Investigating the impact of a mixture of impurities in hydrogen on PEMFC performance

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 1/8 B1806 Investigating the impact of a mixture of impurities in hydrogen on PEMFC performance Ali Al-Sikab*, Jonathan Goh, Graham Smith National Physical Laboratory, Hampton Rd, Teddington, London/United Kingdom TW11 0LW; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract The standard of quality for hydrogen used in fuel cell electric vehicles is defined by ISO 14687:2019 (D), which sets the maximum threshold limit for several impurities in hydrogen. This is due to the sensitivity of fuel cells to impurities in the hydrogen supplied which can affect performance and durability, besides potentially causing irreversible damage to the fuel cell. Despite this, little is understood about the impact of an impurity mixture on the performance of PEMFCs. NPL has tested fuel cell performance at both single cell and stack level under static and simulated drive cycles for up to 300 hours. Performance was then compared between using pure hydrogen (99.995 %), carbon monoxide in hydrogen, and a complex mixture of impurities representative of those found in the ISO standard at their respective threshold limits. This has helped to assess how a fuel cell performs in a worst-case scenario of ISO 14687:2019 (D) compared to normal operation. The results showed a slow drop in performance after 2 hours when the impurities were supplied at the threshold concentrations provided in the standard, which stabilised after 1.5 hours. On the other hand, performance deteriorated immediately and significantly when the concentrations were increased to 4× those in the standard. This experiment also revealed that the impact of the mixture of impurities followed a similar profile to that of CO in hydrogen on PEMFCs. This investigation has provided insight into the significant impact impurity mixtures can have on fuel cell performance and these data can therefore help to inform future revisions of hydrogen quality standards. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 2/8 Introduction Conventional thermochemical methods of energy production combust all compounds found in the fuel leaving little to no sensitivity towards non-fuel compounds. However, when it comes to electrochemical technologies such as polymer electrolyte membrane fuel cells (PEMFCs) energy conversion occurs at a catalyst membrane layer which often operates at low temperatures (80 – 100 °C) and can be extremely sensitive to the purity of the fuel provided.1 Impurities in hydrogen have varying effects on a PEMFC. Some are diluents (nitrogen, helium, methane, argon) that may cause mass transport problems if the amount fraction of hydrogen in the gas supplied to the fuel cell stack becomes too low. 2 Other compounds may adsorb or absorb to the catalyst surface, modifying catalytic active sites or competitively occupying them, therefore impairing performance. Often this effect can be reversed once the impurity desorbs from the catalyst surface; examples of this include carbon monoxide and light-hydrocarbons.3,4 However, some compounds can also adsorb irreversibly to the catalyst surface, thus resulting in permanent functional loss of the catalyst. Examples include sulphur-based compounds such as hydrogen sulphide and non-volatile compounds such as heavy-hydrocarbons.1 Additionally, some impurities such as ammonia can cause PEMFC degradation through the displacement of critical ions in the polymers that exist in the membrane and ionomer, causing an increase in ionic resistivity. To control impurities in hydrogen which can be harmful to PEMFCs, an ISO 14687:2019 (D) hydrogen quality standard exists which designates the maximum permissible concentrations of various classes of compounds for a fuel cell electric vehicle.5 This standard acts as a preventive measure to protect fuel cells from damage due to impure hydrogen, but the conservativeness of the concentration thresholds directly contributes to the expense and complexity of providing hydrogen to end-users. Therefore, consistent evaluation and optimisation of this standard is essential to lowering the cost of hydrogen and helping to realise a sustainable hydrogen economy. 6 This standard is based on a significant body of existing literature; however previous studies have predominantly used binary or ternary gas mixtures. There has been little work done to assess the impact of multiple impurities on performance and importantly there has been no assessment of the impact of worst-case but compliant hydrogen impurity mixture on fuel cell performance. To investigate this, a complex mixture of impurities representing the ISO 14687:2019 (D) standard was fed to both single cells and short stacks with the performance compared to operation with pure hydrogen. 