Dealloyed Pt Nanoparticle Fuel Cell Electrocatalysts: Stability and Aging Study of Catalyst Powders, Thin Films, and Inks

For the entire study, a single homogeneous batch of a carbon-supported (Pt–Cu/C) material was prepared, which served as the precursor of the dealloyed active phase of the electrocatalyst. The precursor was synthesized using a liquid metal salt impregnation method followed by freeze-drying and thermal annealing. This method had been applied repeatedly for the synthesis of binary^{30, 40, 41, 44–46} and ternary⁴² Pt alloys. In short, the solid Cu precursor [, Sigma Aldrich] was dissolved in deionized water (millipore) and added to commercially purchased, high surface area, carbon-supported Pt nanoparticles (catalyst part no. TEC10E30E supplied by Tanaka Kikinzoku Inc., Japan). The mixture was then ultrasonicated (Branson Sonifier 150 with sonifier horn) and frozen in liquid nitrogen. The frozen sample was then freeze-dried (Labconco, 12 L tray dryer, 0.05 mbar) in vacuum overnight at room temperature. The powders obtained were then annealed (Lindberg Blue flow furnace) to maximum temperatures of for 7 h under a 4% hydrogen atmosphere (argon balance) and slowly cooled down to room temperature by natural convection. The catalyst precursor used for this study has a composition of and had a final Pt weight loading at around 22 wt%. This precursor would thus be referred to as " (800/7)." The precursor powder was generally stored under flowing dry nitrogen.

A catalyst ink was prepared by mixing weighed amounts of the catalyst precursor powder in 10 mL of a degassed aqueous solution containing 5 wt% Nafion (Sigma Aldrich) solution and 2-propanol (Sigma Aldrich). The suspension was ultrasonicated for 15 min at room temperature (Branson 150) before of aliquot was dispensed onto the glassy carbon (GC) surface of a rotating disk electrode (RDE) (PINE Instrument). The ink was then dried for 10 min in a atmosphere resulting in a thin film on the GC surface. Film thicknesses were estimated to range between 0.1 and , using the dry density of Nafion and neglecting the volume of the catalyst. The dried films were kept under nitrogen until the electrochemical characterization commenced.

The electrochemical cell used was a custom-made, three-compartment glass cell. The working electrode was commercially obtained with a 5 mm fixed diameter GC surface. The counter electrode was a piece of Pt gauze (Sigma Aldrich). The reference electrode was a mercury–mercurous sulfate electrode (Princeton Applied Research, AMETEK) held in place by a Luggin–Haber capillary. All electrode potentials reported were converted into the reversible hydrogen electrode (RHE) scale. A commercial dual-channel potentiostat (BioLogic) and a rotator (PINE Instrument) were used to conduct the rotating disk experiment. The electrolyte used was 70% redistilled (Sigma Aldrich) and diluted to a 0.1 M concentration. All measurements were conducted at room temperature, and the working electrode was immersed into the electrolyte under potential control at 0.05 V until actual measurements commenced. Cyclic voltammetry (CV) was performed in a deaerated electrolyte under atmosphere. The voltammetric response of the electrocatalysts was first measured during the initial three CV scans between 0.05 and 1.2 V at a scan rate of 100 mV/s to obtain the initial rapid Cu dissolution profiles. Then, the catalyst films were further pretreated using 200 CV scans between 0.05 and 1.2 V at a scan rate of 500 mV/s during which a large amount of Cu was lost from the alloy nanoparticle precursors. Thereafter, the platinum electrochemical surface area (Pt-ECSA) of each catalyst was determined by cycling the pretreated catalysts at 100 mV/s between 0.05 and 1.0 V. The Pt-ECSA of each catalyst was determined using the mean integral charge of the hydrogen adsorption and desorption areas with the double-layer current corrected for, using , assuming 1 H atom is adsorbed to 1 Pt atom. Linear sweep voltammetry (LSV) measurements were conducted in oxygen-saturated electrolyte by sweeping the potential from 0.06 V anodically to the open-circuit potential at a scan rate of 5 mV/s. The ORR activities of the dealloyed, activated catalysts were corrected for mass transport limitation using Eq. 5 in Ref. ²². This equation is the simplest mass transport correction and does not explicitly include a check of whether the -intercept of the Koutecky–Levich plot ( vs ) vanishes at high enough electrode potentials. Mass activities and specific activities were then established at 900 mV at room temperature using the relation. The carbon-supported 30 wt% platinum electrocatalyst used as a synthesis precursor is also used as activity and ECSA baseline in the aging experiments.

