Overcoming the Electrode Challenges of High-Temperature Proton Exchange Membrane Fuel Cells

Abstract

Proton exchange membrane fuel cells (PEMFCs) are becoming a major part of a greener and more sustainable future. However, the costs of high-purity hydrogen and noble metal catalysts alongside the complexity of the PEMFC system severely hamper their commercialization. Operating PEMFCs at high temperatures (HT-PEMFCs, above 120 °C) brings several advantages, such as increased tolerance to contaminants, more affordable catalysts, and operations without liquid water, hence considerably simplifying the system. While recent progresses in proton exchange membranes for HT-PEMFCs have made this technology more viable, the HT-PEMFC viscous acid electrolyte lowers the active site utilization by unevenly diffusing into the catalyst layer while it acutely poisons the catalytic sites. In recent years, the synthesis of platinum group metal (PGM) and PGM-free catalysts with higher acid tolerance and phosphate-promoted oxygen reduction reaction, in conjunction with the design of catalyst layers with improved acid distribution and more triple-phase boundaries, has provided great opportunities for more efficient HT-PEMFCs. The progress in these two interconnected fields is reviewed here, with recommendations for the most promising routes worthy of further investigation. Using these approaches, the performance and durability of HT-PEMFCs will be significantly improved.

1 Introduction

With increasing global warming concerns and on-going depletion of fossil fuels, the need for clean and renewable energy sources is greater than ever [1]. However, their storage and redistribution remain a challenge. As a result, the development of non-polluting electrochemical energy conversion technologies, such as lithium-ion batteries and hydrogen fuel cells, has intensified in recent years [2, 3]. Hydrogen fuel cells are broadly acknowledged to be better suited to heavy-duty and high-power applications, whereas lithium-ion batteries are more suitable for light-duty and lower-power applications [4]. Based on the ions (anions/cations) and/or nature (liquid/solid) of the electrolyte, fuel cells can be divided into alkaline polymer electrolyte fuel cells (APEFCs), proton exchange membrane fuel cells (PEMFCs), phosphoric acid fuel cells (PAFCs), molten carbonate fuel cells (MCFCs), and solid oxide fuel cells (SOFCs). Low-temperature (LT)-PEMFCs (< 120 °C, optimum ~ 80 °C) use Nafion as the solid electrolyte membrane and water as the proton carrier, while high-temperature (HT)-PEMFCs (120–200 °C) typically use phosphoric acid (PA) as the proton conductor. APEFCs with different cationic groups to conduct OH⁻ groups are attracting attention due to their high power density and potential to use platinum-group metal (PGM)-free catalysts [5]. MCFCs, operating at approximately 600–700 °C, use a mixture of a molten carbonate salt suspended in a chemically inert porous ceramic as an electrolyte. SOFCs, solid-state electrochemical devices that convert chemical energy into electrical energy, are operated at even higher temperatures (700–1 000 °C) [6]. Compared to other types of fuel cells, PEMFCs have distinct advantages, such as (i) easy start-up without gas preheating, (ii) broad selectivity of construction materials, including elastomers, for sealing, and (iii) slow degradation processes due to the low temperatures [7, 8].

Following decades of development, LT-PEMFCs are in the early stages of commercialization, with Japan, the UK, the US, China, and the Republic of Korea aiming to deploy large volumes of fuel cell electric vehicles within the next decade, with 200,000 by 2025 and 1 million by 2030 [9, 10]. Despite these recent developments, major roadblocks still lie ahead. As LT-PEMFCs operate optimally at approximately 80 °C with water as a mixture of liquid and vapor, their heat and water balance requires a complex and costly management system [3, 11, 12]. In addition, LT-PEMFCs are sensitive to contaminants and therefore require high-purity and costly hydrogen (US$10 kg⁻¹ by electrolysis) [13,14,15]. More importantly, the lack of an infrastructure for hydrogen transport and refueling is one of the major roadblocks hindering commercialization of hydrogen fuel cell vehicles and other applications [16, 17]. While methanol is a far less expensive alternative to hydrogen (US$0.3 kg⁻¹) and is significantly easier to store and transport than hydrogen, its utilization in LT-PEMFCs through reformed hydrogen is challenging, as its carbon monoxide content needs to be below 10 ppm (1 ppm = 1 μmol mol⁻¹) to avoid substantial poisoning of the active sites [18,19,20,21,22,23]. Furthermore, its temperature needs to be lowered from its reforming temperature (> 200 °C) to the operating temperature of LT-PEMFCs (~ 80 °C) (Fig. 1a) [24,25,26]. The reformate gas is first cooled by a condenser (cooling I), with the methanol in the reformate gas separated from the gas by the first cold trap (cold trap I) and washed in a scrubbing bottle. Then, the reformate gas flows into the methanation chamber to carry out CO clean-up by lowering the CO concentration. Downstream from the methanation chamber, the reformate gas with a significantly reduced CO concentration is further cooled down by the second condenser (cooling II), with vapor separation in the second cold trap (cold trap II). Due to the complexity of these components with cooling, vapor formation, and CO cleaning necessary, there is an urgent demand for fuel cells operating at higher temperatures with better carbon monoxide tolerance, without the need for CO clean-up and heat utilization for evaporation [27]. Fuel cells operating at elevated temperatures typically offer higher energy efficiency with more fuel options and can use low-PGM and PGM-free catalysts. Operating a PEMFC between 100 and 300 °C allows (i) increased oxygen reduction reaction (ORR) and hydrogen oxidation reaction (HOR) kinetics with opportunities for low-PGM and PGM-free catalysts, (ii) higher tolerance to catalyst poisoning from carbon monoxide, sulfur oxide, nitrogen oxide, and hydrogen sulfide with opportunities for low-purity hydrogen, (iii) easier heat dissipation without liquid water issues, and (iv) better thermodynamic quality of the produced heat [28,29,30,31,32,33,34,35]. In HT-PEMFCs, the fuel can consist of a methanol and water mixture, which is pumped into an evaporator before it is directed to the reformer (Fig. 2b). The steam reforming process then converts this fuel mix to a hydrogen-rich gas with a higher CO content than LT-PEMFC, due to the superior CO tolerance of HT-PEMFCs. Finally, this system allows the use of the heat of the HT-PEMFC to heat up the evaporator [36]. To meet these requirements, liquid water-free HT-PEMFCs have been developed by using proton-conducting acids with high boiling points (e.g., PA, phosphates) within a proton-conducting polymer membrane that is thermally stable up to 400 °C [37].

Fig. 1

Copyright 2020, Elsevier. b Operation of liquid-cooled HT-PEMFC systems with a methanol reformer. Adapted with permission from Ref. [36]. Copyright 2016, Elsevier

a Operation of LT-PEMFC system with a methanol reformer. Adapted with permission from Ref. [27].

Fig. 2

Schematic representation of the structure of HT-PEMFCs. The relative sizes and distances are not to scale, and the catalyst layer, binder, filler, MPL, and carbon fibers possess significantly different porosities and sizes

High-temperature proton exchange membranes (HT-PEMs) have tremendously improved, with major progress achieved in their stability, durability, and proton conductivity. The proton conductivity of different classes of HT-PEMs are summarized in Table 1 [37,38,39,40]. While progresses in membrane technologies are crucial, they must occur in conjunction with progresses in electrodes. For instance, to perform well in HT-PEMFCs, a membrane with an extremely high PA retention will require an electrode doped with a high PA content and overall outstanding proton-conducting compounds and fillers to efficiently conduct protons from the membrane to the electrode. Furthermore, while operating HT-PEMFCs above 200 °C is under acute scrutiny, this will likely result in hydration/dehydration issues for PA, causing its decomposition and reduction in its proton conductivity within the membrane and the electrode [41]. While membrane advancements have been investigated in depth and are under active development, significantly less effort has been dedicated to the design of highly efficient catalysts and catalyst layers in highly concentrated and HT acidic environments. Although PGM catalysts are currently the most efficient catalysts for HT-PEMFCs, their low active site utilization and unsatisfactory stability need to be addressed, as these make the assessment of new membranes in real devices inherently challenging and reduce the durability of HT-PEMFCs. Aiming to bring the opportunities of HT-PEMFCs into the spotlight, this review first outlines the advantages of PA-doped membranes for HT-PEMFCs from a durability and stability standpoint and then focuses on the most promising strategies to develop both efficient catalysts and catalyst layers to meet the requirements of the harsh HT-PEMFC environment and therefore overcome the electrode challenges of HT-PEMFCs.

Table 1 Membrane proton conductivity for HT-PEMs

2 Proton Conductors in HT-PEMFC Membranes

The core component of a PEMFC is the membrane electrode assembly, which is composed of a proton exchange membrane sandwiched between two catalyst layers, a microporous layer (MPL) and a gas diffusion layer (GDL), with a typical HT-PEMFC scheme shown in Fig. 2. The proton exchange membrane (PEM) determines the working mode and performance of the PEMFC. A wide range of electrolytes and membranes has been proposed for operation in HT-PEMFCs. For a high-performance polymer electrolyte, stable mechanical properties and high proton conductivity (> 10⁻² S cm⁻¹) at low humidity (or under anhydrous conditions) and at high temperature are ultimately necessary [42]. Following these principles, three main types of proton carriers have emerged: (i) water/hydrated protons, (ii) solid acids, and (iii) PA as the proton conductor, with a variety of polymer matrices proposed to host these proton carriers.

2.1 Water-Based Proton Conductors

The perfluorosulfonic acid (PFSA)-based membrane (Nafion, DuPont) is one of the most broadly used membranes using hydrated protons as the proton conductor and is ideal for LT-PEMFCs below 120 °C. Modifications of the PFSA membrane have been explored to increase its operating temperature to 140 °C using ionic liquid or PA doping [43]. The progress in and limitations of these technologies have been recently outlined in high-quality reviews, with the key findings briefly summarized here [34, 37, 39, 44]. PA and proton ionic liquids experience leaching for PFSA membranes, a problem that needs to be resolved via either modification of the PFSA membrane or use of alternative materials [45]. Similar to PFSA-based membranes, sulfonated hydrocarbon polymers (SHPs) have attracted attention due to their outstanding mechanical and thermal properties and are potential PEM candidates for HT-PEMFCs. However, both the modified PFSA and SHP-based membranes suffer from dehydration under the harsh operating conditions and require a continuous water supply for membrane humidification, making the system impractical for most real-life applications. Sulfonated polymer and their composite-based membranes have also been explored in HT-PEMFCs (100–150 °C). These membranes include neutral polymer backbones of materials such as poly(ethylene-alt-tetrafluoroethylene) (ETFE) grafted with poly(4-vinyl pyridine), acrylic monomers, acrylic acid, and associated polymers due to their high thermal and chemical stability [46,47,48]. Inorganic membranes using solid acids or metal phosphates are promising for HT-PEMFCs and have been explored for the past two decades. Among others, caesium dihydrogen phosphate solid acid (CsH₂PO₄), for instance, has good conductivity (410 mS cm⁻¹) between 160 and 250 °C [49, 50]. Furthermore, metal pyrophosphates (MP₂O₇, M = Sn, Ti, Si, Ge, Ce, and Zr) exhibit relatively high proton conductivity between 100 and 400 °C under water-free conditions [51]. However, the synthesis of a dense, flexible, and heat-resistant electrolyte membrane remains a key requirement for practical application of these materials due to their poor film forming capability. Some progress has been made through the development of organic‒inorganic structures using, for instance, Sn_0.95Al_0.05P₂O₇ and sulfonated polystyrene-b-poly(ethylene/butylene)-b-polystyrene (sSEBS) and inorganic membranes using SnP₂O₇–LaP₃O₉ [52, 53]. More recently, polybenzimidazole (PBI) with gadolinium-doped cerium pyrophosphate membranes has also shown promise, with higher performance than PBI membranes with reduced PA doping amount, while their peak power density decreases above 160 °C [54]. Altogether, although these approaches are interesting, the currently unsatisfactory stability and conductivity of these membranes have led to low performance in HT-PEMFCs compared to PA, and these membranes are therefore not further discussed in this review, with further information available elsewhere [39]. Therefore, this review focuses primarily on PA-doped membranes, as these are currently the most promising and commercially available membranes for HT-PEMFCs.