1. Scientific Approach This study used a complex mixture of impurities that were selected based on the ISO 14687:2019(D) standard and with consideration of the potential interaction of the impurities under pressure, Table 1. For example, ammonia, formaldehyde, and formic acid would react with each other in a mixture so the latter two were excluded despite studies showing an impact of formic acid on fuel cell performance.7 It was also necessary to define specific compounds to represent classes of impurity specified in the standard; for example, non-methane hydrocarbons, total sulphur and halogenated compounds. A mixture of 0.2 μmol mol-1 and 0.8 μmol mol-1 CO in H2 was also prepared for reference experiments. ISO 14687:2019 (D) Standard Nominal impurity mixture used in Experiments Impurity Concentration (μmol mol-1) Impurity Concentration (μmol mol-1) Carbon Monoxide 0.2 Carbon Monoxide 0.2 Sulphur Compounds* 0.004 Hydrogen Sulphide 0.004 Methane 100 Methane 100 Ammonia 0.1 Ammonia 0.1 Other Hydrocarbons** 2 Toluene 0.3 EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 3/8 Table 1: The maximum threshold for a range of chemical impurities that can be supplied from a hydrogen refuelling station to a fuel cell electric vehicle (illustrated) according to ISO 14687:2019 (D). *On a sulphur atom basis. **Nonmethane hydrocarbons on a carbon atom basis. ***On a halogenate ion basis. Testing was conducted on both a single cell and short-stack PEMFCs. The single cell is a standard design comprising of a 25 cm2 active area cell, with single channel graphitic flow field plates assembled in a counter-flow configuration. This standard design was necessary to give a general understanding of how the impurity mixture may affect the cell voltage of the simplest PEMFC. A short-stack PEMFC with identical catalyst coated membranes was also tested to better approximate a fuel cell in automotive systems. As well as galvanostatic measurements and standard electrochemical characterisations, dynamic load cycling (DLC) was performed over 300 hours to monitor the fuel cell under simulated driving cycles over a longer duration. This would demonstrate how the impurities may impact performance levels of fuel cells under close to real life conditions. 2. Experiments ASTM Type I deionised water (Elga Purelab, UK) was used for inlet gas humidification. Pure hydrogen was supplied by a hydrogen generator (Hogen S-Series 2, US) with a nominal purity of 99.9995 % and compressed, dry air compliant with ISO 8573-1 class 1.2.1 was used for the cathode supply. Nitrogen for purging was produced by a generator (Apex GasGen Nevis, UK) with nominal purity of 99.9995 %. The ISO impurity mixture was prepared in-house, contained within 10 L cylinders in the desired ratios and diluted in pure hydrogen to achieve the concentrations specified in Table 1. Single cell testing was performed in a fully automated fuel cell test station (G60, Greenlight Innovation, CA) coupled with a potentiostat/frequency response analyser (Gamry Reference 3000, US). Temperature monitoring was realised using T-type thermocouples embedded on the gas inlets/outlets, in the anode/cathode flow field plates and at the inlet/outlet of the miniature heat exchanger on the cell. Dilution of impurities for testing was carried out using hydrogen flow meter/controllers (Bronkhorst EL-FLOW® Select, max flow 300 mL min-1, NL). Further specifications for the single cell are given in Table 2. Short-stack testing was performed on a fully automated fuel cell test station (G100, Greenlight Innovation, CA). Further specifications for the short stack are given in Table 2. T-type thermocouples were embedded in the gas inlets as well as the inlet/outlet of the water jacket. A dilution system was built to facilitate gas mixing into the anode. The system comprised of two low flowrate hydrogen flow meter/controllers (Bronkhorst EL-FLOW® Select, max flow 300 mL min-1, NL) and one high flowrate flow meter/controller (Bronkhorst EL-FLOW® Select, max flow 30 L min-1, NL) that diluted the impurities in pure hydrogen before supplying to the G100. FlowView™ software (ver. 1.23) was used to dynamically control the flows in a master (pure hydrogen) and slave (impurities) configuration. Halogenated Compounds*** 0.05 Dichloromethane 0.025 Water 5 - - Oxygen 5 - - Helium 300 - - Nitrogen 300 - - Argon 300 - - Carbon Dioxide 2 - - Formaldehyde 0.2 - - Formic Acid 0.2 - - Single Cell Active Area 25 cm2 Nominal Current Density 1.0 A cm-2 EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 4/8 Table 2: Material specifications for single cell and short stack used for all experiments. A baseline galvanostatic measurement at the nominal current density was first carried out on the single cell with pure hydrogen. A dosage response experiment then followed this and the impurity mixture was supplied sequentially to a cell by blending it into pure hydrogen at three different concentrations, the nominal concentration (as specified in Table 1), 2× and 4x the nominal concentration. A 1-hour recovery period was introduced in between each concentration where pure gases were restored. Following this a dynamic loading cycle (DLC) experiment on two separate cells was done with pure hydrogen and 2× the nominal concentration of the impurity mixture, during simulated drive cycles. The DLC protocol was adapted from the EU harmonised single cell PEMFC testing protocol which is based on urban and extra-urban drive cycles.8 The DLC protocol