Platinum supported on titanium–ruthenium oxide is a remarkably stable electrocatayst for hydrogen fuel cell vehicles

Significance

In this study, we showcase titanium–ruthenium oxide (TRO)—a remarkably stable support material, and Pt/TRO, a derivative electrocatalyst that is also extremely stable. These materials have been tested to establish their catalytic activity and stability using accelerated tests that mimic conditions and degradation modes encountered during long-term fuel cell operation. We have evaluated Pt/TRO using these protocols, and provide concrete evidence that this material is far more stable than Pt/C and, would meet the requirements for use in an automotive fuel cell stack. Our cost analysis indicates the use of ruthenium is not a factor given that >90% of catalyst cost resides in the platinum metal; moreover, the exceptional stability of Pt/TRO removes the needs for overdesign or replacement.

Abstract

We report a unique and highly stable electrocatalyst—platinum (Pt) supported on titanium–ruthenium oxide (TRO)—for hydrogen fuel cell vehicles. The Pt/TRO electrocatalyst was exposed to stringent accelerated test protocols designed to induce degradation and failure mechanisms identical to those seen during extended normal operation of a fuel cell automobile—namely, support corrosion during vehicle startup and shutdown, and platinum dissolution during vehicle acceleration and deceleration. These experiments were performed both ex situ (on supports and catalysts deposited onto a glassy carbon rotating disk electrode) and in situ (in a membrane electrode assembly). The Pt/TRO was compared against a state-of-the-art benchmark catalyst—Pt supported on high surface-area carbon (Pt/HSAC). In ex situ tests, Pt/TRO lost only 18% of its initial oxygen reduction reaction mass activity and 3% of its oxygen reduction reaction-specific activity, whereas the corresponding losses for Pt/HSAC were 52% and 22%. In in situ-accelerated degradation tests performed on membrane electrode assemblies, the loss in cell voltage at 1 A · cm⁻² at 100% RH was a negligible 15 mV for Pt/TRO, whereas the loss was too high to permit operation at 1 A · cm⁻² for Pt/HSAC. We clearly show that electrocatalyst support corrosion induced during fuel cell startup and shutdown is a far more potent failure mode than platinum dissolution during fuel cell operation. Hence, we posit that the need for a highly stable support (such as TRO) is paramount. Finally, we demonstrate that the corrosion of carbon present in the gas diffusion layer of the fuel cell is only of minor concern.

Carbon has traditionally been the most common material of choice for polymer electrolyte fuel cell (PEFC) electrocatalyst supports (1) due to its low cost, high abundance, high electronic conductivity (30 S ⋅ cm⁻¹), and high Brunauer, Emmett, and Teller (BET) surface area (200–300 m² ⋅ g⁻¹), which permits good dispersion of the platinum (Pt) catalyst (2–4). The (in)stability of the carbon-supported platinum electrocatalyst is a key issue that currently precludes widespread commercialization of PEFCs for automotive applications. Carbon is known to undergo electrochemical oxidation to carbon dioxide, as shown in Eq. 1. The standard electrode potential for this reaction is 0.207 V vs. standard hydrogen potential (SHE; all potentials henceforth reported vs. SHE, unless otherwise stated). Under normal PEFC operating conditions, the electrode potential is <0.1 V at the anode and between 0.6 V and 0.8 V at the cathode. Despite the fact that the cathode potential is usually significantly higher than the standard potential for carbon oxidation (with an effective overpotential of 0.4–0.6 V), the actual rate of carbon oxidation is very slow due to intrinsic kinetic limitations; in other words, a very low standard heterogeneous rate constant (5).

