Non-Platinum Group Metal Oxygen Reduction Catalysts for a Hydrogen Fuel Cell Cathode: A Mini-Review

Abstract

Although platinum-based catalysts are highly effective for the oxygen reduction reaction (ORR) in proton exchange membrane fuel cells (PEMFCs), their high cost and scarcity limit large-scale commercialization. As a result, platinum group metal-free catalysts—particularly Fe-N-C materials—have received increasing attention as promising alternatives. Despite significant progress, no platinum-group metal-free (PGM-free) catalyst has yet matched the performance and durability of commercial Pt/C in acidic media. Recent advances in synthesis strategies, however, have led to notable improvements in the activity, stability, and active site density of Fe-N-C catalysts. This review highlights key synthesis approaches, including pyrolysis, MOF-derived templates, and cascade anchoring, and discusses how these methods contribute to improved nitrogen coordination, electronic structure modulation, and active site engineering. The continued refinement of these strategies, alongside improved catalyst screening techniques, is essential for closing the performance gap and enabling the practical deployment of non-PGM catalysts in PEMFC technologies.

1. Introduction

To limit the effect of climate change in the future, it is essential that major changes in sectors producing massive amounts of greenhouse gases are made, with the transportation sector being one of these areas. A total of 98% of the vehicles registered in the US in 2022 were gasoline vehicles [1]. These engines produce a vast amount of air pollution and require gasoline, which is fuel made from non-renewable crude oil and other petroleum liquids [2]. A typical internal combustion-powered vehicle emits about 4.6 metric tons of carbon dioxide per year, and 400 g of CO₂ per mile is emitted by a passenger vehicle [3]. The greenhouse effect is the process where heat is trapped near the Earth’s surface due to greenhouse gases such as carbon dioxide and methane [4]. The greenhouse effect that greenhouse gases produce has led to a rapidly increasing surface temperature, which has caused sea levels to rise, extreme weather patterns, and habitat destruction [5]. Additionally, crude oil reserves are set to run out in approximately 50 years [6]. Carbon dioxide is at the highest levels ever recorded in the atmosphere and is a leading contributor to the climate change crisis. The reduction of emissions that would result from the reduced consumption of petroleum is integral to maintaining the integrity of our environment [7].

A fuel cell electric vehicle (FCEV) operating on hydrogen only emits water vapor and does not directly emit any CO₂ [3]. Since PEMFC-powered vehicles rely solely on hydrogen gas, the production method used to obtain it is the determining factor in the CO₂-equivalent emissions of a PEMFC-powered vehicle [8]. Another limiting factor of PEMFCs is the amount of hydrogen available for use in the transportation sector and the limited infrastructure for refueling stations [9]. Increasing the amount of hydrogen produced annually would allow for the expansion of the emission-free vehicle market. The switch to emission-free vehicles would result in increased vehicle efficiency and reduce the sheer volume of petroleum used in the transportation sector. Competition with internal combustion engine vehicles lies in the cost, accessibility, and durability of PEMFC-powered vehicles [10].

Previous and current generations of commercial FCEVs use Pt/C and Pt-Co/C catalysts, respectively. As of 2021, the commercial Toyota Mirai Pt-Co/C catalyst reported a Pt loading of 0.33 mg_Pt cm⁻² for the cathode [11]. Pt-Co/C catalysts, while highly efficient under optimized conditions, are susceptible to performance losses due to carbon monoxide (CO) poisoning and ionomer interaction issues. Their effectiveness depends heavily on precise control of the Pt shell thickness, crystal structure, and the architecture of the carbon support, which must balance accessibility for reactants with protection against ionomer poisoning. Although innovations like mesoporous carbon supports (e.g., HSC-f) have improved Pt utilization and reduced loading requirements, these systems still face limitations related to CO tolerance and mass transport under certain conditions [12]. This reliance on high Pt loading is especially problematic in high-temperature PEMFCs (HT-PEMFCs), where loadings often exceed 1.0 mg Pt/cm² due to challenges like electrolyte flooding and anion adsorption, which further highlights the need for more CO-tolerant and cost-effective alternatives [13].