was run for 100 hours or 305 cycles, followed by cell characterisation and a recovery shutdown period of 10 hours, after which the cell was characterised again to capture irreversible losses. These steps were repeated 3 times for a total of 300 hours of DLC testing. For the short-stack PEMFC, reference baseline data were collected with both pure hydrogen and CO in hydrogen at 0.2 μmol mol-1 and 0.8 μmol mol-1 over a period of 4 hours. Between each test, the short-stack was shut down during which the anode was purged with compressed, dry air for 1 hour to oxidise any absorbed CO. This baseline data reflected nominal performance of the cell, besides its tolerance and recovery when poisoned with a standard impurity such as CO. Following this, the nominal impurity mixture was supplied to the cell and then 4× the nominal concentration of the mixture. 3. Results Figure 1 shows the impact of increasing concentrations of the nominal impurity mixture on a single cell PEMFC versus the voltage decay with pure hydrogen, measured over 2 hours. The results showed that at the nominal concentration of the impurity mixture the cell degradation was significant (5 – 10 mV h-1) compared to pure hydrogen. After a period of recovery, the cell was then supplied with twice the nominal concentration of the impurity mixture and the cell degradation had an almost identical profile to that of the nominal concentration. The concentration of the impurity mixture was then increased to 4 times the nominal concentration, and this produced a markedly different cell voltage response. The voltage decay was immediate and consistent over the two hours, achieving a decay rate of almost 50 mV h-1. Membrane electrode area 3-layer MEA (HyPlat, South Africa) • Membrane: GoreselectTM 735.18 (18 μm) • Catalyst area: Pt/C – corrosion resistant carbon, load cycling resistant electrocatalyst at 0.1 mgPt cm-2 (anode) and 0.4 mgPt cm-2 (cathode) Sub-gasket Silicone gaskets (250 μm thickness) on either side of the CCM provided sealing. Gas Diffusion Layer Freudenberg H23C6 – c.a. 220 μm thickness with hydrophobic treatment and microporous layer Short Stack PEMFC Active Area 220 cm2 Nominal Current Density 0.7 A cm-2 Number of cells 8 Membrane electrode area 3-layer MEA from HyPlat (South Africa) • Membrane: GoreselectTM 735.18 (18 μm) • Catalyst area: Pt/C – corrosion resistant carbon, load cycling resistant electrocatalyst at 0.1 mgPt cm-2 (anode) and 0.4 mgPt cm-2 (cathode) Sub-gasket PET 80 ± 5 μm Gas diffusion layer Freudenberg H14Cx483 – c.a. 186 μm thickness with hydrophobic treatment and microporous layer EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 5/8 Figure 1: Cell voltage decay of single cell PEMFC when exposed to pure hydrogen and different concentrations of the impurity mixture. During the recovery period following exposure to increasing concentrations of the impurity mixture, a polarisation curve was measured after exposure to pure H2 for 1 hour, Figure 2. At 1 A cm-2, the polarisation curves showed higher degradation of cell performance at 2× the nominal concentration (15 mV) than at the nominal concentration (7 mV) which was not immediately obvious from the galvanostatic data. The cumulative effect of irreversible losses incurred with repeated exposure to the impurity mixture at the nominal concentration and 2× nominal concentration would certainly factor into the gradual losses noted in cell performance. However, the steep drop-off (108 mV) in cell performance after exposure to 4× nominal concentration of the impurity mixture alludes to a clear limiting effect on PEMFC performance at that concentration. Figure 2: Comparison of single cell PEMFC polarisation curve data when exposed to pure hydrogen and different concentrations of the impurity mixture. Table 3 summarises the effect observed for the single cell PEMFC with pure hydrogen and with the range of impurity mixture concentrations. At all concentrations of the impurity mixture tested, cell voltage decay was significant considering the short durations used for testing and although average voltage decay was similar at the nominal and 2× nominal concentrations, the effect of the latter appeared to manifest more slowly albeit marginally. While this minor difference can plausibly be attributed to experimental error, it still shows that impact of the impurity mixture at 2× above the ISO threshold is not significantly worse than at the ISO threshold limits. Nominal concentrations tested Test duration / h Voltage decay / mV h-1 Comments EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 6/8 Baseline (pure H2) 3 0 – 1 - Nominal Impurity Mixture 3 5 – 10 Slow decay stabilising after 1.5 h 2× Nominal Impurity Mixture 2 5 – 10 Slow decay stabilising after 2 h 4× Nominal Impurity Mixture 2 30 – 50 Immediate decay Table 3: Summary of experiments and performance losses on single cell PEMFC. The impact of the impurity mixture on average cell voltage of a PEMFC short stack was also measured. To establish a reference, the short stack was first poisoned with an equivalent concentration of CO in H2 at 0.2 μmol mol-1 to reflect the threshold in the ISO standard, and then also 