). When an overpotential is applied, hydrogen bubbles occur at the electrode, and metal ions are not deposited in the bubble generation zone, resulting in the formation of a porous dendritic structure [39]. For the preparation of the Pd-Ag electrode, Ni mesh (Nilaco Corp., Mesh #100, Japan) with a size of a cm2 was pretreated to remove the native oxide layer with isopropyl alcohol (IPA; Daejung Chemicals, Korea), ethanol (95%, Daejung Chemicals, Korea), and 5 M H2SO4 (95%, Daejung Chemicals, Korea) with sonication for 20 min at each step. Then, a Pd-Ag electrode layer was electrodeposited on the pretreated Ni mesh. In the electrochemical deposition of Pd-Ag on Ni mesh, the Ni mesh was set as the working electrode, and Hg/HgO and a Pt plate of cm2 were used as the reference and counter electrodes, respectively. The constant cathodic overpotential of −4 V was applied for 5 min in an electrolyte containing 5 mM PdSO4 (Sigma-Aldrich, Korea), 5 mM AgNO3 (Sigma-Aldrich, Korea), and 1 M H2SO4. After the electrodeposition, the electrode was rinsed with D.I. water and blown with N2.
2.2. Preparation of Pt-Zn Bifunctional Electrode
For the Pt-Zn electrode substrate, a carbon paper with a gas diffusion layer (GDL) (39BB, Sigracet, U.S.A.) of cm2 was used. For the spray coating of Pt-Zn, Pt/C (40 wt.%, Alfa Aesar, Korea), Zn powder (99%, Daejung Chemicals, Korea), and Nafion solution (D521, 5 wt.%, Dupont Co., U.S.A.) were dispersed in a mixture of IPA and D.I. water with a tip sonicator for 90 min. And then, the dispersed solution was sprayed on the carbon paper at 60°C.
2.3. Characterization of Each Electrode
The prepared electrode was analyzed by field emission scanning electron microscope (FESEM; JEOL, JSM 7000F) to confirm the dendritic structure of the Pd-Ag electrode and the morphology of the Pt-Zn electrode. An X-ray diffractometer (XRD, Bruker, D8 ADVANCE) was used to analyze the crystallinity of the electrodeposited Pd-Ag bifunctional electrode.
2.4. System Assembly and Electrochemical Analysis
The cell hardware consisted of end plates, current collectors, and graphite channels, as shown in Figure 2. These two electrodes have been used in carbon-neutralized direct methanol fuel cells. In the system, a Sustainion membrane (X37-50 Grade RT, Dioxide Materials, U.S.A.) was employed to separate the dendritic Pd-Ag electrode from the Pt-Zn electrode, which also worked as an anion-exchange membrane (AEM). In the fuel-cell mode, 2 M CH3OH containing 0.5 M KOH solution was supplied to the Pd-Ag electrode at a flow rate of 20 sccm, and air was continuously fed into the flow channels on the Pt-Zn side at a flow rate of 50 sccm. In the next step of operation, 5 M KOH solution was supplied to the Pt-Zn electrode to oxidize the Zn. During the cell operation, both the fuel-cell mode and the CO2 reduction mode were conducted using a current sweep method at a scan rate of 1 mA/s without applying any external overpotentials. To control the reaction rate, constant currents of 20 mA/cm2 and 10 mA/cm2 were applied in fuel-cell mode and CO2 reduction mode, respectively. The performance of the Pd-Ag electrode in the conventional electrochemical CO2 reduction was also investigated when the cathodic overpotential was applied in the range between -1.0 V and -2.0 V. All electrochemical measurements were conducted using an electrochemical workstation (WMPG1000, WonATech, Korea) at room temperature (R.T.).