2.2 PA-Doped PBI Membranes

2.2.1 PA-Doped PBI Membranes

The strong and extensive hydrogen bonding network of PA causes its high viscosity and its Grotthuss mechanism with high intrinsic conductivity, which renders PA the most stable and most efficient proton carrier for HT-PEMFCs to date (Fig. 2) [55, 56]. While the proton conduction mechanism in HT-PEMFCs does not involve solely PA as the proton carrier, the PA/phosphate approach has thus far yielded superior performance and durability. PA needs to be immobilized within a polymer structure, which has directed scientific attention to phosphonic acid polyelectrolytes [57]. Therefore, polymers such as PBI, polysulfones, poly(phenylene oxide) (PPO), polyether ketones, poly(vinylpyrrolidone) (PVP), and their composites (e.g., polytetrafluoroethylene-poly(ethersulfone)-poly(vinyl pyrrolidone) (PES-PVP/PTFE)) have emerged as promising candidates for PA doping and operation in HT-PEMFCs [41, 58,59,60,61,62,63,64]. As an aromatic heterocyclic polymer, PBI possesses a glass transition temperature of 427 °C and excellent thermal and chemical stability [65]. PBI membranes with a molecular weight ranging from 40–65 kDa (1 kDa = 1 000 g mol⁻¹), following doping with PA (PA-PBI membranes), have been widely adopted for HT-PEMFCs [41, 57, 66, 67]. Recent advances in membrane technology have brought about HT-PEMFCs using PA-PBI that have achieved outstanding performance and stability of over 18 000 h at 160–180 °C [68]. While doping a high PA content within the PBI membrane is crucial to ensure the long-term stability and proton conductivity of the membrane electrode assembly, the mechanical stability of PBI membranes weakens when doped with an excessively high PA content [69,70,71]. Therefore, improvements in the PBI mechanical strength and the PA doping and retention ability are necessary to reach a higher proton conductivity. These topics have been recently discussed in an excellent review [72]. Modification of the PBI membrane, or membrane alternatives, has been investigated by using PBI blends and inorganic additives to improve its performance and proton conductivity and increase its operating temperature to beyond 200 °C.

2.2.2 PBI-Blend Membranes, Ionically Cross-Linked Membranes, and Silica-Based Composite Membranes

PBI blending and ionic cross-linking of proton-conducting membranes can improve their thermal, dimensional, and chemical stability and their PA retention [73]. Ionically cross-linked PBI-blend membranes mixed with sulfonated or phosphonated acidic polymers exhibit proton conductivities above 100 mS cm⁻¹, with excellent mechanical strength and low swelling [74]. In-depth investigation suggests that the chemical and radical stability results from acid–base cross-linking. Aromatic polyether backbones attached to polar pyridine moieties create complexes with PA, which can enhance the PA uptake. Pyridine-based PBI, using PA as both a solvent and a polycondensation agent, can achieve a proton conductivity of 200 mS cm⁻¹ at 200 °C [75]. Low-cost poly(arylene pyridine)s also demonstrate promise, with excellent durability at 160 °C for 100 h [76]. Furthermore, rigid polymeric backbones with high acid absorption through the interaction with the basic backbone create clusters of acid molecules for proton transport. Pyridine units as basic moieties in a rigid aromatic polymer were demonstrated to provide a basic polymer to adsorb PA. Phenyl phosphinoxide moieties included in the polymer structure improve the PA solubility and create more interaction with the doping agent [77]. Solubilizing side chain substituents can also improve the PA solubility and uptake ability [78]. Phosphonated mesoporous silica-based composite membranes have also been explored. These membranes benefit from a high density of ion exchange groups in the composite membrane and the mesoporous morphology of the fillers to conduct protons above 100 °C [79]. Furthermore, promising results have been achieved by using composite membranes based on a phosphotungstic acid (PWA)-doped polyethersulfone-PVP (PES-PVP) matrix for HT-PEMFCs [80]. The addition of PWA (5%) to PA could improve the performance of the HT-PEMFC from 250 to 400 mW cm⁻² [80].

2.2.3 Additives to Enhance the Stability and Conductivity of PBI-Based Membranes

The PA retention and proton conductivity properties of PBI membranes have been further improved by adding additives, such as carbon materials (carbon nanotubes (CNTs), graphene), heteropoly acids (HPAs), and oxides (silica, TiO₂, Al₂O₃, ZrO₂, SnO₂) [62, 81, 82]. Organic/inorganic hybrid nanocomposites (OPBI) have been prepared with surface functionalized silica nanoparticles (NPs) to improve the thermal and mechanical stability of the PBI membrane [83]. PBI and mesoporous silica composite membranes (APBI-DIL4.5-MDA1.5) have a high proton conductivity (220 mS cm⁻¹) and good mechanical stability at 180 °C [84, 85]. Nanocomposite membranes (based on PBI and porous SiO₂ NPs) in particular have a 329 wt% (wt% means the weight percentage) PA uptake with a 244 mS cm⁻¹ proton conductivity at 200 °C [86]. The incorporation of PWA-functionalized mesoporous silica into a PA-PBI membrane substantially enhances its durability in HT-PEMFCs at 200 °C [87]. Composite membranes with 15 wt% PWA-meso-silica have high proton conductivity and exceptional durability, with reduced PA uptake and good mechanical strength. PA stabilization by the PWA-meso-SiO₂ filler enables an ultrahigh stability over 2 700 h in HT-PEMFCs (< 30 mV h⁻¹) at 200 °C and 0.2 A cm⁻² [88]. The dispersion of microsized anatase titanium dioxide into a PBI membrane can increase its acid absorption capability and ionic conductivity in HT-PEMFCs [89]. Furthermore, the incorporation of hydrophilic and acidophilic NPs into a soluble PBI solution enhances its PA absorption ability at 120 °C [90].

2.3 Proton Conductors above 200 °C

Operating HT-PEMFCs above 200 °C could improve their carbon monoxide tolerance even further. However, operations at these temperatures require a new class of membranes, as the stability of the PA/PBI membrane decreases above 180 °C due to the loss of water content and formation of pyrophosphoric acid (H₄P₂O₇) produced by the condensation of PA [91]. Pyridine units containing copolymers and terpolymers have been explored via cross-linking with carboxylic acid side groups and hydrophobic methyl groups to create more robust and stabler polymer electrolyte membranes for HT-PEMFCs. This approach was demonstrated in HT-PEMFCs, achieving outstanding stability for over 350 h [92]. These membranes exhibit outstanding properties compared to PBI polymer membranes in terms of proton conductivity, mechanical strength, and manufacturing properties. A high-performance SnP₂O₇–polymer composite ceramic membrane, with closely packed P₂O₇ creating proton bonding sites and transport pathways, was also investigated, achieving 0.87 W cm⁻² at 240 °C with minimal performance loss in 25% carbon monoxide [93, 94]. Other membrane breakthroughs used poly(2,3,5,6-tetrafluorostyrene-4-phosphonic acid) and phosphonated polymers to reach peak power densities of 1.13 W cm⁻² at 160 °C and 1.74 W cm⁻² at 240 °C under H₂/O₂ conditions. Recently, Atanasov revealed the properties of several state-of-the-art HT-PEMFC phosphonated polymers as proton conductors, such as poly(2,3,5,6-tetrafluorostyrene-4-phosphonic acid) (PWN) with a 70% degree of phosphonation (PWN70), ion-pair coordinated polymer (PA-QAPOH), and PA-doped tin pyrophosphate (TPP/Nafion) composite membranes, and compared their properties with those of PA-PBI (Fig. 3a) [95]. The PWN membranes based on phosphonated poly(pentafluorostyrene) (PPFS) benefit from an efficient and versatile polymer phosphonation method, have high resistance to heat and radicals, and achieve high proton conductivity and performance in hydrogen fuel cells [96]. Tin pyrophosphate (TPP) has received significant interest in the last decade due to its potential as a proton conductor for ionomer electrolytes in HT-PEMFCs [97]. The ion-pair coordinated membranes (PA-QAPOH) use the strong interaction between bisphosphate–ammonium ionic pairs to increase the water tolerance and reduce evaporation of PA at elevated temperatures [98]. The proton conductivity of the ion-pair coordinated polymer (PA-QAPOH) is higher than that of PA-PBI and keeps increasing before decreasing above 220 °C. The maximum conductivity at a higher temperature for the ion-pair polymer could be caused by the higher interaction of the ion pair. The PA-doped tin pyrophosphate (TPP/Nafion) composite membrane shows lower conductivity than the PA-PBI and PA-QAPOH membranes in the temperature range of 120–200 °C, with no decrease in conductivity up to 300 °C. The proton conductivity of PWN70 increases from 10⁻⁴ to 0.2 mS cm⁻¹ as the temperature increases from 120 °C to 220 °C, but the proton conductivity is two orders of magnitude lower than that of TPP/Nafion, PA-PBI and PA-QAPOH (Fig. 3b, c). Overall, TPP/Nafion, PA-PBI and PA-QAPOH electrolytes are unimpacted by changes in relative humidity (RH) as they do not require liquid water to conduct protons, unlike PWN70 which conductivity increases by several orders of magnitude from with higher RH [95].

Fig. 3

Reproduced with permission from Ref. [95]. Copyright Springer Nature 2021

a Chemical structures of PA-PBI, PA-QAPOH, PA-QASOH, PA-doped SnP₂O₇, and PWN70 proton conductors. b Anhydrous proton conductivity of proton conductors as a function of temperature. c Proton conductivity as a function of RH at 80 °C.

Despite substantially outperforming PBI- and metal phosphate-based HT-PEMFCs, the poor stability of the PWN70 ionomer electrolyte above 160 °C must be addressed [95]. Recently, Cheng et al. developed a PA/PBI/SiO₂ composite membrane that can stably operate at temperatures of 210–250 °C [99, 100]. Altogether, these studies bring new opportunities for the practical application of HT-PEMFCs at higher performance and open new routes for commercialization of PEMFCs. Overall, while the development of HT-PEMs has tremendously improved their performance and stability, the most practical membranes rely on PA/phosphates as proton conductors. These environments significantly impact the selection of catalyst materials and the design of the membrane electrode assembly.