0.8 μmol mol-1 (for comparison with the 4× nominal impurity mixture). Figure 3 shows the impact each of these experiments had over 4 hours on the short stack in comparison to when pure hydrogen was supplied. Almost identical voltage decay profiles were observed for 0.2 μmol mol-1 of CO and the nominal impurity mixture (1 mV) and for 0.8 μmol mol-1 and 4× nominal impurity mixture (15 mV) over 4 hours. This indicates a dominating effect of CO on the degradation rate of the short stack. It could be the case that the concentrations of the other compounds in the impurity mixture are simply too low to have an observable voltage decay effect within this test time frame or that the CO is adhering to the catalyst surface at such a fast rate it could be preventing the other impurities from having an observable impact on cell performance. The results of this work suggest that the threshold for the non-CO impurities in the ISO standard are perhaps too strict and that further work is required to better differentiate the impact of impurities on fuel cell performance both in terms of impact time scale and dominating impurities. Figure 3: Comparison of stack voltage loses after being exposed to pure hydrogen and equivalent concentrations of CO in hydrogen and the impurity mixture in hydrogen. Figure 4 shows the polarisation curves measured on the single cell PEMFC at the beginning of test (BoT) and after every 100 hours of DLC testing with pure hydrogen and 2× nominal impurity mixture concentration. Both cells showed the most significant voltage decay after the first 100 h, with 76 mV on the reference cell and almost 100 mV on the cell exposed to the impurities. The cells contrasted the most after 300 hours with the reference cell voltage at 1 A cm-2 only decreasing by 19 mV while the degradation of the impurity mixture cell maintained a steady decline of 70 mV after 200 hours and 300 hours. The evenly-spaced polarisation curves measured for the 2× nominal impurity mixture point to cumulative losses to the cell, in contrast to the narrowing gap between polarisation curves measured at nominal cell voltage with pure hydrogen. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 7/8 Figure 4: Polarisation curve data of single cell PEMFC after undergoing 300 h DLC procedure with, left: pure hydrogen and right: 2× nominal impurity mixture in hydrogen. The losses attributed to the impurity mixture in Figure 4 (right) were quite significant over 300 hours of DLC. These were measured immediately after completion of each 100-hour DLC step and captured the reversible and irreversible losses sustained. Following a 10-hour shutdown period, cell performance recovered to a degree and polarisation curves measured then primarily reflected irreversible losses from the previous 100 hours of DLC, Figure 5. Cell voltage appeared to recover more as the test progressed further, after 300 hours increasing from 0.371 V to 0.463 V at 1 A cm-2. This indicates that a significant proportion of the performance losses measured following the DLC test with the hydrogen impurity mixture used were reversible. Parallels can be drawn with the observation and interpretation from Figure 3, where CO poisoning appeared to have a dominating impact on cell performance over 4 hours of galvanostatic testing. Since CO adsorbs reversibly onto the Pt catalyst and can be oxidised with exposure to oxygen (air), it is feasible that the concentration of CO adhered to the catalyst surface was reduced during the 10-hour shutdown, resulting in the recovered performance even after 300 hours. Figure 5: Comparison of polarisation curves measured immediately after FC-DLC and irreversible loses (following 10 h shutdown) of single cell PEMFC when exposed to 2× nominal impurity mixture in hydrogen. References 1. Narasimha, M. et al. Impact of Designated Contaminants in Hydrogen on the Performance of a Proton Exchange Membrane Fuel Cell. Journal of Propulsion Technology (2023). EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476311 B1806 Page 8/8 2. Cheng, X. et al. A review of PEM hydrogen fuel cell contamination: Impacts, mechanisms, and mitigation. J Power Sources (2007). 3. Pei, P. et al. Effects of carbon monoxide on proton exchange membrane fuel cells and elimination techniques. Int J Hydrogen Energy (2024). 4. Kong, E. H. et al. Dynamics of direct hydrocarbon PEM fuel cells. Sci Rep (2024). 5. Bacquart, T. et al. Hydrogen fuel quality for transport – First sampling and analysis comparison in Europe on hydrogen refuelling station (70 MPa) according to ISO 14687 and EN 17124. Fuel Communications (2021). 6. Al-Sikab, A. & Smith, G. Measurement Challenges in the Hydrogen Sector. (2024). 7. Zhang, X. et al. Influence of Formic Acid Impurity on Proton Exchange Membrane Fuel Cell Performance. J Electrochem Soc (2010). 8. De Marco, G. et al EU Harmonised Test Protocols for PEMFC MEA Testing in Single Cell Configuration for Automotive Applications. Publications Office (2015). Keywords: EFCF2025, H2, LowTemp. Fuel Cells & Electrolysers, impurities, poisoning, ISO 14687:2019 (D), FC-DLC Remark: This work is licensed under Creative Commons Attribution 4.0 International