Loading Impact of a PGM-Free Catalyst on the Mass Activity in Proton Exchange Membrane Fuel Cells

All fuel cell tests were conducted using 5 cm² active area MEAs, prepared by decal transfer. For cathode catalyst layers, a Pajarito Powder catalyst, PMF011904 (≈0.5 wt% of Fe) was used, 1 g portion of which was ball-milled prior to ink preparation in a 20 mL ZrO₂ milling-vessel with 10 ZrO₂ beads of 10 mm diameter (corresponding to 30:1 beads-to-catalyst mass ratio) for 30 min at 200 rpm, using a planetary ball-mill (Pulverisette 7 premium line Fritsch, Germany). PGM-free catalyst inks were prepared by mixing the ball-milled catalyst with a low-equivalent-weight ionomer (700 EW, Asahi Kasei, Japan) dispersed in water and 1-propanol, using a Thinky planetary mixer (500 rpm for 10 min). The ionomer-to-carbon weight ratio (I/C-ratio) was 0.67/1. The inks were coated onto virgin PTFE foil (50 μm thick, APSOparts, Germany) using a Mayer-rod and an automated coater (RK Print Coat Instruments Ltd., UK). To achieve the targeted cathode catalyst loadings of 0.4, 1.0, 2.0, and 4.0 mg_cat cm⁻² _MEA, gap bar coaters (RK PrintCoat Instruments, UK) were used with nominal wet-film thicknesses of ≈70, ≈175, ≈350, and ≈700 μm, respectively. The cathode catalyst coatings were dried at room temperature.

Anode catalyst layers based on a platinum catalyst supported on Vulcan XC72 carbon (20 wt% Pt/C, Tanaka, Japan) were prepared to obtain a loading of 0.1 mg_Pt cm⁻² _MEA with an I/C-ratio of 0.65/1 using the same ionomer; ink preparation and coating were done according to our previous study. ²²

MEAs with an active area of 5 cm² were assembled by hot-pressing a 50 μm thick membrane (Nafion 212^®, Fuel Cell Store, USA), sandwiched between anode and cathode decals, at 135 °C and 4 kN force for 10 min. While the targeted cathode catalyst loadings were 0.4, 1.0, 2.0, and 4.0 mg_cat cm⁻² _MEA, the precise catalyst loadings were determined by weighing the decals before and after hot-pressing. All tested MEAs had a maximum cathode loading deviation of ±10% from the above given target loadings. This is consistent with the maximum ±10% variation in the catalyst mass normalized cathode capacitances calculated from cyclic voltammograms (see Table I). Two nominally identical MEAs were prepared for each of the cathode catalyst loadings, and the error bars shown in the later figures represent the minimum and maximum values of independent measurements with two MEAs. The thickness of the cathode electrodes was measured at four different points for each MEA using a Mitutoyo dial gauge (accuracy of ±3 μm, series 543); the resulting average thickness and standard deviation for the different MEAs are reported in Table II.

Table II. Thicknesses and electrical resistances of the MEAs with the four different catalyst loadings: cathode catalyst layer thickness (t_cathode, using a Mitutoyo dial gauge series 543), electrical in-plane resistivity of the cathode catalyst layer (${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}},$ determined from in-plane four-point-probe acc. to Eq. 1), and estimated electrical through-plane resistance (${R}_{{{\rm{e}}}^{-},{\rm{cath}}},$ acc. to Eq. 2).

nominal catalyst loading [mgcat cm−2 MEA]	actual catalyst loading for two MEAs [mgcat cm−2 MEA]	tcathode [μm]	${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}$ [Ω cm]	${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ [mΩ cm2]
0.4	0.40 ∣ 0.33	n.a.	n.a.	n.a.
1.0	0.90 ∣ 0.84	27 ∣ 24	2.2 ∣ 2.4	5.9 ∣ 5.7
2.0	2.02 ∣ 2.03	59 ∣ 63	2.7 ∣ 2.9	15.9 ∣ 18.3
4.0	3.60 ∣ 3.70	121 ∣ 141	2.8 ∣ 4.5	33.4 ∣ 63.4

Ex situ analysis of the electrical resistance of the cathode catalyst layers was conducted using a four-point probe that was placed onto the cathode catalyst layer of the MEA at room temperature and at ambient relative humidity (≈45% RH). These measurements were conducted by applying the current through the two outer pins and measuring the voltage between the two inner pins of the probe head. The in-plane resistivity of the cathode catalyst layer (${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}$) was calculated for each MEA according to Eq. 1: ²³

$$\begin{eqnarray}&&{\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}\,=\,t\,\cdot \,\displaystyle \frac{U}{I}\,\cdot \,\displaystyle \frac{W}{L}\end{eqnarray} \tag{ 1 }$$

