Investigation of Pt, Pt₃Co , and Pt₃Co / Mo Cathodes for the ORR in a Microfluidic H₂ / O₂ Fuel Cell

All modification procedures started with 3–5 nm unsupported (E-TEK) nanoparticles based upon manufacturer specification. To obtain the nanoparticles, the mixture of as-received unsupported nanoparticles, phosphomolybdic acid , and Millipore water was sonicated (3510 Branson) for 15 min before refluxing at under constant stirring and argon (Ar) flow for 1 h. After cooling the nanoparticle mixture to room temperature, 25 mL was added in a dropwise fashion under sonication. The solution was reheated to under constant stirring and Ar flow for 10 min and was then cooled to room temperature. The resulting nanoparticles were filtered from the solution and rinsed with Millipore water to remove all unwanted residuals.

To obtain the acid pretreated nanoparticles, as-received was loaded into a gold (Au) boat, which was carefully filled with 0.5 M so as not to dislodge the catalyst particles on the bottom of the boat. The catalyst was then cycled 40° times between 0 and 1.40 V vs reversible hydrogen electrode (RHE) at a scan rate of 50 mV/s.³² After each modification, both the resulting catalysts, and -at, were filtered from the solution and rinsed with Millipore water to remove all unwanted residuals

XPS studies were performed on all -based catalysts before and after modification procedures to confirm the changes in the surface structure and binding energies (BEs). A combined electrochemical and ultrahigh vacuum (UHV) XPS system was used so that electrochemical and XPS measurements could be conducted without sample exposure to the ambient environment during transfer between the electrochemical and UHV chambers.^{32, 38} For the XPS characterization, 5.0 mg/mL suspensions of desired catalyst nanoparticles in Millipore water were prepared. After sonication for 30 min, of catalyst suspension was deposited onto a gold disk, and the XPS measurements were taken. Using the M-Probe ESCA software, version 1.36 (Fisons Surface Science), the XPS spectral peaks were fitted using a mixed Gaussian–Lorentzian line shape and Shirley baselines. The BE scale was calibrated using (at 84.0 eV) and (at 932.68 eV).³⁹

Cyclic voltammogram (CV) measurements were carried out using a three-electrode electrochemical cell with a rotating disk electrode (RDE, Autolab, EcoChemie) with a Au tip of 3 mm diameter (polished down to ), a Ag/AgCl (, BAS) reference electrode (in a separate compartment), and a Pt wire (Alfa Aesar) counter electrode. Measurements were performed with a potentiostat (Autolab PGSTAT 100, EcoChemie). A 5.0 mg/mL suspension of the desired catalyst nanoparticles, either Pt black or -based alloys, in Millipore water was prepared. After sonication for 30 min, of the catalyst suspension was deposited on the RDE. The geometric surface area of the RDE was calculated to be , and this value was used for the calculation of the current densities for the CVs. The CVs were measured in 0.5 M which was bubbled with nitrogen for 30 min before testing. For the -based alloys, voltammograms were measured from 0 to 1.2 or 1.4 V vs RHE at a scan rate of 50 mV/s. For the comparative Pt black, the voltammogram was measured from 0 to 1.4 V vs RHE at a scan rate of 20 mV/s. The RDE was stationary for all measurements.

Four different cathode catalysts were investigated: as-received (, E-TEK), Pt black (Pt, Alfa Aesar), (-at), and . For all studies, Pt black was used as the anode catalyst. Each catalyst ink was prepared by mixing 12 mg of the desired catalyst, 0.4 mg Nafion (5 wt % solution, Aldrich) as a binder, Millipore water, and isopropyl alcohol. The catalyst ink was sonicated (Branson 3510) for 2 h to obtain a uniform mixture, which was then painted on a Toray carbon paper gas diffusion layer (EFCG "S" type electrode, E-TEK) to create a gas diffusion electrode (GDE). The electrodes (geometric surface area: ) were then hot-pressed (Carver 3851-0) at 340 psi and to ensure that the catalyst layer was compacted into the gas diffusion layer. All GDEs, both anode and cathode, had the desired catalyst and Nafion.