3 Catalysts in HT-PEMFCs

Over the past decades, considerable improvements in HT-PEMs have increased their viability. Figure 4 compares the typically reported peak power densities of LT-PEMFCs and HT-PEMFCs as a function of their operating temperature [95, 107, 108]. HT-PEMFCs have achieved impressive progress, with values of 1.2 W cm⁻² at 160 °C and 1.5–1.7 W cm⁻² in the 200–240 °C range in H₂/O₂ and 0.6–0.9 W cm⁻² in H₂/air [95]. These studies thus far have predominantly focused on the membrane, while the exploration of highly active catalysts and catalyst layers has received less attention. To design a high-performance membrane electrode assembly for HT-PEMFCs, one should consider not only the catalyst activity and the coupling of protons, electrons, and reactants at the triple-phase boundaries but also the stability at high temperatures, resistivity to phosphate ion poisoning, and PA flooding. Therefore, developing catalysts and catalyst layers compatible with PA-doped membranes should tremendously promote HT-PEMFCs for practical applications.

Fig. 4

Performance comparison of different types of fuel cells at different operating temperatures in H₂/O₂ (black) and H₂/air (red). Fuel cell data are reproduced with permission for LT-PEMFCs from Ref. [108] and △ [107] and for □ HT-PEMFCs from Ref. [95]

3.1 PGM Catalysts for HT-PEMFCs

The synthesis of PGM catalysts has significantly progressed in recent years, with major achievements regarding the Pt utilization and interactions between the catalyst and its support. However, the HT-PEMFC challenges need to be mitigated, such as phosphate poisoning and accelerated carbon corrosion at elevated temperatures, alongside the opportunities this environment offers such as phosphate-promoted ORR.

3.1.1 Highly Active PGM Catalysts

HT-PEMFCs currently rely on PGM catalysts for both the ORR and HOR. As the cathodic ORR is approximately 4–6 times slower than the anodic HOR, great efforts in the past several decades have focused on developing efficient Pt catalysts using low amounts of Pt for the ORR. Several innovative strategies have successfully reduced the Pt utilization via engineering of effective and stable PGM catalysts (Fig. 5) [109]. These strategies include (i) controlling the size of the NPs to increase the mass activity (Fig. 5a), (ii) engineering strong support-particle interactions (Fig. 5b), (iii) controlling the composition to tune the electron structure and reduce the PGM loading (Fig. 5c), and (iv) engineering the shapes and facets to yield high-activity active sites (Fig. 5d).

Fig. 5

Copyright 2018, Wiley

a Size control. b Tuning the support. c Composition control. d Shape/geometry engineering, with (d) reproduced with permission from Ref. [110].

Nanoscale particle size: reducing the size of noble metal particles, despite the specific activity following the order of polycrystalline Pt > unsupported Pt black particles (∼30 nm) > high surface area (HSA) carbon-supported Pt NP catalysts, has been extensively explored to increase the mass activity (Fig. 6a, b) [111]. Decreasing the particle size exposes more Pt sites, thereby increasing the mass activity. The development of nanotechnologies has brought great opportunities to increase the ORR activity while reducing the Pt loading from 4.0 mg cm⁻² to 0.2 mg cm⁻² [112]. The influence of the particle size on the ORR kinetics was intensively investigated in the first decade of the twenty-first century, and one of the most active and stable catalysts for PEMFCs is now widely accepted to be carbon-supported Pt NPs with 2–3 nm particle sizes [111, 113]. Further decreasing the NP size (< 2 nm) may reduce both the stability and mass activity (Fig. 6a, b). However, pioneering work suggests that subnanometre Pt particles may follow a different trend due to the quantum size effect, with the ORR catalytic activity increasing by removing one atom from a 13 Pt atom cluster (Pt₁₃), in which the Pt₁₂ cluster exhibits a twofold improvement in the catalytic activity [114]. The optimal performance in terms of the ORR electrochemical activity of clusters with between 12 and 20 atoms is reached for Pt₁₉ (Fig. 6c, d) [114]. Further downsizing the particle size to the atomic scale, namely single-atom catalysts, has attracted extensive interest [115, 116]. Pt single atoms anchored on N-doped carbon show good carbon monoxide/methanol tolerance for the ORR. The acidic single cell with this cathode catalyst delivers a peak power density of 0.68 W cm⁻² at 80 °C with 0.09 mg_Pt cm⁻² for LT-PEMFCs, offering great potential opportunities [115].

Fig. 6

Reproduced with permission from Ref. [111]. Copyright 2011, American Chemistry Society. c Mass–specific activity as a function of the number of Pt atoms within Pt_x (x = 12 − 24) clusters, synthesized by using either triphenylpyridylmethane (Pt_xDPAG4-PyTPM) or polyamidoamine (Pt_xPAMAM G4-OH). d Atomic structure of Pt NPs (13–19 atoms). Reproduced with permission from Ref. [114]. Copyright 2015, Wiley. e Three-dimensional volcano plot of PGM metal alloy catalysts for the ORR. Reproduced with permission from Ref. [117]. Copyright 2009, Springer Nature. f Volcano plots and free-energy diagrams for the ORR on PGM transition metal alloys. Reproduced with permission from Ref. [118]. Copyright 2009, Springer Nature

a Specific activity (SA) and b mass activity (MA) at 0.9 V_RHE as a function of the electrochemical surface area (ECSA) for HClO₄ and H₂SO₄, with 100 active sites and 30% of the particle covered by the support.

Pt NP interaction with the support: as the size of the Pt NPs decreases, their interaction with the support is enhanced, which allows improvement in the ORR performance by exploring the support-NP interaction to tune the catalytic activity and stability. Carbon materials, such as amorphous carbon, activated carbon, CNTs, carbon nanofibers, graphene, porous carbons, and heteroatom-doped carbon materials, have been intensively studied due to their high surface area, high electronic conductivity, and good stability [119]. PGM NP sintering requires a strong interaction between the NPs and the graphitized carbon supports to prevent carbon corrosion. Thus, CNTs are particularly promising supports for HT-PEMFCs [120, 121]. Functionalization of CNTs with poly(2,2-(2,6-pyridine)-5,5-bibenzimidazole) (PyPBI) or with PVP, pyridine or hydroxypyridine moieties could provide a strong π–π interaction and Pt–N bonding that improve the power density output and stability [121,122,123,124]. Highly graphitized carbons are preferred due to their enhanced electronic conductivity and stability against corrosion [125]. Heteroatom-doped carbon materials, such as N-doped CNTs and iron–nitrogen codoped mesoporous CNFs, have good properties to promote Pt catalysts [126, 127]. Pt NPs loaded on N-doped mesoporous carbon (Pt/NMC) have a mass activity of 0.384 A mg_Pt⁻¹ at 0.9 V versus the reversible hydrogen electrode (RHE) in acidic media, which is 8.1 times higher than that of Pt NPs on Vulcan XC-72R and 5.9 times higher than that of commercialized Pt/C [128]. Metal oxides (e.g., TiO₂, ruthenium-titanium mixed oxide, Ta₂O₅), metal nitrides, oxynitrides, carbide and carbonitrides (TiC, WC, Mo₂C, VC, NbC, Fe₂C, Co₆Mo₆C₂) have attracted enormous interest due to the strong interaction of Pt NPs with these supports to tune both the activity and stability [129,130,131,132,133,134,135,136,137]. However, the reaction of metal oxides with PA at high temperatures to form phosphates could be an issue for catalysts due to their nonconductive properties. Among the materials, transition metal carbides are used as supports due to their chemical stability in acidic media, resistance to poisoning, excellent mechanical durability, and high electronic conductivity. Tungsten carbide-supported Pt NPs, for instance, have demonstrated, with their enhanced ORR activity, increased mass activity and specific activity approximately twofold (at 0.85 V vs. RHE) and sevenfold higher than those of Pt/C, respectively [138]. Pt NPs supported on MoO_x-modified CNTs exhibit higher performance in PBI membrane-based HT-PEMFCs at 140 °C [139]. Finally, graphitized carbon functionalized with metal oxides, metal nitrides, oxynitrides, carbides, and carbonitrides could offer new opportunities.

Pt alloys: the turnover frequency of Pt NPs can be increased by introducing other elements to tune the electronic structure of the Pt sites and optimize the adsorption energy of oxygen intermediates. Furthermore, the introduced metal can reduce the utilization of Pt atoms, especially in the inner core of the Pt NPs. This new class of ORR catalysts is commonly known as PGM alloys [140, 141]. PGM alloys with ORR activity orders of magnitude higher than that of commercial Pt/C have been developed after intensive screening of the optimum composition of PGM alloys. Some well-known systems include PtNi, PtCo, and PtPd [142,143,144,145]. The Pt₃Ni(111) surface is 10 times more active for the ORR than the corresponding Pt(111) surface and 90-fold more active than the current state-of-the-art Pt/C catalysts for PEMFCs; Pt₃Ni is one of the most active ORR catalysts (Fig. 6e, f) [142]. Pt₃Co, Pt₃Ni, Pt₃Y, and Pt₃Sc have promising activity for the ORR, while Pt₃Y and Pt₃Sc can meet both the high activity and outstanding stability requirements [118]. The activity of polycrystalline Pt₃Sc and Pt₃Y electrodes is 1.5–1.8 and 6–10 higher than that of pure Pt, respectively [118]. Among different types of Pt alloys, ordered alloys, core–shell alloys, and Pt-skin alloys have attracted enormous interest due to their superior activity and stability and could be promising for HT-PEMFCs [146,147,148,149,150,151,152,153,154,155,156]. For instance, the use of ordered Pt₃Co intermetallic cores with a 2–3-atomic-layer-thick Pt shell increases the mass activity and specific activity by 200% and 300%, respectively, compared to disordered Pt₃Co alloy NPs and Pt/C. A minimal loss of catalytic activity occurs after 5,000 potential cycles, while the ordered core–shell structure remains virtually intact. The Pt-rich shell and stable intermetallic Pt₃Co core arrangement improves the activity and stability [143]. N-doped carbon-encapsulated ordered PtFe intermetallic compounds have a 1.3-fold enhancement in the peak power density and high stability due to effective prevention of agglomeration, coalescence, and Oswald ripening of the particles [157]. Alternatively, Pd-based core–shell alloys can replace Pt catalysts in HT-PEMFCs. The performance of the PdNiCu@PdIr/C catalyst is higher than that of PdNi@PdIr/C and PdCu@PdIr/C catalysts and reaches 95% of the performance of the commercial PtCo/C catalyst [158]. With improvements in PGM ORR catalysts, new catalysts have been developed by controlling the size, alloy composition, crystalline structure, and morphology to achieve the ORR activity goal set by the US Department of Energy (0.44 A mg_PGM⁻¹ for 0.1 mg_PGM cm⁻²) [159]. One-dimensional (1-D) nanostructures (such as nanowires, nanorods and nanotubes), two-dimensional nanosheets and nanoplates, and three-dimensional nanostructures (such as nanoframes and nanoporous structures) have attracted great attention due to their unique anisotropic structure, which is beneficial to many catalytic reactions [160,161,162,163,164,165,166,167,168,169,170]. Pt₃Ni nanoframes, prepared by erosion of crystalline PtNi₃ polyhedra with surface and three-dimensional molecular accessibility, enhance the mass activity and specific activity 36- and 22-fold, respectively [171].

