Sputtered Platinum Thin-ﬁlms for Oxygen Reduction in Gas Di ﬀ usion Electrodes: A Model System for Studies under Realistic Reaction Conditions

: The development of catalysts for the oxygen reduction reaction in low-temperature fuel cells depends on e ﬃ cient and accurate electrochemical characterization methods. Currently, two primary techniques exist: rotating disk electrode (RDE) measurements in half-cells with liquid electrolyte and single cell tests with membrane electrode assemblies (MEAs). While the RDE technique allows for rapid catalyst benchmarking, it is limited to electrode potentials far from operating fuel cells. On the other hand, MEAs can provide direct performance data at realistic conditions but require specialized equipment and large quantities of catalyst, making them less ideal for early-stage development. Using sputtered platinum thin-ﬁlm electrodes, we show that gas di ﬀ usion electrode (GDE) half-cells can be used as an intermediate platform for rapid benchmarking at fuel-cell relevant current densities (~1 A cm − 2 ). Furthermore, we demonstrate how di ﬀ erent parameters (loading, electrolyte concentration, humidiﬁcation, and Naﬁon membrane) inﬂuence the performance of unsupported platinum catalysts. The speciﬁc activity could be measured independent of the applied loading at potentials down to 0.80 V RHE reaching a value of 0.72 mA cm − 2 at 0.9 V RHE in the GDE. By comparison with RDE measurements and Pt / C measurements, we establish the importance of catalyst characterization under realistic reaction conditions. as a model catalyst, we demonstrate the inﬂuence of di ﬀ erent testing parameters (electrolyte, scan rate, membrane and humidiﬁcation) on the performance of the unsupported catalysts. By investigating the e ﬀ ect of di ﬀ erent catalyst loadings in both GDE and RDE, we identify key design principles and establish the importance of benchmarking under realistic reaction conditions.


Introduction
Fuel cell technologies, which convert chemical energy directly into clean electricity, are expected to play a key role in environmentally friendly energy conversion schemes [1]. Polymer exchange membrane fuel cells (PEMFCs) are very promising both for transportation and stationary applications [2,3]. Platinum-based catalysts typically catalyze both the oxidation of hydrogen at the anode and the reduction of oxygen at the cathode of PEMFCs; as a result, the overall cost of the PEMFC technology is closely linked with the platinum loading [4]. The slow kinetics of the oxygen reduction reaction (ORR) causes significant losses in cell potential. Consequently, the loading at the cathode accounts for the majority of platinum usage in PEMFC. In order to reduce the Pt loading, most scientific and industrial

Preparation of Pt Thin-Film Electrodes
For the Pt thin film deposition, the magnetron electrode was equipped with a planar target of 99.95 % Pt (Junker Edelmetalle, Waldbüttelbrunn, Germany). It was located at the superior part of the recipient. The RF generator (Advanced Energy, Fort Collins, CO, USA) had a driving frequency of 13.56 MHz. The reactor chamber was evacuated to a base pressure of 5 × 10 −3 Pa. An argon plasma was ignited in the chamber at a working pressure of 5 Pa. The platinum loadings of the sputtered Pt TFs were estimated by inductively coupled plasma mass spectrometry (ICP-MS) measurements (Aurora Elite, Bruker, Billerica, MA, USA). For this purpose, the catalysts were digested in aqua regia freshly mixed with 30% HCl (Suprapur, Merck, Darmstadt, Germany) and 65% HNO 3 (Suprapur, Merck, Darmstadt, Germany) in a volumetric ratio of 3:1, respectively. A Dektak 3ST profilometer (Veeco, Plainview, NY, USA) was used to measure the film thickness. Corresponding to the Pt loadings of 11.7 µg Pt cm −2 , 16.2 µg Pt cm −2 , 29.4 µg Pt cm −2 and 66.2 µg Pt cm −2 , the films had a thickness of 15 nm, 20 nm, 37 nm and 83 nm, respectively.

