Greenhouse Gas Implications of Extending the Service Life of PEM Fuel Cells for Automotive Applications: A Life Cycle Assessment

A larger adoption of hydrogen fuel-cell electric vehicles (FCEVs) is typically included in the strategies to decarbonize the transportation sector. This inclusion is supported by life-cycle assessments (LCAs), which show the potential greenhouse gas (GHG) emission benefit of replacing internal combustion engine vehicles with their fuel cell counterpart. However, the literature review performed in this study shows that the effects of durability and performance losses of fuel cells on the life-cycle environmental impact of the vehicle have rarely been assessed. Most of the LCAs assume a constant fuel consumption (ranging from 0.58 to 1.15 kgH2/100 km) for the vehicles throughout their service life, which ranges in the assessments from 120,000 to 225,000 km. In this study, the effect of performance losses on the life-cycle GHG emissions of the vehicles was assessed based on laboratory experiments. Losses have the effect of increasing the life-cycle GHG emissions of the vehicle up to 13%. Moreover, this study attempted for the first time to investigate via laboratory analyses the GHG implications of replacing the hydrophobic polymer for the gas diffusion medium (GDM) of fuel cells to increase their durability. LCA showed that when the service life of the vehicle was fixed at 150,000 km, the GHG emission savings of using an FC with lower performance losses (i.e., FC coated with fluorinated ethylene propylene (FEP) instead of polytetrafluoroethylene (PTFE)) are negligible compared to the overall life-cycle impact of the vehicle. Both the GDM coating and the amount of hydrogen saved account for less than 2% of the GHG emissions arising during vehicle operation. On the other hand, when the service life of the vehicle depends on the operability of the fuel cell, the global warming potential per driven km of the FEP-based FCEV reduces by 7 to 32%. The range of results depends on several variables, such as the GHG emissions from hydrogen production and the initial fuel consumption of the vehicle. Higher GHG savings are expected from an FC vehicle with high consumption of hydrogen produced with fossil fuels. Based on the results, we recommend the inclusion of fuel-cell durability in future LCAs of FCEVs. We also advocate for more research on the real-life performance of fuel cells employing alternative materials.


Introduction
The deployment of fuel-cell electric vehicles (FCEVs) is typically included in public [1] and private [2] strategies to decarbonize the transportation sector. This inclusion is supported by life-cycle assessment (LCA) studies, which show the potential greenhouse gas (GHG) emission benefit of replacing internal combustion engine vehicles (ICEVs) with their fuel-cell counterpart. The promising environmental advantages of FCEVs compared to crucial mechanism affecting the durability of GDMs. In fact, the loss of surface carbon of the MPL induced by both a gaseous stream and water production has been proven more detrimental than chemical degradation caused by the acidic environment [13]. The related performance losses can be limited with a further coating employing binder species able to enhance the adhesion between MPL and GDL, and to consequently avoid the detachment of the surface carbon particles [19]. A precaution concerning operating conditions could be the running of the fuel cell at medium-low relative humidities (not higher than 60%), thus reducing the water content in the device. Moreover, working points far from the concentration polarization region (i.e., high current-density values) should be preferred to limit the overproduction of water from the redox process, and to achieve a higher overall efficiency [13].
Considering the effect of the GDM degradation on the durability performance of an FC is, therefore, fundamental for a reliable evaluation of the environmental impact of the vehicle. This work investigates the role of GDM degradation via LCA for the first time. First, an extensive review of the scientific literature is performed to uncover how fuel-cell durability and performance losses are considered in the LCA studies of FCEVs. Then, supported by the experimental results acquired in previous work by the authors [13], the role of these parameters in affecting the LCA of an FCEV is investigated. For the aforementioned reasons, only the experimental results from mechanically stressed samples were used for the assessment. To assess how the mechanical degradation of GDM can be affected by the choice of the hydrophobic agent, the GHG implications of replacing PTFE with fluorinated ethylene propylene (FEP) are also investigated. In previous experiments, FEP proved to be capable of reducing mass-transfer resistances at a high current density, thus enhancing water management and output power [19,20]. Moreover, thanks to a lower melting point, the use of FEP allows us to decrease the sintering temperature of the coated GDM from 350 • C (for PTFE-based GDMs) down to 260 • C [21]. Despite these clear environmental benefits, the overall GHG consequences of replacing PTFE with FEP in the GDM of an FCEV have not been assessed before. The results of the present work are intended to provide new information for policymakers, scientists, and car manufacturers on the GHG emissions of FCEVs and suggest potential ways to reduce them.

Literature Review
A systematic review of how the durability of PEMFC is treated in LCA studies was performed. The research was limited to studies published after 2015 that performed a life-cycle assessment of passenger cars. In particular, the parameters reviewed were: the lifetime of the FCEVs, durability of the FCs, type of GDL used in the FCs, performance losses of the FCs with time, replacement of the FCs during the service life of the vehicles, and fuel economy of the vehicles. The results of the review are presented in Section 3.1.

