Cost-effective and durable Ru-sputtered Pt/C-based membrane–electrode assembly for passive direct methanol fuel cells

Direct methanol fuel cells (DMFCs) have emerged as a potential power source, owing to several associated advantages such as high energy density as compared with other fuel cell types. One of the other fuel cell types, solid oxide fuel cells (SOFCs) has several advantages, such as high fuel flexibility and easy water management, but because of the use of hydrogen as a fuel not only the volumetric energy density is low, but also the storage of fuel is difficult.^1–4 The feature of high energy density is attributed to the fact that DMFCs use liquid methanol solution as a fuel. Here, methanol as a fuel possesses suitable properties such as high energy density and the ability to exist in a liquid state at room temperature.^5–11 However, functioning of DMFCs require a highly complicated system to manage the concentration of methanol and the water product. To overcome these limitations, passive DMFCs have been developed and studied. Passive DMFCs employ both methanol solution and air that are supplied to the anode and cathode, respectively, by free convection. Moreover, no pumps or other auxiliary parts are required to supply and manage methanol, water, and air. Such highly simplified systems of passive DMFCs serve as an ideal and attractive candidate for portable applications and these are expected to replace the current market-leading portable energy sources, such as rechargeable batteries.

However, passive DMFCs have some technical problems as compared with normal DMFCs.^12–16 For example, one of the biggest challenges to overcome is the crossover of methanol through the electrolyte membrane in passive DMFCs. This is attributed to the non-polar nature of methanol that causes it to easily permeate the proton-exchangeable electrolyte membrane. The transferred methanol causes a decrease in the fuel cell voltage owing to mixed oxidation and reduction at the cathode. Methanol crossover also leads to carbon monoxide (CO) poisoning of the cathode catalyst and water flooding. Catalyst poisoning occurs due to the CO generated by the methanol oxidation reaction (MOR). The CO formed as a result of methanol oxidation easily binds to the Pt catalyst, thereby occupying the reaction site of Pt and eventually blocking it. This results in a rapid reduction in the performance; moreover, such a process is irreversible. To alleviate these issues of DMFCs, several studies are being conducted and are tabulated in Table I. For instance, Munjewar et al. reported and reviewed the major issues and barriers to commercialization of passive DMFC.¹⁷ Yuan et al. reported a reduction in methanol crossover by 63% using a novel methanol-blocking membrane prepared by layer-by-layer assembly of poly (Diallyldimethylammonium chloride) (PDDA) and graphene oxide Nafion® nanosheet.¹⁸ Similarly, Yan et al. fabricated a Nafion® sandwich membrane with a 68% reduction in methanol permeability and which could be used in a highly concentrated solution.¹⁹ Many other research groups have focused on electrolytes and reported self-made membranes for low methanol permeability.^20–22 Kho et al. reported a change in internal temperature and open circuit voltage (OCV) with increasing concentration of methanol over time.²³ Wagner et al. demonstrated the influence of CO poisoning on the Pt and Pt–Ru anodes in a polymer electrolyte membrane membrane fuel cell (PEMFC) through a change in the electrochemical impedance spectra (EIS).²⁴ The influence of methanol crossover on power and energy density was also verified by Gurau et al.²⁵ Zainoodin et al. investigated the degradation mechanisms in MEA with a porous carbon nanofiber anode layer under various operational modes.²⁶ In addition, a lot of research is being done to replace precious metal catalysts and supports. Patel et al. applied high conductivity TiN as a support for the Pt_Ru catalyst and verified a 56% performance improvement.²⁷ Chang et al. developed highly active, anti-poisoning catalyst that promotes the oxidation of methanol to CO₂.²⁸ Sebastian et al. synthesized highly active non-platinum group metal catalyst for DMFC and investigated its performance and durability with variation of fuel concentration and temperature.^29,30 Osmieri et al. reported the performance and durability of non-noble cathode catalyst and demonstrated Fe-N-C catalysts as potential candidates for Pt.³¹ Research on various alternative catalysts has been actively reported, and in particular, many studies using graphene have been reported.^32–34 

In the present study, a novel fabrication method of MEAs for DMFCs is suggested and their electrochemical properties are evaluated. The MEA with only Pt/C catalyst, which is normal MEAs for PEMFCs, was first synthesized and Ru was then sputtered onto the sprayed Pt/C catalyst layer directly. In this way, the fabrication of MEA is highly simplified and the use of Ru is highly reduced. Here, the thickness and nano-morphology of sputtered Ru were controlled by regulating the inert gas pressure inside a chamber, target–substrate distance, and the sputtering power. After conducting a morphological study, the Ru-sputtered MEA was mounted onto the passive DMFC and its electrochemical polarization characteristics as well as the EIS were observed to further investigate the electrochemical characteristics. In addition, both the performance and the catalytic durability were assessed and their potential to be used for passive DMFCs confirmed.