Overall, readers should keep in mind that while insightful, some of these pioneering demonstrations have been either purely theoretical or only investigated by using aqueous model systems with 3-electrode cells and a rotating disk electrode (RDE) suppressing mass transport at ambient temperature. As the HT-PEMFC environment is far harsher than diluted perchloric acid or that in LT-PEMFCs, the performance of these materials needs to be further elucidated with concentrated PA, at elevated temperatures and finally in HT-PEMFCs.

3.1.2 PGM Catalysts for HT-PEMFCs

To further the research in this direction, PGM catalysts have been synthesized with the unique challenges of HT-PEMFCs in mind, such as increased phosphate poisoning resistance, phosphate-promoted ORR, and higher carbon corrosion resistance at elevated temperatures.

Phosphate poisoning challenges of PGM catalysts: PGM catalysts for the ORR have been intensively studied in the past decades for LT-PEMFCs. However, the high temperature and PA environment of HT-PEMFCs cause severe underperformance of the most promising PGM catalysts. The impact of the concentrated PA electrolyte on the performance, durability, and gas diffusion cannot be assessed in an RDE configuration using diluted PA (0.1–1 M PA, 1 M = 1 mol L⁻¹). The differences between the RDE and HT-PEMFC environments have recently come under focus [172]. To summarize, in the RDE environment, the active sites are saturated with dissolved O₂ in a liquid acidic electrolyte at typically ambient temperature, whereas in the HT-PEMFC membrane electrode assembly, the reaction only occurs at the triple-phase interface between the gas, catalytic sites, and charge transfer species (protons and electrons) [172]. PA adsorption on the PGM active sites hinders the utilization of active sites for the ORR in the cathode and reduces the performance and durability [41, 66, 67, 173,174,175,176]. As a result, Pt catalysts usually exhibit poor performance and stability in PA environments and HT-PEMFCs even at high Pt loadings [176]. For instance, the ORR activity and kinetic current density of Pt(100), Pt(110), Pt(111), and PtSn(111) decrease even with small amounts of PA (1 mM to 100 mM) in the perchloric acid solution (Fig. 7a, b). Specifically, 100 mM PA in HClO₄ reduces the half-wave potential of Pt(100), Pt(110), Pt (111), and PtSn(111) in HClO₄ solution at ambient temperature by 38 mV to 93 mV [173]. As a result, the performance of most PA-PBI-based HT-PEMFCs is typically half that of LT-PEMFCs, despite a much higher Pt loading (1 mg_Pt cm⁻² vs. 0.2–0.3 mg_Pt cm⁻²) [41].

Fig. 7

Reproduced with permission from Ref. [173]. Copyright 2010, Royal Society of Chemistry. c Phosphate-ion adsorption on oleylamine-doped Pt/C. Reproduced with permission from Ref. [177]. Copyright 2013, American Chemistry Society. d CNT/ABPBI/Pt@IL synthesis and H⁺ and ORR pathways. Reproduced with permission from Ref. [178]. Copyright 2018, Wiley

PA poisoning and mitigation strategies of PGM catalysts. a Performance of Pt(111) in 0.1 M HClO₄ with 0, 1, 10, 50 and 100 mM PA. b ORR kinetic current density of Pt(100), Pt(110), Pt(111), and PtSn(111) in 0.1 M HClO₄ with/without 100 mM PA (recolorized for this review).

Increasing the phosphate tolerance: preventing the adsorption of PA on Pt surfaces can prevent phosphate poisoning-related limitations [179]. N-doped carbon-encapsulated Pt-Fe ordered intermetallic NPs (O-Pt-Fe@NC/C) exhibit higher activity and poisoning resistance than Pt/C in a PA environment, as the N-doped carbon creates a protective layer between the Pt-Fe surface and phosphate anions [157]. Surfactant organic molecules and oleylamine similarly enhance the ORR activity of Pt NPs while preventing the adsorption of phosphate anions (Fig. 7c) [179]. Similarly, the addition of small amounts of oleylamine-modified Pt NPs can triple the ORR activity by preventing PA adsorption [177]. Modifying the support can improve the performance of PGM catalysts in HT-PEMFCs. For instance, polymers, such as PVP and poly[2,2-(2,5-pyridine)-5,5-bibenzimidazole] (para-PyPBI), have been used to functionalize the carbon supports of Pt NPs and improve the power density [178]. Protic ionic liquid modifications have been explored for a new class of CNT Pt NP catalysts, as the strong affinity of the ionic liquid with PA avoids active site poisoning and improves the HT-PEMFC performance and durability (Fig. 7d) [178]. Phosphate poisoning of PGM electrocatalysts can be reduced by altering the crystallinity and alloy composition. Modifying the electronic structure of Pt₂Ni₁ by argon heat treatment following specific anion adsorption has seen promising developments, improving the PA poisoning resistance and enhancing the ORR activity and stability [180]. As the ORR is a proton-coupled electron process with sluggish kinetics, improving the proton transport mechanism could greatly improve its activity. Despite PA adsorption poisoning the Pt sites and reducing the activity of PGM catalysts, PA conducts protons and could also be used to facilitate the ORR. Therefore, using phosphates to promote the ORR activity of PGM catalysts has been explored to improve the performance of HT-PEMFCs.

Silica-promoted ORR: silica (SiO₂), while inactive to the ORR, strongly bonds to acids such as PA and can both prevent acid leaching and improve the proton conductivity. The introduction of SiO₂ reduces the loss of PA from both the membrane and catalyst layers and prevents phosphate poisoning of Pt. Most recently, Cheng and Wang’s group prepared a CNT@x-SiO₂-Pt cathode catalyst by modifying CNTs with SiO₂ nanoclusters (Fig. 8) [181]. This catalyst achieves a record power density of 0.765 W cm⁻² in H₂/O₂ and 0.486 W cm⁻² in H₂/air without back pressure at 160 °C. Furthermore, this catalyst achieves 1.061 W cm⁻² in H₂/O₂ at 240 °C (with 4 wt% SiO₂ loading). At 160 °C, the strong adsorption of phosphate on SiO₂ hinders phosphate poisoning of Pt, enhances the distribution of PA in the catalyst layer, and supplies local protons for the ORR (Fig. 8c). Most importantly, above 200 °C, the SiO₂ nanoclusters transform into phosphate silicate and supply a stable proton chain, which enables much higher HT-PEMFC performance than previously reported under similar conditions. Furthermore, the enhanced proton conductivity is also consistent with the smaller ohmic resistance and faster charge transfer of CNT@x-SiO₂-Pt catalysts than commercial Pt/C and CNT-Pt at 0.6 V. Therefore, introducing SiO₂ next to the Pt NPs can maintain stable operation and power density of HT-PEMFCs. This new discovery provides a new pathway to develop efficient HT-PEMFC catalysts and resolves the phosphate poisoning issues for commercialization of HT-PEMFCs [181].

Fig. 8

Reproduced with permission from Ref. [181]. Copyright 2021, Springer Nature

a Pt-SiO₂ structure and b transmission electron microscopy (TEM) image of Pt-SiO₂. c to e Performance of cathode catalysts in HT-PEMFCs. c Pt/C, CNT-Pt and Pt-SiO₂ at 160 °C, and d Pt-SiO₂ from 160 °C to 240 °C. e Stability of Pt-SiO₂Pt/C, CNT-Pt, and commercial Pt/C at 160 °C without back pressure (H₂/O₂).

PGM catalyst stability at high temperature: the chemical and structural stability of PGM catalysts in PEMFCs is impacted by (i) metal dissolution, including the dissolution of Pt and its alloys (e.g., Ni, Co), (ii) Ostwald ripening and agglomeration and (iii) support corrosion and particle detachment (Fig. 9). These degradations can often occur simultaneously, as the dissolved metal forms nanoclusters can promote carbon corrosion and weaken the interaction of the metal particles with the supports, which has been well studied for LT-PEMFCs [182]. However, operations at elevated temperatures (100–250 °C) accelerate the chemical and structural degradation, with an increase in Pt NP size after operations in HT-PEMFCs [32]. Therefore, catalysts with higher stability must be designed for successful application of HT-PEMFCs, with a few promising approaches. The addition of PWA-meso-silica into PBI membranes has been reported to increase the stability of the Pt catalyst in the electrodes due to the formation of phosphosilicate preventing PA leaching. In these configurations, Pt NPs confined by SiO₂ nanoclusters have a higher stability due to the strong interaction between Pt and SiO₂, the Pt nanoconfinement hindering aggregation, and the graphitized CNTs offering a stable support at high temperatures (Fig. 8). This paves the way for further catalyst development. Therefore, a combination of spatial nanoconfinement within nanopores enhancing the interactions between particles and the carbon support, stable core shells/Pt skin and engineered intermetallic structures could improve the Pt stability issue in HT-PEMFCs [183]. These approaches offer new opportunities to improve the stability of PGM catalysts in HT-PEMFCs (Fig. 9).

Fig. 9

Degradation mechanisms and mitigation strategies for PGM catalysts

3.2 PGM-Free Catalysts

Currently, high PGM catalyst loadings are needed to improve the viability of HT-PEMFCs for practical applications due to the phosphate chemisorption of PGM cathode catalysts. The high cost of PGM catalysts and poor expression of catalytic sites in the PA environment drive the development of PGM-free catalysts for HT-PEMFCs [174].

3.2.1 Highly Active PGM-Free Catalysts

PGM-free catalysts: extensive efforts have been dedicated to developing low-cost PGM-free cathode catalysts for PEMFCs over the past two decades. Their performance and efficiency have been significantly improved, narrowing the half-wave potential gap from 100 to 12 mV [184, 185]. As cathode potentials below 0.7 V are typically used to generate high power densities, the synthesis of metal–nitrogen–carbon (M–N–C) catalysts free of reduced metal, metal carbide, or metal oxide particles is preferred due to the poor activity and stability of these compounds under acidic conditions. Single-atom M–N–C catalysts have emerged as one of the most promising alternatives due to their high intrinsic activity for the ORR [185]. Therefore, there has been a focus on the synthesis of single atom-based catalysts for LT-PEMFCs [186,187,188,189]. Specifically, nitrogen-coordinated Fe/Co/Mn/Sn single-atom catalysts have been generally recognized as the most promising alternatives to Pt catalysts [187,188,189,190].

A number of good reviews have been published to summarize the advancements and challenges in PGM-free catalysts for LT-PEMFCs, with the main findings summarized here [184, 189, 191, 192]. The major active center of these catalysts under acidic conditions is the nitrogen-coordinated Fe/Co/Mn/Sn atom embedded in the carbon matrix (M–N_x) [187,188,189,190, 193, 194]. Among Fe/Co/Mn/Sn–N_x-based catalysts, Fe–N_x single-atom catalysts outperform Co–N_x, Sn–N_x and Mn–N_x single-atom catalysts [195,196,197]. The half-wave potential of Fe–N_x single-atom catalysts has been enhanced by 100 mV in recent years, narrowly reaching that of Pt in acidic environments [188, 190, 198, 199]. Ballard Power Systems recently demonstrated and commercialized a 30 W stack using solely PGM-free catalysts for the cathode [200]. Dual Fe–Co–N_x sites have even recently displayed outstanding onset potential and half-wave potential (1.06 and 0.863 V, respectively) [201]. Figure 10 highlights the latest advances in these single-atom or single-atom configuration PGM-free catalysts to replace Pt/C.