2.5. Product Analysis
The products were sampled after the operation for 1 hour in fuel-cell mode and 30 min in CO2 reduction mode, respectively. These products were analyzed quantitatively and qualitatively using gas chromatography (GC; YL6500, YOUNGIN Chromass, Korea) with a thermal conductivity detector (TCD) and flame ionization detector (FID). In the gas chromatography, H2, CO, and CO2 were observed with the packed column for TCD. The column temperature of GC-TCD was maintained at 170°C, and the inlet of the gas chromatography and detector were maintained at a temperature of 220°C when using TCD. The FID using GS-CARBONPLOT column was used to confirm that there were no other products except H2, CO, and CO2. The column temperature of GC-FID was increased from 60°C to 200°C at a rate of 15°C/min, and the inlet of the GC and detector were maintained at the temperature of 220°C. The faradaic efficiency (F.E.) of the product was obtained by the ratio of each partial charge of the products to the total charge, as shown by the following equation: where is the charge, is the amount of the product (number of moles, mol), is the number of electrons required to obtain 1 mol of the product, and is the Faraday constant (96485 C/mol).
3. Results and Discussion
For the carbon-neutralized direct methanol fuel cell, a two-electrode system with channels was constructed, in which the electrodes were a Pd-Ag bifunctional electrode and a Pt-Zn bifunctional electrode. The Pd-Ag bifunctional electrode has morphology of dendritic structures, and the Pt-Zn electrode is spherical structure as presented in Figure 1. In our system, there are two modes: fuel-cell mode and CO2 reduction mode. The red line represents the operation of the methanol fuel-cell mode, and the blue line represents the operation of the CO2 reduction mode. In the step of fuel-cell mode operation, the methanol fuel is preferentially oxidized to CO2 at the Pd surface of Pd-Ag electrode in an alkaline condition of the electrolyte and releases electrons. On the Pt-Zn electrode, the air supplied through the flow channels receives electrons from methanol oxidation and is reduced to OH- ions at the Pt surface. In the next-step operation of the CO2 reduction mode, the Zn is oxidized to Zn(OH)42- in a highly alkaline solution. In contrast, the CO2 produced by methanol oxidation is reduced to CO at the Ag surface of Pd-Ag without any external potential.
To operate the methanol oxidation/CO2 reduction cell, the high-surface-area dendritic Pd-Ag layer was electrochemically deposited on the bare Ni mesh, as shown in Figure 3. A uniform dendritic Pd-Ag layer was electrodeposited on the Ni mesh, and a porous structure, which is advantageous for reaction activity and mass transfer, was formed because of the dynamic hydrogen bubble evolution reaction. This high-surface-area dendritic Pd-Ag electrode provides a lot of reaction-active sites to occur methanol oxidation and CO2 reduction. As seen in the inset of Figure 3(b), the dendritic structure of electrodeposited Pd-Ag electrodes exhibits the shape of small branches with stems of a few micrometers in length compared with the bare Ni mesh, as shown in Figure 3(a). In the XRD analysis on the fabricated Pd-Ag bifunctional electrode, the Ag (111), Ag (200), Ag (220), and Ag (311) peaks are presented at 38.4°, 45.3°, 63.9°, and 77.2° whereas Pd (111), Pd (220), and Pd (311) are also existed at 39.2°, 66.1°, and 79.5°, as shown in Figure 3(c). On the other hand, the Pt-Zn electrode was coated on carbon paper using the general spray coating method, as shown in Figure 3(d). The spherical Pt-Zn catalyst was uniformly coated without agglomeration. The weight ratio of Pt and Zn was found to be 51.1 to 48.9 by EDS.

The two fabricated electrodes (Pd-Ag electrode and Pt-Zn electrode) were used in carbon-neutralized direct methanol fuel cells. Here, it is important to choose a membrane that can withstand the strong alkaline solution and organic solution. Thus, the Sustainion® AEM, which has high stability in organic and highly alkaline solutions, is used to move OH- ions and separate the two electrodes in both fuel-cell mode and CO2 reduction mode. During the fuel-cell operation, the methanol fuel (2 M CH3OH+0.5 M KOH) is supplied to Pd-Ag electrode and methanol is oxidized to CO2 at the Pd surface of Pd-Ag and releases electrons. On the Pt-Zn electrode, the humidified air supplied through the flow channels is reduced to OH- ions at the Pt surface of Pt-Zn electrode. The OH- ions also moved toward the Pd-Ag electrode to maintain the balance of ions, as presented in Figure 4(a). The electrochemical reactions in fuel-cell mode are described by the following equations:

(a)