Two GDEs, anode and cathode, were placed on opposite sides of a 0.15 cm thick poly(methyl methacrylate) sheet, such that the catalyst-covered sides interfaced with the 3 cm long and 0.33 cm wide precision-machined window. The window had an inlet and an outlet on either side such that the aqueous sulfuric acid (, GFS Chemicals, Inc., double distilled by Vycor) electrolyte flows between the electrodes. Two 0.1 cm thick graphite windows served as the current collectors. The hydrogen and oxygen gas flow chambers were precision-machined into polycarbonate sheets. This multilayered assembly was held together using binder clips (Highmark). Fuel cell studies were conducted using a potentiostat (Autolab PGSTAT-30, EcoChemie) at room temperature with hydrogen and oxygen gas (laboratory grade, S.J. Smith), each at a flow rate of 50 sccm. Electrolyte flow rates were varied from 0.0 to 0.9 mL/min using a syringe pump (Harvard apparatus). The concentration of was varied from 0.25 to 3.0 M (approximate pH range of 0.6 to −0.48). Fuel cell polarization curves were obtained by steady-state chronoamperometric measurements at different cell potentials. All studies were conducted four times to ensure the immediate chemical and electrocatalytic stability of the cathodic materials under the acidic fuel cell operating conditions. Because all catalyst loadings were identical, , the exposed geometric electrode surface area was used to calculate the current and power densities for the microfluidic fuel cell tests. The aqueous electrolyte stream exited the fuel cell through a plastic tube (Cole Parmer, 1.57 mm i.d.) and was collected in a beaker. The polarization losses on the anode and the cathode were independently analyzed using multimeters (Fluke) by placing a Ag/AgCl reference electrode (, BAS) in the collection beaker.^{33, 34, 37} No potential drop was observed along the plastic tubing connecting the fuel cell and the reference electrode.³⁷

Electrochemical impedance spectroscopy (EIS) measurements were performed on the acidic microfluidic fuel cell operated with the five different cathode catalysts. AC impedance spectra were obtained using a frequency response analyzer module controlled by a potentiostat (Autolab PGSTAT-30, EcoChemie). The spectra were measured at a constant voltage mode by decreasing frequencies from 10 kHz to 30 mHz at 9 points/decade. The modulating voltage was 10 mV root-mean-squared. The measurements were performed at cell potentials ranging from 100 to 800 mV. The resulting Nyquist plots isolated the effects of the internal cell resistance and fuel cell electrode charge-transfer resistance . Each measurement was performed three times.

By cycling electrodes across a potential range, catalyst dissolution rates were accelerated, leading to rapid performance losses due to a reduced electrochemical surface area. The rate and extent of these losses depended on the potential ranges and the cycle patterns (i.e., square waves vs cycles). Here, cathodes were aged in a three-electrode electrochemical cell via cyclic voltammetry under acidic conditions. Each cathode was cycled over a range of fuel cell operating potentials (0.10–1.00 V vs RHE) in 0.5 M for 1000-cycle sets at a 50 mV/s scan rate. After each set of 1000 cycles, the electrodes were rinsed with Millipore water to remove any residual acid. Between cycling sets, the cathodes were tested in an acidic microfluidic fuel cell to investigate the effects of aging on the electrode performance.