GDMs' Preparation
To investigate how the GDM affects the durability of FCs, two cells with GDMs coated with different polymers were tested in the lab. Two fluorinated polymers, namely PTFE and FEP, were used as hydrophobic agents for both GDLs and MPLs. In previous works, FEP was shown to improve the water management of a common PEMFC at a high current density, allowing the device to achieve higher output power densities [19,20]. For the experimental study, commercial carbon fiber GDLs (SAATI Group) were immersed in a 1 wt.% polymeric suspension for 20 min and then subjected to a 30 min thermal treatment. During the latter, a temperature of 350 • C was reached for PTFE and 260 • C for FEP [20]. Once treated, 40 µm MPLs were deposited onto GDLs using the bladecoating technique with a linear velocity of 0.0154 m s −1 , corresponding to a shear rate of about 350 s −1 . The MPLs were prepared from inks containing both carbon black (Vulcan XC72R from Cabot, Cabot Italiana S.p.A., Ravenna, Italy) and multi-wall carbon nanotubes (CNTs, NTX1, Nanothinx, Patras, Greece) as the carbonaceous phase, the above-mentioned hydrophobic fluorinated polymers, isopropyl alcohol as the dispersant, and water as the solvent. Inks were stirred at 8000 rpm for 10 min by the UltraTurrax T25 homogenizer (IKA GmbH, Staufen, Germany). After deposition, the resulting GDMs were treated for 30 min at the same temperature (350 • C and 260 • C for PTFE-based and for FEP-based samples, respectively) reached in the GDLs' hydrophobization stage to eliminate solvents and dispersants and consolidate the deposited carbon layers [20]. The other components of the tested fuel cell were fixed so that only the influence of the hydrophobic polymer agent on performance, durability, and sustainability was assessed. Specifically, graphitic bipolar plates (Ballard Fuel Cells), with a single serpentine in the anodic compartment and a triple one at the cathode, were adopted. A catalyst-coated membrane (Baltic Fuel Cells) was employed as the membrane electrode assembly with Nafion-212 electrolyte and Pt/C catalytic layers with a catalyst loading of 0.3 and 0.6 mg cm −2 at the anode and cathode, respectively.

PEMFCs' Performance Losses
The effect on durability of PTFE-and FEP-based GDMs was preliminarily evaluated in a previous work [13], where the fuel cell was kept at a constant current density (0.5 A cm −2 , which is a common working point for conventional devices [18]) for 1000 h at 60 • C and an 80-100% (anode-cathode) relative humidity. Polarization experiments were performed every 100 h feeding 0.25 NL min −1 of pure hydrogen and 1 NL min −1 of air to the anode and cathode, respectively. Such tests, however, are only slightly representative of the resistance against degradation during the service life of the device. Therefore, an ad-hoc accelerated stress test (AST) was performed to quicken the degradation of GDMs and avoid testing the fuel cell for thousands of hours. Among the possible degradation mechanisms that affect the performance of GDMs (i.e., chemical, mechanical, and thermal), the mechanical is the most detrimental, leading to the detachment of the surface carbon of the MPLs. For this reason, a mechanical AST was carried out by assembling the prepared GDMs in a dummy cell [13]. For the test, a Teflon film was used as a separator between the anode and cathode where the GDMs were assembled. To avoid distortion of the analysis and prevent potential electrochemical stress on the samples, hydrogen and catalysts were not used. Pressurized air was supplied continuously for 1000 h to each side of the cell, with a twofold flow rate compared to that employed for conventional tests (0.5 NL min −1 at the anode and 2 NL min −1 at the cathode).
After performing the AST, the stressed GDMs were assembled in a lab-scale fuel cell to obtain the polarization curves. The fuel cell was operated at 60 • C and 80-100% (anodecathode) relative humidity with hydrogen (0.25 NL min −1 ) and air (1 NL min −1 ) supplied to the anode and cathode, respectively. The polarization curves were then compared to those obtained with fresh (i.e., not stressed) components to estimate the performance loss induced by the AST.