During operation of automotive PEFC stacks, fuel/air mixed fronts are known to occur during stack startup and shutdown. Air usually fills the flow channels when the stack is nonoperational. During startup, the hydrogen fed into the stack displaces the air from the anode flow channels, leading to a mixed fuel–oxidant front that lasts for timescales corresponding to the residence time in the flow channel. These mixed-reactant fronts result in significant electrode polarization; under these conditions, the PEFC cathode can experience potentials as high as 1.5–2.0 V (6–8), which corresponds to a significantly higher overpotential for the carbon oxidation reaction (1.3–1.8 V as opposed to 0.4–0.6 V during traditional operation). The electrochemical reaction rate constant, which increases exponentially with overpotential, is significantly enhanced during this period. Under these conditions, carbon corrosion is exacerbated. In a second mechanism, fuel starvation at the anode catalyst sites as a consequence of fuel overutilization or flooding (lack of fuel access to catalyst site) also exacerbates carbon corrosion. In this case, carbon is oxidized to provide protons and electrons in place of the (absent) fuel (9, 10). From our previous work and that of others, it has been suggested that the presence of platinum accelerates the oxidation of the carbon support; the CO₂ generation rate was observed to be higher for catalysts with higher platinum loading, as measured with online infrared-based CO₂ measurement (11) and online mass spectroscopy (12–14).

The adverse consequences of carbon corrosion include (i) platinum nanoparticle agglomeration/detachment; (ii) macroscopic electrode thinning/loss of porosity in the electrode; and (iii) enhanced hydrophilicity of the remaining support surface. The former results in loss of catalyst active surface area and lower mass activity resulting from reduced platinum utilization (9), whereas the latter two result in a lower capacity to hold water and enhanced flooding, leading to severe condensed-phase mass transport limitations. Clearly, both consequences directly impact PEFC cost and performance, especially in the context of automotive stacks. It is of great value to automotive original equipment manufacturers (OEMs) to eliminate or mitigate the issue of support corrosion.

Any viable alternative noncarbon support must possess high surface area and electron conductivity, in addition to being highly corrosion resistant across the anticipated potential/pH window (15). Tin oxide (16) and tin-doped indium oxide (ITO) have been proposed as an alternate electrocatalyst support; however, the stability of tin oxide in acidic environments is uncertain and it has hitherto been difficult to synthesize ITO nanoparticles with adequate surface area. TiO₂ and WO₃ are particularly attractive as catalyst supports in PEFCs because they have very good chemical stability in acidic and oxidative environments (17). However, undoped titania is a semiconductor and its electron conductivity is very low, as is that of WO₃ (18). Substoichiometric titanium oxides (Ti₂O₃, Ti₄O₇, Magnéli phases) obtained by heat treatment of TiO₂ in a reducing environment (i.e., hydrogen, carbon) have electron conductivity similar to graphite (∼1,000 S ⋅ cm⁻¹) as a consequence of the presence of oxygen vacancies in the crystalline lattice (19–21). However, the heat treatment process reduces the surface area of these materials, precluding the preparation of supported electrocatalysts with good Pt dispersion (22). The literature contains several studies suggesting that Magnéli phase materials are suitable support materials for use in PEFCs; however, long-term experiments are needed to ascertain whether these materials are stable for extended periods without loss of electronic conductivity (oxidation of the substoichiometric oxide), especially when placed in an oxidizing environment in the PEFC cathode (23, 24).

Doping of titania with transition metals (e.g., Nb and Ta) has been used to enhance its electrical conductivity. Some of these doped titania supports have been catalyzed and evaluated for electrochemical stability and activity in a PEFC. Most of the doped titania materials have electronic conductivities in the interval 0.1–1 S ⋅ cm⁻¹ (substantially lower than carbon catalyst supports), but showed activities for the oxygen reduction reaction that were comparable to commercial Pt/C catalysts when evaluated in a rotating disk electrode (RDE) (25–33).

Mixed-metal oxides such as TiO₂–RuO₂ and SiO₂–RuO₂ have been shown to be an excellent alternative to carbon-based electrocatalyst supports due to their high electrochemical stability across a wide potential window (34). Ex situ stability tests (in an RDE) performed on these materials indicated no changes in the double-layer pseudocapacitance for 10,000 cycles (under an aggressive start/stop stability protocol). The initial fuel cell performance obtained with Pt-catalyzed TiO₂–RuO₂ (Pt/TRO) was close to that obtained with a commercial catalyst benchmark.