As shown in Fig. 1(a), two types of MEAs were synthesized for comparison. First, the standard MEA with only Pt catalyst was fabricated by spraying Pt/C-based ink onto the Nafion® 117 membrane (Du Pont Co., USA). The Pt/C catalyst ink was made using commercial Pt black (40 wt.% Pt, Johnson Matthey, UK), isopropyl alcohol (IPA, Daejung chemistry Co., Republic of Korea), deionized water, and 5 wt.% Nafion® solution (Sigma Aldrich Co., USA). The electrochemically reactive area was precisely controlled and defined using a mask with a hole of area 1 × 1 cm², through which the as-prepared Pt/C ink was sprayed. During spraying, the pristine Nafion® was placed on a specially designed hot-vacuum plate to strongly fix and prevent it from wrinkling. The temperature of the plate was 80 °C. Both anodic and cathodic catalyst-coated layers (CCLs) were sprayed with the same ink. The second MEA was prepared by simply sputtering Ru onto the as-prepared MEA described above. A 100-nm thick Ru catalyst layer was sputtered onto both the anodic and cathodic CCLs, and the depositing area was precisely controlled using the same mask with a hole of area 1 × 1 cm². Here, the DC sputtering power and inert argon gas pressure were 200 W and 12.0 Pa, respectively. The digital camera images of the as-prepared two MEAs are indicated in Fig. 1(b), where the CCLs with and without sputtered Ru are clearly distinguishable.

After fabricating the two MEAs, a four-step activation process was performed to secure the high protonic conductivity of the electrolyte membrane. This process is intended to recover the sulfonic acid groups in the Nafion® membrane.³⁶ The MEAs were first boiled for 1 h in 5 vol.% H₂O₂ solution, then in deionized water, followed by boiling in 0.5 M H₂SO₄ solution for 1 h, and then again in deionized water for 1 h.

The electrochemical performances of the as-prepared MEAs were measured using the passive DMFC as shown in Fig. 1(c). It consisted of a methanol chamber, two gaskets, two current collectors, and two gas-diffusion layers (GDLs); and finally one endplate was used to assemble all components tightly using four bolts. A polytetrafluoroethylene (PTFE) film, Au-coated stainless steel, and Sigracet 39BC (SGL Carbon Co., Germany) were used as a gasket, GDL, and current collector, respectively. The current collector was obtained by sputtering Au onto a laser cut, 0.1-mm thick stainless steel. The Ar pressure and the DC power for Au-sputtering were 0.67 Pa and 200 W, respectively. The volume of methanol chamber was 5 × 5 × 5 cm³. A 4-M methanol solution was supplied as a fuel into the cell.

The surface structure and cross-sectional images of the as-prepared MEAs were obtained and investigated using a field-emission scanning electron microscope (FE SEM; Zeiss Supra 55VP, Carl Zeiss, Germany). Energy-dispersive X-ray spectroscopy (EDS, Bruker Co, USA) confirmed the presence of Pt and Ru on the catalyst layer. The performance of the fuel cell was measured at room temperature using a Solartron 1287A and 1260A impedance analyzer (Solartron Analytical, UK). The anodic and cathodic electrodes of the MEA were simply exposed to 4 M methanol fuel and air, respectively, to supply these by natural convection, and not forced nor controlled. Polarization characteristics were investigated first by measuring the I–V curve. The measurement started from the OCV of the fuel cell and swept toward 0.1 V. The voltage sweep speed was 10 mV/s; corresponding current density was measured and the measurement was stopped at 0.1 V vs. RHE (reversible hydrogen electrode). Right after measuring the I–V curve, the corresponding EIS were measured and visualized using the Nyquist plot. A sinusoidal voltage input with an amplitude of 30 mA and a frequency ranging from 20,000 to 0.1 Hz at 0.2 V vs. RHE was applied to the fuel cell and resulting current response was monitored. After measuring the I–V curve and EIS, the time evolution of the OCV was monitored for 1 hour and the above measurement process was repeated, thereby measuring the degradation in the performance of the catalytic activity of the MEA by MOR. Finally, the passive DMFC with the MEA mounted onto it was left without any modifications for a day following which I–V was measured again. Two MEAs were tested using the above test procedure.