From Lab Results to Real-World Performance
In this section, we attempt to link the fuel consumption of single-cell laboratory tests to real-world operations. Considering the hydrogen flow rate that is needed to power an FC over a certain mileage, the corresponding current can be computed from Equation (1): where . m H 2,cons [kg h −1 ] is the stoichiometric mass flow rate of hydrogen (i.e., the amount of fuel consumed), MW H 2 [kg mol −1 ] the molecular weight of hydrogen, and F [C mol −1 ] the Faraday's constant. The current value I [A] is needed to find the corresponding working voltage for the lab-scale system assembled with PTFE-and FEP-treated samples. The voltage loss induced by the accelerated stress test is then considered: to keep a constant power output (V·I) while the cell is aging, the current must be increased. Higher currents lead to higher hydrogen consumption levels. Figure 1 shows the degradation performances of the two cells tested in the lab: since FEP-based GDMs can limit the degradation rate (µV h −1 ) compared to PTFE, they are also supposed to reduce the hydrogen consumption.
Clean Technol. 2022, 4, FOR PEER REVIEW working voltage for the lab-scale system assembled with PTFE-and FEP-treated samples. The voltage loss induced by the accelerated stress test is then considered: to keep a constant power output (V•I) while the cell is aging, the current must be increased. Higher currents lead to higher hydrogen consumption levels. Figure 1 shows the degradation performances of the two cells tested in the lab: since FEP-based GDMs can limit the degradation rate (μV h −1 ) compared to PTFE, they are also supposed to reduce the hydrogen consumption. The experimental data were used to get the real-world degradation rates for the GDMs and estimate the total lifetime of the FC. Table 1 reports the voltage values obtained from the polarization curves reported in Figure 1 (i.e., the voltage at 0.5 A cm −2 from a single-lab-scale FC assembled with both fresh (Vi) and aged (V1000) materials), the current density needed to keep the power constant, the corresponding hydrogen flow rates, and the lifetime. The lifetime ( [h]) was calculated with Equation (2) as the ratio between the voltage difference and the degradation rate [μV h −1 ]. The values corresponding to the limit of the ohmic zone (0.45 V for PTFE and 0.52 V for FEP) were considered as final voltages (Vf in Figure 1). The method adopted here draws on the work by Chen et al. [22], who defined three load areas (low, transient, and high) for different operating conditions. Start-stop and idling conditions are allocated to the transient area, while the high load region is related to high-power operations, such as acceleration and high speed. Operating the FC beyond the high load area is considered too demanding for both materials and distribution equipment [22]. Based on the described approach, the reduction in hydrogen consumption (in terms of flow rate) is estimated to be around 3% when running a real lab-scale fuel cell The experimental data were used to get the real-world degradation rates for the GDMs and estimate the total lifetime of the FC. Table 1 reports the voltage values obtained from the polarization curves reported in Figure 1 (i.e., the voltage at 0.5 A cm −2 from a single-lab-scale FC assembled with both fresh (V i ) and aged (V 1000 ) materials), the current density needed to keep the power constant, the corresponding hydrogen flow rates, and the lifetime. The lifetime (t l [h]) was calculated with Equation (2) as the ratio between the voltage difference and the degradation rate DR [µV h −1 ]. The values corresponding to the limit of the ohmic zone (0.45 V for PTFE and 0.52 V for FEP) were considered as final voltages (V f in Figure 1). The method adopted here draws on the work by Chen et al. [22], who defined three load areas (low, transient, and high) for different operating conditions. Start-stop and idling conditions are allocated to the transient area, while the high load region is related to high-power operations, such as acceleration and high speed. Operating the FC beyond the high load area is considered too demanding for both materials and distribution equipment [22]. Based on the described approach, the reduction in hydrogen consumption (in terms of flow rate) is estimated to be around 3% when running a real lab-scale fuel cell assembled with FEP instead of PTFE after the accelerated stress test. Two cases were considered to link the single-cell data to the real-world performance of an FC stack composed of several cells: best-case (low-consumption) and worst-case (high-consumption). They are defined by the combination of the power density output and initial fuel consumption. Four references were used to define the extreme scenarios: the minimum and maximum power densities (0.64 and 0.91 W cm −2 ) came from Evangelisti et al. [23] and Benitez et al. [24], and the minimum and maximum initial fuel consumptions (0.58 and 1.15 kgH 2 /100 km) from Huss et al. [25] and Frank et al. [26], respectively. Both PTFE-and PEF-based vehicles are assumed to start from the same initial fuel consumption, and they increase it with time in response to the performance losses measured in the lab. Assuming a constant performance degradation, the baseline FC (PTFE-based) reaches the critical ohmic zone limit after 4109 h (Table 1). This performance is within the same order of magnitude of the 5000 h for a 2020 FCEV reported by Hill et al. [9]. On the other hand, the FEP-based FC reaches the critical ohmic zone limit after 6387 h, lasting approximately 50% longer than its PTFE-based counterpart. If the baseline vehicle has a typical service life of 150,000 km, corresponding to an average velocity of about 36.5 km h −1 (e.g., [23,27]), the FEP-based FC reaches its end of life after 233,180 km. Therefore, the initial hydrogen consumption of the single cell is calculated from the ratio between the molar flow rate needed to power the FC at the initial current density (computed from the reference power density) and the average velocity. From the initial hydrogen consumption, the number of cells needed (N c ) can be calculated as reported in Equation (3), considering that the minimum and maximum initial consumptions of the reference stack correspond to 0.58 [25] and 1.15 [26] kgH 2 /100 km, respectively: The end-of-life consumption (i.e., at 150,000 km for the PTFE-based vehicle and 233,180 for the FEP-based one) was calculated considering the increase in the hydrogen molar flow rate needed to keep the power constant (Equations (4) and (5)):  Table 2 summarizes the results for the low-and high-consumption cases in terms of the number of cells, the total area of GDMs needed, and fuel consumption. An electrode active area of 100 cm 2 was assumed to calculate the total area of the GDMs.

Global Warming Potential
This section of the article focuses on the methodology adopted for the LCA, in accordance with ISO standards 14040 [28] and 14044 [29].

Goal and Scope
The goals of this LCA are twofold: (1) to assess the GHG implications of including the performance losses of FCs in the LCA of an FCEV; (2) to assess the GHG implications of replacing PTFE with FEP in the GDMs of FCs for FCEVs. Two vehicles are compared across the entire life cycle (i.e., from cradle to grave): one where the GDM is coated with PTFE (i.e., reference vehicle), and one where the GDM is coated with FEP. A different GDM affects the emissions of a vehicle throughout its life cycle: from the production phase (emissions affected using different materials), through the use phase (emissions affected by the different hydrogen consumption levels when operating the vehicle), to the end-of-life phase (emissions delayed by the differential durability of the FCs). The system boundary of the study is illustrated in Figure 2. The functional unit of this study is one kilometer traveled by the vehicle (i.e., one vehicle-kilometer (vkm)). The LCA is performed with the software SimaPro 9.1, and ecoinvent 3.6 (cut-off system model) is used as the background database. Climate change is the only environmental impact category analyzed in the study, and the impact is quantified over a time horizon of 100 years. The characterization factors for the 100-year global warming potential (GWP100) are from the 5th assessment report of the Intergovernmental Panel on Climate Change [30]. All GHG emissions arising from the life cycle of the vehicle are assumed to occur at year zero. Therefore, the GHG emissions occurring at a delayed time (e.g., end-of-life emissions delayed thanks to the extension of the service life of the vehicle) are not discounted based on the shorter time they will warm the planet in the time horizon considered.

Global Warming Potential
This section of the article focuses on the methodology adopted for the LCA, in accordance with ISO standards 14040 [28] and 14044 [29].