Here, t is the measured cathode electrode thickness, U is the measured voltage, I is the applied current, and W/L is a geometrical factor derived from the ratio between the width of the tested sample (≈2.24 cm) and the spacing between the two inner pins (1 mm), yielding W/L = 4.53. Under the reasonable assumption that the electrical resistance of the cathode catalyst layer is isotropic, i.e., that in-plane and through-plane resistivity are identical, the through-plane areal resistance (${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$) for each cathode catalyst loading was calculated according to the following equation:

$$\begin{eqnarray}&&{R}_{{{\rm{e}}}^{-},{\rm{cath}}}\,=\,t\,\cdot \,{\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}\,=\,{t}^{2}\,\cdot \,\,\displaystyle \frac{U}{I}\,\cdot \,\displaystyle \frac{W}{L}\end{eqnarray} \tag{ 2 }$$

Here, the calculated through-plane electrical resistance ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ is expressed in units of mΩ cm², so that it can later be used for the correction of the voltage loss originating from the electrical resistivity of the cathode catalyst layer.

Electrochemical measurements in 5 cm² single-cell PEMFCs were performed using an in-house manufactured cell hardware and commercial graphite flow fields with 7 channels and a single serpentine (0.5/0.5 mm land/channel widths; manufactured by Poco Graphite, USA). ²⁴ Gas diffusion layers (GDLs) were supplied by Freudenberg (H14C10, Germany), and the GDL compression was adjusted by setting the GDL strain to 20.0 ± 1.0% by adjusting the thickness of the nearly incompressible PTFE-coated fiberglass gaskets (compressing by 7% at the applied force), assembled at a torque of 12 Nm. Fuel cell tests were performed on automated test stations (G60, Greenlight Innovation, Canada) equipped with a load-bank and an additional potentiostat (Reference3000, Gamry, USA) to conduct electrochemical impedance spectroscopy (EIS) as well as cyclic voltammetry (CV).

MEAs with platinum-based cathode catalysts are generally subjected to a several hours long conditioning procedure under H₂/air. ²⁵ However, since PGM-free catalysts are reported to exhibit significant degradation within the first hours of operation, ^8,17 no MEA conditioning was performed in the present study. The mass activity of the PGM-free catalysts was determined from H₂/O₂ polarization curves at an absolute pressure of 170 kPa_abs (set at the cell inlet), a relative humidity 90%, and at T_cell = 80 °C, using differential flow conditions (2000 nccm and 5000 nccm on anode and cathode, respectively; nccm is defined at standard conditions of 273.15 K and 101.3 kPa). Polarization curves were recorded in galvanostatic mode, holding each current point for 5 min and recording the averaged cell potential over the last 30 s; the polarization curves were recorded starting from open circuit voltage (OCV) and progressing to higher current densities. After each current measurement point, the high frequency resistance (HFR) was determined by electrochemical impedance spectroscopy (EIS) with 10% current perturbation (in the frequency range of 1–100 kHz with 10 points per decade).

After recording a first H₂/O₂ polarization curve, cyclic voltammograms (CVs) were recorded at T_cell = 40 °C and at ambient pressure, with fully-humidified 5% H₂ in N₂ supplied to the anode (at 200 nccm) and dry N₂ supplied to the cathode (at 50 nccm). Three different scan rates were used (50, 100 and 150 mV s⁻¹), recording three cycles for each scan rate.

After recording the CVs, a second H₂/O₂ polarization curve was recorded to investigate if the preceding tests may have already degraded the catalyst. This is followed by an impedance analysis under blocking conditions to determine the cathode electrode proton conduction resistance: following the procedure introduced by Liu et al ., ²⁶ EIS was conducted at 0.2 V (acquired between 500 kHz and 0.2 Hz, with a voltage amplitude of 3.5 mV) with H₂ on the anode and N₂ on the cathode side (1000 nccm each) at T_cell = 80 °C, at 90% relative humidity, and at 170 kPa_abs. After these measurements, three more consecutive H₂/O₂ polarization curves were recorded in the same way as described above.