(b)
Based on the standard reduction potential presented above, the theoretical open-circuit voltage (OCV) of the overall cell reaction is 1.211 V.
In the next step of the CO2 reduction mode, the Zn is oxidized to Zn(OH)42- in 5 M KOH solution. Here, the electrolyte should be maintained above pH 14 to be dissolved in the electrolyte in the form of Zn(OH)42- ions after Zn oxidation. In contrast, the CO2 produced by methanol oxidation was spontaneously reduced to CO at the Ag surface of Pd-Ag, as shown in Figure 4(b). The following equations are the electrochemical equations in CO2 reduction mode:
Therefore, the theoretical OCV in CO2 reduction mode is 0.71 V based on the standard reduction potential.
Here, the performance of the carbon-neutralized direct methanol fuel cell using a bifunctional electrode was evaluated. In the fuel-cell test, the Sustainion® AEM was wetted with water prior to assembly between the cathode and anode at R.T. First of all, the polarization and power density curves in fuel-cell mode obtained during the operation of a methanol fuel cell are presented in Figure 5(a). The OCV was measured to be 0.731 V, and the cell potential decreased with increasing current density. This is mostly attributed to the ohmic polarization caused by resistance at the membrane-electrode assembly interface and interfacial electron transfer between the current collectors and the catalyst layers. Additionally, this ohmic polarization can be caused by the partially Zn oxidation because the water and OH- ions are diffused into Pt-Zn electrode. Nevertheless, with Pd-Ag and Pt-Zn bifunctional electrodes, the maximum power density of 12.11 mW/cm2 was obtained at a current density of 48.56 mA/cm2 for 2 M methanol and 0.5 M KOH solutions. The performance of fuel-cell mode in our system was compared with other DMFCs as shown in Table 2. When compared with other DMFCs, our system has good performance despite using Pt-Zn catalyst for ORR and operating at R.T.

(a)

(b)
To control the reaction rate and analyze the products in fuel-cell mode, the fuel-cell mode was operated at a constant current condition of 20 mA/cm2 for 6 hours. The methanol and humidified air were continuously supplied in the fuel-cell mode to provide enough fuel for a 6-hour reaction. During the constant current operation, the potential of the cell was maintained at 0.421 V, which occurred by methanol oxidation and ORR, as shown in Figure 5(b). The products of the fuel-cell mode after 1-hour reaction were captured and investigated using GC analysis. The product from fuel-cell mode was CO2, and the total F.E. was confirmed to be nearly 100%, indicating that there were no liquid products. The major product obtained from the fuel-cell mode was mostly CO2, and CO was not produced, which was confirmed by GC analysis. At this point, 1.1 mmol of CO2 was produced, and the F.E. of CO2 was 98.81%.
After the operation with fuel-cell mode, the second step of spontaneous CO2 reduction mode was conducted. In the operation with CO2 reduction mode, a high alkaline solution was supplied to the Pt-Zn bifunctional electrode for the Zn oxidation. Figure 6(a) presents the polarization and power density curves of the CO2 reduction mode. The OCV and the maximum power density were found to be 0.677 V and 11.76 mW/cm2, respectively. The cell potential decreased rapidly in the range of 4 mA/cm2 to 30 mA/cm2. This is due to ohmic resistance resulting from the formation of Zn(OH)2 on the surface of Zn despite maintaining the pH of the electrolyte over 14 to suppress Zn(OH)2. The Zn(OH)2 is a water-insoluble by-product produced at the Zn surface. Once the precipitate of Zn(OH)2 has been produced at Zn surface, it is difficult to dissolve it into Zn(OH)42- even in a strong alkaline solution. These precipitates of Zn(OH)2 inhibit that the Zn oxidation reaction at the Zn surface has higher resistance than metallic Zn. In the area after 30 mA/cm2, the concentration polarization has been dominated by the amount of Zn, which means Zn in Pt-Zn electrode was mostly oxidized.

(a)

(b)
To control the reaction rate and confirm the products in the CO2 reduction mode, the CO2 reduction mode was driven at a constant current condition of 10 mA/cm2 for 30 min. During the constant current mode, the potential of the cell was maintained at 0.377 V, which means the amount of Zn was enough to react for 30 min, as shown in Figure 6(b), and the products of the CO2 cell mode were analyzed using GC analysis. Here, CO and H2 were observed on GC analysis, and 0.74 mmol of CO was produced. The H2 production reaction is the competitive reaction with the CO2 reduction reaction, and the potential gap between Zn oxidation and HER is positive value, which means HER also occurs spontaneously. Nevertheless, in this system, the amount of H2 produced was small because the concentration of protons in the electrolyte was low, which resulted from the catholyte in the CO2 reduction mode being an organic solution containing an alkaline solution and the anolyte being a highly alkaline solution. The total conversion of CO2 in this system was 67.27%. Also, the F.E. calculated by equation (2) of CO and H2 was 89.11% and 9.83%, respectively. The total F.E. was also confirmed to be nearly 100%, indicating that there were no liquid products. After operating two modes, each bifunctional electrode was confirmed by FESEM as shown in Figure 7. Compared with Figure 3, the morphologies of both electrodes have been little after the cell operation.