Goal and Scope
The goals of this LCA are twofold: (1) to assess the GHG implications of including the performance losses of FCs in the LCA of an FCEV; (2) to assess the GHG implications of replacing PTFE with FEP in the GDMs of FCs for FCEVs. Two vehicles are compared across the entire life cycle (i.e., from cradle to grave): one where the GDM is coated with PTFE (i.e., reference vehicle), and one where the GDM is coated with FEP. A different GDM affects the emissions of a vehicle throughout its life cycle: from the production phase (emissions affected using different materials), through the use phase (emissions affected by the different hydrogen consumption levels when operating the vehicle), to the end-oflife phase (emissions delayed by the differential durability of the FCs). The system boundary of the study is illustrated in Figure 2. The functional unit of this study is one kilometer traveled by the vehicle (i.e., one vehicle-kilometer (vkm)). The LCA is performed with the software SimaPro 9.1, and ecoinvent 3.6 (cut-off system model) is used as the background database. Climate change is the only environmental impact category analyzed in the study, and the impact is quantified over a time horizon of 100 years. The characterization factors for the 100-year global warming potential (GWP100) are from the 5th assessment report of the Intergovernmental Panel on Climate Change [30]. All GHG emissions arising from the life cycle of the vehicle are assumed to occur at year zero. Therefore, the GHG emissions occurring at a delayed time (e.g., end-of-life emissions delayed thanks to the extension of the service life of the vehicle) are not discounted based on the shorter time they will warm the planet in the time horizon considered. Two scenarios may be considered, and to explore those, different sensitivity analyses were performed on crucial fuel-cell parameters. The scenarios differ in the end-of-life stage: in scenario A, both vehicles are assumed to be disposed of after 150,000 km in accordance with the typical lifespan of vehicles reported in the literature [23,24,31,32]. In scenario B, the reference vehicle is assumed to be disposed of after 150,000 km, while for the FEP-based vehicle this occurs after 233,180 km. The latter value may be calculated from our experimental results, considering the lifetime of the FEP-coated GDM results to be 1.55 times longer than the GDM coated with PTFE (i.e., 6387 h vs. 4109 h, as reported in Table 1). Scenario A aims to investigate how the performance losses of the FC affect the LCA of the vehicle, and how the different GDMs affect the production phase and the hydrogen consumption of the vehicle during the operational phase. On the other hand, scenario B aims to assess the potential implication of extending the service life of the vehicle by improving the durability of the FC using FEP. Regarding vehicle production, the manufacture of all the vehicle components but the FC is assumed to be independent of the polymer used in the GDMs. In our work, sensitivity analyses on fuel-cell Two scenarios may be considered, and to explore those, different sensitivity analyses were performed on crucial fuel-cell parameters. The scenarios differ in the end-of-life stage: in scenario A, both vehicles are assumed to be disposed of after 150,000 km in accordance with the typical lifespan of vehicles reported in the literature [23,24,31,32]. In scenario B, the reference vehicle is assumed to be disposed of after 150,000 km, while for the FEP-based vehicle this occurs after 233,180 km. The latter value may be calculated from our experimental results, considering the lifetime of the FEP-coated GDM results to be 1.55 times longer than the GDM coated with PTFE (i.e., 6387 h vs. 4109 h, as reported in Table 1). Scenario A aims to investigate how the performance losses of the FC affect the LCA of the vehicle, and how the different GDMs affect the production phase and the hydrogen consumption of the vehicle during the operational phase. On the other hand, scenario B aims to assess the potential implication of extending the service life of the vehicle by improving the durability of the FC using FEP. Regarding vehicle production, the manufacture of all the vehicle components but the FC is assumed to be independent of the polymer used in the GDMs. In our work, sensitivity analyses on fuel-cell parameters, the initial fuel consumption, and hydrogen production technologies were performed with the aim of analyzing the variability of results due to the variation of input parameters. Regarding cell parameters, two different power densities (i.e., 0.64 W cm −2 and 0.91 W cm −2 ) were investigated; these values were selected to assess the best and worst cases, as explained in Section 2.4. The same approach was adopted for the initial fuel consumption (0.58 and 1.15 kgH 2 /100 km), using the lowest and highest values found in the literature (see Table 3). The four combinations of power densities and initial fuel consumptions examined correspond to four different numbers of cells, and therefore, four different areas of the GDMs (see Table 2 for the values considered). Three hydrogen production pathways are considered for the operation phase of the vehicles: (i) electrolysis powered by the average European electricity mix; (ii) electrolysis powered by renewable electricity; (iii) steam methane reforming. The potential impacts are assessed for all the possible combinations of power densities, initial fuel consumption, and hydrogen production pathways.

Life-Cycle Inventory: Vehicle Product System
The only difference in the production stage of the compared vehicles is the fabrication of the GDM.

Gas Diffusion Medium
The production of the GDM can be divided into three main steps: (i) synthesis of the following monomers: tetrafluoroethylene (TFE) used for PTFE production, and TFE and hexafluoropropylene (HFP) for FEP production; (ii) polymerization; (iii) preparation of the GDM: i.e., coating of the carbon cloth with the selected polymer, and thermal treatment. In our work, the synthesis of the TFE monomer available on ecoinvent 3.6 was used for the assessment, while the inventory for HFP production was based on the information reported by Rodriguez et al. [33]. In their report, HFP is produced via combined pyrolysis of TFE and HFC-23. Input and output flows of the HFP manufacturing process are specified in the Supplementary Material (see Table S2). Mass-based allocation was performed to divide the energy inputs between the two co-produced monomers (TFE and HFP). The number of co-products was based on the annual production of the factory and on additional references [33,34]. Where precise information about the FEP manufacturing process was not available (i.e., emitted GHGs), conservative assumptions were made (see Section S2.1). The efficiency of the PTFE and FEP polymerization processes (see Tables S1 and S2) was based on Rodriguez et al. [33]. The consumption of water, initiators, and surfactants was neglected since they were assumed to be the same for the two polymers. As for the coating process, the immersion of the carbon fibers into a polymeric suspension and the subsequent thermal treatment were modeled. The amount of polymer on the carbon cloth material (34 wt.%) was calculated by weighting the GDL before and after the coating treatment. The electrical energy necessary for the thermal treatment was calculated from the different temperatures required in the lab to treat the two polymers: i.e., 350 • C for PTFE and 260 • C for FEP (see the Supplementary Information for the calculations performed to obtain the energy consumption).