RDE measurements were performed in 0.1 M HClO₄ (Kanto Chemical Co.,Inc., Japan) at 25 °C, using a glassy-carbon disk working electrode (Pine, USA, A = 0.196 cm²) coated with the catalyst. A rotating ring disk electrode shaft (Pine, USA) with a Pt-ring was employed, whereby the ring electrode was used to calibrate the static reversible hydrogen electrode (RHE) that was used as reference electrode prior and after the actual RDE experiments: by conducting hydrogen oxidation/evolution scans on the Pt ring electrode (at 400 rpm) while bubbling pure H₂ through the electrolyte, the precise value for 0 V vs RHE was determined from the transition potential between oxidation and reduction currents; this was used to determine the offset of the static RHE reference electrode from 0 V vs RHE (typically less than 2 mV). The static RHE reference electrode consisted of a Pt wire immersed in H₂-saturated 0.1 M HClO₄ (Kanto Chemical Co., Inc., Japan); ²⁷ a gold wire was used as counter electrode.

The catalyst inks for RDE experiments were prepared by dispersing a commercial PGM-free ORR catalyst from Pajarito Powder (PMF11904, USA) in dimethylformamide (anhydrous, 99.8%, Sigma-Aldrich, Germany) and bath sonicating the dispersion for 30 min, followed by the addition of a defined amount of a 5 wt% Nafion® ionomer solution (Sigma-Aldrich, Germany) and another bath sonication for 5 min. The ionomer/catalyst ratio was chosen to be 0.44 g g⁻¹ and the concentration of the catalyst ink was chosen such that upon deposition of 10 μL of ink onto the glassy carbon disk, the desired catalyst loadings were obtained (ranging from 40–600 μg_cat cm⁻² _geo).

Rotating disk measurements were carried out at 25 °C in O₂-saturated HClO₄, using an Autolab potentiostat (Metrohm, Switzerland) in a potential window of 0.05–1.1 V_RHE, a scan rate of 5 mV s⁻¹, and at 1600 rpm. The ORR data were recorded during the 3^rd positive-going scan. Impedance spectroscopy was measured to determine the non-compensated high frequency resistance (HFR) between the working and the reference electrode. All RDE data were i) iR-corrected using the determined HFR and ii) corrected by the capacitive currents determined in the absence of O₂ (i.e., under Ar); as the commonly used oxygen mass transport correction is not valid for these thick catalyst films, ²⁸ the RDE data are shown only for current densities of ≤0.6 mA cm⁻², which is ≤20% of the diffusion-limited current density for a 2-electron reduction process, so that oxygen concentration gradients within the electrolyte boundary layer and within the electrode can be considered negligible.

Prior to the measurements, the cell was not subjected to any H₂/air conditioning that is conventionally used for MEAs with Pt-based ORR catalysts, in order to avoid any possible catalyst degradation. In the case of PGM-free ORR catalysts, conditioning with air on the cathode might likely result in hydrogen peroxide production in the catalyst layer; this in turn is hypothesized to constitute a major catalyst degradation pathway via hydroxyl radicals that are produced in a Fenton reaction. ⁸ Since this study focuses on the loading impact of a PGM-free catalyst on the ORR mass activity and the H₂/O₂ performance, the confirmation of the catalyst loading is essential and was achieved via an analysis of the mass normalized currents obtained by cyclic voltammetry, which are not compromised by any possible mass transport resistances in the thick catalyst layers.

Hence, CVs were recorded and normalized by the mass of the catalyst in the cathode electrode, whereby the reasonably good areal overlap of the mass-normalized CVs indicates that the nominal cathode catalyst loadings are in good agreement with the actual catalyst loadings (Fig. 1). This can be examined more quantitatively by comparing the catalyst mass-normalized or specific charge (q_spec) between 0.2 V to 0.75 V (marked by the gray shaded region in Fig. 1) and the specific capacitance (C_spec) at 0.4 V (marked by the dashed line in Fig. 1). The average values for q_spec and C_spec are 38 C g⁻¹ _cat and 95.5 F g⁻¹ _cat, respectively, and differ by only ±7% across all MEAs (Table I).

Zoom In Zoom Out Reset image size

By further inspection of the CVs with the 0.4 and 1.0 mg_cat cm⁻² _MEA loadings, a slight current increase is observed at ≈0.6 V, which in the literature is usually assigned to a quinone/hydroquinone redox couple. ²⁹ Another redox couple is evident at ≈0.8 V for the higher MEA loadings (2.0 and 4.0 mg_cat cm⁻² _MEA) and can be assigned to a Fe³⁺/Fe²⁺ redox couple (at 0.77 V according to the Fe Pourbaix diagram ³⁰ ) that is typical for non-heme Fe–N₄ moieties of Fe–N–C catalysts, exactly as reported in Ref. 31, meaning a moiety different from those in biological heme groups. These slight differences in the shape of the CVs between low- and high-loaded electrodes could derive from a difference in degradation between low and high loadings during the electrochemical tests preceding the CVs (see the section entitled "Stability of the catalyst during sequential H₂ /O₂ polarization curves"), which could cause a slight increase in capacitance and the little changes observed in the CV profiles.