Physical Characterization
The morphology of the sputtered Pt thin-film was investigated with scanning electron microscopy (JSM 7500F, JEOL, Tokyo, Japan) with a field-emission gun, a semi-in-lens conical objective lens and a secondary electron in-lens detector for high-resolution and high-quality image observation of structural features of the deposited films at a maximum specified resolution of 1.0 nm at 15 keV. The technique enables imaging the surface without any preparative coatings. The GDE was directly placed onto the SEM unit holder. The phase identity and crystallite size of the Pt catalyst was investigated by means of x-ray diffraction (XRD) using a Bruker D8 Advance Diffractometer (Bruker, Billerica, MA, USA), with measurements performed over a 2Theta range from 20 • to 80 • , step width 0.05 • and 5 s per step. Cu was used as the anode target.

Gas Diffusion Electrode (GDE)
Electrochemical measurements were carried out using a GDE setup [28]. In contrast to RDE, benchmarking is limited to low overpotentials, the GDE setup allow testing of ORR activity at high current densities under realistic mass transport conditions. The GDE half-cell consists of two cell components (see Figure 1): (i) a lower component made of body of stainless steel with a flow field and gas supply (ii) an upper cell body of polytetrafluoroethylene (PTFE) with electrolyte and counter electrode (platinum mesh) and reference electrode (reversible hydrogen electrode, RHE). For cleaning, the Teflon upper part was soaked in mixed acid (H 2 SO 4 :HNO 3 = 1:1, v:v) overnight. Subsequently, it was rinsed thoroughly by ultrapure water, and boiled three times. Surfaces 2019, 2 FOR PEER REVIEW 4

Electrochemical Measurements
All electrochemical measurements were performed using a computer-controlled potentiostat (ECi 200, NordicElectrochemistry, Copenhagen, Denmark). The measurements were performed using 4 M HClO4 and 1 M HClO4 aqueous solutions at room temperature. The high electrolyte concentrations reduce the solution resistance between working electrode, counter electrode and reference electrode. Prior to the measurements, the working electrode was purged from the backside (through the gas diffusion layer) with O2 gas and the catalyst was conditioned by potential cycles between 0.1 and 1.0 VRHE at a scan rate of 100 mV s −1 until a stable cyclic voltammogram could be observed (ca. 30 cycles). The ECSA of the catalyst was determined by conducting CO stripping voltammetry. The working electrode was held at 0.05 V during a CO purge through the GDL for 2 min followed by an Ar purge for 10 min. The ECSA was determined from the CO (QCO) oxidation charge recorded at a scan rate of 50 mV s −1 and the respective Pt loading (LPt) using a fixed conversion coefficient of 390 µC cm −2 Pt [16]; ECSA = QCO / (LPt × 390 µC cm −2 Pt). To determine the ORR activity, linear sweep voltammetry (LSV, anodic scan) was conducted by purging the electrode with O2 from below and scanning the potential at a scan rate of 50 mV s −1 or 100 mV s −1 . The polarization curves were corrected for the non-faradaic background by subtracting the cyclic voltammograms (CVs) recorded in Ar-purged electrolyte at the identical scan rate. Furthermore, the resistance between the working and reference electrode (~10 Ω) was determined using an AC signal (5 kHz, 5 mV) and thereafter compensated for using analogue positive feedback scheme of the potentiostat. The resulting effective resistance was 1 Ω or less for each experiment.

Structural Characterization of Pt Thin-Films
Representative SEM measurements of Pt thin films (TF) are displayed in Figure 2a-c. It can be seen that small Pt domains with a size of ca. 5 nm agglomerate to form cauliflower-like structures with a size between 50 nm and 150 nm. The cauliflower morphology is related to a fractal structure. The resulting surface activity and surface area is determined by this structure. The cross-section in Figure 2b shows the structure of the top-surface layer to be porous and the attaching part to the microporous layer of the gas diffusion electrode to be denser. XRD of Pt TF shows broad diffraction peaks of the indices (111), (200) and (220) crystal structure (see Figure 2d). The average crystallite size The geometric surface area of the working electrode was 3 mm. The GDE was used with and without membrane between electrolyte and working electrode. When the membrane was used, it was pressed during assembling of the cell to the catalyst layer. For measurements using humidification, the gas was bubbled through a gas humidifier.