Vehicle Manufacture and End of Life
The GWP impact of these phases was based on results found in the literature. The average impacts reported in four studies [23,24,31,35] were considered, and sensitivity analysis using the different values reported in the four studies was also performed. According to the literature, the overall impact of manufacturing and disposing of a passenger car ranges from 1.20 t CO 2 eq. [31] to 15.5 t CO 2 eq. [24], with an average value of 14.5 t CO 2 eq. Since the GHG emissions from vehicle manufacturing were sourced from the literature, the same impact was considered for the two compared vehicles (i.e., only the GDM impact on the operational phase of the vehicle was assessed).

Life-Cycle Inventory: Use Phase
In the use phase, only the impacts of hydrogen consumption are considered. The amount of hydrogen consumed was experimentally calculated as explained in Section 2.4. The fuel consumptions shown in Table 2 were used for the assessment. Hydrogen production via different pathways was modeled based on Bieker et al. [36]. In their study, the electricity consumption for hydrogen production via electrolysis (including compression and dispensing) amounted to 1.68 MJ per MJ of hydrogen [36]. The life-cycle carbon intensity of EU electricity was set by Bieker et al. as 199 g of CO 2 eq. kWh −1 [36], assuming that the electricity mix during the service life of the vehicle will progressively decarbonize. The vehicle was assumed to start operating in 2021, and the Stated Policy Scenario of the IEA's World Energy Outlook was used for the estimate [37]. The carbon intensity of the EU renewable energy was estimated to be 23 g of CO 2 eq. kWh −1 , considering a mix composed of 33% of solar energy and 67% of wind energy [36]. Finally, the carbon intensity of hydrogen production via steam reforming was assumed to be 113 g of CO 2 eq. MJ −1 of hydrogen [36]. A lower heating value for hydrogen of 120 MJ kgH 2 −1 and an electrolyzer efficiency of 57 kWh kgH 2 −1 were assumed for the study [36]. Maintenance and tire emissions were neglected since previous studies showed their role in the life-cycle impacts to be trivial [9].

Literature Review
Results of the literature review are summarized in Table 3, where the primary assumptions made in the different studies are reported. If the reviewed studies made a sensitivity analysis of the parameters here investigated, for readability purposes, the variation of the values is presented in this section but not in the table. Since 2015, 19 studies were found in the literature that assessed the life-cycle environmental performances of FCEVs. All studies reported the fuel economy of the FCEV, with a range from 0.58 [25] to 1.15 kgH 2 /100 km [26]. These extreme values were the ones used in the present analysis to calculate the impacts for the low-and high-consumption cases. The minimum consumption value was obtained from version 5 of the European Commission JEC Well-to-Wheels analysis. Low consumption was ascribed to the lower driving resistance of the vehicle segment compared to the FCEVs available on the market [25]. The other end of the fuel economy spectrum was calculated from the fuel efficiency of FC light-duty vehicles available in the GREET model (1.38 MJ km −1 ) [26,38]. Most assessments used secondary data for the average fuel consumption of the vehicles. Often, the initial fuel economy provided by the car manufacturer was used for the analysis; therefore, the fuel consumption considered was lower than that experienced in real-life conditions. For instance, the FCEVs monitored in the framework of the European project H2ME had an average fuel consumption of 1.11 kgH 2 /100 km in Germany and 1.26 kgH 2 /100 km in Denmark [39]. These values are considerably higher than the average consumption assumed in the reviewed studies: 0.92 kgH 2 /100 km. Real-life consumption results are also higher than those calculated from standardized driving cycles: for instance, Hwang et al. obtained a consumption of 0.96 kgH 2 /100 km when considering the New European Driving Cycle (EUDC) [7]. Overall, the fuel economy seems to depend more on the driving conditions than the FC system [32]. In their study, Simons and Bauer forecasted an overall improvement in the fuel efficiency of only 0.2% in 10 years [32].
Although every study reported the amount of hydrogen consumed per 100 km, less information was provided regarding the durability of the FC and its effect on the fuel economy of the vehicle. The most covered parameter related to durability was the lifetime of the vehicle, which ranged from a maximum of 190,000 km [40,41], retrieved from the 2017 budget of the Fuel Cell Technology Office of the US Department of Energy [42], to a minimum of 120,000 km. The latter value was provided by Ahmadi et al. [10], who accounted for the performance losses of the FC due to repeated stops and starts. A degradation rate of 2.65 × 10 −5 Vh −1 (Chen et al. [22]) from an initial cell voltage of 0.78 V was considered, and the FC was assumed to be usable for 12,000 h or until the voltage dropped below 0.5 V (i.e., 10,600 h of operation with the aforementioned degradation rate). In the other studies reviewed, the service life of the FC either was not mentioned or was assumed to be the same as the vehicle. The exceptions were the studies from Hill et al. [9] and Kannangara et al. [43]. Hill et al. assumed the service life of the FC in 2020 to be 5000 h, corresponding to a lifetime of the vehicle of more than 225,000 km [9]. The value for the durability of the FC was validated by stakeholder experts, and future projections were also developed to account for changes in technology [9]. According to the projections, the service life of FCs is expected to reach 10,000 h in 2050. This value for 2050 is lower than the one Ahmadi et al. assumed for 2020. Kannangara et al. assumed the service life of the FC to be 150,000 km, and a lifetime vehicle mileage of 150,000-200,000 km. In the case where the operating mileage is above 150,000 km, the FC stack components are assumed to be replaced [43]. Sensitivity analyses including the replacement of the fuel cell and/or different mileages for the vehicle were carried out in other studies, too (e.g., [9,31,44]). Several studies included the environmental impacts of the maintenance of the vehicle, but none provided information regarding the maintenance of the FC. The main reason for this deficiency is the uncertainty related to real-world FC performance [24]. Nevertheless, the capacity of the FC is not expected to deteriorate by more than 15% in 10 years [32].
As mentioned earlier, Ahmadi et al. provided the only study to assess the GHG impact of the performance loss of the FCEV [10]. From an initial consumption of 1.1 kgH 2 /100 km, based on the 2018 Toyota Mirai performance specifications, they found a reduction of the fuel economy (averaged over the 120,000 km lifetime of the vehicle) of 10-15 km per kg of hydrogen, depending on the driving pattern. As a consequence, the lifetime GHG emissions of the vehicle increased on average by 3200 kg CO 2 eq.
As for the fuel-cell components, most LCA studies used secondary data for the inventory of the GDM. The most referenced papers are the ones from Simons and Bauer [32], Miotti et al. [31], and Evangelisti et al. [23]. Simons and Bauer [32] modeled a GDM composed of a macroporous layer of woven carbon cloth hydrophobically treated with PTFE (15% loading), and a microporous layer of carbon (or graphite) particles mixed with PTFE binder. Similarly, Miotti et al. [31] considered a GDL composed of a 0.28-mm non-woven carbon substrate macroporous layer with a 0.04-mm microporous layer of PTFE and carbon black. The thicknesses are expected to reduce in the future, but the authors also consider the use of PTFE in 2030 and 2050. Finally, Evangelisti et al. [23] considered, for the baseline scenario, a GDM composed of a carbon cloth material coated with 10 wt.% PTFE and 5 wt.% of carbon black for the microporous layer. A complete LCA of the fuel cell system was also provided by Usai et al. [45]. In their case, the GDM results were lighter and thinner (0.21 mm with a density of 0.45 g cm −3 ) than those in the aforementioned studies. Yet, the GDL (produced from polyacrylonitrile) was again assumed to be treated with PTFE (15 wt.%) to enhance its water management.