The stability and aging of the carbon-supported (800/7) alloy nanoparticle catalyst precursor under ambient conditions and in air were studied by weighing four 5 mg samples of the material into four 20 mL scintillation vials. Filter disks were placed over the mouths of the vials and were kept in place using open-top screw caps. Filter disks were used to facilitate air exchange while preventing contamination from dusts or other solid impurities. All four samples were kept in a well-vented part of a clean room for the entire experimental duration. The first vial of sample was used for electrochemical characterization immediately without being aged in air (air aged for 0 days in Fig. 3). The other three vials of the sample were consecutively used for electrochemical characterization after being exposed to air for 5, 12, and 26 days (air aged for 5, 12, and 26 days, respectively). Each powder sample was made into a catalyst ink and immediately dispensed onto the GC surface of an RDE before being analyzed electrochemically. The vials of ink were then returned to the aging storage location for use in the thin-film and ink stability experiments.

Zoom In Zoom Out Reset image size

Each of the four ink samples used for the powder stability experiments (Power Stability and Aging Studies section) was returned to their aging location after film dispensing. The following day, the ink was sonicated in a water bath for 15 min at room temperature before being dispensed and cast into a thin film onto the GC surfaces of five different RDE shafts. The five RDE thin films were dried and kept in the aging location. The first thin film was analyzed electrochemically immediately upon drying (age of film: day 2 in Fig. 4), while the other four RDEs were analyzed consecutively for the next four days (age of film: days 3–6 in Fig. 4).

Zoom In Zoom Out Reset image size

Each catalyst ink sample was electrochemically characterized for a total of six consecutive days, with the first day being when the powder sample was made into an ink. After the first day (age of ink: day 1) and for the next five days (age of ink: day 2–6), each ink sample was sonicated in a water bath for 15 min at room temperature immediately before being dispensed and cast into a thin film on the GC surface of an RDE. All dried thin films were immediately tested electrochemically to study the stability of the catalyst ink solution. Experimental measurement errors were evaluated in occasional repeat experiments. A relative experimental error of 5–10% in the ECSA values was determined in repeat experiments of the ink stability behavior.

In studying the aging effects on the (800/7) catalyst precursor material, four freshly prepared, weighted powder samples were taken out of their storage box and exposed to a controlled ambient pressure (0.2 atm oxygen partial pressure), an ambient temperature of about 25°C, and for 0, 5, 12, and 26 days from the time of synthesis. The powders were exposed to ambient conditions such that no solid contamination (dust particles, etc.) were able to deposit onto the powder sample. These precursor powder samples are thereof referred to as "Pt–Cu– days–P," where , 5, 12, 26. After the controlled aging experiment, each powder sample was used to prepare a catalyst ink, which was immediately dispensed onto the GC disk of an RDE. Each thin-film catalyst went through an identical electrochemical pretreatment (dealloying) protocol, and the resulting Pt-enriched Pt–Cu/C thin-film catalyst was analyzed for its ECSA and its electrochemical reactivity for the ORR.

Figure 1 shows the final CV profiles of the four aged powder samples (dashed lines) and compares them with that of a standard pure Pt/C electrocatalyst, which had been stored under . The CV profiles of all Pt–Cu catalysts resemble one another but exhibit a significantly smaller under-potentially deposited hydrogen () profile than Pt/C between 0.05 and 0.3 V, consistent with the previous work.^{23, 24, 30, 41} The smaller profile is probably due to the larger Pt–Cu particle sizes after the annealing process. In addition, the shapes of the profiles of Pt–Cu/C are similar to that of Pt/C even though the relative heights of the peaks are slightly altered. The shapes of the regions of all air-aged Pt–Cu/C samples are also identical despite the different exposure durations to air. This indicates that the contact with air does not alter the relative affinity of the different Pt facets present on the surface of the nanoparticles to hydrogen adsorption. The massive Cu dealloying process during the activation of the catalysts is likely to offset any differences in Cu oxide formation during the aging.

Zoom In Zoom Out Reset image size

Figure 2 compares the LSV profiles and the Tafel plots of the Pt–Cu/C and Pt/C catalysts. All Pt–Cu/C catalysts show a relatively well-defined flat plateau in the region of the diffusion-limiting current between 0.1 and 0.7 V. The value of the transport-limited current is consistent with earlier studies.^{32, 33, 39–41, 47} Onset potentials from a mixed kinetic-diffusion control region to a diffusion-limiting region occurred at around 0.7 V. All Pt–Cu/C catalysts exhibited a significant shift in the mixed region toward more positive potentials compared to the pure Pt/C. The potential shift is about and is again consistent with the delayed onset of oxide formation observed from the CVs during the anodic scan. Tafel plots (inset of Fig. 2) indicate significant intrinsic activity improvements of all Pt–Cu/C catalysts regardless of the duration of their exposure to air. However, the significant differences in the Tafel slopes at high current densities (hcd's) suggest the influence of mass transport effects reflected in the diffusion-limited regions of the air-aged Pt–Cu/C samples.