Electrochemical Measurements
All electrochemical measurements were performed using a computer-controlled potentiostat (ECi 200, NordicElectrochemistry, Copenhagen, Denmark). The measurements were performed using 4 M HClO 4 and 1 M HClO 4 aqueous solutions at room temperature. The high electrolyte concentrations reduce the solution resistance between working electrode, counter electrode and reference electrode. Prior to the measurements, the working electrode was purged from the backside (through the gas diffusion layer) with O 2 gas and the catalyst was conditioned by potential cycles between 0.1 and 1.0 V RHE at a scan rate of 100 mV s −1 until a stable cyclic voltammogram could be observed (ca. 30 cycles). The ECSA of the catalyst was determined by conducting CO stripping voltammetry. The working electrode was held at 0.05 V during a CO purge through the GDL for 2 min followed by an Ar purge for 10 min. The ECSA was determined from the CO (Q CO ) oxidation charge recorded at a scan rate of 50 mV s −1 and the respective Pt loading (L Pt ) using a fixed conversion coefficient of 390 µC cm −2 Pt [16]; ECSA = Q CO /(L Pt × 390 µC cm −2 Pt ). To determine the ORR activity, linear sweep voltammetry (LSV, anodic scan) was conducted by purging the electrode with O 2 from below and scanning the potential at a scan rate of 50 mV s −1 or 100 mV s −1 . The polarization curves were corrected for the non-faradaic background by subtracting the cyclic voltammograms (CVs) recorded in Ar-purged electrolyte at the identical scan rate. Furthermore, the resistance between the working and reference electrode (~10 Ω) was determined using an AC signal (5 kHz, 5 mV) and thereafter compensated for using analogue positive feedback scheme of the potentiostat. The resulting effective resistance was 1 Ω or less for each experiment.

Structural Characterization of Pt Thin-Films
Representative SEM measurements of Pt thin films (TF) are displayed in Figure 2a-c. It can be seen that small Pt domains with a size of ca. 5 nm agglomerate to form cauliflower-like structures with a size between 50 nm and 150 nm. The cauliflower morphology is related to a fractal structure. The resulting surface activity and surface area is determined by this structure. The cross-section in Figure 2b shows the structure of the top-surface layer to be porous and the attaching part to the microporous layer of the gas diffusion electrode to be denser. XRD of Pt TF shows broad diffraction peaks of the indices (111), (200) and (220) crystal structure (see Figure 2d). The average crystallite size estimated by the Scherrer equation is 7.2 nm, which fits to the Pt domain size identified with the electron microscope.   Figure 3 shows the base cyclic voltammograms in argon-saturated electrolyte as well as COstripping measurements in order to determine the electrochemically active surface area. Figure 3a displays in detail the cyclic voltammograms of Pt TF with increasing Pt characteristic features for higher loadings, as observed in the hydrogen under potential deposition (0.1-0.4 VRHE) region as well as the region where O-containing species are adsorbed (0.7-1.0VRHE). The linear sweep voltammogram of the CO oxidation charge transfer is shown in Figure 3b for the different loadings. The sweep shows two CO oxidation peaks, one peak between 0.75 and 0.77 VRHE and a second around 0.82-0.85 VRHE, in contrast to the CO oxidation sweep measured in the RDE with one peak at 0.73 VRHE. The integrated CO oxidation charge is plotted against the Pt loading with the ideal linear slope crossing 0 and the second lowest loading; see Figure 3c. The oxidation charge increases with loading. However, it can be seen that the CO oxidation charge as function of the loading deviates from an ideal straight line for the highest loading. This could indicate that there is a mass transport problem due to blocking of CO as a reactant or trapped product; however, no CO oxidation peaks were visible in sub sequential sweeps. Thus it can be assumed that the ECSA measurement with CO adsorption is valid.