XPS studies were performed on the as-received samples and the new samples, -at and , to characterize the surface chemical composition and to confirm the success of the modification procedures. The comparative XPS survey spectra of the commercial alloy before and after cyclic pretreatment under acidic conditions are shown in Fig. 3. For the commercial sample, a nominal atomic Pt:Co ratio of 3:1 was determined from the XPS peak analysis (Fig. 3a), confirming the quality of the as-received sample. Cobalt (Co) peaks arise at 800 and 780 eV, corresponding to a peak and a peak, respectively.^{32, 40} The strong oxygen peak at 531 eV, along with the positive shift in the BE of the Co peaks, indicates a cobalt oxide surface species. Metallic Co is present as a shoulder in the peak at 778 eV. Further quantitative analysis of the Co signals shows that is cobalt oxide and is metallic Co (Table I). After the acidic electrochemical pretreatment, the XPS spectrum shows a disappearance of the Co 2p and the O 1s peaks (Fig. 3b), but the platinum and the metallic Co peaks remain, which indicates that the surface cobalt oxides have dissolved.³² This dissolution of the metal oxide species under acidic conditions causes a decrease in the oxygen reduction activity of the Pt–Co alloy.² However, the dissolution of the surface cobalt oxides from the nanoparticle surface leads to the formation of a core-shell catalyst particle or a skeleton structure with a Pt surface and a Co-rich layer beneath, enhancing the activity of the Pt–Co catalyst.^{21, 24} Here, the formation of a core-shell structure is further supported by the CV and XPS data. CVs of the as-received and pretreated sample (Fig. 5a) show the removal of the Co surface species of the pretreated sample, as it now resembles the CV of pure Pt (Fig. 5b). Further XPS peak analysis of the acid pretreated sample confirms the removal of all cobalt oxide species, yet Co is still present as metallic Co, beneath a protective Pt skin, which agrees with the CV (Fig. 5a), which lacks Co peaks; only those of Pt remain (Table I). More detailed XPS and electrochemical analyses of the , -at, and Pt catalysts may be found in a previous publication by Duong et al.³²

Zoom In Zoom Out Reset image size

Table I. Quantitative surface analysis of as-received before and after electrochemical pretreatment in 0.5 M .

		-at
Pt:Co ratio	3:1	36:1
Cobalt oxide concentration (%)		0
Metallic cobalt concentration (%)		100

Zoom In Zoom Out Reset image size

The addition of Mo was added to the catalysts to test the activity and stability, similar to the addition of Mo to Pd-based catalysts.^{35, 36} XPS survey spectra (Fig. 4) and CVs (Fig. 5c) were also performed to verify the successful molybdenum (Mo) deposition onto the . Though difficult to discern from the broader survey spectra shown in Fig. 4a, small dual Mo peaks exist at 231 and 236 eV, confirming the deposition (Fig. 4b). Based upon the position of the BE peaks, Mo exists on the catalyst surface both in metallic form ( peak at 231 eV) and in oxide form (possibly peak at 236 eV).⁴⁰ From the peak analysis, nominal atomic ratios of 3:1 and 37:1 are determined for Pt:Co and Pt:Mo, respectively. Previous work by Raghuveer et al. and Sarkar et al. on the oxygen reduction activity of Pd–Co–Mo and Pd–Mo, respectively, demonstrated that catalytic performance decreased if alloys contain greater than 10 atom % Mo.^{35, 36} This decrease in the catalytic activity would be due to a reduction in electrochemically active surface sites, as the ORR does not occur on Mo. Therefore, a low deposition of the transition metal is favorable and is reflected in our catalyst with a to 3 atom % Mo.

Zoom In Zoom Out Reset image size

The Mo addition to the catalyst was also apparent in the CVs of and in 0.5 M (Fig. 5c). Distinctive oxidative and reductive peaks are observed around 0.5 V vs RHE on the catalyst and indicate the presence of Mo species. This observation agrees well with the aforementioned XPS measurements and previously reported results.⁴¹ Furthermore, the catalyst immediately demonstrates improved stability compared to the as-received catalyst under acidic conditions. Over 100 cycles, the performance remains consistent with only marginal decreases in the Pt and Co oxide as well as Mo peaks, indicating minimal dissolution.