In this work we further assess the efficacy of titanium–ruthenium mixed oxides as a stable catalyst support, with an emphasis on membrane electrode assembly (MEA) testing. We evaluate support and catalyst stability using accelerated degradation tests (both ex situ and in situ). We assess the impact of stringent accelerated degradation tests on MEA performance. We also evaluate the oxygen reduction reaction (ORR) activity of the Pt/TRO catalyst. We compare both the durability and the activity of this support and catalyst against an established commercial catalyst benchmark, Pt/high surface-area carbon (HSAC; Tanaka TEC10E50E; 50% Pt on HSAC).

TRO (Ti:Ru mol ratio 1:1) support material was prepared by precipitation of ruthenium hydroxide on commercial TiO₂ nanopowder dispersed in deionized water (Aeroxide P25, BET surface area 50 m² ⋅ g⁻¹; Acros Organics) (35). The powder was calcined at 450 °C for 3 h in air to yield anhydrous, electron-conducting TRO. This material had a BET surface area of 33 ± 2 m² ⋅ g⁻¹ and an electron conductivity of 21 S ⋅ cm⁻¹; both values are lower than typically reported for Vulcan carbons (∼200 m² ⋅ g⁻¹ and 30 S ⋅ cm⁻¹) but are reasonable for a catalyst support material (2–4). Platinum nanoparticles were deposited on the catalyst support by the reduction of hexachloroplatinic acid precursor with formic acid (36). The resultant Pt nanoparticles had diameters ranging between 4 and 6 nm (Fig. S1). The relatively high particle size resulted in lower values for the electrochemically active surface area (ECA). Detailed characterization of the support and catalyst are provided in our previous work (34).

The electrochemical stability of the support (TRO) and catalyst (Pt/TRO) were evaluated using accelerated stress test protocols similar to those developed by the Fuel Cell Technical Team of the US Drive Partnership in collaboration with the US Department of Energy, with some minor differences as described below (37). The idea behind establishing such protocols was to have set of standardized testing methods to evaluate fuel cell performance and durability across laboratories, thereby allowing a proper comparison of the results obtained at different facilities and permitting proper evaluation of technologies/materials resulting from various funded projects. In this study, we used two different protocols that measure (i) the stability of the support due to start/shutdown voltage spikes, either stand-alone support or catalyzed support, the latter to investigate the impact of platinum catalyst on the support corrosion rate (start–stop protocol) (38), and (ii) Pt catalyst degradation due to dissolution/Ostwald ripening as a consequence of load cycling—excursions to near the open-circuit potential—during normal fuel cell operation (load-cycling protocol) (38, 39). These protocols effectively imitate and induce, in an accelerated fashion, the degradation mechanisms that occur during extended normal fuel cell vehicle operation.

It is generally accepted by automotive OEMs that the fuel cell stack in an automobile should operate for at least 5,000 h and 60,000 startup/shutdown cycles without any significant voltage loss (38). To evaluate the stability of the support using the start–stop protocol, the working electrode potential was cycled in a triangular waveform between 1.0 and 1.5 V at a scan rate of 500 mV/s (triangular wave form) for 5,000 cycles (Fig. 1A). Cyclic voltammograms (CV) were recorded initially (baseline) and after 100, 200, 500, 1,000, 2,000, and 5,000 cycles to characterize the support by estimating the electrode pseudocapacitance (or, in an equivalent method, the current at 0.4 V in the capacitive region of the CV).

Fig. 1.

Start–stop and load-cycling protocols used to evaluate the stability of the support and catalyst, respectively.

Changes in fuel cell load occur as a consequence of the varying power demands that are incurred during a typical drive cycle. Although somewhat buffered by hybridization strategies, some level of load cycling is inevitable. To evaluate the stability of the platinum catalyst under load cycling, the cathode potential was cycled in a rectangular waveform from 0.95 V (near the open circuit voltage; approaching no-load conditions) to 0.6 V (close to the maximum power; approaching full load conditions) for 10,000 cycles (Fig. 1A; note: the US Drive load-cycling protocol is slightly different, and involves potential cycling from 0.65 to 1 V) CVs were recorded initially, and after 100, 200, 500, 1,000, 2,000, 5,000, and 10,000 cycles. The stability of the catalyst was evaluated from the measured change in ECA and in electrode polarization.

