Multiple variations of cathode catalyst materials were subjected to a comprehensive in-situ characterization including an accelerated stress test focused on degrading the cathode catalyst layer. The following cathode catalyst layer properties were varied during production: I/C ratio, Pt/C wt%, carbon support type, ionomer EW and catalyst type. Each material was subjected to a base version of an accelerated stress test over 30,000 rectangular potential cycles between 0.6 V and 0.95 V, with 3 s dwell time at each potential. As variations to this AST, each material was additionally tested with an upper potential limit of 1.15 V, while the I/C ratio and carbon support material variations were further degraded at varying relative humidities of 40%, 70% and 100%. Table 1 gives an overview of all conducted experiments, split in material and AST variations. It should be noted that test number 2 and 31 are the same dataset, as it serves as a reference in two variations. However, for better readability of the whole data collection they have been listed as separate entries. The same accounts for test numbers (3/14), (11/33), (12/18), (22/35) and (27/41), resulting in a total of 36 unique datasets.Table 1 Measurements overview including variations in material composition and AST boundary conditions.The decal transfer route has been used as a CCM production process due to its high reliability and reproducibility with different materials in this study35. The cathode catalyst layers were produced by screen printing on a decal substrate with a target Pt loading of 0.38 mg cm−2. Catalyst powders were provided by two different suppliers: Pt and PtCo catalysts on high surface area carbons (50 wt% Pt/C) provided by supplier 1 (SUP1), Pt catalysts on high surface area carbon with Pt/C ratios varying between 20 and 60 wt% provided by supplier 2 (SUP2) and Pt catalysts (50 wt% Pt/C) deposited on both high and low surface area carbon supports (SUP2). Aquivion (liquid dispersion, 25% in water, Sigma-Aldrich Chemie GmbH) with an EW of 720, 790, 830 and 980 g mol−1 was used as the ionomer. The ionomer and carbon content were adjusted to achieve I/C ratios of 0.5, 0.8 and 1.2. A mixture of organic solvents (50 Vol.% ethylene glycol, 50 Vol.% propylene glycol methyl ether) was used for dispersion. For the anode catalyst layers, a Umicore Elyst 0390 catalyst with 20% platinum on low surface area carbon was used to achieve a loading of 0.05 mg cm−2. After printing, each catalyst layer was dried in a convection dryer at 150 °C for 10 min. Each layer was afterwards transferred via hot-pressing at 180 °C and 5 MPa (referred to the printed area of 20 cm2) for 15 min onto a GORE SELECT membrane with 18 µm thickness. As gas diffusion layer, a Freudenberg H23C9 was used on both anode and cathode side. The ink dispersing and details of the screen-printing process are further described in a previous publication36. Similar screen printing procedures have also been investigated by other groups37,38,39,40.All tests were carried out in a fully automated in-house developed test bench together with a “Baltic ISE qCf Liquid Cooling high amp zero gradient” test cell with 12 cm2 active area. The same test cell and test bench were used throughout all tests. Variations in relative gas humidity were realized by dynamically mixing dry and humidified gas streams, while the cell temperature was controlled by a Lauda Eco Silver RE1050 cryostat. For electrochemical measurements a Höcherl and Hackl PLI1206ZVSV5 electric load was used in combination with a Zahner Zennium potentiostat. During all measurements, the cells were compressed at a constant clamping pressure of 1.35 MPa.Table 2 gives an overview of the characterization protocol applied to each material variation. A full characterization was performed at beginning of test (BoT) and end of test (EoT), while a shortened in-between characterization was performed at distinctive aging intervals. The applied protocol resulted in an approximate testing time of 120 h per material, leading to a total of over 4,000 h testing time for all 36 tests. The