Evaluation of double layer capacitance and carbon support corrosion under variable conditions

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 1/10 A1112 Evaluation of double layer capacitance and carbon support corrosion under variable conditions Amina Alimbekova*, Lius Daniel, Jari Ihonen VTT – Technical Research Centre of Finland, Biologinkuja 3-5, 02044 Espoo/Finland; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract The performance and longevity of proton-exchange membrane fuel cell (PEMFC) depends on, among, other factors, health, and durability of the catalyst layer. The state of health and degradation rate of PEMFC membrane electrode assemblies (MEAs) is typically assessed using diagnostic markers such as electrochemical surface area (ECSA), gas crossover and double layer capacitance. Recently, galvanostatic charge and discharge techniques have been developed and successfully replicated for both single cell and a fuel cell stack. These methods utilize potential-time differentials and galvanostatic current to calculate ECSA and double layer capacitance (Cdl) and other parameters. However, analyses of double layer capacitances using this method under varying conditions — such as relative humidities, carbon support type — remain limited. Additionally, Cdl as a time-resolved diagnostic marker has not been widely used for stacks when derived from the galvanostatic methods. Here is presented a framework to deconvolute Cdl contributions from Pt, carbon, and ionomer using galvanostatic method, establish degradation patterns and propose a predictive model for remaining useful life. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 2/10 Introduction Fuel cells are considered an effective hydrogen-based solution for electrification in industrial sectors and heavy-duty or marine transportation. Proton exchange membrane (PEM) fuel cell systems have been developed to meet performance requirements for these applications. However, the service life remains a recognized limitation in the widespread adoption of fuel cells. Understanding, monitoring and diagnosing the physicochemical characteristics of a PEM fuel cell is one of the critical factors in widening the bottleneck of fuel cell use.[1], [2] Electrocatalysts for PEMFC based on platinum are widely used due to their effective electrocatalytic behavior. The high surface-to-weight ratio of platinum particles within carbon support increases the surface area available for reactions. However, the platinum particles do not maintain their structure over time, so changes in the morphology of the catalyst layer lead to a reduced electrochemical response.[3], [4] The morphology changes, like Pt coarsening, coalescence, detachment, and migration, occur during due to operating parameters (relative humidity, temperature) and cycles that involve variable loads, high potentials, potential cycling, and short stops.[5], [6], [7] Carbon blacks are extensively utilized as catalyst supports due to their optimal stability and high porosity. The degradation of the carbon support occurs predominantly at the cathode, owing to the thermodynamic tendency for the electrochemical oxidation of carbon (E0≥ 0.207 V vs RHE). While carbon corrosion progresses slowly under standard operating conditions, it becomes appreciable during transient states. This phenomenon is observed when the cathode encounters elevated anodic overpotentials, exceeding 1.2 V, particularly during start-up and shut-down procedures involving air presence.[8] Carbon corrosion significantly decreases the performance and stability of the platinum catalyst due to Pt detachment and degradation of the Pt-support.