The support and catalyst (Pt/HSAC, Pt/TRO) were examined with both protocols described above. The experiments were performed both ex situ on supports/catalysts deposited onto a glassy carbon RDE, and in situ in a fully assembled fuel cell. The experiments were always performed with the working electrode placed in a nitrogen environment to minimize side reactions.

The durability ex situ experiments were performed in an RDE setup at 60 °C using 0.1 M perchloric acid as the electrolyte, a glassy carbon rod counter electrode, and a hydrogen reference electrode. Both CV (at a scan rate of 50 mV/s) and linear polarization (scan rate of 10 mV ⋅ s⁻¹, various rotation rates) were performed at room temperature for ORR evaluation.

The in situ experiments were performed in a 25 cm² single fuel cell. MEAs were prepared using a Nafion 211 membrane, with anode catalyst loading of 0.4 mg ⋅ cm⁻² Pt/HSAC and cathode catalyst loading of 0.35 mg/cm². The experiments were performed at 80 °C, passing hydrogen (0.5 L ⋅ min⁻¹) through the anode (counter and pseudoreference electrode) side and nitrogen through the cathode/working electrode (0.5 L ⋅ min⁻¹). The gases were humidified at either 100% relative humidity (RH) or 40% RH before entry into the cell. The 100% RH operating point was chosen to maximize carbon corrosion during the accelerated test (at high voltage, carbon corrosion requires water). The 40% RH condition was chosen as a possible operating point for the fuel cell stack in an automobile. CV and V–I polarization curves were obtained at the beginning and end of the potential cycling tests for each of the MEAs tested. The V–I polarization curves were obtained at 100% and 40% relative humidities, using hydrogen as fuel and air as oxidant.

Initially, several preliminary experiments were performed to ascertain whether the in situ and ex situ approaches yielded similar results. Without a doubt, both methods yielded near identical results in terms of induced loss in ECA upon exposure to said protocols. Subsequently, both catalysts were exposed to the start–stop protocol ex situ, and the impact of this test on the catalyst was studied using CV. The results are shown in Fig. S2. Pt/TRO did not show any sign of surface modification or instability, as observed for Pt/HSAC, and the H₂ adsorption peak potential did not shift unlike in Pt/HSAC. Both observations indicated the superior stability of the TRO support upon potential cycling. The ratio of ionomer to support (well studied for Pt/C) was then optimized for the Pt/TRO catalyst via an ex situ RDE study (Fig. S3). An optimal ionomer-to-catalyst ratio (I/C) value of 0.58 g · g⁻¹ was obtained (contrast with 0.43 for Pt/HSAC). The ECA; ORR mass and specific activities; number of electrons transferred during the ORR; and the Tafel slopes for the ORR were measured for both catalysts at their optimal I/C ratios. Figs. S4–S7 and accompanying discussion in SI Text describe the results in more detail. Briefly, the Pt/TRO had lower ECA and mass activities, but a higher specific activity than Pt/HSAC due to the larger platinum particle size (4–6 nm) in Pt/TRO. The number of electrons transferred during the ORR was estimated from a Levich plot to be 3.9 for Pt/HSAC and 3.2 for Pt/TRO; the Tafel slopes estimated from Koutecky–Levich analysis were 80 and 94, respectively. The ECA, mass activity (i_m), and specific activity (i_s) of Pt/TRO and Pt/HSAC were then estimated for both catalysts upon exposure to the start–stop protocol ex situ (Fig. 2). The TRO support showed much better stability than carbon. The loss in ECA, mass activity, and specific activity after 5,000 start–stop cycles were, respectively, 16%, 18%, and 3% for Pt/TRO. In comparison, Pt/HSAC was much more severely degraded; its ECA dropped by 39%, mass activity dropped by 52%, and specific activity dropped by 22%. These ex situ studies suggested that TRO was a very stable support and that Pt/TRO was indeed a much more stable electrocatalyst than Pt/HSAC, albeit perhaps less active due to the larger platinum particle size.

Fig. 2.

Comparison of ECA and ORR specific (i_s) and mass (i_m) activity for Pt/HSAC and Pt/TRO catalysts before (BoL) and after (EoL) the start–stop protocol (ex situ experiments, performed on a RDE, in 0.1 M HClO₄ electrolyte saturated with N₂ at 60 °C).