Fig. 10

Half-wave potential of various M–N–C catalysts evaluated in acidic electrolytes

Despite impressive improvements in performance, the stability of Fe–N–C catalysts in LT-PEMFCs is quite poor. These catalysts suffer from (i) flooding of the porous structure in the catalyst layer due to the liquid water, (ii) high mass transfer resistance due to the thicker catalyst layer resulting from the high catalyst loading, (iii) demetallation of Fe-based active sites via oxidation of atomically dispersed Fe by H₂O₂ radicals in active FeN_x moieties (< 1.0 V vs. RHE) and leaching of Fe NPs (> 0.3 V vs. RHE), and (iv) electrochemical carbon oxidation reaction and corrosion caused by starts-stops (> 0.207 V vs. RHE) and H₂O₂ or H₂O₂-derived radicals (e.g., hydroperoxyl). Carbon oxidation at high potential (> 0.9 V) destroys atomically dispersed active site Fe–N_x–C_y species. Stable operando potential windows below 0.7 V and operational strategies are suggested to avoid degradation of Fe–N–C catalysts in acidic media [186].

3.2.2 PGM-Free Catalysts for HT-PEMFCs

The HT-PEMFC environment presents interesting opportunities to utilize the active sites of PGM-free catalysts potentially better than in LT-PEMFCs. First, M–N_x single-atom catalysts have high tolerance for phosphate poisoning, which renders them attractive for HT-PEMFCs using phosphate or PA as proton carriers (Fig. 11) [174, 218]. In addition, increasing the operating temperature avoids liquid water flooding within the nanopores of the single-atom catalysts [31,32,33]. Therefore, operating these catalysts in HT-PEMFCs may finally fully unlock their potential [192] (Table 2).

Fig. 11

Reproduced with permission from Ref. [218]. Copyright 2018, Elsevier. c Pt/C and d PANI–Fe–N–C in a 0.1 M HClO₄ electrolyte with 0, 0.05, 0.15 and 0.30 M H₃PO₄. Reproduced with permission from Ref. [174]. Copyright 2014, American Chemistry Society

Performance of Pt/C and Fe–N–C catalysts in different acidic environments. a Pt/C and b BP-FeNC in HClO₄, H₂SO₄, H₃PO₄ and HCl.

Table 2 Overview of PGM-free catalysts in LT-PEMFCs. Data are extracted from different references so their significant digits are left unchanged

PGM-free catalysts were reported in HT-PEMFCs for the first time in 2015 [215]. Following this technological breakthrough, PGM-free catalysts have come under increasing investigation over the past six years for HT-PEMFCs, with improvements in performance, efficiency, and durability (summarized in Table 3). Most importantly, the stability of PGM-free catalysts is far superior in HT-PEMFCs to that in LT-PEMFCs, as summarized in Fig. 12.

Table 3 Overview of PGM-free catalysts in HT-PEMFCs. Data are extracted from different references so their significant digits are left unchanged

Fig. 12

Performance and durability of PGM-free catalysts in a LT-PEMFCs and b HT-PEMFCs. Note: this graph was produced by using the data from Tables 2 and 3

PGM-free catalysts were first reported in HT-PEMFCs using hollow microspheres with a graphitic layer of encapsulated Fe₃C NPs with a high surface area (374 m² g⁻¹) as the active sites (Fig. 13a, b) [215]. Most notably, this catalyst achieves a much higher stability in HT-PEMFCs than in LT-PEMFCs, confirming that the higher tolerance of PGM-free catalysts captured in low-concentration PA environments using a RDE (Fig. 11) occurs in real devices. Designing catalysts with etched silica templates can enhance oxygen and PA diffusion to the active sites in HT-PEMFCs (Fig. 13c, d) [216]. The durability of PGM-free catalysts in HT-PEMFCs was further increased by using a Fe–N–C catalyst (BP-FeNC) (Fig. 13e, f), with less than 32% performance degradation in 400 h [218]. The performance (0.06–0.18 W cm⁻²) of these pioneering catalysts could be improved in HT-PEMFCs through increased active site density, reduced carbon corrosion, and phosphate-promoted ORR. Furthermore, more systematic investigations of the PA doping procedure, catalyst layer design, and membrane assembly procedure should be conducted.

Fig. 13

Reproduced with permission from Ref. [215]. Copyright 2015, Royal Society of Chemistry. c Silicate-derived synthesis of Fe/N/C and d polarization at 150 °C, dry H₂/O₂. Reproduced with permission from Ref. [216]. Copyright 2018, Wiley. e BP-FeNC structure and f polarization of 60 wt% Pt/C (1.6 mg_Pt cm⁻²), 20 wt% Pt/C (0.9 mg_Pt cm⁻²), and BP-FeNC (7.8 mg_cat cm⁻²) at 160 °C, dry H₂/O₂. Reproduced with permission from Ref. [218]. Copyright 2018, Elsevier

PGM-free catalysts in HT-PEMFCs. a Synthesis of Fe/C and b polarization of Pt/C and Fe/C-700 cathodes in LT-PEMFCs (80 °C, Nafion membrane, 100% RH) and HT-PEMFCs (160 °C, PBI membrane, dry gases).

Active site density and stability: iron-based single-atom catalysts suffer from a low surface-active site density, with active sites encapsulated within the carbon structure [217]. For instance, the ORR half-wave potential and onset potential of microporous carbon materials become more positive as the Fe content is increased from 0 to ~ 0.6 wt%, but a further increase in the Fe content does not improve the ORR activity due to the encapsulation of the active sites inside the structure [226]. The active site loading is generally limited in the range of 1–3 wt%(Fe) due to the undesired aggregation at high loading. Due to strong Fe clustering at high Fe content, the utilization efficiency of the Fe–N–C catalysts drops from ~ 0.4 to ~ 0.1 when the Fe loading increases from 0.3 to 2.8 wt% [199]. Enhancing the loading and utilization is one of the main solutions to further boost the performance of these noble metal-free catalysts. To enhance the Fe–N₄ site utilization, a chemical vapor deposition approach was developed by using transmetallation of Zn–N₄ sites into Fe–N₄ sites, achieving an active site density of 1.92 × 10²⁰ sites per gram with 100% site utilization even with a Fe loading of 2.0 wt% [227]. Most recent research shows that strong interactions between active sites can alter the intrinsic ORR activity when the intersite distance is smaller than approximately 1.2 nm, while synergistic effects continuously boost the performance until the distance between adjacent Fe sites becomes smaller than 0.7 nm [228]. A new synthesis strategy successfully reduced the catalyst loading to 0.3 mg_Fe cm⁻² by increasing the active site density (Fe loading of (7.7 ± 1.3) wt%) [221]. This procedure synthesizes two-dimensional thin graphene with a thickness of 0.4–0.6 nm (Fig. 14a, c). The iron single-atom catalysts embedded in nitrogen-coordinated graphene (FeSA-G) have a surface atomic Fe density of 3.5 sites nm⁻², which is among the highest for iron single-atom catalysts. These catalysts are highly active and stable for the ORR with high phosphate tolerance, which is particularly appealing for HT-PEMFCs. The FeSA-G catalyst in combination with the SiO₂ NP-doped PA-PBI (PA/PBI/SiO₂) membrane developed by Cheng’s group achieves outstanding performance and stability in HT-PEMFCs, which is attributed to its high resistance to phosphate poisoning and enhanced ORR kinetics (Fig. 14d, f) [221]. This is consistent with Li’s group reporting negligible adsorption of PA on a Fe–N–C catalyst [218]. Most importantly, the increase in power density from 160 to 230 °C, reaching similar performance to that of Pt at 230 °C, demonstrates the potential of this catalyst for operation at even higher temperatures (Fig. 14d, e). Most recently, high loading of Fe–N–C with 7 wt% atomic iron sites using a sacrificial metal (Zn) in the initial synthesis step achieved an active site density of 7.8 × 10¹⁹ sites g⁻¹ by using the ex situ CO chemisorption method [229].

Fig. 14

Reproduced with permission from Ref. [221]. Copyright 2019, Wiley

a TEM, b atomic force microscopy (AFM), and c aberration corrected (AC)-scanning transmission electron microscopy (STEM) images of FeSA-G catalysts. Polarization and power density of HT-PEMFCs using 0.3 mg_cat cm⁻² FeSA-G and 1 mg_Pt cm⁻² cathodes d at 160 °C and e at 230 °C. f One-hundred-hour stability of the HT-PEMFC at 0.6 V for FeSA-G and 0.5 V for Pt.

Reduced carbon corrosion: carbon corrosion is more severe for high-loading iron single-atom and nitrogen defect catalysts within the carbon matrix in HT-PEMFCs due to their elevated temperature and concentrated PA environment (Table 2) [207, 213]. Therefore, supporting high-density single atoms on conductive and corrosion-resistant carbon substrates can fulfill the requirement of HT-PEMFCs. Among the carbon-based supports, CNTs are thermally stable and enhance the electrocatalytic activity of the supported electrocatalysts due to their good graphitization [230, 231]. Supporting a monolayer of iron single atoms on graphitized carbon could expose all the active sites and boost the electrochemical stability [232,233,234]. However, depositing iron single atoms homogenously on CNTs with high density is challenging, as metal atoms tend to agglomerate. The use of a silica-protective layer allows preferential formation of atomic iron active sites instead of Fe NPs [235, 236]. The silicate template-based method involves multiple steps, while the loading of atomic Fe–N_x on the CNTs is less than 2.0 wt%. The active site density of PGM-free catalysts in HT-PEMFCs has been increased via a facile template-free method to prepare high-density iron single atoms supported on highly crystallized CNTs (Fig. 15a) [225]. Iron-based single-atom catalysts are synthesized through one-pot pyrolysis of well-ground hemin porcine with dicyandiamide and acid-oxidized CNTs. The surface of the CNTs (< an 8 nm diameter) supports a continuous layer of amorphous carbon, with a high density of isolated iron atoms on the CNT surface. The 3.5 wt% atomic iron loading single-atom catalyst has an onset potential of 0.95 V and a half-wave potential of 0.801 V for the ORR in O₂-saturated 0.1 M HClO₄ solution. This performance is comparable to that with 12.5 μg_Pt cm⁻² and attributed to the high density of atomic sites and conductive CNT graphene networks. The as-synthesized iron single-atom catalyst has a 0.266 W cm⁻² peak power density and excellent stability (negligible performance loss at 0.5 V over 100 h) at 240 °C in HT-PEMFC cathodes (Fig. 15c, e) due to the highly graphitized CNTs. This supplies a facile and practical route for efficient PGM-free catalysts for HT-PEMFCs. The Cheng group most recently demonstrated that the in situ growth of atomic layers of iron single atoms on carbon supports could boost the turnover frequency by using the sublayer to tune the activity of the top-layer atomic sites, enhancing the activity and durability for the ORR [237]. Interestingly, tuning the CO content (0–10%) within the hydrogen stream has also been reported to actually possibly reduce the carbon corrosion rate occurring during start-stop [238], while carbon corrosion can also be mitigated through careful control of HT-PEMFC operation [239].