Electrochemical Characterization
Our results from CO-stripping experiments therefore, show that the decrease in ECSA for higher Pt loadings is a result of decreasing Pt utilization. This is supported by the cross-section SEM micrograph displayed in Figure 2b with a higher loading. The graph shows a higher porosity at the near surface region, compared to the region near the Microporous Layer (MPL). Increasing the loading/film thickness leads to lower Pt utilization due to lack of long-range porosity for higher  Figure 3 shows the base cyclic voltammograms in argon-saturated electrolyte as well as CO-stripping measurements in order to determine the electrochemically active surface area. Figure 3a displays in detail the cyclic voltammograms of Pt TF with increasing Pt characteristic features for higher loadings, as observed in the hydrogen under potential deposition (0.1-0.4 V RHE ) region as well as the region where O-containing species are adsorbed (0.7-1.0V RHE ). The linear sweep voltammogram of the CO oxidation charge transfer is shown in Figure 3b for the different loadings. The sweep shows two CO oxidation peaks, one peak between 0.75 and 0.77 V RHE and a second around 0.82-0.85 V RHE , in contrast to the CO oxidation sweep measured in the RDE with one peak at 0.73 V RHE . The integrated CO oxidation charge is plotted against the Pt loading with the ideal linear slope crossing 0 and the second lowest loading; see Figure 3c. The oxidation charge increases with loading. However, it can be seen that the CO oxidation charge as function of the loading deviates from an ideal straight line for the highest loading. This could indicate that there is a mass transport problem due to blocking of CO as a reactant or trapped product; however, no CO oxidation peaks were visible in sub sequential sweeps. Thus it can be assumed that the ECSA measurement with CO adsorption is valid.  Figure 4 shows a comparison between the linear sweep voltammograms during the oxygen reduction reaction for both RDE and GDE. As stated in our previous work [28], the GDE enables the measurement of realistic mass transport conditions and high current densities. In contrast, the RDE is mass transport limited due to low solubility of oxygen in the condense phase/liquid electrolyte, but it's well defined hydrodynamics allow extraction of intrinsic kinetic currents [16,17]. Therefore, it is only possible to measure the kinetic current densities at very low overpotentials.

Performance of Pt Thin-Films in the GDE Setup
The ORR activity from RDE measurements in acidic electrolyte is typically reported at 0.9 VRHE for less active platinum catalysts and for highly active catalysts it is often only possible to measure at even higher potentials up to 0.95 VRHE [33]. However, fuel cells typically operate at a working potential of in the range of 0.6-0.8 VRHE, as indicated in Figure 4. At these potentials, the current in the RDE is under full mass transport limitation. While in the GDE the current density is not free of mass transport limitations, GDE measurements reveal more realistic current densities [28].
Our GDE setup allows measuring a maximum oxygen reduction current density of ~1 A cm −2 using only 16.2 µgPt cm −2 ; this is over two orders of magnitude higher than what is observed in the RDE measurements. The maximum current density is reached at ~0.3 VRHE. At potentials below 0.3 VRHE in the GDE it can be seen that the current decreases. This correlates to the region where hydrogen under potential adsorption occurs. Therefore, it is presumed that the decrease in the current density observed between 0.1 VRHE and 0.3 VRHE is due to blocking by protons on the Pt surface which would Our results from CO-stripping experiments therefore, show that the decrease in ECSA for higher Pt loadings is a result of decreasing Pt utilization. This is supported by the cross-section SEM micrograph displayed in Figure 2b with a higher loading. The graph shows a higher porosity at the near surface region, compared to the region near the Microporous Layer (MPL). Increasing the loading/film thickness leads to lower Pt utilization due to lack of long-range porosity for higher loadings of sputtered Pt thin-films. Consequently, the ECSA measurements have to be carefully done for each loading, especially with loadings above 30 µg Pt cm −2 . The ECSA of the Pt TF GDE is determined to be 20 m 2 g −1 Pt for the low-loaded 16.2 µg Pt cm −2 gas diffusion electrodes. Figure 4 shows a comparison between the linear sweep voltammograms during the oxygen reduction reaction for both RDE and GDE. As stated in our previous work [28], the GDE enables the measurement of realistic mass transport conditions and high current densities. In contrast, the RDE is mass transport limited due to low solubility of oxygen in the condense phase/liquid electrolyte, but it's well defined hydrodynamics allow extraction of intrinsic kinetic currents [16,17]. Therefore, it is only possible to measure the kinetic current densities at very low overpotentials.