In situ accelerated degradation tests were then performed on MEAs. Fig. 3 shows the polarization curves obtained (at 100% RH) on MEAs prepared with Pt/TRO and Pt/HSAC before and after exposure to the start–stop protocol, in situ. There are two significant observations to note. First, despite the larger Pt particle size, and concomitantly lower ECA and mass activity as ascertained by ex situ RDE tests, the Pt/TRO electrocatalyst yielded an initial MEA performance that was slightly lower (especially at lower current densities) to that obtained with an established benchmark (please note that this benchmark performance is very much in line with industry standards and is among the best available today). This result indicated that the Pt/TRO catalyst was very much viable in terms of catalytic activity and performance. Second, and even more significantly, whereas the Pt/HSAC MEA revealed a very significant (and most likely catastrophic) loss in performance, the Pt/TRO showed minimal loss in performance upon exposure to 1,000 start–stop cycles. The loss in cell voltage at 1 A · cm⁻² at 100% RH was only ∼15 mV for Pt/TRO, whereas the corresponding loss was too high to permit operation at 1 A ⋅ cm⁻² for Pt/HSAC; this MEA failed at a current density of ∼0.4 A ⋅ cm⁻². The 40% RH data, shown in Fig. 4, revealed a similar trend in terms of stability—exceptional stability for Pt/TRO as opposed to very poor stability for Pt/HSAC. These observations were attributed to the much higher stability of the TRO support compared with HSAC. MEAs prepared with each catalyst were then exposed to the load-cycling protocol. Though a significant loss in ECA was observed in each case, consistent with platinum dissolution and agglomeration, there was minimal impact on performance (see Fig. S8 and accompanying discussion in SI Text. This result suggested that the stability of the support was much more important, from the context of cell and stack failure, than the stability of the platinum particles that are loaded onto the support.

Fig. 3.

Comparison of fuel cell performance at 100% RH obtained with Pt/HSAC and Pt/TRO before and after exposure to the start–stop protocol (1,000 cycles). Experiment performed in a 25-cm² single fuel cell at 80 °C and 100% RH. Catalyst loadings: anode, 0.40 mg Pt ⋅ cm⁻²; cathode, 0.35 mg Pt ⋅ cm⁻²; I/C ratios, optimal for each electrode; polymer electrolyte membrane, Nafion 211. Closed symbols: beginning of life (BoL); open symbols: end of life (EoL).

Fig. 4.

Comparison of fuel cell performance at 40% RH obtained with Pt/HSAC and Pt/TRO before and after exposure to the start–stop protocol (1,000 cycles). Experiment performed in a 25-cm² single fuel cell at 80 °C and 40% RH. Catalyst loadings: Anode: 0.40 mg Pt ⋅ cm⁻²; cathode: 0.35 mg Pt ⋅ cm⁻²; I/C ratios: Optimal for each electrode; Polymer electrolyte membrane: Nafion 211. Closed symbols: beginning of life (BoL); Open symbols: end of life (EoL).

Finally, we measured the carbon dioxide concentration in the cathode exit stream during the accelerated degradation test (start–stop protocol) and found extremely low levels of CO₂ (between 3 and 10 ppm) in the case of Pt/TRO (Fig. 5). In contrast, the CO₂ emission levels from a conventional Pt/HSAC catalyst were ∼200 ppm. Of course, the Pt/TRO is carbon-free, and no CO₂ emission would emanate from this material. This observation was, however, a clear indicator that the main source of carbon being oxidized to carbon dioxide in an MEA was the carbon catalyst support, and not the gas diffusion layer (GDL) or the graphite flowfields. Both MEAs in this study used identical GDLs and flowfields. The small amount of CO₂ observed in the MEA prepared with Pt/TRO arose almost certainly from the corrosion of carbon in the microporous layer of GDL. We believe this to be a unique method to quantify the corrosion rate (in situ) of the carbon in the GDL microporous layer. We propose that the Pt/TRO catalyst can be used in the future in conjunction with carbon dioxide monitoring to measure the corrosion rate of candidate GDLs.