Fig. 15

Reproduced with permission from Ref. [225]. Copyright 2021, Elsevier

a STEM, b high-resolution TEM and c AC-STEM images of FeSA/HP. d Polarization and power density of HT-PEMFCs with a 4 mg_cat cm⁻² FeSA/HP cathode at 160, 200, 220 and 240 °C. e Performance of the HT-PEMFCs with Pt/C and FeSA/HP cathodes at 240 °C. f Stability of the HT-PEMFC using the Pt/C cathode at 0.6 V and the FeSA/HP cathode at 0.5 V and 240 °C.

PA-promoted ORR: in HT-PEMFCs, PA is the proton carrier in both the membrane and the catalyst layer. PA adsorption next to the active sites could locally supply protons to promote the ORR, leading to an enhanced performance of the HT-PEMFCs, rather than poisoning of the catalysts. Regulation of the proton transport mechanism could improve the ORR activity, as the ORR is a proton-coupled electron process [240,241,242]. Atomically dispersed bimetallic FeCu atoms anchored on nitrogen-doped CNTs (FeCu/N-CNTs) have been designed as a new electrocatalyst for the ORR in PA-based HT-PEMFCs [223]. The FeCu/N-CNT catalysts were synthesized through pyrolysis of dicyandiamide in the presence of hemin porcine and copper(II) acetylacetonate after a subsequent acid treatment (FeCu(X:Y), with X:Y being the iron-to-copper atomic ratio) (Fig. 16a-b) [223]. The Fe–N and Cu–N sites with distances of 4–5 Å (1 Å = 1 × 10⁻¹⁰ m) synergistically promote the ORR activity under acidic conditions. In addition to high tolerance to phosphate, these FeCu/N-CNT catalysts have enhanced ORR activity in the presence of PA or so-called phosphate-promoted ORR activity. The ORR activity is closely correlated with the Fe:Cu ratio, with the best results obtained on FeCu(4:1), achieving an onset potential of 0.960 V versus RHE and a half-wave potential of 0.811 V versus RHE. The half-wave potential increases by 18 mV with increasing amount of PA (Fig. 16d). The Cu–N atomic site could strongly absorb the phosphate ions and local protons to assist the ORR at the adjacent Fe–N active site. The HT-PEMFCs with either the FeCu(1:0) or FeCu(4:1) cathode reach 0.24 and 0.30 W cm⁻² peak power densities, with the latter identical to that with the Pt cathode (Fig. 16e). Furthermore, these catalysts achieve the highest stability to date for PGM-free catalysts for LT-PEMFCs and HT-PEMFCs, with retention for over 100 h (Fig. 16f). These new discoveries supply a new pathway to design phosphate-promoted ORR catalysts, offering new opportunities to overcome the PA poisoning in HT-PEMFCs.

Fig. 16

Reproduced with permission from Ref. [223]. Copyright 2021, Elsevier

a High-resolution TEM and b AC-STEM images of FeCu-N-CNTs. c Linear scan voltammetry curves of FeCu/N-CNTs with different Fe/Cu atomic ratios in 0.1 M O₂-saturated HClO₄. d FeCu(4:1) in O₂-saturated 0.1 M HClO₄ with the addition of PA. e Polarization and power density of PA/PBI composite membrane cells using 4 mg_cat cm⁻² FeCu(1:0), FeCu(4:1) and 1 mg_Pt cm⁻² cathodes at 230 °C. f Stability of the cells with a Pt/C cathode at 0.6 V and a FeSA/HP cathode at 0.5 V and 240 °C.

Overall, the traditional synthesis of single-atom catalysts through pyrolysis of a mixture of carbon/nitrogen/metal sources, which encapsulates the active sites inside the carbon structure and poorly graphitizes the carbon, requires careful reconsideration to meet the challenges of M–N_x single-atom catalysts for the ORR in HT-PEMFCs. The development of novel methods based on supporting a single layer of high-density bimetallic or multimetallic atomically dispersed catalysts on conductive and corrosion-resistant carbon supports would be a promising approach to solve the challenges of poor intrinsic activity and stability and low active site density and utilization. In this sense, two-dimensional graphene could be a useful support because every carbon honeycomb is utilized in the conductive support, avoiding encapsulation. In addition, the graphene sheets supply a planar platform to build triple-phase boundaries on the surface and outperform microporous carbon structures. However, they are still unexplored at this stage. Furthermore, investigation of the best acid doping procedure and cell activation procedure, improvement in the membrane and GDL properties, and design of macropores in the catalyst should be explored, with a strong focus on the uniquely complex challenges of PGM-free catalysts.

4 Highly Efficient and Stable Catalyst Layers for HT-PEMFCs

While LT-PEMFCs and HT-PEMFCs may appear similar aside from the higher operating temperature, the proton transport mechanisms within the electrodes are fundamentally different. As a result, LT- and HT-PEMFC require catalyst layers with drastically different architectures and properties. Specifically, in LT-PEMFCs, proton transport occurs via the liquid water generated by the electrochemical reaction/provided by the humidified gas and via the proton-conducting ionomer in humidified conditions [243]. In contrast, HT-PEMFCs rely on the viscous acidic electrolyte, doped within the membrane electrode assembly prior to operation, to reliably and evenly conduct protons through their entire lifecycle [244]. While the requirements for the catalyst layer in LT-PEMFCs have been recently outlined in an excellent review by Suter et al., these requirements differ for HT-PEMFCs and should be investigated further [244]. In HT-PEMFCs, the catalyst layer structure must allow diffusion of the viscous acid to the active sites while also stabilizing it to avoid leaching and depletion. Figure 17 succinctly summarizes the differences in the electrode properties of LT- and HT-PEMFCs. The conventional RDE test system does not sufficiently reflect the real conditions and the intrinsic activity in the PA environment for HT-PEMFCs (Fig. 17a, b). The liquid water-free PA environment of HT-PEMFCs makes establishing a robust triple-phase boundary at all active sites challenging and reduces the catalyst utilization (Fig. 17c, d).

Fig. 17

a Typical three-electrode configuration and b single fuel cell to evaluate the ORR activity of PGM and PGM-free catalysts. Ability to reflect the conditions of the catalysts in three-electrode systems and fuel cells for c LT-PEMFC and d HT-PEMFC applications

As summarized in Fig. 17d, the viscous acid environment causes complex challenges for HT-PEMFCs, leading to higher catalytic loadings yet lower current densities than in LT-PEMFCs. This indicates that HT-PEMFCs currently suffer from a weaker ability to reflect the inherent catalytic activity and a poorer active site utilization than LT-PEMFC. A few groups have designed half-cells enabling high-loading gas diffusion electrodes (GDEs) at ambient temperature in liquid electrolytes [245,246,247] and in concentrated PA and HT environments [248,249,250,251] to assess catalysts at higher current densities. Such innovative configurations should be more heavily used by researchers working on catalysts for HT-PEMFCs to evaluate the catalyst performance at elevated temperatures in concentrated acids. Thus, the development of catalysts for HT-PEMFCs should be systematically evaluated in a practical HT-PEMFC to reflect the real conditions. Furthermore, the structure of the catalyst layer determines the catalytic site utilization. The following section discusses how to optimally design the membrane electrode assembly for HT-PEMFCs, with a special focus on optimizing the active site utilization and creating more efficient triple-phase boundaries to boost the expression of catalytic sites through improvement in the catalyst layer and MPL properties.

4.1 Preparation, Assembly, and Activation of HT-PEMFCs

4.1.1 Acid Doping

A high-quality membrane electrode assembly should have a high membrane conductivity, a high proton conductivity, and subsequently high performance. This can be achieved via an efficient PA doping process, optimized contact between the membrane and electrode to allow excellent proton conductivity, and optimized distribution of the electrolyte in both the membrane and electrode. Doping the viscous acid within the membrane electrode assembly constitutes a crucial step of HT-PEMFC preparation. The acid doping temperature and duration heavily impact the membrane proton conductivity and interactions [252, 253]. Careful consideration of the catalyst layer substrate and acid doping substrate is necessary, as both heavily impact the quality of the electrolyte/electrode interaction interface and performance. To date, doping PA on the membrane and depositing the catalyst layer on a GDL provide the highest power densities and current densities (Fig. 18) [254, 255]. Finally, physically removing the excess acid before operation (dabbing, wiping) improves the HT-PEMFC performance and stability by mitigating the formation of phosphate compounds within the flow fields and endplates [256].

Fig. 18

Preparation and conditioning of HT-PEMFC membrane electrode assemblies with gradual acid diffusion. The relative sizes and distances are not to scale, and the catalyst layer, MPL, and carbon fibers possess significantly different porosities and sizes

4.1.2 Membrane Electrode Assembly Fabrication

Once the PA is suitably doped within the membrane/electrode, binding the electrode to the membrane is necessary. LT-PEMFCs are typically hot-pressed at approximately the membrane glass transition temperature (127 °C) to bind the electrodes to the membrane and create a membrane electrode assembly [257]. However, this step may be redundant, as the heat of the LT-PEMFC during operation equally binds the materials together [258]. An important clarification is proposed here for the hot-pressing process of HT-PEMFCs. As PBI membranes are thermally stable up to 400 °C, hot pressing (130–200 °C) does not create a physical bond between the membrane and the electrode [259]. While heat and pressure might accelerate acid diffusion from the membrane to the electrode, this process might expel too much free-flowing acid and negatively affect the performance [260, 261]. Nevertheless, as the membrane mechanical strength can vary and decreases with increased doping amount, polysulfone sheets can strengthen the weakest edges of the membrane, in which case a hot-pressing step has been reported to bind these sheets to the membrane (200 °C) [262]. Recently, a new membrane electrode assembly fabrication method has been proposed by depositing a thin layer of PBI (35 µm) onto the GDE anode [263]. This approach doubles the power density and drastically reduces the ohmic resistance, highlighting that more innovative methods should be considered in the future.