Performance of Pt Thin-Films in the GDE Setup
The ORR activity from RDE measurements in acidic electrolyte is typically reported at 0.9 V RHE for less active platinum catalysts and for highly active catalysts it is often only possible to measure at even higher potentials up to 0.95 V RHE [33]. However, fuel cells typically operate at a working potential of in the range of 0.6-0.8 V RHE , as indicated in Figure 4. At these potentials, the current in the RDE is under full mass transport limitation. While in the GDE the current density is not free of mass transport limitations, GDE measurements reveal more realistic current densities [28].
Our GDE setup allows measuring a maximum oxygen reduction current density of~1 A cm −2 using only 16.2 µg Pt cm −2 ; this is over two orders of magnitude higher than what is observed in the RDE measurements. The maximum current density is reached at~0.3 V RHE . At potentials below 0.3 V RHE in the GDE it can be seen that the current decreases. This correlates to the region where hydrogen under potential adsorption occurs. Therefore, it is presumed that the decrease in the current density observed between 0.1 V RHE and 0.3 V RHE is due to blocking by protons on the Pt surface which would limit the full oxygen reduction to water. In this region, the catalyst reduces oxygen to hydrogen peroxide, which corresponds to a two-electron transfer compared to a four-electron transfer for the reduction of oxygen to water [34]. It is notable that the maximum current density is reached in the same potential region as the maximum in the extended Koutecky-Levich analysis [23] as well as the potential of zero total charge (PZTC) [35]. The maximum oxygen reduction rate, therefore, correlates with the minimum ion coverage; most likely a competition between protons and oxygen species at the surface and, to a lesser extent, perchlorate anions.
Surfaces 2019, 2 FOR PEER REVIEW 7 limit the full oxygen reduction to water. In this region, the catalyst reduces oxygen to hydrogen peroxide, which corresponds to a two-electron transfer compared to a four-electron transfer for the reduction of oxygen to water [34]. It is notable that the maximum current density is reached in the same potential region as the maximum in the extended Koutecky-Levich analysis [23] as well as the potential of zero total charge (PZTC) [35]. The maximum oxygen reduction rate, therefore, correlates with the minimum ion coverage; most likely a competition between protons and oxygen species at the surface and, to a lesser extent, perchlorate anions.

Benchmarking Different Pt Loadings in the GDE Setup
The Pt TF catalyst was tested with different Pt loadings on the GDE, as shown in Figure 5. The obtained linear sweep voltammograms are plotted as kinetic current densities normalized by the catalyst geometric surface area, the specific activities (normalized by the roughness factor) and mass activities. The activities normalized by geometric surface area show the expected increase of current density over the full potential range. At 0.3 VRHE, the maximum activity is achieved and reaches 0.75 A cm −2 for 11 µgPt cm −2 and nearly 1 A cm −2 for the highest loading 66 µgPt cm −2 . The onset potential for the ORR is shifting to lower overpotentials. The specific activities were calculated using the roughness factor determined by CO oxidation; it can be seen that the onset potential stays constant with a value of 0.71 mA cm −2 Pt at 0.90 VRHE. The specific activity starts to differ in the region from 0.75 VRHE-0.8 VRHE for the different loadings. Furthermore, the lowest Pt loading leads to the highest specific activity measured. The Pt TF catalyst with 11 µgPt cm −2 achieves a specific activity of 94.0 mA cm −2 Pt at 0.65 VRHE. While the GDE with the highest loading of 66 µgPt cm −2 reaches only a value of Figure 4. Comparison of the current density-potential curves as measured in a GDE setup and a conventional RDE setup. The activity measurement in the GDE setup with oxygen saturated flow field was conducted at 298 K, 4 M HClO 4 , Nafion 117 membrane, 0% RH and 50 mV s −1 . The measurement of the activity in the RDE setup was conducted at 1600 RPM, 298 K, 0.1 M HClO 4 and 50 mV s −1 . The catalyst was a Pt thin film prepared by magnetron sputtering with a Pt loading of 16.2 µg Pt cm −2 for both RDE and GDE. The right axis shows the current density enhancement as compared to the diffusion limited current observed in RDE.