For all four loadings, the oxygen mass activity of the PGM-free cathode catalyst was determined from the first recorded H₂/O₂ polarization curve (referred to as initial in the subsequent figures). For MEA loadings from 0.4 mg_cat cm⁻² _MEA to 2.0 mg_cat cm⁻² _MEA, the H₂/O₂ performance increases substantially (Fig. 2a), as one would expect. However, no significant gain in H₂/O₂ performance at high current densities is observed anymore between 2.0 mg_cat cm⁻² _MEA and 4.0 mg_cat cm⁻² _MEA, suggesting a substantial decrease in catalyst utilization for the thick high-loaded MEAs. A similar behavior was observed by Banham et al ., ⁸ who showed a significant gain in the high-current density H₂/O₂ performance for their MEAs with an Fe–N–C ORR catalyst when increasing the loading from 1.0 mg_cat cm⁻² _MEA to 2.5 mg_cat cm⁻² _MEA, but no further gain when increasing the loading from 2.5 mg_cat cm⁻² _MEA to 4.0 mg_cat cm⁻² _MEA. These authors had concluded that it would be necessary to optimize the porosity of the electrodes as well as the ionomer content in the cathode catalyst layer to further improve the high-current density H₂/O₂ performance for the MEAs with the high cathode catalyst loading of 4.0 mg_cat cm⁻² _MEA. With regards to the evaluation of the ORR mass activity that is determined in the low current density region under O₂, the transport resistances should be sufficiently small to still allow for its quantification.

Zoom In Zoom Out Reset image size

For a typical PEMFC with Pt/C catalyst based MEAs, the high frequency resistance (HFR) represents the sum of the membrane proton conduction resistance (${R}_{{\rm{membrane}}}$) and the electrical contact resistance between the GDL and the flow field (${R}_{{\rm{contact}}\,{\rm{DM}}/{\rm{FF}}}$). However, here we observe an increase of the HFR when increasing the cathode catalyst loading from 1.0 to 4.0 mg_cat cm⁻² _MEA (i.e., from ca. 70 to ca. 90 mΩ cm²; see Fig. 2b), and no significant HFR difference between 0.4 and 1.0 mg_cat cm⁻² _MEA (i.e., ca. 70 mΩ cm²). For the used membrane (Nafion^® 212), the expected resistance under the described operating conditions of 80 °C and 90% RH is ${R}_{{\rm{membrane}}}$ ≈50 mΩ cm², ³² while the contact resistance between the gas diffusion layers and the flow fields (${R}_{{\rm{contact}}\,{\rm{DM}}/{\rm{FF}}}$) was measured by a previously described method, ³³ yielding a value of ≈10 mΩ cm² for 20% GDL compression. This would add up to an estimated HFR of ≈60 mΩ cm², which means that the PGM-free catalyst based MEAs in this study show a higher than expected HFR value, particularly for the thick high-loaded electrodes. A similar HFR increase with increasing cathode catalyst loading was observed by Baricci et al. ¹⁸ . One reasonable explanation for this behavior could be the higher electrical resistance of the rather thick catalyst layers based on PGM-free catalysts compared to the conventional, only ≈10 μm thick Pt/C electrodes, for which the through-plane electrical resistance is negligible and generally ignored. ³⁴ In fact, in one of the previous studies of our group (Landesfeind et al. ³⁵ ), an increase in HFR was reported for cases where the electrical resistance across a porous electrode becomes significant compared to the ionic resistance. In this case, the determination of the ionic resistance from the Nyquist plot obtained under blocking conditions is less straightforward. This will be discussed in more detail in the following section.