4.2 PA Distribution in the Catalyst Layer

The structure and properties of the catalyst layer are crucial for diffusion of the protons from the membrane to the active sites and the gases from the GDL to the active sites. While the majority of the studies discussed here focus on PA as the viscous acid used, readers are encouraged to draw analogies for other viscous proton conductors, such as protic ionic liquids, while their properties may differ [43]. Although improvements in the catalyst utilization of HT-PEMFCs have been observed through better catalyst layer designs, the electrochemical performance and active site utilization remain lower than those in LT-PEMFCs, with only a minority of studies reporting reductions in catalyst loading and higher performance (Fig. 19) [264,265,266]. While both LT-PEMFCs and HT-PEMFCs have exhibited improvements in the volumetric power density with lower loadings, insufficient efforts have been dedicated to this area for HT-PEMFCs [264]. Advanced engineering catalyst deposition techniques have shown promise in reducing the catalyst loading and controlling the catalyst layer porosity. For instance, HT-PEMFCs prepared by electrospraying the Pt catalyst layer achieve good performance with loading lower than 0.1 mg_Pt cm⁻² at the cathode, while the electrode suffers from a significantly slower conditioning time (500 h) due to its very high PA diffusion limitations [267]. Ultrasonic spraying allows reduction in the loading from 1.208 to 0.350 mg_Pt cm⁻² and significantly improves the performance due to the enhanced catalyst ink distribution, causing a higher platinum utilization due to the efficient stirring and forced convection during the ultrasonic spraying process [265]. Reactive spray deposition technology, using a one-step jet-flame direct deposition process, also shows potential for loading reduction while suffering from proton and gas transport issues [268]. Although these advanced catalyst deposition techniques can reduce the catalyst loading and control the catalyst layer porosity, these methods are not widely applicable due to their high costs and limited availability [265, 267, 268]. Instead, dedicated approaches to improve the active site utilization via better membrane electrode assembly and GDE design should be investigated, with the main findings highlighted in the following sections.

Fig. 19

Influence of the PGM catalyst loading on the performance of a LT-PEMFCs [264] and b HT-PEMFCs [265,266,267,268]

4.2.1 PA Diffusion in the Electrode

The viscous acid slowly diffuses from the membrane to the catalyst layer to form triple-phase boundaries over the first 50–100 h, or activation period. After the activation, the performance usually stabilizes with a reduction and stabilization of the ohmic resistance [36, 215, 222, 224, 269,270,271,272]. Furthermore, more catalytic sites may become available in the electrodes during the activation period, as excess acid may evaporate/leach out [273]. As activation occurs under load (0.1–0.2 A cm⁻²), faster activation procedures should be investigated to assess the catalytic activity before poisoning/degradation/corrosion occurs. While lowering the catalyst layer hydrophobicity activates the cell faster (low PTFE content/introduction of wetting agents) by enhancing PA diffusion, long-term durability studies are necessary, as this may accelerate acid leaching and reduce the number of triple-phase boundaries [260]. As these two challenges are inherently linked, they have been addressed simultaneously in the literature. Most studies use a viscous acid as the proton conductor. This can cause challenges in the establishment of triple-phase boundaries between the electrode, gas, and viscous electrolyte, as the gas diffusivity in viscous PA is low [274, 275]. Therefore, in this configuration, most triple-phase boundaries form when active sites are in close contact with the free-flowing viscous electrolyte. However, further electrode coverage/flooding will be detrimental to the electrochemical reactions, as gas diffusion limitations will occur. This has been partially mitigated via the addition of additives to the viscous acid to improve the oxygen permeability by increasing its solubility or diffusivity, as reviewed by Myles [276]. Localized PA accumulation within the catalyst layer pores can cause gas diffusion limitations by hindering gas diffusion to the active sites. This effect has been modeled by Kazdal, describing electrochemically active and inactive active sites depending on their PA coverage. To summarize, only partially covered particles create triple-phase boundaries, while fully immersed particles face gas diffusion limitations and may be inactive [277]. Furthermore, the pore size needs to be accurately controlled within both the catalyst and catalyst layers, as active sites within submicron-sized pores may not be reached by the viscous acid within the catalyst layer [224]. These challenges are summarized in Fig. 20a, highlighting viscous acid emerging from the membrane, diffusing through the catalyst layer, and eventually leaching through the MPL.

Fig. 20

a Schematic of the HT-PEMFC electrode with the viscous acid diffusing and leaching from the acid-doped membrane. b to e Catalyst ink binder and filler. b Hydrophobic binder and free-flowing acid. c Acid-doped PBI and free-flowing acid. d Phosphate filler, PBI and free-flowing acid. e Combination of the binder and filler from (b) to (d). The relative sizes and distances are not to scale, and the catalyst layer, binder, filler, MPL, and GDL possess significantly different porosities and sizes

Hydrophobic binders: acid flooding issues have been mitigated via introduction of hydrophobic binders such as polyvinylidene fluoride (PVDF) and polytetrafluoroethylene (PTFE) and accurate control of the pore sizes within the catalyst layer (Fig. 20b) [254, 260, 275, 278,279,280,281]. A more uniform binder distribution using hydrophilic surfactants and a well-controlled ink sintering process (temperature, duration) is crucial, as binder clustering creates hydrophobic cavities with reduced PA content, reducing the active site utilization in these areas [267, 280, 282,283,284,285,286,287,288]. Innovative approaches have been proposed to improve the triple-phase boundary quality throughout the electrode, from relatively crude approaches (hydrophobicity gradients) to more complex approaches (in situ combination of hydrophobicity and electron/proton conductivity). A gradual increase in the hydrophobic binder content (deposition of a catalyst ink with increasing hydrophobicity) from the MPL to the membrane surface improves the electronic conductivity and acid diffusion while increasing the catalyst utilization and alleviating mass transfer issues [289,290,291,292]. More complex approaches involve (i) depositing hydrophobic NPs on carbon black to locally combine electronic conductivity with hydrophobicity, (ii) functionalizing hydrophobic nanospheres (200 nm PVDF) for interaction with PA and conduction of protons, and (iii) increasing the Pt hydrophobicity using pulse deposition [293,294,295].

Proton-conducting fillers: proton-conducting ionomers/metal fillers have been introduced into the catalyst layer to stabilize the electrolyte in solid form and create a better electrode/gas/solid electrolyte triple-phase boundary [88, 296,297,298,299,300]. A PBI dispersion was added to the catalyst ink to create a solid, acid-saturated and proton-conducting ionomer with easy binding and proton transfer between the acid-doped PBI membrane and the catalyst layer (Fig. Figure 20c) [218, 225, 278, 301]. The PBI content impacts the quality of the triple-phase boundary within the catalyst layer. Specifically, high PBI contents reduce the electrocatalytic activity and worsen gas diffusion by covering the catalyst particle, leading to increased oxygen diffusion resistance. In contrast, low PBI contents increase the ohmic resistance and reduce the electrocatalytic activity by reducing the proton pathways to the active sites [302, 303]. PBI swelling during PA doping (up to 200%) can block active site accessibility by causing mass transport limitations, which can be mitigated by the introduction of an inorganic filler (TiO₂) [300]. Conductive polymer binders composed of phosphate groups with methyl phosphonic acid (PPO-MPA)/6-oxohexyl phosphonic acid (PPO-HPA) have been introduced to HT-PEMFC catalyst layers. PA has been replaced with polymeric acids such as poly(vinylphosphonic acid) (PVPA) in the catalyst layer and the membrane, as this compound easily binds to PBI via acid–base reactions and provides impressive performance and durability [304]. Quaternary ammonium (QA) bisphosphate has been used as a solid polymer electrolyte within electrodes [94]. Other fillers have been introduced to create proton-conducting pathways through phosphate formation. Proton-conducting fillers using CsHSO₄, metal pyrophosphates such as zirconium hydrogen phosphate Zr(HPO₄)₂, Al₂O₃ to form aluminum triphosphate and lanthanum phosphate have been incorporated into the catalyst layer (Fig. 20d) [296,297,298,299,300]. ZrO_x and Ni(0) introduced into the PGM electrode interact with PA to create more active sites [305]. Carbon nanofiber (CNF) composites doped with Zr/Ni/Gd have been proposed to create phosphate metal species on which the electrode material (Pt) is then deposited [306]. These proton-conducting fillers improve the catalyst layer proton conductivity and hinder acid (PA) leaching. Similarly, covalent organic frameworks (COFs) have high PA retention capability and good proton transfer ability [307]. Overall, approaches combining all the improvements achieved in the literature should be explored to optimize the triple-phase boundary (Fig. 20e), while accurate pore control can improve gas diffusion.

Binderless catalyst layer: interestingly, acceptable performance has been reported by using catalyst layers without any binder or filler, creating a PA interface between the membrane and catalyst layer using the capillary force within the catalyst layer [283, 295, 308, 309]. Promising durability was achieved at 0.2 A cm⁻² in HT-PEMFCs [295], with 900 h stability at 0.2 A cm⁻², and the high loading used (0.96 mg_Pt cm⁻²) may have created a barrier to PA leaching. This method was further refined, achieving impressive performance over 3,000 h at 0.2 A cm⁻², with a performance loss of 4% and a voltage decay of 5 μV h⁻¹ in the last 1 500 h. This study used 0.1 mg_Pt cm⁻² loading via electrospraying, which created a uniform catalyst layer with low porosity [309]. Therefore, within binderless electrodes, a triple-phase boundary in the catalyst layer can be established by acid transfer from the acid-doped membrane to the electrodes and can therefore be tailored by using catalysts with varied Pt/C, achieving impressive performance with low-loading cathodes (0.095 mg_Pt cm⁻²) in HT-PEMFCs [310]. Therefore, this promising approach should be investigated in more detail by the research community at lower loadings.

4.2.2 Hindering Acid Leaching

Even with the addition of hydrophobic binders, polymer ionomers, and phosphate fillers, a portion of free-flowing liquid acid remains within the membrane electrode assembly, which cannot be contained/stabilized. Its leaching from the catalyst layer will eventually reduce the number of triple-phase boundaries and reduce the HT-PEMFC performance and durability. Physically containing this acid within the catalyst layer by reducing both the catalyst layer and MPL porosities and the MPL hydrophobicity has been proposed.

Catalyst layer: aside from the efforts summarized previously regarding PA stabilization within the catalyst layer, other studies have focused on suppressing the formation of macropores within the catalyst layer, as these lead to acid build-up and breakthrough from the catalyst layer. As investigated by Halter, build-up of pressure with the catalyst layer pores and cracks as low as 40 mbar (1 bar = 1 × 10⁵ Pa) is sufficient for PA breakthrough from the membrane to the flow fields [311]. This was achieved by blending CNTs into the ink and using advanced deposition techniques, such as ultrasonic spraying, to achieve uniform deposition [312,313,314]. However, excessive reduction in the catalyst layer porosity would reduce the active site utilization by hindering PA diffusion into the depths of the catalyst layer.

MPL: the MPL (the top layer of the GDL, deposited on the carbon fibers) is composed of densely packed carbon black NPs with a hydrophobic binder. While MPLs with wide cracks enhance water management in LT-PEMFCs, these are unsuitable for HT-PEMFCs because they accelerate acid leaching [315, 316]. In contrast, highly hydrophobic, crack-free MPLs with coarse graphitized carbon particles function as a barrier to PA and are widely used for HT-PEMFCs [252, 317,318,319]. Replacing the carbon black NPs of the MPL with materials better suited for the harsh PA environment has been explored [287, 320,321,322]. For instance, carbon nanospheres are resistant to carbon corrosion in the PA environment and increase the thermal and electrochemical stability [287, 320]. Silicon carbide has higher electrochemical and thermal resistance than carbon black, but as its conductivity is lower, it may not be the best candidate [321]. Finally, reticulated polyaniline nanowires, with a fine porous structure and densely packed carbon NPs, improved the performance by 36% by enhancing oxygen diffusion and reducing the resistance [322]. Altogether, improvement in the catalyst layer and GDEs is necessary to sustain sufficient triple-phase boundaries and avoid acid flooding and leaching.