The performance of an acidic microfluidic fuel cell operated with hot-pressed Pt electrodes, both anode and cathode, was investigated as a function of electrolyte flow rates (0.0, 0.3, 0.6, and 0.9 mL/min). Hydrogen and oxygen flow rates were 50 sccm each to ensure an adequate fuel and oxidant supply.³³ As shown in Fig. 6a and 6b, the polarization curves suggest that cell performance is limited by the cathode and is weakly dependent on the electrolyte flow rate. In prior work, we described in detail performance variations as a function of flow rate in this microfluidic fuel cell, specifically with respect to water management at the cathode.³³ At high current densities, performance improvements are observed when utilizing a flowing electrolyte compared to a stationary electrolyte because the dynamic stream facilitates water removal on the cathode. Individual electrode polarization curves confirm that any performance enhancements are due to improved water management at the cathode as the anode performance appears flow rate independent (Fig. 6b). Small enhancements are observed with increasing flow rate until 0.6 mL/min after which performance plateaus. At this point, the microfluidic configuration shifts from a mass-transfer limited regime, due to water removal from the cathode, to a kinetically limited regime, limited by ORR kinetics at the cathode. Cell peak power densities confirm an optimal electrolyte flow rate of 0.6 mL/min after which cell performance levels off (Fig. 6c). Furthermore, identical trends were observed with fuel cells operated with and cathodes, confirming that in the current configuration, ORR kinetics may be observed without significant interference from mass-transfer limitations at an electrolyte flow rate . This flow rate dependence is also a function electrolyte concentration. Thus, while a flow rate of 0.6 mL/min is a transition point for 0.5 M , this transition point flow rate would be reduced for lower concentrations and be elevated for higher concentrations. Also, the effects of hot-pressing electrodes on transport processes within the GDE and consequent impact on the fuel cell performance are yet to be determined and will be addressed soon.

Zoom In Zoom Out Reset image size

Analyses of the four Pt-based cathodes were performed in a microfluidic fuel cell with a flowing acidic electrolyte of varying concentrations. Figure 7a shows comparative polarization and power density curves for the fuel cell operated with each cathode, and all cathodes demonstrated good oxygen reduction activity and chemical stability. Fuel cells operated with either or cathodes exhibited enhanced performance over those with either -at or Pt cathodes, and the performance shifts are due to changes in the oxygen reduction activity of the cathode catalysts (Fig. 7b). The core-shelled -at cathode demonstrates slightly greater ORR activity than the Pt cathode, suggesting that the electrocatalytic properties of the core-shell nanoparticle are very similar to pure Pt despite the 25% reduction in Pt content. The cathode exhibits slightly greater ORR activity than the cathode and significantly greater ORR activity than the Pt cathode, suggesting that the Mo addition did not adversely affect catalytic activity.

Zoom In Zoom Out Reset image size

Further in situ studies were performed at varying concentration from 0.25 to 3.0 M. The purpose of these studies is to analyze cathode performance and stability as a function of acid pH. The performance of the different cathodes is analyzed at peak power density to simulate the high current densities typically observed in working fuel cells. As evident in Fig. 7b, the anode polarization curves are identical from experiment to experiment; thus, changes in cell performance may be attributed exclusively to different cathodes. Thus, while varying electrolyte concentration impacts the fuel cell flow rate dependence, comparative analyses may be performed on different cathode materials. Acidic microfluidic fuel cells operated with or cathodes significantly outperformed fuel cells operated with Pt or -at over the range of concentrations (Fig. 8). Subtle effects on the fuel cell performance due to changes in acid concentration for Pt and -at are shown in Fig. 8a, and those due to changes in acid concentration for and are shown in Fig. 8b. At concentrations , cell performance was limited by the increased cell resistance associated with the low electrolyte conductivity at these concentrations. At concentrations , cell performance leveled out and slightly decreased as increased sulfate/bisulfate poisoning reduced the active catalytic surface area available for reaction.^{42, 43} Furthermore, when operating the fuel cell with either or cathodes at high acid concentrations, when performance exceeds , enough water was produced that the transport processes become insufficient and the cathode flooded. For milder acidic conditions , the -at and Pt cathodes showed similar catalytic activities (Fig. 8a). At harsher conditions , the Pt cathode showed a slightly higher activity than the -at cathode, which may be due to a greater number of active sites on the Pt cathode surface compared to the number of sites available on the -at cathode surface. As shown in Fig, 8b, cathodes slightly outperform cathodes over the entire range of acidic conditions, suggesting that the Mo addition improves ORR activity. IR-correction analyses confirmed that all cathode performances were pH-independent for the pH range of 0.6 to −0.48 (0.25 to 3.0 M ).