Benchmarking Different Pt Loadings in the GDE Setup
The Pt TF catalyst was tested with different Pt loadings on the GDE, as shown in Figure 5. The obtained linear sweep voltammograms are plotted as kinetic current densities normalized by the catalyst geometric surface area, the specific activities (normalized by the roughness factor) and mass activities. The activities normalized by geometric surface area show the expected increase of current density over the full potential range. At 0.3 V RHE , the maximum activity is achieved and reaches 0.75 A cm −2 for 11 µg Pt cm −2 and nearly 1 A cm −2 for the highest loading 66 µg Pt cm −2 . The onset potential for the ORR is shifting to lower overpotentials. The specific activities were calculated using the roughness factor determined by CO oxidation; it can be seen that the onset potential stays constant with a value of 0.71 mA cm −2 Pt at 0.90 V RHE . The specific activity starts to differ in the region from 0.75 V RHE -0.8 V RHE for the different loadings. Furthermore, the lowest Pt loading leads to the highest specific activity measured. The Pt TF catalyst with 11 µg Pt cm −2 achieves a specific activity of 94.0 mA cm −2 Pt at 0.65 V RHE . While the GDE with the highest loading of 66 µg Pt cm −2 reaches only a value of 38.6 mA cm Pt −2 at 0.65 V RHE , corresponding to a "loss" of 59% at 0.65 V RHE . The reason for the potential-dependent specific activity is most likely related to mass transport, which can be distinguished due to oxygen transport, proton transport, H 2 O transport or electronic conductivity. In order to distinguish between the different influences, several operation parameters were changed and discussed later in this paper.
Surfaces 2019, 2 FOR PEER REVIEW 8 38.6 mA cmPt −2 at 0.65 VRHE, corresponding to a "loss" of 59% at 0.65 VRHE. The reason for the potentialdependent specific activity is most likely related to mass transport, which can be distinguished due to oxygen transport, proton transport, H2O transport or electronic conductivity. In order to distinguish between the different influences, several operation parameters were changed and discussed later in this paper. Comparing the mass activity for the different loadings, the highest activity can again be achieved with the GDE with the lowest loading of 11 µgPt cm −2 . The activity of the GDE with 11 µgPt cm −2 reaches 17 A mg −1 Pt at 0.65 VRHE and 55 A mg −1 Pt at 0.3 VRHE. The highest loaded GDE achieves only a mass activity of 5 A mg −1 Pt at 0.65 VRHE and 19 A mg −1 Pt at 0.3 VRHE. At 0.65 VRHE this is a loss of 75% and at 0.3 VRHE the "loss" is 70% of the mass activity compared to the lowest loading.

Systematic Change of Operation Parameters
The Pt TF GDE was evaluated with different operational parameters to understand the activity determining factors of the unsupported catalyst layers (see Figure 6). First, Figure 6a shows the ORR polarization curves of the GDE with a separating membrane, as compared with a non-separated GDE setup with 4 M perchloric acid. The separating membrane increases the geometric activity over the full potential window and at 0.65 VRHE by 104%. To clarify the cause of this drastic increase in activity, we performed a set of measurements with 1 M HClO4 instead of 4 M HClO4 (see Figure 6a). Reducing the molarity of the electrolyte greatly enhanced the activity. Interestingly, the polarization curve for Pt in 1 M HClO4 resembles the one obtained in 4 M HClO4 using a Nafion membrane. This suggests that specific anion adsorption is inhibiting the rate of ORR at high electrolyte concentrations. While perchloric acid is generally considered a non-adsorbing electrolyte, it is known to inhibited ORR on Pt at high concentrations [36]. Thus, in the absence of a separating membrane, an electrolyte concentration of 1 M HClO4 would be preferable. While no differences were observed in the pure kinetic region (until ~0.8 VRHE), the membrane slightly enhanced the activity as compared to 1 M Comparing the mass activity for the different loadings, the highest activity can again be achieved with the GDE with the lowest loading of 11 µg Pt cm −2 . The activity of the GDE with 11 µg Pt cm −2 reaches 17 A mg −1 Pt at 0.65 V RHE and 55 A mg −1 Pt at 0.3 V RHE . The highest loaded GDE achieves only a mass activity of 5 A mg −1 Pt at 0.65 V RHE and 19 A mg −1 Pt at 0.3 V RHE . At 0.65 V RHE this is a loss of 75% and at 0.3 V RHE the "loss" is 70% of the mass activity compared to the lowest loading.