Fig. 2(a) depicts the cross-sectional image of the as-fabricated Ru-sputtered MEA. The upper part of the Nafion® membrane is an anode and the lower part is a cathode. The thickness of the anodic and cathodic CCLs region was similar (28.1 μm and 28.7 μm, respectively), from which both the amounts of anodic and cathodic CCL are precisely controlled. The corresponding loading of Pt was 1.0 mg/cm² for both anode and cathode. Fig. 2(b) and (c) show the surface images of the standard MEA and Ru-sputtered MEA, respectively. It can be seen that both images have a porous surface, suggesting that the CCL region could act as an electrode and methanol/air could penetrate through it. From Fig. 2(b) and (c) it is seen that the average size of pores inside the CCL region was reduced slightly after deposition of Ru. It is speculated that the sputtered Ru particles blocked the pores. Nevertheless, the difference in pore sizes in Fig. 2(b) and (c) did not appear significant and it was expected not to disturb the redox reaction in both electrodes. Fig. 2(d) and (e) show the EDS images of Pt and Ru on the surface of the Ru-sputtered MEA, corresponding to Fig. 2(c). Overall, uniformly scattered green and yellow dots indicate that the fabrication method used led to well-dispersed catalyst layers. In addition, the sputtered Ru onto the Pt/C CCL did not induce the decrease in EDS intensity of Pt, implying that the sputtered Ru did not completely block the pores of the CCL. It also corresponds with the observation of the nano-morphology in Fig. 2(b) and (c) that sufficient pore and resulting reactants pathways were secured.

The electrochemical performances of the passive DMFCs with standard and Ru-sputtered MEA are shown in Fig. 3(a) and (b), respectively. Here, 4 M methanol solution, which is relatively higher than the normal concentration for DMFCs, was used owing to two reasons: further simplification of the system and decreased power consumption of the system.³⁵ In Fig. 3(a) and (b), three different polarization characteristics were measured and compared: pristine, after operating for an hour, and after 1-day case. According to Fig. 3(a), the OCV and peak power density of pristine MEA were 0.52 V and 3.01 mW/cm², respectively. Considering the theoretical OCV of a DMFC to be 1.199 V, OCVs of both normal and Ru-sputtered MEAs were significantly low. According to the Nernst potential, which represents the theoretical OCV of the fuel cell, such a low OCV could be explained by a change in the concentration, if there exist no sealing and current leakage problems. In this case, the methanol crossover would lower the OCV, thus affecting the performance at low and high current density range. The passage of methanol from the anode to the cathode would cause its oxidation (MOR) at the cathode and lower the OCV. Therefore, a relatively thicker membrane for DMFCs is used than for PEMFCs, as evident from the use of Nafion® 117 (178 μm thick) in the present study to reduce this phenomenon.³⁷ Activation loss is dominant in the low current density region, resulting in a severe voltage drop due to MOR. In the following middle current density region, the ohmic loss is dominant and the voltage does not fall steeply unlike the voltage in the low current density region. Interestingly, no concentration losses were observed in all three cases, which is attributed to the use of high concentration methanol solution and low operating current. The results of the polarization curve after 1 hour and 1-day show, i.e., decreased OCV and increased activation loss increased as depicted in both Fig. 3(a) and (b), reflecting the influence of methanol crossover and the poisoning of Pt catalyst by MOR-generated CO. However, the OCV and peak power density for the Ru-sputtered MEA were 0.417 V and 3.28 mW/cm², respectively, as shown in Fig. 3(b). The OCV is lower than that of the standard MEA. It is believed that such low OCV is the result of additional sputtered Ru onto the CCL and this could block the diffusion of the methanol. However, such Ru sputtering does not significantly affect the final electrochemical performance of the MEA, as the final power density is significantly higher than that of the DMFC with the standard MEA (0.61 mW/cm² vs. 1.42 mW/cm² for “after 1 hour” and 0.60 mW/cm² vs. 1.42 mW/cm² for “after 1 day” case in Fig. 3(a) and (b)). In the subsequent measurements, the OCV after 1 hour was rather higher than that under pristine condition owing to the temporary decrease in methanol crossover. However, low OCV was still recorded 1 day after the measurement. In both tests, the current–voltage curves exhibited similar performances, which was lower than that of the pristine Ru-sputtered MEA. However, the overall electrochemical performance of the Ru-sputtered MEA was higher than that of the standard MEA, indicating that the MEA-fabrication method described in the present study exhibits the potential to convert the normal CO-poisoning resistant MEA by simply applying additional sputtering process. Moreover, this Ru-sputtering method is beneficial as it could modify the current commercialized MEAs for PEMFCs using pure hydrogen as a fuel to the MEA for DMFCs by sputtering Ru onto them.