Zoom In Zoom Out Reset image size

The Tafel slopes at the low current density (lcd) region are close to the theoretical Tafel slopes of 60 mV/dec and that observed from bulk and single crystals reported previously.^{22–24, 35} In the hcd region, the Tafel slopes are slightly larger than those obtained from the Pt reference samples. The reason for the change in the Tafel slopes between lcd and hcd had been in much discussion. Gasteiger et al. and Paulus et al. explained that the changes in the Tafel slope are due to the changes in the oxide species adsorption on the catalyst surface and to an increase in inaccuracy in the RDE mass transport corrections near the diffusion-limited current density, respectively.^{22, 23}

The Pt mass-based transport-corrected kinetic current density at 0.9 V (Pt mass-based activity) , the real surface area normalized current density at 0.9 V (specific activity) , and the ECSA of the four aged powder samples and the pure Pt/C sample are compared in Fig. 3. From the Pt-ECSA comparison plot (Fig. 3c), it is seen that Pt-ECSA increases as the duration of exposure to air increases, until the 12th day, and then decreases from the 12th to the 26th day. This observation can be attributed to a suboptimal interaction between the Nafion solution and the metal particles and carbon support pores (together controlling the catalyst utilization) of the Pt–Cu–0 day–P, leading to nonoptimized three-phase boundaries on the surface of the Pt–Cu nanoparticles. This, in turn, leads to some Pt–Cu nanoparticles not being in contact with the acidic electrolyte and oxygen, thus not participating in the ORR, lowering the observed catalytic activities.^{48, 49}

Both the and plots (Fig. 3a and 3b) show that all Pt–Cu/C samples exhibit significantly higher catalytic activities than Pt/C, with the highest being and being of the Pt–Cu–5 days–P material. These ORR activity values lie clearly above the technological target performance of the fuel cell cathode catalysts as set out by the Department of Energy for 2010 ⁵⁰ and are consistent with the high ORR activity reported earlier.^{30, 42, 43} The activities increased from Pt–Cu–0 days–P to Pt–Cu–5 days–P and remained about constant for Pt–Cu–12 days–P and Pt–Cu–26 days–P. Referring to Fig. 3a, it is not surprising that improved as Pt-ECSA increases because more Pt surface is available for ORR. In addition, (Fig. 3b) is observed to also follow the same trend. This is so despite an increase in Pt-ECSA for the samples air-aged from 0 to 12 days. However, when Pt–Cu nanoparticles were exposed to air for more than 12 days, a drop in both Pt-ECSA and absolute catalytic activities was observed. Because each of the powder samples originated from a single Pt–Cu/C parent source, any changes in electrochemical characteristics must be due to the effects of the oxidation of the nanoparticles in air. The Pt–Cu/C powder samples were thus stable in air for at least 12 days without adversely affecting their electrochemical properties.

Figure 4 shows the Pt mass-based ORR activity at 0.9 V , the specific activity at 0.9 V , and the Pt-ECSA of aged Pt–Cu/C catalyst thin films prepared from the four Pt–Cu/C powder samples described in the previous section. The nomenclature for the catalyst thin films used hereforth follows "Pt–Cu– days films" with denoting the number of days the precursor powder was exposed to air; i.e., "Pt–Cu–26 days film" would refer to a film made from the powder "Pt–Cu–26 days–P." Within seconds, a set of five ink-cast, dry films was prepared on five identical commercial GC electrodes from each aged precursor powder. The time elapsed between the preparation of the liquid catalyst ink and the casting of the catalyst thin film was on the order of a few hours. After their preparation, the individual catalyst thin films were aged under identical ambient conditions from up to 6 days, and one thin film each was catalytically tested on consecutive days.

From the Pt-ECSA plot of Fig. 4c, the active Pt surface area of the aged films generally decreases with time, evidencing a significant impact of dry age and air exposure on the precursor films. The significant drop in Pt-ECSA of the films ( to 50%) suggests that either interactions between air and surfaces of Nafion embedded Pt–Cu alloy nanoparticles or the irreversible aging processes, such as cracking, over the prolonged drying duration had altered the mesoscopic and microscopic film network structure such that the achievable charge after identical dealloying of the Pt–Cu nanoparticles decreased. We hypothesize that the continued film cracking would impact the structure and functionality of the proton-conducting network, which is critical in the discharge of protons at the catalytically active surface.