Systematic Change of Operation Parameters
The Pt TF GDE was evaluated with different operational parameters to understand the activity determining factors of the unsupported catalyst layers (see Figure 6). First, Figure 6a shows the ORR polarization curves of the GDE with a separating membrane, as compared with a non-separated GDE setup with 4 M perchloric acid. The separating membrane increases the geometric activity over the full potential window and at 0.65 V RHE by 104%. To clarify the cause of this drastic increase in activity, we performed a set of measurements with 1 M HClO 4 instead of 4 M HClO 4 (see Figure 6a). Reducing the molarity of the electrolyte greatly enhanced the activity. Interestingly, the polarization curve for Pt in 1 M HClO 4 resembles the one obtained in 4 M HClO 4 using a Nafion membrane. This suggests that specific anion adsorption is inhibiting the rate of ORR at high electrolyte concentrations. While perchloric acid is generally considered a non-adsorbing electrolyte, it is known to inhibited ORR on Pt at high concentrations [36]. Thus, in the absence of a separating membrane, an electrolyte concentration of 1 M HClO 4 would be preferable. While no differences were observed in the pure kinetic region (until~0.8 V RHE ), the membrane slightly enhanced the activity as compared to 1 M HClO 4 in the mass-transport-dominated region (0.3-0.8 V RHE ), possibly due to enhanced proton transport or reduced flooding of the hydrophilic catalyst layer. At high current densities, oxygen transport is critical. Recently, Kongkanand el. al showed that local O 2 resistance in the ionomer film is essential for the performance supported catalysts [4]. While unsupported catalysts are generally free of ionomer, the degree of wetting would influence the oxygen transport in the catalyst layer. It is, therefore, possible that the membrane reduces flooding as a result of less penetrating electrolyte, as compared to measurements without the Nafion membrane. However, from the presented data, it cannot be concluded if adsorbed water on the surface hinders the reaction. The specific activities and ECSAs for the RDE and GDE benchmarking measurements on Pt TFs are summarised in Table 1, as well as compared to commercial Pt/C in the RDE, GDE and MEA. The specific activity of Pt/C in the GDE cell at 60° C and 0.9 VRHE was found to be comparable to the specific activity of the MEA with commercial Pt/C and decreased at room temperature. In general, the specific activity at 0.9 VRHE of GDE and MEA of Pt/C was found to be smaller compared to the RDE measurements. This is also the case for Pt TF where the specific activity in the RDE was benchmarked with 1.82 mA cm −2 Pt at 0.9 VRHE and in the GDE with 0.72 mA cm −2 Pt with membrane or 0.71 mA cm −2 Pt without membrane. The decrease in specific activity for commercial Pt/C is nearly 50% compared to a loss approximately 60% for Pt TF. In the higher current density region, at 0.65 VRHE, the specific activity with Pt TF is significantly higher than with Pt/C. In contrast, the current density at 0.65 VRHE by mass is only slightly increased for Pt TF -with 24.1 A/mgPt for Pt/C and 28.6 A/mgPt for Pt TF. This indicates that unsupported Pt catalysts have an advantage in activity compared to commercial Pt/C catalysts, especially without humidification at low temperatures. In contrast, at higher loadings, where most devices operate, the low Pt utilization inhibits the performance of the unsupported Pt catalysts. The investigations show that with the help of simple benchmarking of different catalyst systems in the GDE, new ways of developing possible applications can be explored. It turns out that different catalyst systems also require different environmental conditions and can play to their strengths under different operating conditions. Secondly, the scan rate was varied to 50 mV s −1 compared to the 100 mV s −1 , the results showed negligible decrease in activity (see Figure 6b). This indicates that the current densities are defined by faradaic currents instead of capacitive currents. Capacitive currents increase as the current is proportional to the applied scan rate. Furthermore, it is a sign for a reaction in equilibrium, where mass transport should not be the significant driver limiting the reaction. If mass transport would limit the reaction rate, as the reactant supply drops, the geometric current density should decrease with lower scan rate.
The third operation parameter studied was the humidification of oxygen. It was previously shown that a change of oxygen humidification can have a significant impact on the gained activity of platinum nanoparticles supported on carbon (Pt/C) [28]. However, there is nearly no change in the activity of Pt TF after humidifying the oxygen with 100% relative humidity (RH); see Figure 6c. This indicates that proton transport is not a limiting step with unsupported Pt catalysts. This is in contrast to Pt/C in GDE, where we observed a doubling in activity with humidification at room temperature, indicating a limiting behaviour; see Figure 6d. In our previous study, Pt/C catalyst was mixed with Nafion [28]. The Nafion ionomer has to be humidified to work as a sufficient proton conductor. In the unsupported Pt TF catalyst, protons seem to be transported on the metallic surface by either Pt-OH or Pt-H. The transport of protons on platinum was first investigated by McBreen et al. [37] and recently, Zenyuk et al. suggested that the transport mechanism might be potential dependent above 0.7 V, where Pt-O and Pt-OH are present on the surface. At potentials below the point of zero charge, the proton is believed to be transported by surface migration [38]. This speculation was already introduced earlier for oxygen reduction catalysts by Debe et al. for Pt on non-conductive organic whiskers [39], the so-called nano-structured thin films (NSTF). This is further supported by the fact that the catalyst layer is also more hydrophobic compared to Pt/C. Thus, at room temperature, sufficient wetting of catalyst layer can be obtained only from water generated by the ORR reaction. Thus, our results support the claim that proton transport on bare Pt is not a limiting step.
The specific activities and ECSAs for the RDE and GDE benchmarking measurements on Pt TFs are summarised in Table 1, as well as compared to commercial Pt/C in the RDE, GDE and MEA. The specific activity of Pt/C in the GDE cell at 60 • C and 0.9 V RHE was found to be comparable to the specific activity of the MEA with commercial Pt/C and decreased at room temperature. In general, the specific activity at 0.9 V RHE of GDE and MEA of Pt/C was found to be smaller compared to the RDE measurements. This is also the case for Pt TF where the specific activity in the RDE was benchmarked with 1.82 mA cm −2 Pt at 0.9 V RHE and in the GDE with 0.72 mA cm −2 Pt with membrane or 0.71 mA cm −2 Pt without membrane. The decrease in specific activity for commercial Pt/C is nearly 50% compared to a loss approximately 60% for Pt TF. In the higher current density region, at 0.65 V RHE , the specific activity with Pt TF is significantly higher than with Pt/C. In contrast, the current density at 0.65 V RHE by mass is only slightly increased for Pt TF -with 24.1 A/mg Pt for Pt/C and 28.6 A/mg Pt for Pt TF. This indicates that unsupported Pt catalysts have an advantage in activity compared to commercial Pt/C catalysts, especially without humidification at low temperatures. In contrast, at higher loadings, where most devices operate, the low Pt utilization inhibits the performance of the unsupported Pt catalysts. The investigations show that with the help of simple benchmarking of different catalyst systems in the GDE, new ways of developing possible applications can be explored. It turns out that different catalyst systems also require different environmental conditions and can play to their strengths under different operating conditions.