To further investigate the science behind the electrochemistry of the Ru-sputtered MEA, EIS at 0.2 V vs. RHE was measured and analyzed, as shown in Fig. 4(a) and (b). When the EIS fitted into the equivalent circuit (indicated as an inset in Fig. 4(a) and (b)), an ohmic resistance could be calculated at the left intercept of the half circle on the x-axis. Here, ohmic resistances were recorded to be 1.0 Ω·cm² and 1.7 Ω·cm² in standard MEA and Ru-sputtered MEAs, respectively. It is believed that a slight difference between the two ohmic resistances could be attributed to the contact resistance between the sputtered Ru layer on Pt/C and GDL. Since the compatibility between the carbon (Pt/C) and another carbon (GDL) is higher than that between metal (mostly Ru) and carbon (GDL), it could tolerate the increased ohmic resistance. Note the ionic resistance is well known as the major factor for the increase in the ohmic resistance; however, if the clamping force is lower than the conventional system such as in the present study, the electrical resistance can no longer be ignored. However, such increased ohmic resistance of Ru-sputtered MEA does not affect the final electrochemical performance, when the charge transfer resistance is considered. The charge transfer resistance, which is directly related to the catalytic activity inside the MEA, was 7.5 Ω·cm² for only Pt/C case and approximately 30.2 Ω·cm² for Ru-sputtered Pt/C case. This indicated that the sputtered Ru positively affected the activation of the whole catalyst inside the fuel cell. Moreover, according to Fig. 4(b) that depicts the impedance spectra obtained after 1 hour of operation, the ohmic resistance did not change in both MEAs, whereas the charge transfer resistances for both MEAs increased, implying deterioration of the catalyst of the MEA. However, the Ru-sputtered MEA exhibited a relatively little increase in charge transfer resistance as compared to that of the standard MEA. It indicates clearly that the deterioration of the catalyst was effectively increased by sputtering Ru onto Pt/C. Moreover, the amount of the deposited Ru was 80% lower than that of the sprayed Pt, showing great possibility and potential of reducing the fabrication cost of MEAs for DMFCs.

To further confirm the efficacy of Ru-sputtering method described in the present study, the catalytic durability was monitored in real-time by measuring the variations in OCVs for two MEAs, as shown in Fig. 5. The initial OCV of standard MEA was 0.448 V and was continuously reduced to 0.37 V for an hour. In other words, the fuel cell performance and corresponding catalytic activity decreased by 17.4%. However, in the case of Ru-sputtered MEA, the initial OCV was 0.431 V, which reduced to 0.382 V after an hour, a 11.2% decrease. It indicated that the polarization characteristics, the corresponding EIS, and the real-time variations in the OCV proved the positive functionality of Ru sputtering onto the Pt/C and its effect on increasing both the performance and the catalytic durability.

In the present study, a novel method of reducing the use of Ru while increasing the electrochemical performance as well as the catalytic durability of DMFCs using high concentration (4 M) methanol solution fuel was suggested. An 80% reduction in the use of Ru was achieved by simply sputtering it onto the Pt/C on an electrolyte membrane. Moreover, the electrochemical performance and catalytic activity increased, as confirmed by voltage–current characterization and EIS analysis. In addition, a real-time monitoring of the OCV variation was conducted to confirm these results. It is believed that this study could contribute to reducing the use of the novel metal in DMFC, simplifying the fabrication process, and finally promoting the commercialization of the passive DMFC technology.

This work was supported financially by an NRF grant funded by the Ministry of Science and ICT (2017R1C1B5076732).