Fig. 5.

Evolution of carbon dioxide in the cathode exit stream during the in situ support durability test (start–stop protocol) for Pt/HSAC and Pt/TRO.

Both in situ and ex situ experiments confirmed that TRO (TiO₂–RuO₂) is an exceptionally stable catalyst support, and that Pt/TRO is an exceptionally stable electrocatalyst that yields initial (and final) fuel cell performance slightly lower (especially at lower current densities) to the best benchmark commercial Pt/HSAC catalyst. We report herein an alternate catalyst support to carbon that demonstrates both excellent (equivalent to benchmark) initial performance, and such exceptional stability upon exposure to an extremely stringent accelerated test that has been confirmed by automotive OEMs to accurately induce degradation and failure mechanisms seen during extended normal operation with regular start–stop cycles.

A Note on Choice of Benchmark Used

One might argue that HSAC is perhaps not the best support to use as a comparator, and that more stable carbon supports exist. We believe the latter is indeed true. However, the Pt/HSAC catalyst used here represents an internal industry benchmark in terms of performance. We used it to illustrate the exceptional initial (and as it turns out) final performance of the Pt/TRO electrocatalyst. Furthermore, we have also subjected many other, more stable, carbon supports and derivative electrocatalysts to similar test protocols—all of them show unacceptable losses in performance upon exposure to the start–stop protocol.

A Note on Costs and Viability

An argument is frequently made that the presence of ruthenium in the catalyst support is undesirable because of its high cost relative to carbon (which is very inexpensive). This project was performed with active participation from Nissan Technical Center, North America, an automotive OEM that is well aware of the cost factor. Our cost modeling has revealed that regardless of the support used, over 95% of the cost of the catalyst originated from the use of platinum as the active electrocatalyst. Moreover, when the much-enhanced stability of TRO is taken into account, the Pt/TRO electrocatalyst is more than viable from a cost perspective, compared with Pt/HSAC (or any other equivalent carbon support). We would be happy to share this analysis with any interested reader.

Conclusions

The electrochemical stability of a unique electrocatalyst—platinum supported on a highly stable TRO support—was investigated using both ex situ (RDE) and in situ (MEA) tests. The accelerated test protocols used to evaluate stability were carefully designed in conjunction with Nissan Technical Center, North America. The start–stop protocol mimicked the potential transients that are observed during fuel cell stack startup and shutdown and that contribute to severe electrocatalyst support corrosion, whereas the load-cycling protocol mimicked potential transients seen during full-load to no-load transitions that are sometimes encountered during fuel cell operation, and that contribute to platinum dissolution.

Ex situ studies were performed in a rotating ring electrode using electrocatalyst inks with individually optimized ionomer to catalyst ratios of 0.58 g ⋅ g⁻¹ for Pt/TRO and 0.43 g ⋅ g⁻¹ for Pt/HSAC. The initial ECSA and mass activities were lower for Pt/TRO than for Pt/HSAC, which was attributed to the larger platinum particle size on TRO. The lower ECSA also resulted in a higher specific activity for Pt/TRO (we note although that for practical purposes, the mass activity is more relevant). The number of electrons transferred for the ORR was estimated from Levich plots to be 3.2 for Pt/TRO and 3.9 for Pt/HSAC. The Tafel slopes, derived from a Koutecky–Levich analysis, were 94 mV per decade and 80 mV per decade respectively, for Pt/TRO and Pt/HSAC.

The Pt/TRO was much more stable that Pt/HSAC by all measures. After 5,000 start–stop cycles, the loss in ECSA was 16% for Pt/TRO compared with 39% for Pt/HSAC. The loss in ORR mass and specific activities were 18% and 3% for Pt/TRO, compared with 52% and 22% for Pt/HSAC. These ex situ screening studies indicated that the Pt/TRO was an electrocatalyst with exceptionally high stability, but lower activity as a result of larger platinum particle size.