Conclusions
In this paper, we present the electrochemical characterization of platinum thin-films in a GDE setup. Unlike the commonly used RDE method, this half-cell method facilitates investigations under triple phase boundary conditions. Moreover, the GDE setup allows achieving current density values comparable to operating current densities in actual PEMFCs.
Herein, we investigate sputtered Pt with very low loadings, achieving high current densities (close to 1 A cm −2 ) in GDE measurements. Furthermore, we show that different parameters such as loading, electrolyte concentration, humidification and Nafion membrane influence the performance of the Pt catalyst layer. By carefully measuring the activity with different loadings in the GDE, we observe that specific activities at potentials down to 0.75-0.80 V RHE can be determined independently of the applied loading. The values obtained in the GDE were two-and-half times lower than those measured with the RDE technique using the same catalyst. Our results confirm that activity values obtained in liquid half-cells are not directly transferable to real devices. This is mainly due to transport resistance at the triple phase interface and not a result of higher loadings or different testing parameters.
Interestingly, our results show that at fuel-cell relevant overpotentials, the activity is highly dependent on the loading and use of the membrane. Furthermore, we observed that humidification had no influence on the activity of catalyst layer for the unsupported catalyst. This is in contrast to Pt/C, where the use of humidification nearly doubled the catalytic activity. These findings highlight the potential impact of GDE half-cell benchmarking. While RDE remains a valuable tool for observing trends in intrinsic catalytic performance, the properties of the catalyst layer have to be investigated under more realistic conditions. Using sputtered platinum as a model catalyst, we show that the catalyst layer can be efficiently evaluated in the fuel cell operation region using a GDE half-cell setup. It is our hope that this enables accelerated development of fuel cell catalysts by shortening expensive and time-consuming MEA optimization.

Conflicts of Interest:
The authors declare no conflict of interest.