In contrast, Fe-N-C catalysts have emerged as a promising class of PGM-free materials for the oxygen reduction reaction (ORR) [14,15]. These catalysts typically feature atomically dispersed Fe-N₄ moieties coordinated within a nitrogen-doped carbon matrix. While they offer notable cost and performance advantages, challenges remain in achieving long-term durability in acidic PEMFC conditions, clearly identifying the active sites, and mitigating structural degradation over time. These structure–performance limitations are discussed in greater detail in the Synthesis Strategies Section of this review. These catalysts exhibit strong intrinsic tolerance to CO poisoning, making them more robust under practical fuel cell operating conditions where impurities are present, including trace CO potentially introduced through reformate use or ambient air exposure in real-world environments [16]. Moreover, Fe-N-C materials offer a significant cost advantage and can be synthesized with tunable porosity and active site distribution, supporting efficient mass transport and catalytic activity without the need for scarce and expensive noble metals [14]. This makes Fe-N-C catalysts an attractive and sustainable alternative to Pt-Co/C systems for future fuel cell applications.

This review provides a focused assessment of Fe-N-C catalysts specifically within the context of FCEV deployment, emphasizing their structural characteristics, synthesis strategies, and tolerance to air-derived impurities like CO. By directly comparing these materials to commercial Pt-based catalysts and highlighting recent advances in Fe-N-C performance under acidic PEMFC conditions, this work aims to bridge a critical gap in the literature and support the development of durable, cost-effective PGM-free alternatives for transportation applications.

2. Fundamental Aspects of Fuel Cells and ORR

Proton exchange membrane fuel cells (PEMFCs) directly convert the chemical energy in hydrogen to electricity with pure water and heat as byproducts. Specifically, a hydrogen fuel cell is an electrochemical cell that uses a spontaneous redox reaction to produce current. The two sandwiched electrodes with an electrolyte, shown in Figure 1, are the anode and the cathode. All potentials reference the Standard Hydrogen Electrode (SHE), which is defined as 0 V under standard conditions (1 atm H₂, 1 M H⁺, 25 °C). The half-cell and general redox reactions are as follows:

$$\begin{array}{c}\mathrm{Anodic} \mathrm{Reaction} (\mathrm{Hydrogen} \mathrm{Oxidation} \mathrm{Reaction})\\ {\mathrm{H}}_2\to 2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{E}_a^o=0\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\\ \mathrm{Cathodic} \mathrm{Reaction} (\mathrm{Oxygen} \mathrm{Reduction} \mathrm{Reaction})\\ {\mathrm{O}}_2+4{\mathrm{H}}^{+}+4{\mathrm{e}}^{-}\to 2{\mathrm{H}}_2\mathrm{O}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{E}_c^o=1.23\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\\ \mathrm{Overall} \mathrm{Reaction}\\ 2{\mathrm{H}}_2+{\mathrm{O}}_2\to 2{\mathrm{H}}_2\mathrm{O}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{E}^o=1.23\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\end{array}$$

where the standard electromotive power is 1.23 V, and $E_a^o$, and $E_c^o$ are the anodic and cathodic standard electromotive force, respectively.

Catalysts facilitate the overall reaction between hydrogen fuel and oxygen from the air to produce water vapor. Without a catalyst, the reaction would not be able to produce a usable amount of electricity [17]. As the cathode has sluggish reaction kinetics and catalyst stability issues, our focus will be on the cathode-side reaction (oxygen reduction reaction) and its catalyst.

Oxygen Reduction Reaction

With the oxygen molecule converting into water with an overall four-electron transfer, it is highly exothermic. However, since 1.23 V is required to enable this reaction, it is hard to enable this reaction with sluggish kinetics without a suitable catalyst. A catalyst is needed to lower the amount of potential required to make this reaction occur. The oxygen reduction reaction can occur in a direct four-electron transfer or a series of two-electron transfers and is shown below for acidic media [17]:

$$\begin{array}{c}\mathrm{Direct} \mathrm{Pathway}\\ {\mathrm{O}}_2+4{\mathrm{H}}^{+}+4{\mathrm{e}}^{-}\to 2{\mathrm{H}}_2\mathrm{O}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{\mathrm{E}}^{\mathrm{o}}=1.23\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\\ \mathrm{Series} \mathrm{Pathway}\\ {\mathrm{O}}_2+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2{\mathrm{O}}_2\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{\mathrm{E}}^{\mathrm{o}}=0.69\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\\ {\mathrm{H}}_2{\mathrm{O}}_2+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to 2{\mathrm{H}}_2\mathrm{O}\hspace{1em}\hspace{1em}\hspace{1em}{\mathrm{E}}^{\mathrm{o}}=1.77\mathrm{V} \left(vs\mathrm{S}\mathrm{H}\mathrm{E}\right)\end{array}$$

The direct pathway is the most favorable as it involves the breaking of the O-O bond of the oxygen molecule and provides the most released free energy. For complete oxygen reduction, the catalyst should be able to generate H₂O from O₂ through the direct pathway with minimal H₂O₂ production. The series pathway discharges nearly half of the free energy compared to the direct pathway because of the high energy required to break the O-O bond. The series pathway introduces harmful free radical species into the electrolyte while also providing lower energy conversion efficiency [17].

3. Synthesis Strategies and Material Design of Non-PGM ORR Catalysts

Expensive platinum-based catalysts are currently used on both electrodes of a proton exchange membrane fuel cell (PEMFC). On top of being expensive and precious, platinum-based catalysts are not considered perfect for both electrodes (the cathode and the anode). The cathode suffers from sluggish ORR kinetics, which leads to declining activity, methanol crossover, and CO poisoning effects [18,19,20]. The cathode reaction involves the reduction of an oxygen molecule and has been a major barrier to the development of economically enticing technology [19]. This poor ORR activity and low stability, coupled with the high price of platinum catalysts, limit the mass commercialization of PEM fuel cells with the current technology. Over the last few decades, much research has been undertaken to find a suitable alternative that offers the same, if not better, integrity and stability as the standard platinum-based catalysts [21]. Finding a suitable alternative to platinum-based catalysts has been a great challenge since no catalyst has currently surpassed platinum-based catalysts in integrity, stability, and oxygen reduction (ORR) performance [22,23]. Many platinum-free metal-based catalysts have been investigated to find a catalyst that can balance between ORR activity, electron conductivity, stability, and O₂ transport [24,25]. The most common categories of PGM-free catalysts for acidic ORR include iron–nitrogen–carbon (Fe-N-C) [26], carbon-based [27], and other molecular catalysts [28,29]. Studies showing Fe-N₄ moieties or Fe-N₄ coordination environments have been shown to have particularly high ORR activity, and well-defined Fe-N bonding sites serve as highly active and stable centers [30,31]. Beyond the density of Fe-N_x sites, their accessibility and exposure at the catalyst surface play a critical role in determining ORR activity. These structures are common across both Fe-N-C materials and molecular macrocycles, offering a strong link between heterogeneous and homogeneous catalyst design [32]. Q. Wang et al. demonstrated that lower catalyst loading led to increased H₂O₂ yield, indicating that inner Fe sites are less utilized, likely due to mass transport or site accessibility limitations [33]. This suggests that catalytic performance may be limited not by the presence of Fe-N_x moieties but by their availability at the reaction interface. In contrast, 2D catalysts, such as g-FePc [28], may suffer less from these limitations due to their atomically thin structure, which could facilitate greater exposure of active sites and more uniform utilization.

3.1. Iron–Nitrogen–Carbon (Fe-N-C) Catalysts

Among the transition metal–nitrogen–carbon catalysts, Fe-N-C catalysts have been shown to be the most promising for the oxygen reduction reaction due to their high activity and economic value [34]. Various synthesis methods of Fe-N-C composites can differ in metal loadings and active site densities. Some promising synthesis approaches include pyrolysis, Fe-doped MOF precursors, wet impregnation, ball-milling, cascade anchoring strategies, co-doping, and others. Fe-N-C synthesis strategies each offer trade-offs in activity, stability, and scalability. Pyrolysis is essential for Fe-N_x site formation, though it can also lead to the generation of inactive Fe species. In MOF-derived approaches, the MOF acts as a template that, upon pyrolysis, yields materials with high surface area and atomic dispersion; however, these methods can be complex and often result in low yields. Wet impregnation is scalable but may suffer from poor site uniformity. Emerging methods like cascade anchoring improve site accessibility but involve multi-step synthesis. Understanding these trade-offs is essential for the rational design of future Fe-N-C catalysts, particularly as the field moves toward optimizing site density, durability, and integration into practical MEA architectures.