Though ex situ tests are a useful way to screen new candidate materials, the true viability of an electrocatalyst can only be ascertained from in situ tests in a MEA; these were therefore performed. MEA tests demonstrated both the high performance and, more importantly, the exceptional stability of the Pt/TRO electrocatalyst. Polarization curves (obtained at 100% and 40% RH) on MEA prepared with Pt/HSAC revealed a very significant loss in performance upon exposure to 1,000 cycles of the start–stop protocol. The MEA with Pt/TRO showed very little change in performance upon exposure to 1,000 cycles of the start–stop protocol, both at 100% RH and at 40% RH. The voltage loss was on the order of 10–15 mV at 1 A ⋅ cm⁻², which is eminently acceptable given the harshness of this accelerated test. On the contrary, the voltage losses were too high to permit operation at anywhere close to 1A ⋅ cm⁻² (or even 0.5 A ⋅ cm⁻²) for Pt/HSAC at either 100% or 40% RH, a result that would equate to catastrophic failure if it were to occur in an automotive fuel cell stack.

Despite the larger Pt particle size in Pt/TRO, and concomitantly lower ECA and mass activity as ascertained by RDE tests, the Pt/TRO electrocatalyst yielded initial MEA performance slightly lower (especially at lower current densities) to that obtained with the benchmark Pt/HSAC electrocatalyst. Because the beginning-of-life and end-of-life performances were nearly identical, we can state that we have identified in Pt/TRO an electrocatalyst that meet durability targets for automotive fuel cell stacks. We note that by using appropriate processing methods to lower Pt particle size, we can further enhance the activity of the Pt/TRO, and we are working toward this now.

When the catalysts were subjected to the load-cycling protocol for 10,000 cycles, there was a significant loss in ECA observed in each case, consistent with platinum dissolution and agglomeration, and this was in line with expectations. Though TRO is a corrosion-resistant catalyst support, it was not designed to mitigate Pt dissolution (which would be rather difficult given the facility of this process). However, most interestingly, there was minimal detrimental impact on performance (for either catalyst) despite the large losses seen. This result suggests that the stability of the support is far more important than the stability of the platinum particles that are loaded onto the support in terms of avoiding stack failure. The fuel cell stack is likely to be much more forgiving of platinum dissolution and agglomeration during load cycling than of support corrosion and related effects arising from startup and shutdown cycles. In conjunction with the fact that hybridization methods will inevitably be used to minimize load cycling, we posit that identifying a corrosion-resistant support is a key priority, and we offer TRO as one such outstanding corrosion-resistant support that also yields highly active electrocatalysts with excellent performance.

Finally, we measured the carbon dioxide concentration in the MEA cathode exit stream during the accelerated start–stop degradation tests for both catalysts. The CO₂ emission levels from the Pt/HSAC catalyst was ∼200 ppm. In contrast, extremely low levels of CO₂ (between 3 and 10 ppm) were observed in the case of Pt/TRO. This CO₂ was clearly a result of carbon corrosion in the gas diffusion layer and/or the bipolar plates (TRO has no carbon). This finding provides a quantitative measure of the extent of carbon corrosion arising from sources other than carbon support, and confirms that the carbon in the gas diffusion layer does not corrode anywhere near the same extent as the carbon electrocatalyst support.

For further information on detailed materials and methods, see SI Text, which contains the following: RDE electrode preparation and protocols used in the experiments; transmission electron microscope micrographs of the Pt/TRO catalyst; cyclic voltammograms obtained on 40% Pt/TRO and 50% Pt/HSAC catalysts tested ex situ (RDE) using the start–stop protocol; optimization of ionomer (Nafion) to catalyst weight ratio for 40% Pt/TRO; cyclic voltammograms in nitrogen-degassed 0.1 M perchloric acid and ORR polarization curves obtained in oxygen-saturated 0.1 M perchloric acid for 40% Pt/TRO and 50% Pt/HSAC catalysts; comparison of ECA and activity toward the ORR of 40% Pt/TRO and 50% Pt/HSAC catalyst; Levich plot, Koutecky–Levich plot, and Tafel slope for 40% Pt/TRO; Levich plot, Koutecky–Levich plot, and Tafel slope for 50% Pt/HSAC; and fuel cell performance of 40% Pt/TRO and Pt/HSAC catalyst upon exposure to the load-cycling protocol (80 °C and 40% relative humidity) for 10,000 cycles.