Zoom In Zoom Out Reset image size

To test the long-term durability of the catalysts studied, accelerated cathode aging studies were performed in a three-electrode electrochemical cell where the electrodes were cycled multiple times over a range of fuel cell operating potentials, 0.1–1.00 V vs RHE, in 0.5 M . Potential cycling is a widely used technique in fuel cell durability studies, both ex situ (i.e., RDE and GDE) and in situ (i.e., PEM-based fuel cells), as potential cycling under acidic conditions has been shown to expedite catalyst and electrode degradation.^{3, 29–31, 44–52} First investigated by Kinoshita et al. , potential cycling reduces the electrochemical active surface area of catalytic particles via metal dissolution into the acid and growth of larger particles at the expense of smaller particles, analogous to Ostwald ripening.^{31, 45} Reduced catalyst surface area decreases absolute activity and lowers fuel cell performance. The rate and extent of electrochemical surface area losses depend on potential ranges and on cycle patterns (i.e., square waves vs cycles). Previous studies have shown, at potentials greater than 0.8 V vs RHE, increased dissolution rates, and, consequently, reduced catalyst and fuel cell performances are observed for the Pt and Pt–M alloy catalysts in both in situ and ex situ studies.^{31, 47–52}

After each 1000-cycle set, cathode performance was investigated in an acidic microfluidic fuel cell. Peak power densities for the fuel cell operated with each cathode studied as a function of cathode aging are shown in Fig. 9a. Over the course of the 10,000 cathode aging cycle, the peak performance of Pt, , -at, and decreased by 11.8, 33.2, 1.9, and 8.3%, respectively (Table II). Both modified cathode catalysts, -at and , demonstrated superior durability, compared to Pt and , in acidic environments under fuel cell operating potentials. Over the course of 10,000 cycles, the performance of the as-received activity reduces to that of the -at, which seems to indicate a complete loss of cobalt oxide species from the catalytic surface of the core-shelled particle. Despite a modest overall improvement in catalytic performance, the enhanced stability and reduced Pt content make the core-shelled a viable alternative to pure Pt as a fuel cell cathode catalyst.

Zoom In Zoom Out Reset image size

Table II. Peak power density losses in an acidic microfluidic fuel cell as a function of cathode aging before and after 10,000 cycles. Accelerated aging studies performed in a three-electrode electrochemical cell in 0.5 M cycling between 0.10 and 1.00 V vs RHE at a 50 mV/s scan rate. For all studies, a fresh Pt black anode was used.

Cathode catalyst	Peak power density		Decrease (%)
	0 cycles	10,000 cycles
Pt	121.6	107.2	11.8
	176.0	117.6	33.2
-at	124.4	122.0	1.9
	177.6	162.8	8.3