[9], [10] Performance and durability of MEA can be described by parameters, such as electrochemical surface area (ECSA, m²⋅g⁻¹ [Pt]), hydrogen crossover current density (iH2, mA⋅cm⁻²), and double-layer capacitance (Cdl, mF⋅cm⁻²). Catalytic layer's activity is indicated with ECSA and can be affected by structural and morphological changes. The Faraday current that corresponds to the hydrogen permeability of the MEA over time is iH2, which can show reduction of thickness and aging. The interface condition of the catalytic layer, that shows the distribution of water and ionomer on platinum particles and carbon supports is described with Cdl. It correlates with electrode's surface area, carbon corrosion extent, and ionomer quantity and distribution. [11] Assessing MEA parameters and choosing MEAs with high uniformity for fuel cell stack assembly can enhance the stack's consistency, performance, and longevity. Electrochemical measurements are a common method to assess MEA and fuel cell performance in situ. Common electrochemical techniques for assessing MEA parameters in a single cell include the polarization curve method, cyclic voltammetry (CV), linear sweep voltammetry (LSV), electrochemical impedance spectroscopy (EIS), and galvanostatic methods.[2], [11], [12] CV and LSV techniques involve potentiodynamic control which is impractical for fuel cell stacks. The potential of individual cells cannot be controlled independently due to the heterogenous cells within a stack. Cells behave non-uniformly, due to differences in Pt loadings and operational history, and the distribution of the sweep rate imposed by a potentiostat will not be uniform.[13] Additionally, these experiments, including EIS, require specialist high-accuracy hardware that is difficult to integrate into test systems.[14] EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 3/10 A galvanostatic control method has recently been developed to assess these parameters at the stack level. [14], [15] This study aims to extract capacitive responses using a galvanostatic approach to extract Cdl. The total Cdl, will then be deconvoluted by observing and quantifying the contributions from ionomer, platinum and carbon support. By focusing on the linear voltage region before Faradaic reactions, separated from Pt hydrogen adsorption/desorption peaks, a robust analysis could be achieved. The goal is to develop in situ stack-level tracking of Cdl to propose a predictive model for MEA lifetime. Validating and optimizing the galvanostatic technique with CV, EIS at a single cell and then stack level will provide a comprehensive method for assessing fuel cell degradation and enhancing longterm performance. 1. Galvanostatic charge based method The galvanostatic charging method (GCM) involves applying a constant current to the fuel cell while recording voltage signals. The observed curve features voltage changes that occur quickly or slowly, i.e., have a larger or a smaller absolute slope respectively. The larger slope can be attributed to a few electrochemical processes occurring, while the smaller indicates electrochemical reactions that are associated with large charges. For the fuel cells, voltage change is usually slow both from approximately 0 – 0.3 V and 0.7-1.0 V, corresponding to the hydrogen adsorption-desorption and Pt oxidation-reduction processes respectively when compared to the conventional CV. At the same time, the voltage changes are fast around 0.4-0.6 V that are linked to the double layer charging currents. The galvanostatic charging-discharging plots can be transformed into voltage-differential capacity as first shown by Stevens et al. (2003) [16]. These can be equivalent to CV, as differential capacity (C/V) is comparable to the normalized currents from CV, when divided by the scan rate 𝐴 𝑉 𝑠−1 [=]𝐴 𝑠 𝑉 [=]𝐶 𝑉. However, as Lee et al. (2012) [15] notes, PEMFC in H2/N2 mode experience hydrogen crossover, which occurs in the similar voltages (~0.4 – 0.6 V), which is not capacitive and should be considered alongside the double layer capacitance. Thus, they outline the main electrochemical contributions to the differential charge dQ at a current iGa during time duration dt are hydrogen adsorption-desorption (QH/Pt), double layer charging (Qdl), and oxidation of crossover hydrogen gas (QH2). 