LCA
In this section, first, the results for scenario A (i.e., both vehicles disposed of after 150,000 km) are presented, followed by the results for scenario B (i.e., a vehicle with PTFEbased GDMs disposed of after 150,000 km and a vehicle with FEP-based GDMs disposed of after 233,180 km).

Scenario A
Performance losses of the fuel cell affect the hydrogen consumption of the vehicle throughout its service life, and therefore, its carbon footprint. Compared to a case where hydrogen consumption remained constant until the end-of-life (i.e., the typical assumption adopted in the LCAs of FCEVs found in the literature), the reference vehicle consumed from 183 to 362 kg of additional hydrogen. This consumption corresponds to an additional emission of 0.2-4.9 t CO 2 eq. If compared to the life cycle of the vehicle, including production and end-of-life, this equals an increase of 1-13%. The variation in the increased emissions depends on the initial fuel consumption and the pathway for hydrogen production. The change in the GHG performance is less marked in the case of hydrogen produced via electrolysis powered by renewable energy (i.e., less than 3 g CO 2 eq. per km). This is due to the lower GHG impact of hydrogen production, which reduces the impact of using more hydrogen. In the case of electrolysis powered by the EU electricity mix, the additional emissions amount to 14-27 g of CO 2 eq. km −1 (up to 4 t CO 2 eq. for the whole service life), depending on the initial fuel consumption considered. The highest impact variation is noticed when hydrogen is produced via steam methane reforming. In this case, performance losses lead to an increase in the emissions of up to 4.9 t CO 2 eq. over the entire service life of the vehicle. The higher savings in this case are due to the high carbon intensity of hydrogen produced from fossil fuels. Table 3. Literature review results. * Plug-in hybrid fuel cell battery vehicle. ** The value has been recently updated to 0.91 [46].

Reference
Year Focusing on GDM manufacturing, the production of FEP-based GDMs results in a GWP higher than PTFE-based GDMs. This is mainly because PTFE is a TFE homopolymer, while FEP also contains HFP. The latter is produced via pyrolysis of HFC-23 and TFE, and therefore, requires an additional process with additional energy consumption. Nevertheless, the GWP impact variation in the manufacture of FEP and PTFE is almost negligible if compared to the impacts of the use phase of the vehicle. When the total GDL area of the FC is considered, the impact variation is in the order of 0.001 g CO 2 eq. km −1 . On the other hand, substituting PTFE-with FEP-based GDMs allows a saving of from 0.1 to 3.1 g CO 2 eq. per km during the use phase of the vehicle, corresponding to a GWP reduction of from 22 to 459 kg CO 2 eq. over the whole lifetime mileage (i.e., 150,000 km). As for the previous comparison with a hypothetical vehicle without performance losses, the higher GHG savings when using FEP (i.e., 3.1 g CO 2 eq. km −1 , corresponding to an overall saving of 459 kg CO 2 eq.) are obtained when hydrogen is produced via steam methane reforming. Conversely, the lowest savings from the use of FEP compared to the PTFE-based reference vehicle are obtained when hydrogen is assumed to be produced via renewable electrolysis (i.e., less than 0.3 g CO 2 eq. per km). In summary, substituting PTFE-with FEP-based GDMs allows the GHG emissions to be reduced by 10% due to performance losses. Figure 3 shows the GWP reduction when using PEF-coated GDMs over the whole vehicle life cycle. The difference over the entire life cycle almost coincides with that of the use phase due to the negligibility of the impact of GDM production. The GWP reduction corresponds to a relative decrease of approximately 1.6% of the use-phase impacts.
Clean Technol. 2022, 4, FOR PEER REVIEW Figure 3. Global warming potential reduction per km (left y-axis) and over the entire life cy y-axis) when substituting PTFE-with FEP-based GDMs in an FCEV. The reduction is calc the difference between the GWP for the PTFE-based FCEV and the GWP for the FEP-bas Floating bars represent the range of values obtained for each production pathway by va initial fuel consumption and the cell power densities.