characterization protocol includes the methods described below. Further explanation on the displayed error bars is provided in the “Technical Validation” section.Table 2 Overview of the applied test protocol, including a Break-In procedure, comprehensive beginning of test characterization, voltage cycling and shortened in-between characterization, and an end of test characterization.Step 1 – Break-InAs an initial conditioning procedure, each cell was first operated galvanostatic for 2 h at a constant current of 1.5 A cm−2, at H2/air (with 2 and 5 normal liter per minute (NLPM) respectively), 100% RH, 80 °C, and 2 bara gas pressure. During this conditioning step, the potential was recorded every 30 s. Afterwards, the cell was dynamically cycled between OCV (10 s), 0.6 V (60 s) and 0.4 V (60 s) for a total of 6 h in which the final 10 s in each cycle step were recorded. Figure 1 gives an overview of the acquired data during both conditioning steps. In case a cell could not reach the requested 1.5 A cm−2 by obtaining a voltage >0.4 V, the current density was reduced in steps of 0.2 A cm−2 until the voltage requirement was met.Fig. 1Break-In procedure for initial cell conditioning: 2 h of galvanostatic steady state operation at 1.5 A cm−2 (left) followed by 6 h of potentiostatic cycling between OCV, 0.6 V and 0.4 V (right). The reproducibility is indicated by transparent error bands.Step 2 – Limiting Current MeasurementsLimiting current measurements were carried out according to Beuscher and Baker et al.41,42,43,44 to obtain oxygen diffusion resistances. The cell was operated at 80 °C and 100% RH with 2 NLPM H2 on the anode and a O2 + N2 mix with a total flow of 5 NLPM on the cathode. A total of 16 limiting current values were extracted as the maximum current of voltage sweeps from 0.40 to 0.15 V, at diluted oxygen concentrations of 1, 1.5, 2 and 2.5%, and gas pressures of 1.5, 2.0, 2.5 and 3.0 bara. An exemplary data set of test run 34 is shown in Fig. 2.Fig. 2Measured limiting currents vs. O2 concentration and pressure (left). Calculated total mass transport resistance in each pressure and resistances split into pressure dependent and independent part (right). All values shown at both BoT and EoT of test run 34.The total mass transport resistance RO2,total was calculated according to Eq. 1 as a function of Faradays constant F, the oxygen concentration cO2, and the measured limiting current ilim. The resistance was then averaged over all concentrations in each pressure level.$${R}_{O2,{total}}=4F\frac{{c}_{o2}}{{i}_{{lim}}}$$
(1)
$${R}_{O2,{total}}={\hat{R}}_{p,{dep.}}\cdot p+{R}_{p,{indep}.}$$
(2)
The obtained total resistances RO2,total for each pressure p were divided into a pressure dependent (molecular diffusion) and independent part (Knudsen diffusion) by linear extrapolation according to Eq. 2. The molecular diffusion is expressed as the slope of the resistance change with pressures in s m−1 bar−1.Step 3, 6, 8 – Current-Voltage (I-V) curvesThe cell current was measured at distinctive points between open circuit voltage (OCV) and 0.2 V, at H2/air (with 2 and 5 NLPM respectively), 80 °C, 2 bara gas pressure and gas humidities of 40%, 70% and 100% RH.Each point was averaged over 30 s, after conditioning for a minimum of 5 minutes, and until a stability of <1% was reached. The high frequency resistance was measured in each operating point between 0.9 and 0.4 V, by recording a shortened impedance spectrum between 500 Hz and 10 kHz. In addition to this shortened spectrum for HFR evaluation, a full spectrum down to 1 Hz was recorded at various operating points, which is discussed in the next paragraph. The recorded HFR data was used to carry out a I-R correction of the performance data. Polarization curves were measured at BoT and EoT at all RHs, and additionally after 1k, 5k and 10k cycles at 100% RH. An exemplary data set of the recorded polarization curves as well as the respective full impedance spectra at 0.1 A cm−2 are shown in Fig. 3.Fig. 3I-V-curves from BoT to EoT at 100%, 70% and 40% RH (left column) and corresponding EIS