𝑑𝑄= 𝑖𝐺𝐴𝑑𝑡 = 𝑑𝑄𝐻/𝑃𝑡 +𝑑𝑄𝑑𝑙 +𝑑𝑄𝐻2 =𝐶𝐻/𝑝𝑡(𝑉)𝑑𝑉+𝐶𝑑𝑙𝑑𝑉+𝑖𝐻2𝑑𝑡 (1) 𝑑𝑄 𝑑𝑉 ⁄= 𝑖𝐺𝑎 𝑑𝑡 𝑑𝑉 ⁄ = 𝐶𝐻/𝑃𝑡(𝑉)+𝐶𝑑𝑙 +𝑖𝐻2 𝑑𝑡 𝑑𝑉 ⁄ (2) 𝑖𝐺𝑎 =(𝑑𝑄𝐻/𝑃𝑡 𝑑𝑉 ⁄+𝐶𝑑𝑙)𝑑𝑉 𝑑𝑡 ⁄ + 𝑖𝐻2 (3) Brightman et al. (2013) [14] used the pseudo-CV to obtain the ECSA values, which is derived from the galvanostatic curves from an 18-cell stack. Hydrogen at the anode was shown to have an effect of hydrogen crossover was observed, thus the hydrogen concentration was reduced and an appropriate galvanostatic current density was selected. However, there remains an inherent error. Pei et al. (2014) [17] analyzed voltage curves in galvanostatic charging using the discrete derivative of voltage, dQ/dt. Then, he calculated iH2 with higher accuracy by linear regression when charging with multiple galvanostatic currents. According to equation (2), when the derivative dQH/Pt/dt is 0, the Cdl and iH2 can be calculated as following: EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 4/10 𝑖𝐺𝑎 = 𝐶𝑑𝑙 𝑑𝑉 𝑑𝑡 ⁄ |dQ dt=0 + 𝑖𝐻2 (4) Then Pei et al. (2018) [18] introduced a charging model using short-circuit resistance (Rsc) from improved LSV findings and enhanced the analytic process of iH2 in galvanostatic charging. There, the hydrogen crossover is represented as a constant current diode with limiting current, IH2, and Rsc, the short-circuit resistance, that describes the insulation of MEA. The galvanostatic charging current is then: 𝐼𝐺𝑎 = (𝑑𝑄𝐻/𝑃𝑡 𝑑𝑉 ⁄+𝐶𝑑𝑙)𝑑𝑉 𝑑𝑡 ⁄ + 𝐼𝐻2 +𝑉 𝑅𝑠𝑐 (5) Ren et al. (2022) [19] introduced the micro-current excitation (MCE) method based on an integral form excitation model. This method addresses challenges associated with the original GCM, including the requirement for a high-precision constant current supply, faster sampling frequency, and the accumulation of errors during data processing. The error revision accounts for the ohmic resistances as an initial voltage jump when measuring the galvanostatic curves. Then Cdl and Re are decouples using a single multiple linear regression algorithm, where the simplified equation assumes ionic current is approximated to the excitation current: ∫𝑖𝐺𝑎 𝑡 0𝑑𝑡 𝑡= 𝑄𝐻/𝑃𝑡 1 𝑡+𝐶𝑑𝑙𝑉−𝑉0−𝑖𝐺𝑎𝑅Ω 𝑡+𝑖𝐻2 +1 𝑅𝑒 ∫𝑉 𝑡 0𝑑𝑡 𝑡 ,𝑉 ∈ [𝑉𝑙𝑜𝑤,𝑉𝑢𝑝𝑝] (6) The components ∫𝑖𝐺𝑎 𝑡 0𝑑𝑡 𝑡,1 𝑡,𝑉−𝑉0−𝑖𝐺𝑎𝑅Ω 𝑡,𝑎𝑛𝑑 ∫𝑉 𝑡 0𝑑𝑡 𝑡 are derived variables which are used to identify MEA parameters: 𝑄𝐻/𝑃𝑡,𝐶𝑑𝑙,𝑎𝑛𝑑 𝑖𝐻2. With this method, MCE analyzes these signals using differential and integral methods to determine MEA parameters. This efficient approach measures ECSA, Cdl, Re, and iH2 for evaluating MEA status and stack consistency. 2. Double layer capacitance model Catalyst layer degradation mainly occurs due to changes in the type or number of active sites. An accelerated stress test (AST), based on parameters like electrochemical surface area (ECSA) or carbon corrosion during potential cycling, is often used to evaluate catalyst layer durability. In a state-of-the-art PEMFC catalyst layer, the active site type may stay the same during an AST if crystallite faces are randomly oriented. Therefore, changes in ECSA during operation or AST can be linked to variations in the number of accessible active sites. Performance degradation due to fewer active sites, thus reduced ECSA, may principally result from carbon support corrosion, ionomer/catalyst interface loss, reduced catalyst loading due structural changes of catalyst nanoparticles. [20] Two primary methods have been employed to study carbon corrosion in fuel cells: analyzing the electrochemical properties of carbon during corrosion and identifying oxidation products in the exhaust gases. The first method qualitatively compares changes in double-layer capacity but lacks quantitative carbon loss analysis. Combining electrochemical measurements with quantitative methods, like gas chromatography, enables extraction of this information. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 5/10 This approach explicitly describes interfaces that contribute to the total Cdl,total within the fuel cell catalyst layer. [21], [22] Therefore, the total double layer capacitance can be modelled as a sum of contributions of different interfaces: 𝐶𝑑𝑙,𝑡𝑜𝑡𝑎𝑙 =𝐶𝑑𝑙,𝑃𝑡 +𝐶𝑑𝑙,𝐶 +𝐶𝑑𝑙,𝑖𝑜𝑛𝑜𝑚𝑒𝑟 (7) Where: • 𝐶𝑑𝑙,𝑃𝑡 is capacitance at the Pt/electrolyte interface, largely influenced by Pt surface area and the local electrochemical environment. • 𝐶𝑑𝑙,𝐶 is capacitance at the carbon/electrolyte interface, associated with the electronic support structure (e.g., Vulcan or graphitized carbon). • 𝐶𝑑𝑙,𝑖𝑜𝑛𝑜𝑚𝑒𝑟 is capacitance due to ionomer phase interfaces, including ionomer surface coverage on both Pt and carbon. Each component responds differently to operating conditions and degradation processes, enabling their selective study. To quantify the influence of multiple operational and material parameters on the Cdl, a full factorial analysis of variance (ANOVA) model will be used. The Cdl values when derived from the measurements done under systematically varied conditions. Three categorical factors to be considered: relative humidity, AST stage (beginning, middle or end of life), and MEA type. The three-way ANOVA model is expressed as: 𝐶𝑑𝑙 𝑖𝑗𝑘 = 𝜇 + 𝛼𝑖+𝛽𝑗+𝛾𝑘+(𝛼𝛽)𝑖𝑗 +(𝛼𝛾)𝑖𝑘 +(𝛽𝛾)𝑗𝑘 +(𝛼𝛽𝛾)𝑖𝑗𝑘 +𝜀𝑖𝑗𝑘 (8) Where: • 𝐶𝑑𝑙 𝑖𝑗𝑘 is the measured Cdl for the i-th RH level, j-th AST level, and k-th MEA type. • 𝜇 is the overall mean Cdl. • 𝛼𝑖,𝛽𝑗,𝛾𝑘 are the main effects of RH, AST stage, and MEA type respectively. • (𝛼𝛽)𝑖𝑗,(𝛼𝛾)𝑖𝑘,(𝛽𝛾)𝑗𝑘 are two-way interaction terms. • (𝛼𝛽𝛾)𝑖𝑗𝑘 are three-way interaction terms. • 𝜀𝑖𝑗𝑘 is the residual error term. This model will allow evaluation not only of individual contributions of RH, AST stage, and MEA type to Cdl, but also their interactions, which may present synergistic of antagonistic effects in degradation or capacitance behavior. The aim of this work is to isolate and interpret the contributions of each component galvanostatically under different controlled variables. Pt degradation protocols (Pt-AST) will be applied to the MEAs for reduce the ECSA and track the corresponding Cdl change (Cdl,Pt). To evaluate the influence of carbon support corrosion on Cdl,total, carbon-specific AST will be used to quantify the Cdl,C. 3. Materials and Equipment The single cell tests are performed on commercial IRD Fuel Cell supplied MEAs (Pt/carbon Hispec® 9100 (AlfaAesar) catalyst coated membrane on a reinforced Nafion membrane) with the area of 25 cm2 and a Pt loading of 0.3 mg/cm2 for both the anode and the cathode, MEA 1. Multi-single cell stack is an assembly of 8 cells connected electrically in series, where current is applied as a stack. Gases are fed in parallel, hydrogen and nitrogen gas (purities 99.999 %). The air underwent compression, filtration, and drying processes, resulting in a humidity level of 0.5 hPa. Total 8 cells (area of one cell 23 cm2) are assembled in the multi-single cell. This configuration allows for measuring different MEAs at the same time under the same operational history. [23], [24] EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 6/10 The fuel cell test station (G60 series, Greenlight Innovation) is used to control gas composition, applied load, flow, and humidity, while the IviumStat2.h potentiostat/galvanostat, equipped with a uMUX 8-channel multiplexer, performs galvanostatic measurements. The working electrode is attached to the cathode current collector, while the reference and counter electrodes are connected to the anode side. The CV, EIS and AST procedure is controlled by MultiEmStat4 HR 8-channel potentiostat / galvanostat / impedance analyzer. 