To further confirm the impact of the electrical resistance of the electrode, four-point-probe in-plane electrical conductivity measurements were conducted, which, in combination with the measured cathode electrode thickness (t_cathode; see 2^nd column in Table II) and Eq. 1, yields the electrical in-plane resistivity of the cathode catalyst layer (${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}};$ see 3^rd column in Table II). As shown in Table II, the cathode catalyst layer thickness is roughly proportional to the cathode catalyst loading of the MEA, whereby owing to the error in the thickness measurement (ca. ±3 μm), no sufficiently accurate thickness values could be obtained for the thinnest electrode with a loading of 0.4 mg_cat cm⁻² _MEA. (for this reason, no thickness values are reported for the 0.4 mg_cat cm⁻² _MEA electrodes in Table II). The variation of the electrical in-plane resistivity values (${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}$) of the MEAs is reasonably small (as one would expect), except for one of the 4.0 mg_cat cm⁻² _MEA loading MEAs, for which an almost 2-fold higher ${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}$ value was obtained. We believe that this high ${\rho }_{{{\rm{e}}}^{-},{\rm{cath}}}$ value is the result of cracks in the catalyst layer that lead to an apparent increase of in-plane resistivity. In general, one can conclude that reliable in-plane four-point probe resistance measurements can only be made with electrodes that are thick enough to allow for a sufficiently precise thickness measurement and that are thin enough to be sufficiently free of cracks. Therefore, the true in-plane resistivity of the here produced catalyst layers is approximately 2.6 ± 0.3 Ω cm (corresponding to the average and the standard deviation of the five measurements in Table II, excluding the measurement of one of the 4.0 mg_cat cm⁻² _MEA loading MEAs).

Assuming an isotropic electrical conductivity of the cathode catalyst layer, an approximate value of the through-plane electrical resistance (${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$) can be calculated from Eq. 2 and is shown in the last column of Table II (for the above given reason, the value of one of the 4.0 mg_cat cm⁻² _MEA loading MEAs is likely compromised due to cracks in the catalyst layer).

It has already been discussed in the literature that for the typically rather thick PGM-free cathode catalyst layers the proton conduction resistance (${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$) plays an important role, especially in the high current density region. ¹⁷ Therefore, AC impedance measurements in blocking condition (i.e., with H₂ on the anode and N₂ on the cathode) were conducted to further quantify the ionic resistance of the cathode catalyst layer. As in the case of MEAs with platinum-based cathode catalysts, the proton conduction resistance through the catalyst layer can be quantified using a transmission line model (TLM) analysis. ²⁶ In general, for Pt/C-catalyst-based electrodes the difference between the low-frequency resistance (LFR) and the HFR represents one third of ${R}_{{{\rm{H}}}^{+},{\rm{cath}}},$ ²⁶ whereas for PGM-free catalysts this may not be the case, depending on the ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}/{R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ ratio. As mentioned in the previous section, the determination of the proton conduction resistance from the Nyquist plot (blocking conditions) is less straightforward if ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ becomes significant compared to ${R}_{{{\rm{H}}}^{+},{\rm{cath}}},$ which is described in Fig. 2 of Ref. 35. On the other hand, in that study it has been shown that if ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}/{R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ ≤ 0.10, the true ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ value is larger by at most 20% compared to the apparent ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ value that would be obtained from the simple TLM analysis approach given by Liu et al., ²⁶ i.e., compared to the value obtained from ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}\,=\,3\cdot \,\left(LFR-HFR\right);$ for ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}/{R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ ≤ 0.05, this difference between the true and the apparent ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ value is at most 10%.

To evaluate the possible error in the ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ value (in through-plane direction) obtained from the simple TLM analysis (i.e., assuming a negligible contribution from ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$), we assume that the electrical resistance of the catalyst layers determined by the 4-point-probe measurements is a reasonably good measure of the catalyst layer through-plane electronic resistance. For this, we first determine the ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values for the differently loaded electrodes by applying the simple TLM analysis to the blocking condition impedance data shown in Fig. 3a. The resulting apparent ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values are given by the gray bars in Fig. 3b for all four loadings. Based on the ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ values in Table II, the ${R}_{{{\rm{e}}}^{-},\,\mathrm{cath}}/{R}_{{{\rm{H}}}^{+},\,\mathrm{cath}}$ ratios for the 1.0-4.0 mg_cat cm⁻² _MEA loaded MEAs are ≈0.05, which means that the actual ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values are expected to be ≈10% larger than the apparent ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values reported in Fig. 3b. Therefore, the simple TLM analysis yields reasonably accurate ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values for the here examined electrodes.