In terms of the ORR activities as a function of film age, there is an overall downward trend in both and for all four sets of five thin films each. The activity values, and , of all samples were significantly above that of our standard Pt/C at the beginning, yet they fell rapidly and approached the initial activity values of pure Pt/C by day 4.

An intraday ORR activity comparison between films from different powders (Fig. 4a and 4b) reveals a monotonously decreasing ORR activity with age, with the "Pt–Cu–12 days–film" exhibiting the highest activity among the different powders. This is in line with the measurement of Fig. 3 where that powder sample ranged among the most active when measured in the form of fresh ink samples. Surprisingly, the absolute values of the mass and specific activities are higher on 2-days old thin films compared to fresh ones ("day 1" in Fig. 3). In view of the combined drop of ECSA in Fig. 4c, this observation is even more surprising. On subsequent days, the experimental mass and specific activities fell rapidly and converged at the end of the experiment. The observed activity trajectories are consistent with the hypothesis that Pt-ECSA decreases with time due to degradation and cracking of the Nafion film network: The number of active precursor Pt–Cu nanoparticles participating in the dealloying and ORR catalysis gradually reduced, causing a drop in activities.

To study and understand the deterioration of the Pt–Cu/C films, the Tafel plots of "Pt–Cu–12 days–films" and "Pt–Cu–26–days films" are presented in Fig. 5 for all days of aging. The two sets of films were chosen because they represented the highest and lowest performing catalytic films over the duration of the experiment, respectively, not taking into account the "Pt–Cu–0 days–films." Figure 5a evidences that the lcd and hcd Tafel slopes of the set of films from the 12 day old power were essentially parallel to those of the initial Pt/C Tafel slopes. However, in Fig. 5b, it is observed that even though the Tafel slopes of this set of films at lcd were essentially parallel to Pt/C, those at hcd were much steeper than that of Pt/C. This increase in the Tafel slope typically indicates mass diffusion transport limitations of protons, water, or reactant molecules possibly caused by changes in the mesoscopic film network structure due to the prolonged drying of the catalyst film.³⁵ So, the detrimental processes of air and prolonged drying rendered the precursor Pt–Cu/C thin films unsuitable for the ORR activity measurement even after just a day.

Zoom In Zoom Out Reset image size

We also investigated the effect of aging and air exposure to liquid precursor powder inks, that is, liquid suspensions of the precursor powder mixed with a Nafion ionomer. As before, the Pt–Cu/C precursor inks were dispensed as thin films on a rotating GC electrode and immediately tested for ECSA and electrochemical ORR activity. Thereafter, the tested ink batch was returned to the controlled aging place and left there. This study covered an aging range of up to 6 days. The nomenclature of the inks used from hereon is "Pt–Cu– days ink," with number of days the original powder was left aging in ambient conditions. Figure 6 shows the experimental , , and Pt-ECSA values as the function of age of the four sets of inks made from the same four aged precursor powders.

Zoom In Zoom Out Reset image size

From the Pt-ECSA plot (Fig. 6c), it is observed that with the exception of the day 6 surface area of the ink made from the fresh precursor powder, the ECSA of each ink does not vary significantly with time. An intraday comparison of the Pt-ECSA values again reveals a maximum for the 12-days old powder and its inks. The liquid precursor ink is apparently a suitable format in which the favorable ECSA characteristics of the dry powder can be preserved for over 5–6 days. As seen above, dry films do not provide this capability. We hypothesize that the increase in Pt-ECSA suggests many participating Pt surface atoms in the hydrogen underpotential deposition process.

Looking closer at the ECSA values as a function of time, a moderate increase in the second day of measurement of the ink is observed for most powders. The trends in Fig. 6c were found reproducible in repeat experiments with experimental surface area values showing a 5–10% relative error. We hypothesize that the wetting of micropores and, hence, slightly enlarging the contact surface between the Nafion ionomer, metal particles and carbon support over the first 24 h of ink aging⁵¹ and that this could result in a moderate increase in ECSA. However, the ECSA increase is unlikely to be responsible for the evolution of the ORR activity of the aging ink because the catalytic ORR activities of the ink samples were as high as four to six times that of the standard pure Pt/C by day 2 and then converged back to that of pure Pt/C by day 6. Importantly, a significant increase in and activities was observed for most inks after 1 day of aging; the activities then remained essentially constant before they dropped on days 5 and 6. The activity of the ink made from the 12-days old powder exhibited a particularly significant activity boost after the first day ("day 2" in Fig. 6) but then gradually converged with the performances of other inks over the period of the experiment. Clearly, the moderate increase in the ECSA value on day 2 cannot be responsible for the significant ORR activity increases. There appear to be other activity controlling characteristics than merely the number of electrochemically accessible surface Pt atoms, which manipulate the observed ORR activity. We hypothesize that the improved three-phase interface between metal, carbon pores, and ionomer on day 2 may create more favorable conditions for the dealloying of the precursor material. The prolonged storage of the Pt–Cu alloy precursor in the ionomer ink may cause a premature molecular oxygen-induced chemical leaching of the Cu surface atoms, which maintains a fresh unoxidized metal surface and may result in a more complete leaching of the precursor during the electrochemical process.^{40, 45}