4. Experimental procedures Figure 1. The experimental procedure studying degradation under AST of different MEAs. The experimental procedure is pictured in Figure 1. All MEAs are activated by load cycling (applied current density of 86.9 mA cm⁻² or 2 A) alternating between H₂/N₂ and H₂/Air modes under cell temperatures of 70 °C and 40 °C. Polarization curves are measured for each set to assess cell activity. The polarization curves are recorded in current-control mode at 70°C with fully humidified gas feedings. The protocol involves decreasing the current density from 695.65 mA cm−2 down to 0 mA cm−2 (i.e., 16 – 0 A). Stoichiometry for the polarization curve set at 5/5 for anode and cathode. The activation process continues until cell activity reaches a plateau. Characterization methods CV, GCM, and EIS are measured at 60 °C with other operational conditions listed in Table 1. For CV tests, hydrogen is given to the anode, acting as a reference/counter electrode and nitrogen – to the cathode. The potential is swept from 0.1 V to 0.8 V at 20 mV s⁻¹ over two cycles per scan, with the second cycle used for analysis. Galvanostatic charge measurements are conducted at constant current densities, with initial and cut-off voltages of 0.1 V and 0.8 V, respectively. The full factorial design of experiment for the ANOVA will be done, where each of different types of MEAs is characterized at RH 30%, 80%, 100% and at three stages of life – beginning, middle, and end (BOL, MOL, EOL respectively). EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 7/10 Table 1. Characterization procedures and experimental conditions. Method Measured electrode Gas supply Flow (SLPM) Pressure (kPa) RH (%) Cell temperature (°C) CV Cathode Cathode: 100% N2 Anode: 100% H2 2 5 30 30 30, 80, 100 60 Scan rate (mV s-1) 20 Potential window (V) 0.1 – 0.8 Galvanostatic charge method Cathode Cathode: 100% N2 Anode: 100% H2 2 5 30 30 30, 80, 100 60 Applied current (mA cm-2) 5, 6, 7, 8, 9, 10 Potential window (V) 0.1 – 0.8 EIS Cathode Cathode: 100% N2 Anode: 100% H2 2 5 30 30 30, 80, 100 60 Frequency, amplitude 100 kHz to 0.1 Hz Amplitude ±15 mV (±30 mV for drier cases) Potential (V) 0.4 Voltage cycling under H2/N2 (anode/cathode) is conducted to age the cathode electrode. Since only very low currents are present during the voltage cycling under H2/N2 (i.e., capacitive and H2 crossover currents), the applied cell voltage matches the cathode potential measured against the reversible hydrogen electrode (RHE) potential. The cell voltage in the applied H2/N2 cycling protocol was modulated with a square-wave profile, varying UPL and LPL as summarized in Figure 2 and Table 2. Dwell times were 30 s at UPL and 12 s at LPL, with transient potential less than 0.1 s (>4Vs−1). After a set amounts of square waves, a “short stop” is introduced, where the potential is set to 0.06 V for 110 s. H2/N2 flows of 0.1/0.1 SLPM (anode/cathode) were used during the cycling segments, with the cell operating at ambient pressure, 80 °C, and 100% RH. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 8/10 Figure 2. Exemplary AST profile that includes voltage cycling and mimicked short-stop. The highlighted parameters are variables in AST. Table 2. Experimental conditions varied in AST within an 8-cell multi-single cell. Cell MEA UPL (V) LPL (V) 1 1 0.9 0.7 2 2 0.9 0.7 3 3 0.9 0.7 4 2 0.8 0.7 5 1 0.9 0.7 6 2 0.9 0.75 7 2 0.8 0.75 8 2 0.75 0.7 References [1] M. Bahrami, J.-P. Martin, G. Maranzana, S. Pierfederici, M. Weber, and S. Didierjean, “Fuel cell management system: An approach to increase its durability,” Appl. Energy, vol. 306, p. 118070, Jan. 2022, doi: 10.1016/j.apenergy.2021.118070. [2] D. Hissel and M. C. Pera, “Diagnostic & health management of fuel cell systems: Issues and solutions,” Annu. Rev. Control, vol. 42, pp. 201–211, 2016, doi: 10.1016/j.arcontrol.2016.09.005. [3] W. Bi and T. F. Fuller, “Modeling of PEM fuel cell Pt/C catalyst degradation,” J. Power Sources, vol. 178, no. 1, pp. 188–196, Mar. 2008, doi: 10.1016/j.jpowsour.2007.12.007. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476231 A1112 Page 9/10 [4] W. Bi, G. E. Gray, and T. F. Fuller, “PEM Fuel Cell Pt∕C Dissolution and Deposition in Nafion Electrolyte,” Electrochem. Solid-State Lett., vol. 10, no. 5, p. B101, 2007, doi: 10.1149/1.2712796. [5] P. J. Ferreira et al., “Instability of Pt∕C Electrocatalysts in Proton Exchange Membrane Fuel Cells,” J. Electrochem. Soc., vol. 152, no. 11, p. A2256, 2005, doi: 10.1149/1.2050347. [6] Y. Shao-Horn, W. C. Sheng, S. Chen, P. J. Ferreira, E. F. Holby, and D. Morgan, “Instability of Supported Platinum Nanoparticles in Low-Temperature Fuel Cells,” Top. Catal., vol. 46, no. 3–4, pp. 285–305, Dec. 2007, doi: 10.1007/s11244-007-9000-0. [7] E. Colombo, A. Casalegno, L. Guetaz, and A. Baricci, “Revealing the critical role of low voltage excursions in enhancing PEM fuel cell catalyst degradation by automotive hydrogen/air potential cycling experiments,” Int. J. Hydrog. Energy, vol. 65, pp. 292– 307, May 2024, doi: 10.1016/j.ijhydene.2024.03.373. [8] C. A. Reiser et al., “A Reverse-Current Decay Mechanism for Fuel Cells,” Electrochem. Solid-State Lett., vol. 8, no. 6, p. A273, 2005, doi: 10.1149/1.1896466. [9] O. Kim et al., “Impact of fuel starvation–induced anode carbon corrosion in proton exchange membrane fuel cells on the structure of the membrane electrode assembly and exhaust gas emissions: A quantitative case study,” J. Power Sources, vol. 615, p. 235032, Sep. 2024, doi: 10.1016/j.jpowsour.2024.235032. [10] N. Macauley et al., “Carbon Corrosion in PEM Fuel Cells and the Development of Accelerated Stress Tests,” J. Electrochem. Soc., vol. 165, no. 6, pp. F3148–F3160, 2018, doi: 10.1149/2.0061806jes. [11] D. Chen et al., “Proton exchange membrane fuel cell stack consistency: Evaluation methods, influencing factors, membrane electrode assembly parameters and improvement measures,” Energy Convers. Manag., vol. 261, p. 115651, Jun. 2022, doi: 10.1016/j.enconman.2022.115651. [12] S. Zhou and R. Jervis, “A Review of Polymer Electrolyte Fuel Cells Fault Diagnosis: Progress and Perspectives,” Chemistry–Methods, vol. 4, no. 1, p. e202300030, Jan. 2024, doi: 10.1002/cmtd.202300030. [13] Y. Chatillon, C. Bonnet, and F. Lapicque, “Differential capacity plot as a tool for determination of electroactive surface area within a PEMFC stack,” J. Appl. Electrochem., vol. 43, no. 10, pp. 1017–1026, Oct. 2013, doi: 10.1007/s10800-0130585-7. [14] E. Brightman, G. Hinds, and R. O’Malley, “In situ measurement of active catalyst surface area in fuel cell stacks,” J. Power Sources, vol. 242, pp. 244–254, Nov. 2013, doi: 10.1016/j.jpowsour.2013.05.046. [15] K.-S. Lee et al., “Development of a galvanostatic analysis technique as an in-situ diagnostic tool for PEMFC single cells and stacks,” Int. J. Hydrog. Energy, vol. 37, no. 7, pp. 5891–5900, Apr. 2012, doi: 10.1016/j.ijhydene.2011.12.152. [16] D. A. Stevens and J. R. Dahn, “Electrochemical Characterization of the Active Surface in Carbon-Supported Platinum Electrocatalysts for PEM Fuel Cells,” J. Electrochem. Soc., vol. 150, no. 6, p. A770, 2003, doi: 10.1149/1.1573195. [17] P. Pei, H. Xu, X. Zeng, H. Zha, and M. Song, “Use of galvanostatic charge method as a membrane electrode assembly diagnostic tool in a fuel cell stack,” J. Power Sources, vol. 245, pp. 175–182, Jan. 2014, doi: 10.1016/j.jpowsour.2013.05.201. [18] P. Pei, Z. Wu, Y. Li, X. Jia, D. Chen, and S. Huang, “Improved methods to measure hydrogen crossover current in proton exchange membrane fuel cell,” Appl. Energy, vol. 215, pp. 338–347, Apr. 2018, doi: 10.1016/j.apenergy.2018.02.002. [19] P. Ren et al., “Micro-current excitation for efficient diagnosis of membrane electrode assemblies in fuel cell stacks: Error analysis and method optimization,” Energy Convers. Manag., vol. 258, p. 115489, Apr. 2022, doi: 10.1016/j.enconman.2022.115489.