(a)

(b)
In addition, the CO2 reduction mode was compared with the conventional electrochemical CO2 reduction system in terms of product and energy. In order to compare with the conventional electrochemical CO2 reduction, a constant potential of -1.0~-2.0 V was applied in the cell. All the products produced by conventional electrochemical CO2 reduction were captured and analyzed by GC, and the F.E. was calculated by equation (2). The F.E. of each product is compared in Table 3, while different overpotentials were applied. Generally, the more overpotential applied at the cathode, the more currents occurred, which means the total amount of reaction in the cell increased. Thus, the amount of CO2 reduction and HER has increased because the reactions at cathode are HER and CO2 reduction. As shown in Table 3, the F.E. of H2 and total CO2 conversion has increased more at higher overpotentials, and F.E. of CO has decreased. For example, when -1.0 V was applied, the CO2 conversion and the F.E. of H2 and CO were 62.35%, 10.51%, and 88.46%, respectively. However, -2.0 V was applied; the CO2 conversion and the F.E. of H2 and CO were 75.11%, 39.67%, and 60.22%, respectively. This is because the amount of HER reaction increased much more than the amount of CO2 reduction. When the high overpotential was applied at the cathode, the HER, which is a competing reaction with CO2 reduction, occurred more easily due to the difference in standard reduction potential. Compared to conventional electrochemical CO2 reduction applied low overpotential, spontaneous CO2 reduction in this system shows remarkable results in the F.E. of CO and total CO2 conversion. Compared with applied high overpotential, although the total CO2 conversion is lower than that of applied high overpotential, the F.E. of CO is higher. In addition, in terms of energy efficiency, the spontaneous CO2 reduction in this system is a much more energy efficient system because this system produces energy, while the conventional electrochemical CO2 reduction system, which should be applied high overpotential, consumes a lot of energy.
Furthermore, compared with other CO2 reduction systems, our study proves to be a promising technology in terms of energy consumption as shown in Figure 8. As described in Figure 8, the energy must be consumed for electrochemical CO2 reduction although other oxidation reactions are used instead of OER. However, our system produces energy and converts CO2 to CO. Moreover, our system shows a high F.E. of up to 90%. Using the unique direct methanol fuel-cell system, which is combined methanol fuel cell with spontaneous electrochemical CO2 reduction system, we generate the power at both methanol fuel-cell mode and CO2 reduction mode and convert CO2 to CO, which is produced in direct methanol fuel cell.

4. Conclusions
In this study, the carbon-neutralized DMFC using two bifunctional electrodes was developed. The system was operated in fuel-cell mode and CO2 reduction mode sequentially. In fuel-cell mode, the maximum power density was 12.11 mW/cm2 when methanol fuel was oxidized to CO2 at the Pd-Ag electrode. In the next step of the CO2 reduction mode, CO2 was reduced to CO at the Ag surface in the Pd-Ag electrode, and electricity was generated simultaneously. In addition, Zn was oxidized to Zn(OH)42- ions in strongly alkaline solutions. In both modes, the F.E. of CO2 and CO was 98.81% and 89.11%, respectively. Also, the total conversion of CO2 in CO2 reduction mode was 67.27%. Compared with conventional direct liquid fuel cells, this system is unique and eco-friendly as it operates spontaneously in both modes and allows the reduction of CO2 to CO. This carbon-neutralized direct methanol fuel cell makes fundamental and technological progress toward green technology, cost-effectiveness, and effective utilization of fuel and CO2.
Data Availability
The data used to support the findings of this study are included within the article.
Disclosure
This study has been presented in 11th Asian Conference on Electrochemical Power Sources (ACEPS’11) at Singapore.
Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.
Acknowledgments
This work was supported by the Basic Science Research Program and Basic Research Laboratory through the National Research Foundation of Korea (NRF), funded by grants from the Ministry of Science (NRF-2022R1A2B5B01001764, NRF-2021R1A4A1024129).