Zoom In Zoom Out Reset image size

Note that despite the fact that the ratio of ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}/{R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ of ≈0.05 is relatively small, this will still result in a significant added contribution of ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ to the HFR, estimated to be on the order of ΔHFR ≈ 0.05·${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ (see Fig. 2 in Ref. 35). For the 2.0 and 4.0 mg_cat cm⁻² _MEA loaded MEAs, this would amount to ΔHFR ≈ 6 mΩ cm² and 33 mΩ cm², respectively, which is reasonably consistent with the HFR measurements shown in Fig. 2b. It should further be noted that the here-observed small impact of ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ on the determination of ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ does not necessarily mean that the impact of ${R}_{{{\rm{e}}}^{-},{\rm{cath}}}$ is negligible for all PGM-free cathode catalyst based electrode layers, as the electronic resistance of some PGM-free catalysts can be significantly higher, in which case a more complex analysis of the impedance spectra is required in order to quantify ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ (currently being examined in our group).

Assuming that the ORR kinetics of the PGM-free catalyst follow Tafel kinetics, the effective proton conduction resistance (${R}_{{{\rm{H}}}^{+},{\rm{cath}}}^{{\rm{eff}}}$) that allows to determine the voltage loss due to the proton conduction resistance across the cathode (${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$) can be determined by the same approach previously described by Neyerlin et al. ³⁶ (based on assuming a negligible electrical resistance of the catalyst layer):

$$\begin{eqnarray}&&{R}_{{{\rm{H}}}^{+},{\rm{cath}}}^{{\rm{eff}}}=\displaystyle \frac{{R}_{{{\rm{H}}}^{+},{\rm{cath}}}}{3+\zeta }\end{eqnarray} \tag{ 3 }$$

Here, ζ represents the dimensionless correction factor that depends on the product of current density and the proton conduction resistance divided by the intrinsic Tafel slope for the ORR (b), i.e., on $\left(i\,\cdot \,{R}_{{{\rm{H}}}^{+},{\rm{cath}}}\right)/b.$ ³⁶ The voltage losses associated with the proton conduction resistance can then be determined from:

$$\begin{eqnarray}&&{\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}=i\,\cdot \,{R}_{{{\rm{H}}}^{+},{\rm{cath}}}^{{\rm{eff}}}\end{eqnarray} \tag{ 4 }$$

Since ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}^{{\rm{eff}}}$ depends on the intrinsic Tafel slope for the ORR, the Tafel slope must be known in order to determine ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}.$ For this reason, Fig. 4 shows the cell voltage correction for the proton transport losses for the differently loaded MEAs, assuming either b = 140 mV dec⁻¹ (open symbols) or b = 70 mV dec⁻¹ way that is generally observed for Pt/C-based cathodes ²² (closed symbols).

Zoom In Zoom Out Reset image size

Figure 4 shows that for the small ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values for the thin electrodes of the low loaded MEAs (i.e., for 0.4 mg_cat cm⁻² _MEA (green) and 1.0 mg_cat cm⁻² _MEA (orange)), the difference in ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$ based on assuming Tafel slopes of either 70 or 140 mV dec⁻¹ is negligible in the whole current density range. On the other hand, for the thicker catalyst layers of the MEAs with 2.0 mg_cat cm⁻² _MEA (blue) and 4.0 mg_cat cm⁻² _MEA (red) that exhibit rather large ${R}_{{{\rm{H}}}^{+},{\rm{cath}}}$ values (see Fig. 3b), the difference in ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$ between the two assumed Tafel slopes becomes more significant at high current densities, amounting to ≈20 mV and ≈30 mV at 1000 mA cm⁻² _MEA for the 2.0 and 4.0 mg_cat cm⁻² _MEA loaded MEAs, respectively. Nevertheless, this difference is insignificant in the low current density region in which ORR mass activities are generally determined. For example, when determining the ORR mass activity of the PGM-free catalyst at the commonly used cathode potential of 0.80 V, the current density even for the highest catalyst loading of 4.0 mg_cat cm⁻² _MEA would only be ≈100 mA cm⁻² _MEA (see Fig. 2a), where the difference in ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$ is only ≈2 mV between the two assumed Tafel slopes of 70 and 140 mV dec⁻¹.

However, when correcting the polarization curves shown in Fig. 2 for ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$ over the entire current density range, a specific Tafel slope value has to be chosen a priori and then would have to be consistent with the transport-corrected H₂/O₂ performance curves. In the following, we have assumed an intrinsic Tafel slope of 140 mV/dec. For comparison, Ahluwalia et al. ³⁷ reported two ORR Tafel slopes for an Fe–N–C catalyst, namely ≈120 mV dec⁻¹ at low current densities and ≈220 mV dec⁻¹ at high current densities, whereby the latter was considered to be the intrinsic Tafel slope based on their kinetic model.