The dramatically enhanced durability of the cathode compared to that of the cathode is a very promising result. By performing EIS measurements on the acidic microfluidic fuel operated with fresh (0 cycles) or aged (10,000 cycles) and cathodes, the root causes of the improved durability may be determined. Figure 9b and 9c shows the representative Nyquist plots of the fuel cell studies conducted with fresh and aged and cathodes, respectively, at a cell potential of 0.4 V, the potential at which peak power densities were observed. For all experiments, the same fresh Pt anode was used; thus, comparative shifts in the impedance spectra can be attributed to the changes in the different cathode materials. The high frequency intercept on the -axis (initial data point closest to the origin in Fig. 9b and 9c) represents the cell resistance , which includes electrolyte solution resistance and any ohmic resistances in the fuel cell components (i.e., catalyst layer, gas diffusion layer, and current collectors).⁵³ As the experimental setup and operating conditions are identical, the differences between the measured cell resistances and the calculated electrolyte solution resistance ( in 0.5 M )⁵⁴ can be attributed to the changes in cathode internal resistance, which varies as a function of electrochemical aging and the catalyst used. In general, as the increases due to greater electrode resistances, the electrode and overall fuel cell performance decrease. An increase in , as a function of aging, is observed for fuel cells operated with the cathode (Fig. 9b). No increase in is observed for fuel cells operated with the cathode (Fig. 9c). This increase in may be attributed to Co leaching from the unsupported nanoparticles under acidic fuel cell operating conditions, adversely affecting adhesion between the catalyst and gas diffusion layers, which increases contact resistances. These qualitative observations suggest that the Mo addition to the nanoparticles prevents leaching of Co species.

The two fuel cell electrode reactions, represented by the semicircular features in each Nyquist plot, can be described by two parallel circuits with different time constants. Here, is a constant phase element capacitance describing the porous, distributed nature of the GDEs and is the overall fuel cell charge-transfer resistance represented by the diameter of the semicircle. In acidic media, hydrogen oxidation reaction kinetics are rapid and contribute minimally to the overall response.⁵⁵ Thus, the sluggish cathodic ORR kinetics are the dominant contribution to the measured . An increase in , as a function of aging, is observed for fuel cells operated with the cathode (Fig. 9b). No increase in is observed for the fuel cells operated with the (Fig. 9c). This increase in can again be attributed to Co leaching for the unsupported nanoparticles, which reduces the ORR activity, thus lowering the cathode and overall fuel cell performance. These qualitative observations suggest that the retains a high ORR activity, despite electrochemical aging, by minimizing Co leaching from the Pt alloy.

To quantitatively analyze the trends shown in the representative Nyquist plots in Fig. 9b and 9c, all experiments were performed in triplicate to account for any cell-to-cell variations resulting from (i) differences in assembly/disassembly and (ii) differences in exact operating parameters due to the nature of the syringe pump used to drive the electrolyte flow. The critical and values as well as the total resistance , as shown in Table III, for fuel cells operated with fresh and aged and cathodes. With the as the cathode, the measured increased from to as a function of electrochemical aging. This corresponds to an increase in in the . In contrast, when using cathodes, the measured increased from to due to aging. This negligible increase in in the confirms that the cathode structure does not degrade significantly over the course of the electrochemical aging.

Table III. Cell resistance , charge-transfer resistance , and total resistance for microfluidic fuel cells operated with fresh (0 cycles) and aged (10,000 cycles) and cathodes . In all studies, hydrogen and oxygen flow rates were 50 sccm each, electrolyte flow rate was 0.6 mL/min, was 0.5 M, and experiments were performed at room temperature. For all studies, a fresh Pt black anode was used.

Cathode catalyst		()	()	()
	Fresh
	Aged
	Difference
	Fresh
	Aged
	Difference

Furthermore, while operating the fuel cell with cathodes, the measured increased from to as a function of aging (Table III). This corresponds to an increase in in the . In contrast, when using the cathodes, the measured changed from to as a function of aging. This negligible increase in in the confirms that the cathode activity does not reduce significantly over the course of the electrochemical aging. The quantitative observations suggest that cathodes outperform cathodes by minimizing transition-metal leaching for the catalyst surface. Also, the effects of Co leaching are twofold: (i) an increase in the internal electrode resistance and (ii) a decrease in the ORR activity. Consequently, the improved performance observed for the cathodes may be attributed to the dual effects of enhanced electrode and catalyst stability, but further work would be needed to confirm this.