Scenario B
In this scenario, an increase in the lifetime mileage of the vehicle with FE GDMs is considered based on the assumption that a slower degradation of the to a longer service life of the entire vehicle. For this reason, the impact of the system (i.e., manufacturing and end of life) is distributed over a longer lifeti 233,180 km) for the FEP vehicle with respect to the reference vehicle (i.e., 150,000 resulting normalized impact of the vehicle system is 96.7 g CO2 eq. km −1 for th based vehicle and 62.2 g CO2 eq. km −1 for the FEP-based vehicle. The impacts of the use phase per driven km are higher in the case of FEP PTFE. Considering an initial fuel economy of 0.58 kgH2/100 km, the impact is h 0.6-6.8 g CO2 eq. km −1 depending on the hydrogen production pathway. When a fuel consumption of 1.15 kgH2/100 km is considered, the difference between the of FEP and PTFE increases: from 1.3 kg CO2 eq. km −1 in the case of hydrogen prod electrolysis with renewable electricity, to 13.4 kg CO2 eq. km −1 in the case of h from steam methane reforming. The FEP-based vehicle has a higher GWP impac during the use phase due to the high fuel consumption in the last kilometers of it life (approximately 80,000 km longer than the reference vehicle). As a result, the fuel consumption during the entire life of the vehicle is higher.
Despite the higher impact of the use phase, the overall life-cycle GWP impa FEP vehicle is lower than that for the PTFE vehicle for all the analyzed fuel pro pathways and cell parameters. The benefit of extending the lifetime mileage of the in fact, overcomes the drawbacks of the higher fuel consumption in the last kilom the service life. Specifically, the introduction of FEP allows the impact to be red an amount ranging from 15 to 36 g of CO2 eq. per km, depending on FC featu hydrogen production pathways. The greatest benefit can be noted in the case of h produced via electrolysis powered by renewable sources (i.e., a GWP impact dec

Scenario B
In this scenario, an increase in the lifetime mileage of the vehicle with FEP-based GDMs is considered based on the assumption that a slower degradation of the FC leads to a longer service life of the entire vehicle. For this reason, the impact of the vehicle system (i.e., manufacturing and end of life) is distributed over a longer lifetime (i.e., 233,180 km) for the FEP vehicle with respect to the reference vehicle (i.e., 150,000 km). The resulting normalized impact of the vehicle system is 96.7 g CO 2 eq. km −1 for the PTFE-based vehicle and 62.2 g CO 2 eq. km −1 for the FEP-based vehicle.
The impacts of the use phase per driven km are higher in the case of FEP than for PTFE. Considering an initial fuel economy of 0.58 kgH 2 /100 km, the impact is higher by 0.6-6.8 g CO 2 eq. km −1 depending on the hydrogen production pathway. When an initial fuel consumption of 1.15 kgH 2 /100 km is considered, the difference between the impacts of FEP and PTFE increases: from 1.3 kg CO 2 eq. km −1 in the case of hydrogen produced via electrolysis with renewable electricity, to 13.4 kg CO 2 eq. km −1 in the case of hydrogen from steam methane reforming. The FEP-based vehicle has a higher GWP impact per km during the use phase due to the high fuel consumption in the last kilometers of its service life (approximately 80,000 km longer than the reference vehicle). As a result, the average fuel consumption during the entire life of the vehicle is higher.
Despite the higher impact of the use phase, the overall life-cycle GWP impact of the FEP vehicle is lower than that for the PTFE vehicle for all the analyzed fuel production pathways and cell parameters. The benefit of extending the lifetime mileage of the vehicle, in fact, overcomes the drawbacks of the higher fuel consumption in the last kilometers of the service life. Specifically, the introduction of FEP allows the impact to be reduced for an amount ranging from 15 to 36 g of CO 2 eq. per km, depending on FC features and hydrogen production pathways. The greatest benefit can be noted in the case of hydrogen produced via electrolysis powered by renewable sources (i.e., a GWP impact decrease by 27-37 g CO 2 eq. km −1 ), as the GHG emissions near the end of life of the FEP vehicle are lower in this case. When hydrogen is sourced from electrolysis powered by the EU electricity mix, the GWP difference between an FEP-and PTFE-based vehicle ranges from 17 to 33 g of CO 2 eq. per km. Finally, in the case of hydrogen from steam methane reforming, the GWP difference between the vehicles ranges from 15 to 31 g CO 2 eq. per km. Figure 4 shows the percentage of GWP reduction that would occur if PTFE-based GDMs were substituted with FEP-based ones. The GHG impact decreases from a minimum of 7.4% (in the case of hydrogen produced via steam methane reforming) to a maximum of 32% (in the case of hydrogen from electrolysis powered by renewable electricity mix).
Clean Technol. 2022, 4, FOR PEER REVIEW km. Figure 4 shows the percentage of GWP reduction that would occur if PTF GDMs were substituted with FEP-based ones. The GHG impact decreases minimum of 7.4% (in the case of hydrogen produced via steam methane reformi maximum of 32% (in the case of hydrogen from electrolysis powered by ren electricity mix). . Global warming potential reduction (%) resulting from the substitution of PTF GDMs with FEP-based ones. The GWP reduction is calculated as the percentage decre respect to the GWP of PTFE-based GDMs (assumed as a reference case). Floating bars illus range of reduction for each hydrogen production pathway, obtained by varying the in consumption and considering different impacts for vehicle production and disposal phase