Figure 5 shows the H₂/O₂ polarization curves in terms of the recorded cell voltage (${E}_{{\rm{cell}}};$ same data as in Fig. 2a), of the HFR-corrected cell voltages ((${E}_{HFR\,{\rm{corr}}.};$ using the HFR data shown in Fig. 2b), and of the cell voltages corrected for the HFR and for ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$ $\left({E}_{HFR+{R}_{{{\rm{H}}}_{{\rm{cath}}}^{+}}{\rm{corr}}.}\right);$ based on the values shown in Fig. 4 for b = 140 mV dec⁻¹).

Zoom In Zoom Out Reset image size

In order to determine the ORR mass activities for the differently loaded MEAs, the current densities of the HFR- and ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$-corrected H₂/O₂ performance curves (dotted lines in Fig. 5) were normalized by the cathode catalyst loading. This is shown in Fig. 6a, indicating that for a given HFR- and ${\rm{\Delta }}{V}_{{{\rm{H}}}^{+},{\rm{cath}}}$-corrected voltage (${E}_{HFR+{R}_{{{\rm{H}}}_{{\rm{cath}}}^{+}}{\rm{corr}}}.$) the mass activity (i_mass, in units of A/g) is essentially independent of the cathode catalyst loading. At the typical transport-corrected reference voltage of 0.80 V, the mass activities for cathode catalyst loadings of 1.0-4.0 mg_cat cm⁻² _MEA range between 30–40 A g⁻¹; at 0.75 V, the mass activities range between 60–100 A g⁻¹ for all MEAs (0.4–4.0 mg_cat cm⁻² _MEA). Thus, there is no significant variation of the ORR mass activity over a 10-fold change in catalyst loading. Jaouen et al. also reported that the mass activity at 0.8 V over a 4-fold change in the loading has just slightly changed for nine PGM-free different catalysts. ³⁸

Zoom In Zoom Out Reset image size

The dependence of the ORR mass activity of the PGM-free catalyst on catalyst loading was also investigated by RDE experiments in 0.1 M HClO₄ at 25 °C, using a wide range of catalyst loadings (40, 80, 160, 290, and 600 μg_cat cm⁻² _geom). After correcting the disk potential for the uncompensated electrolyte resistance (obtaining the iR-free cell voltage, ${E}_{iR-{\rm{free}}}$) and for the capacitive current contributions, the geometric currents were normalized by the catalyst loadings to obtain the mass activities (i_mass). No correction for the oxygen mass transport resistance was performed, since the maximum geometric current densities for the RDE data shown in Fig. 6b are kept at ≤20% of the diffusion-limited current density for a 2-electron reduction process, so that O₂ transport resistances should be negligible (i.e., the RDE data are acquired in the kinetically-controlled region). As can be seen in Fig. 6b, the mass activities in the kinetically-controlled region overlap quite closely for all the catalyst loadings (40–600 μg_cat cm⁻² _geom), suggesting that the catalyst loading does not have any impact on the ORR mass activity in the kinetically-controlled region, analogous to what has been observed in the PEMFC measurements (Fig. 6a). The mass activities at 0.80 V vs RHE range between 1.3–2.0 A g⁻¹, without showing any trend with respect to catalyst loading (excluding the data set for the loading of 160 μg_cat cm⁻² _geom that has very large error bars, the mass activity ranges between 1.3–1.5 A g⁻¹). This is comparable to the mass activities of Fe–N–C catalysts synthesized by several different groups, reported to range between 0.7–3.2 A g⁻¹ at 0.80 V vs RHE and 25 °C (taken in 0.5 M H₂SO₄ at 1600 rpm, at loadings of 800 μg_cat cm⁻² _geom). ³⁹ The ≈25-fold higher ORR mass activities at 0.80 V vs RHE obtained in the single-cell PEMFC measurements at 80 °C (Fig. 6a) compared to the RDE measurements at 25 °C (Fig. 6b) is likely due to the difference in temperature, in which case this would correspond to an apparent activation energy of ≈50 kJ mol⁻¹. This is much higher than expected ^40,41 and requires a further reason to explain the difference. The use of catalyst without ball milling for RDE analysis and ball-milled for PEMFC test could be the second important origin of lower activity in RDE, possibly increased in PEMFC due to exposition of additional active sites to electrolyte/ionomer and to improved dispersion/homogeneity of the catalyst ink. ⁴²