4.2.3 Operando Monitoring of the Acid Distribution

While designing HT-PEMFCs with optimal acid distribution is a challenge, this is worsened by the difficulty in monitoring acid diffusion in real devices. The performance after a long-term stability test (polarization, electrochemical impedance spectroscopy (EIS)) is typically used to evaluate the long-term stability of the new catalyst layer configuration toward enhancing acid diffusion and hindering acid leaching [296,297,298,299,300]. However, performance losses could be caused by catalyst degradation at elevated temperatures [323]. Therefore, while valuable knowledge is obtained using electrochemical techniques, more localized techniques are necessary to elucidate the electrolyte distribution within the electrode. While operando monitoring of LT-PEMFCs has been extensively studied, it has remained somewhat limited for HT-PEMFCs [12]. A few studies have investigated acid doping, diffusion, and leaching in situ and operando, with the main findings summarized below.

Monitoring the acid content in the membrane and electrode: the PA content within the PBI membrane is typically determined by titration or weighing, but these methods lack spatial resolution. Confocal Raman microscopy can evaluate the microscale PA distribution within a PA-doped PBI membrane using the 1572 and 1611 cm⁻¹ peaks as an indicator of the interaction between the AB-PBI host and PA dopant within the membrane [253, 324]. However, PA redistribution occurs under load and is challenging to observe. To date, it has been captured by using a bespoke three-layered membrane embedded with microreference electrodes to measure the PA redistribution in a PBI-based membrane [325]. The transformation of PA into phosphorus acid has been observed by using scanning photoelectron microscopy (SPEM) by simultaneously probing the morphology, surface composition, and chemical state of the catalyst/electrolyte interface [326]. PA anion adsorption on the cathode has been captured by using operando X-ray absorption spectroscopy (XAS) [176]. Furthermore, neutron scattering can monitor the proton mobility within the electrode from atomic to mesoscopic length scales [327]. While these techniques supply valuable insights, more knowledge could be obtained from operando dynamics over time.

Acid diffusion and leaching in HT-PEMFCs: the acid distribution and leaching after operation for thousands of hours have been captured ex operando by using scanning electron microscopy with elemental mapping by identifying the phosphorus distribution [315]. Furthermore, Bevilacqua and Yezerska elucidated the role of the MPL in hindering PA diffusion and the collapse of the catalyst layer following hydrogen starvation using ex operando computed tomography [269, 328]. More advanced operando monitoring requires bespoke systems and new methodologies. For instance, Han detected PA leaching from fuel and oxidant exhausts via spectrophotometric analysis using a color reagent solution (such as molybdenum blue and ascorbic acid) [329]. While this method is insightful and could be broadly applicable for rapid evaluation and monitoring, it lacks localized information. Proton diffusion within operating HT-PEMFCs has been monitored by using quasielastic neutron scattering (QENS) [330]. Boillat and Lin successfully monitored the hydrogen and proton distribution within a large HT-PEMFC using neutron imaging [331, 332]. Furthermore, large heterogeneous current density and heat distributions have been observed by using heat, voltage and current mapping in HT-PEMFCs, which could be attributed to uneven PA distributions over large areas [333, 334]. While acid leaching from bespoke microsized HT-PEMFCs has been observed by using X-ray computed tomography, large-scale information may need to be obtained as well [335, 336]. Lee and coworkers proposed a multipurpose sensor capable of locally monitoring the current density, temperature, pressure, voltage, and flow rate simultaneously for HT-PEMFCs and inserted it at different locations within individual cells (inlet and exhaust) and within stacks [337, 338]. Overall, these large-scale studies highlight that the design and assembly of the cell highly impact PA diffusion and highlight the necessity for further PA monitoring within HT-PEMFC cells and stacks [339].

5 Conclusions and Future Developments

Operating PEMFCs at elevated temperatures is particularly promising because it enables the use of less expensive and more readily available hydrogen fuels, such as reformate hydrogen, syngas, and methanol, faster ORR and HOR kinetics, easier heat and water management, and much higher tolerance for contaminants. The latest advances in PEMs have brought HT-PEMFCs into the spotlight. However, before HT-PEMFCs can truly become commercially viable, key challenges spanning from the HT-PEM to the catalysts, catalyst layer, and membrane electrode assembly properties need to be resolved, with the main findings and most promising research directions thoroughly reviewed here.

HT-PEMs: HT-PEMs with improved acid doping and retention capabilities and conductivities are essential for operations at higher temperature, current density, and gas stoichiometry, all of which demand a better understanding of polymer–acid interactions [340]. New membranes should be developed with high proton conductivity networks, with, for instance, ionic clusters on the membrane surface [341]. In addition, a trade-off may need to be reached between a membrane with intrinsically high proton conductivity and high doping and retention capabilities, as these drive different membrane synthesis approaches [342]. Recently developed cross-linked membranes synthesized based on poly(ethylene imine) (PEI) and poly(ether ketone cardo) should be explored further, as the abundant PEI amino groups provide superior PA absorption capability and conductivity for these membranes [343]. Promising new materials have also emerged that can perform within PEMFCs from − 20 to 200 °C using ultra-microporous Trager’s base (TB)-based membranes or from 80 to 220 °C using QA-bisphosphate ion pairs. These breakthroughs not only open new avenues to operate PEMFCs from lower temperatures but also blur the line between LT- and HT-PEMFCs [98, 344]. Finally, to fully elucidate the HT-PEM performance and durability, long-term stability tests, along with accelerated stress tests, need to be extensively carried out [345]. Efforts to accurately monitor the changes in membrane properties during and after operation should be pursued by, for instance, continuously monitoring the acid leaching rate and membrane thickness loss [346,347,348].

PGM catalysts: PGM catalysts are currently the most broadly investigated and only commercially available catalysts for practical application in HT-PEMFCs. However, PGM catalysts suffer from severe PA poisoning and thermal instabilities in HT-PEMFCs. Efforts should be dedicated to using the proton carrier (PA) in HT-PEMFCs to aid the 4-proton-electron-coupled process by locally supplying protons rather than poisoning the active sites. Therefore, developing PA-promoted PGM catalysts by fine tuning both the PGM NP structures and the local environment at the nanometre scale around the NPs is of great significance. In addition, the high operating temperature and strongly acidic environment catalyze the degradation of carbon-supported PGM catalysts in HT-PEMFCs. To address these challenges in HT-PEMFCs, a combination of multiple strategies to improve the PA resistance and durability should be envisioned. Overall, the development of corrosion-resistant alloy particles and supports as well as spatial nanoconfinement and enhanced phosphate poisoning resistance will offer new opportunities to improve the performance and durability of PGM catalysts in HT-PEMFCs.

PGM-free catalysts: PGM-free catalysts are particularly attractive for replacing PGM catalysts in HT-PEMFCs. In HT-PEMFCs, PGM-free catalysts have increased ORR kinetics, are free of water flooding, and enable easier decomposition of hydrogen peroxide [31, 33, 174, 218]. In addition, PGM-free catalysts are phosphate resistant. Nevertheless, several bottlenecks remain for M–N–C single-atom catalysts in HT-PEMFCs. Strategies to reach a trade-off between the stability and active site density by reducing the thickness of the single-atom cathode catalyst layers without compromising the activity and stability and therefore enhancing the active site utilization are outlined from the findings of this review. To meet the activity and durability challenges of M-N_x single-atom catalysts for the ORR in HT-PEMFCs, novel concepts of designing high-density atomic centers with controlled configurations and coordination environments will bring about new single-atom catalysts with high intrinsic activity. Researchers should focus on single-layer or double-layer high-density bimetallic or multimetallic atomically dispersed catalysts on conductive and corrosion-resistant carbon supports to improve the intrinsic activity and stability and increase the active site density and utilization in HT-PEMFCs.

Catalyst layer: beyond the inherent challenges of the catalyst, an efficient membrane electrode assembly in HT-PEMFCs requires a catalyst layer designed to achieve high utilization of the catalyst active sites, with an optimized structure to build efficient triple-phase boundaries and easy pathways for ion and reactant transport. In HT-PEMFCs, regarding the viscous acid, such as PA or protic ionic liquid, distribution critically influences the HT-PEMFC performance. An uneven distribution leads to PA poisoning and PA flooding, causing poor expression of catalytic sites, while an optimized distribution via in situ triple-phase boundary engineering can enhance the HT-PEMFC performance [178, 349]. Thus, optimizing the catalyst layer structures via innovative design and fabrication by controlling the local environment at the meso-, micro- and nanoscale and constructing pathways at the microscale to realize the full potential of the catalyst materials are of great significance [350]. Innovative techniques should be applied to monitor the electrode configurations and structures to understand the triple-phase boundaries. Altogether, a combination of hydrophobicity and high proton conductivity is necessary for high proton diffusivity and active site utilization within HT-PEMFC catalyst layers and for a highly efficient triple-phase boundary. Finally, while operando characterization of HT-PEMFCs has thus far received limited attention, this area needs to be continuously developed to further the understanding of the complex underlying processes [335]. Inspiration could be drawn from the extensive operando studies conducted on LT-PEMFCs while focusing on the unique HT-PEMFC environment to develop more PA monitoring methodologies [12, 335]. The authors anticipate that operando monitoring of HT-PEMFCs will help drive the development of new catalysts, catalyst layers, electrodes, and membranes and become as prevalent as it currently is for LT-PEMFCs [12]. The enhanced knowledge of the proton transport, catalyst/ionomer distribution, and oxygen transfer will further drive the fabrication of novel electrodes with dedicated channels for proton, gas, and vapor water transport in HT-PEMFCs. For instance, electrodes composed solely of electrospun CNF structures with catalytic sites, acting as both the GDL and catalyst layer, have reached impressive performance and outstanding durability (130 to 1,400 h). Overall, the next breakthrough in HT-PEMFCs will be driven by the combination of advanced operando monitoring techniques with innovative catalysts and catalyst layers [4, 305, 318].

HT-PEMFCs beyond 200 °C: while most studies have focused on increasing the operating temperature of PEMFCs from 80 °C to 120–200 °C, operating PEMFCs above 200 °C will be in principle even more attractive, as it allows utilization of in situ reformed hydrogen without cooling, significantly increased carbon monoxide tolerance, and even faster electrode kinetics. A handful of pioneering studies have ventured into this new direction, facing new challenges. While outstanding PEMFC performance has been reached at 240 °C using poly(2,3,5,6-tetrafluorostyrene-4-phosphonic acid) membranes with a peak power density approaching 1.8 W cm⁻² at 240 °C, their durability is far from satisfactory (10–20 h) [95]. In contrast, new PA/PBI/SiO₂ membranes from Cheng’s group using a phosphate silicate proton conductor operate between 200 and 250 °C for hundreds of hours with a high carbon monoxide content (10%) and achieve 0.237 W cm⁻² at 260 °C in HT-PEMFCs. Considering these outstanding developments, efforts to improve the membrane, catalyst, and catalyst layer need to occur in unison to fulfilling the full potential of PEMFCs at even higher temperatures.