spectra at 0.1 A cm−2 measured from 1–10 kHz (right column). Data shown from cell 11 (I/C 0.8, EW 790 g/mol, 50 wt% Pt/C on HSA carbon, UPL 1.15 V @ 100% RH).Step 4, 7, 9 – Electrochemical Impedance Spectroscopy (EIS) in H2/airAfter each polarization curve, impedance spectra were measured in H2/air conditions, at 2 bara, 80 °C and 100%, 70%, and 40% RH on both anode and cathode. The spectra were measured from 1 Hz – 10 kHz, at 0.1, 0.5 and 1.0 A cm−2 with an amplitude of 10% of the direct current. The transmission line model discussed in the H2/N2 EIS section was not employed for parameter fitting as it assumes a homogeneous electrode distribution, resulting in low accuracy when fitting EIS measured under load. To perform further analysis of the spectra, an adapted model is necessary that allows for parameter variations in the electrode along the through-plane direction.Step 5 – EIS in H2/N2
Impedance spectra were measured in H2/N2 at 200 and 450 mV with a 10 mV amplitude, at 80 °C, atmospheric pressure and 100% RH on both anode and cathode. A transmission line model adapted from Makharia et al.45 was used to evaluate the spectra, which accounts for double layer capacity CDL, protonic catalyst layer resistance Rion, high frequency resistance RHFR, charge transfer resistance RCT and hardware inductivity Linduct.. To achieve better fit accuracy, the capacitors have been replaced with constant phase elements, introducing the exponent φ. Figure 4 shows an example data set of the fitted impedance data and the corresponding parameters.Fig. 4EIS spectra in H2/N2, measured from 1–10 kHz. Ionic resistance and double layer capacity evaluated by fitting with a transmission line model (equivalent circuit shown in inlay).The resulting equivalent circuit is shown in the inset of Fig. 4. The total impedance of the circuit can be described by Eqs. 3, 4:$$Z(\omega )=\sqrt{{R}_{{ion}}{Z}_{{RC}}}\,\coth \,\left(\sqrt{\frac{{R}_{{ion}}}{{Z}_{{RC}}}}\right)+{R}_{{HFR}}+j\omega L$$
(3)
$${Z}_{{RC}}=\frac{{R}_{{CT}}}{1+{R}_{{CT}}{(j\omega )}^{\varphi }{C}_{{DL}}}$$
(4)
Step 10, 11 – Cyclic Voltammetry (CV) and Linear Sweep Voltammetry (LSV)For the CV, the cell was cycled 5 times between 0.05 and 0.95 V at 100 mV s−1 scan rate, with 1 NLPM H2 on the anode and no flow in the cathode. The measurement was performed at 80 °C and 100% RH at atmospheric pressure. Prior to the voltage cycling, the cell was conditioned for 15 minutes in the mentioned conditions. Afterwards, the ECSA was obtained by averaging the H2 adsorption and desorption charge, and the double layer capacity was evaluated at the minimum in the region between 0.3 and 0.6 V46. CVs were conducted in each aging step from 0 to 30k cycles. Figure 5 displays an exemplary CV dataset of test run 2. The LSV was performed in the same conditions, only with 1 NLPM N2 on the cathode, and a potential sweep from 0.1 to 0.5 V with 1 mV s−1 scan rate. The H2 crossover current was evaluated as the current at 200 mV in a linear fit of the LSV curve between 0.3 and 0.5 V47.Fig. 5Cyclic voltammetry and evaluated ECSA and double layer capacity over time for cell 2 (I/C 0.8, EW 790 g mol−1, 50 wt% Pt/C on HSA carbon, UPL 0.95 V @ 100% RH).Step 12 – Accelerated Stress Test (AST)The catalyst degradation was performed according to the protocol proposed by the US Department of Energy (DoE), where the cell is operated at 80 °C, in H2/N2 atmosphere and cycled between 0.6 and a fixed upper potential limit (UPL) for 30,000 cycles, with 3 s dwell time at each potential and a fixed slope time of 0.25 s between potentials. For all material variations, measurements were conducted with an UPL of 0.95 V and 1.15 V. For the I/C ratio and carbon support variation the AST was additionally performed at 100%, 70% and 40% RH. Current and voltage data was recorded at 10 Hz during the AST as shown in Fig. 6.Fig. 6Applied accelerated stress test, voltage cycling between 0.6 and 0.95 V and 1.15 V with 3 s dwell time at each potential (left) and cell current response (right).
In-Situ Characterization of Cathode Catalyst Degradation in PEM Fuel Cells