Based on the PEMFC and RDE data in Fig. 6, it is clear that the loading of the here examined Fe–N–C-based ORR catalyst does not affect its mass activity, which is at variance with our previous study with an Fe-substituted ZrO₂ (Fe_xZr_1−xO_2−δ ) catalyst, where we had observed an increase in ORR mass activity with catalyst loading. ¹⁵ In that study we had hypothesized that with an increase in catalyst loading, which entails an increase in catalyst layer thickness, the probability of intermediate hydrogen peroxide to be further reduced rather than leaving the catalyst layer would also increase. The associated decrease in peroxide concentration within the catalyst layer with increasing catalyst loading was then assumed to results in a positive potential shift of the Nernst potential for the electrochemical reduction of O₂ to H₂O₂, thus in an apparent variation in mass activity. The here reported catalyst is also known to provide a lower peroxide yield (≈3% at 0.4 V_RHE and 800 μg_cat cm⁻² _geom ³⁹ ) compared to Fe_xZr_1−xO_2−δ catalyst (≈10% at 0.4 V_RHE and 576 μg_cat cm⁻² _geom ¹⁵ ). The evident large difference in mass activity and peroxide yield between the here used Fe–N–C catalyst and the Fe_xZr_1−xO_2−δ catalyst in our previous study is the most likely origin for the different relationships between ORR mass activity and catalyst loading. More explicitly, using a very active (and very low peroxide-producing, see ⁴³ ) catalyst, it is unlikely that peroxide leaves the catalyst layer unreacted even at low loadings, and it is hard to detect any further significant decrease in peroxide concentration with a loading increase, i.e., it is unlikely to detect any loading effect on mass activity.

Aside from the lower ORR mass activity compared to Pt/C-based catalysts, the low stability of PGM-free catalysts is another barrier on the way of their utilization as an alternative for PGM-based catalysts in PEMFCs. Their degradation mechanisms have been studied, but they are not completely clear, thus the way to prolong the operating hours for PGM-free catalysts has not yet emerged. Recent comprehensive studies suggested that the main degradation mechanisms of MEAs with PGM-free cathode catalysts is related to a loss of active sites caused by the Fe-catalyzed formation of hydrogen peroxide in the presence of oxygen. ^44,45

As shown in Fig. 7, a significant catalyst degradation is evident over the course of the acquisition of only five H₂/O₂ polarization curves. Comparing the initial polarization curve with the 2^nd polarization curve, which was recorded after the CVs shown in Fig. 1, a clear performance decay is evident for both low loaded (green curves) and high loaded (red curves) MEAs. Between the 2^nd and the 3^rd polarization curve, the impedance measurements to determine the proton conduction resistance were conducted. As the performance loss between the 2^nd and the 3^rd polarization curves does not seem larger than between the 1^st and the 2^nd polarization curves, this impedance measurement under H₂/N₂ does not seem to lead to an additional performance degradation; this would be consistent with the findings by Ahluwalia et al. ⁴⁴ that the degradation of PGM-free catalysts is most pronounced in the presence of oxygen. Over the 4^th and 5^th polarization curve, the performance degradation per curve is decreasing, but it is still significant.

Zoom In Zoom Out Reset image size

Looking at the cell voltage difference between initial and 5^th H₂/O₂ polarization curves at a current density of 100 mA cm⁻² _MEA, higher degradation in the low loaded MEAs (85 mV loss) was observed compared to high loaded MEAs (40 mV loss). These results are in agreement with the literature, where it has been reported that higher loadings show higher stability due to the higher absolute catalyst amount available for the ORR reaction. ⁸ At higher current density (500 mA cm⁻² _MEA) the difference in the degradation of the low and high loaded MEAs is rather small (55 mV vs 45 mV loss). Nevertheless, it cannot be excluded that this may also be related to the different potential ranges in which the differently loaded cathodes operate when recording the H₂/O₂ polarization curves, i.e., extending to much lower voltages in the case of the low loaded MEAs. Therefore, the difference in the average potential might also be another factor if there were a potential dependence of the catalyst degradation, even though Ahluwalia et al. ⁴⁴ suggest that potential is not the main driver for the degradation of PGM-free catalysts.

In summary, while MEAs with lower catalyst loadings exhibit higher degradation at low current density and almost the same degradation at high current density, the presented data do not allow to conclude whether the loading of the PGM-free ORR catalyst impacts catalyst durability. Also, in order to evaluate the loading impact on stability of the catalyst, it is necessary to deconvolute and quantify the various voltage loss terms over the course of a durability experiments, which will be part of future work.