The observed value of about for the 12 day old powder on the second day of the ink represents one of the highest Pt mass activities ever reported for nanoparticle alloy electrocatalysts. This finding is likely to be of great importance for the development of practical ink preparation and membrane electrode assembly (MEA) building recipes.

The Tafel plots of inks made from Pt–Cu–0 days–P and Pt–Cu–12 days–P, tested for 6 consecutive days, are shown in Fig. 7. As with Fig. 5, the lcd region of the Tafel plots of all Pt–Cu/C samples were parallel to Pt/C but shifted toward an hcd expressing the favorable ORR activity of dealloyed Pt–Cu electrocatalysts. The evolution of the Tafel slopes of the two sets of inks over time is consistent with our previous observations and conclusions and sheds light on the evolution of the activity with ink age. The day 1 and day 6 ink made from the "Pt–Cu–0 days–P" powder deviated clearly from the Tafel slope of Pt/C at hcd. The inks aged for durations between 2–5 days were fairly parallel to the standard Pt/C. The same can be said of the ink sample prepared from Pt–Cu–12 days–P.

Zoom In Zoom Out Reset image size

This observation again suggests that catalyst films made from fresh and very old inks exhibit serious diffusion limitations and lower activities. Our studies suggest that the inks appear to require some moderate aging of about 24–48 h after preparation to unfold their full activity. This finding has important implications for the development of optimized ink preparation recipes for fuel cell catalyst inks and pastes. Again, prolonged aging beyond the optimum resulted in a poor film quality and severe diffusion limitations.

Our conclusions about the existence of an optimal ink age are corroborated by visual observations of the stability of the ink suspension. A suspension of poor stability appears clear with precursor powder settling to the bottom of the vial, as shown in Fig. 8. Figure 8a, 8c and 8e shows a "Pt–Cu–26 days–ink" on day 1 when the ink was freshly prepared, day 4, and day 6, respectively. Figure 8b, 8d and 8f reports a micrograph of the corresponding catalyst thin films. On day 1, the freshly prepared ink suspension was stable, evidenced by its uniformly black appearance. The carbon-supported catalyst particles remained suspended even in the absence of external stirring or sonication. The corresponding catalyst thin film was spatially very uniform in thickness across the 5 mm diameter GC electrode (Fig. 8b). The visual quality of the ink and film then remained unchanged over the following 2 days where the ink/film reached their optimum performance. As the aqueous Nafion catalyst ink aged further, the visual appearance of the catalyst ink changed in that small black carbon/catalyst agglomerates formed and stick to the wall of the vial. The ink therefore started to clear up. At the same time, the ink showed increased hydrophobicity evidenced by its increasing affinity to deposit and bind to the hydrophobic Teflon shaft of the RDE, in which the GC electrode surface was embedded. Dispensing of the ink onto the GC without forming spontaneous deposits on the Teflon became impossible (Fig. 8d). The film formed upon drying was nonuniform in thickness and electrode coverage. This increase in hydrophobicity suggested a change in the carbon black particle surface properties. Carbon blacks in the dry state are rather hydrophobic in nature; however, ultrasonication aided by polyelectrolytes can cause water molecules to enter the carbon pores and to make it behave rather hydrophilically. The data in Fig. 6 demonstrated that during days 4 and 5, the inks/films began to drop in catalytic ORR activities. By day 6 of the aging experiment, the borosilicate wall of the scintillation vial was fully covered by a thin layer of black catalyst agglomerates. The ink suspension was now almost completely clear. The catalyst thin film formed was nonuniform with large uncovered islands of the exposed GC surface. In Fig. 6c, a poor film quality did not seem to affect the Pt-ECSA of the samples; however, the nonuniform film exhibited very low and . In addition, Tafel slopes of the 6 day old ink exhibited significant deviations from the standard values indicated suboptimal mass transport.

Zoom In Zoom Out Reset image size