Conclusions
The goal of this study was to investigate the role of fuel-cell performance l the life-cycle GHG emissions of a fuel-cell electric vehicle. Our review of the existi studies showed that performance losses have rarely been included in the assessme most plausible reason for this deficiency is the scarcity of real-world data on b durability of the cells and fuel economy of FC vehicles. As a consequence, t majority of the studies available in the scientific literature assume a consta consumption (ranging from 0.58 to 1.15 kgH2/100 km) for the vehicles througho service life, which ranges in the studies from 120,000 to 225,000 km. Neverthe studies seem to agree that fuel consumption and FC durability are fund parameters in the life cycle impacts of a vehicle. In fact, fuel consumption is typic main source of environmental impacts, and the performance losses of the potentially a sufficient reason to replace the vehicle. Therefore, we investigated how extending the durability of the fuel cell by re PTFE with FEP in the GDM would affect the greenhouse gas performance of the Firstly, experimental results on the GDM durability were converted into parame the life-cycle assessment of FC vehicles. Then, the life-cycle GHG impacts of the d Figure 4. Global warming potential reduction (%) resulting from the substitution of PTFE-based GDMs with FEP-based ones. The GWP reduction is calculated as the percentage decrease with respect to the GWP of PTFE-based GDMs (assumed as a reference case). Floating bars illustrate the range of reduction for each hydrogen production pathway, obtained by varying the initial fuel consumption and considering different impacts for vehicle production and disposal phases.

Conclusions
The goal of this study was to investigate the role of fuel-cell performance losses in the life-cycle GHG emissions of a fuel-cell electric vehicle. Our review of the existing LCA studies showed that performance losses have rarely been included in the assessments. The most plausible reason for this deficiency is the scarcity of real-world data on both the durability of the cells and fuel economy of FC vehicles. As a consequence, the vast majority of the studies available in the scientific literature assume a constant fuel consumption (ranging from 0.58 to 1.15 kgH 2 /100 km) for the vehicles throughout their service life, which ranges in the studies from 120,000 to 225,000 km. Nevertheless, all studies seem to agree that fuel consumption and FC durability are fundamental parameters in the life cycle impacts of a vehicle. In fact, fuel consumption is typically the main source of environmental impacts, and the performance losses of the FC are potentially a sufficient reason to replace the vehicle. Therefore, we investigated how extending the durability of the fuel cell by replacing PTFE with FEP in the GDM would affect the greenhouse gas performance of the vehicle. Firstly, experimental results on the GDM durability were converted into parameters for the life-cycle assessment of FC vehicles. Then, the life-cycle GHG impacts of the different vehicles were assessed based on inventories and parameters found in the scientific literature and LCA databases. Performance losses resulted in increasing the life-cycle GHG emissions of the fuel-cell electric vehicle by up to 13%. The GHG emission savings of using an FC with lower performance losses (i.e., with FEP) were negligible compared to the overall life-cycle impact of the vehicle when the service life of the vehicle was fixed at 150,000 km. Both the GDM component and amount of hydrogen saved accounted for less than 2% of the GHG emissions that arise during the vehicle operation alone. On the other hand, when the service life of the vehicle depends on the operability of the fuel cell, the global warming potential impact per driven km of the FEP-based FCEV reduces by 7 to 32%. Given the significant role that these parameters can play on the overall emissions, we advocate for the inclusion of fuel-cell durability and performance losses in future LCAs of fuel-cell electric vehicles. The range of the results depends on several variables, such as the greenhouse gas emissions from hydrogen production and the initial fuel consumption of the vehicle. Higher GHG emission savings are expected from an FC vehicle with high consumption of hydrogen produced with fossil fuels.
Our study is a first attempt to investigate the environmental implications of using an alternative coating for the GDM of fuel cells to increase their durability, with the results indicating a potential carbon footprint reduction of the fuel-cell electric vehicle. Though extensive sensitivity analysis was performed on the results, further research is necessary to improve the validity of the assessment. In particular, the durability parameters used for the LCA were based on single-cell experiments carried out in a laboratory; primary data on the durability performance and the fuel economy of the alternative fuel cell in a real vehicle are, therefore, needed. Moreover, LCA results show a real environmental benefit of using FEP only when the service life of the innovative vehicle is assumed to be longer, and the impacts are normalized per km driven. However, there is no consensus in the LCA community on the correct methodological approach to compare products with different lifetimes. Moreover, the results do not include the impacts of the maintenance of the vehicle, which could reduce the environmental benefit of extending its service life (i.e., the vehicle could require more maintenance work after 150,000 km). At the same time, the delayed emissions (e.g., the emissions from the disposal of the vehicle) were not discounted. A more thorough analysis of the potential implications of the end-of-life stage of the vehicle (e.g., recyclability of the fuel cell) is, therefore, needed.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cleantechnol4010009/s1, Section S1: Acronyms, Section S2: Life Cycle Inventory of gas diffusion medium (GDM), Section S3: Supplementary results, Section S4: Sensitivity analysis of the effect of the fuel cell degradation rate on GWP, Table S1: Inventory of PTFE-based gas diffusion medium, Table S2: Inventory of FEP-based gas diffusion medium.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.