Pyrolysis: Pyrolysis is one of the most common approaches for synthesizing Fe-N-C composites. This was employed by Ao et al., as shown in Figure 2a, where they first freeze-dried sodium chloride, dicyandiamide, D(+)-glucosamine hydrochloride, and anhydrous iron (III) chloride in a vacuum before heating to 550 °C for 3 h and annealing at 800 °C for another 2 h in an inert argon gas environment. Ao et al.’s synthesis yielded a Fe/N-functionalized 3D porous carbon network, and ultrafine Fe₄N nanoparticles were grown in the carbon framework during pyrolysis [35]. Another pyrolysis approach in which single Fe sites on N-doped porous carbon nanosheets were synthesized using a confined pyrolysis strategy was employed by Z. Yang et al., as shown in Figure 2b [36]. By utilizing a poly(ether imide) polymer as a carbon precursor and 1,10-phenanthroline (Phen) as a space isolation agent, they were able to promote the full Fe transformation to single Fe atoms without forming Fe nanoparticles. Specifically, cyanuric acid and melamine dispersed in ethanol were filtered and then pyrolyzed at 550 °C for 6 h to form C3N4 nanosheets, which were then freeze-dried and heated to 100 °C for 1 h with ferrous chloride, the poly(ether imide) polymer, and Phen to synthesize the final catalyst [36].

While these two examples are discussed in detail, pyrolysis is also widely used in other synthesis strategies, including MOF-derived and support-assisted methods. MOF-derived catalysts, for instance, utilize metal–organic frameworks, like ZIF-8, as precursors, as mentioned in a separate section [37]. Upon pyrolysis, these frameworks decompose to form nitrogen-doped carbon matrices embedded with atomically dispersed Fe-N₄ sites [38,39]. This method benefits from the uniform distribution of metal centers and the inherent porosity of MOFs, which can enhance mass transport and active site accessibility [40]. Support-assisted strategies, on the other hand, involve impregnating nitrogen-rich carbon supports—such as carbon nanotubes or graphene—with iron precursors, followed by pyrolysis [41,42]. This approach allows for better control over the morphology and electronic properties of the resulting catalysts, as the characteristics of the support material can influence the dispersion and coordination environment of the active Fe sites [43]. Highlighting the distinctions between these pyrolysis-based methods—such as MOF-derived, support-assisted, and polymer-confined strategies—helps illustrate the wide range of structural control and active site engineering possible in Fe-N-C catalyst synthesis.

Cascade Anchoring Strategy: Traditional pyrolysis synthesis offers relatively low metal loadings (<5 at.%). To combat this low metal loading, strategies, such as cascade anchoring, can be employed to increase metal loading up to 12.1 wt.%. This was achieved by L. Zhao et al., who were able to produce a scalable synthesis with high metal-N_x loading. As shown in Figure 3, the metal ions are first chelated using glucose as a chelating agent before being anchored onto high-surface-area, oxygen-species-rich porous carbon supports and pyrolyzed at ~500 °C to allow for maximized protection of the metal ions. From their synthesis, they were able to determine that the chelation of metal ions, physical isolation of the chelate complex, and N-species binding at increased temperatures were key factors in achieving high-metal-loading single-atom catalysts (SACs) [44].

Metal–Organic Frameworks: Hybrid crystalline materials, such as metal–organic frameworks (MOFs), have become some common precursors for metal–nitrogen–carbon complexes. Zeolitic imidazolate frameworks (ZIFs), which are a subclass of MOFs, contain N-filled ligands and tetrahedral coordination metal ions and are a type of coordination polymer. As shown in Figure 4, these ZIFs can be used as sacrificial templates to prepare nitrogen-doped carbon materials due to their nitrogen-rich composition. As employed by D. Zhao et al., metalated ZIFs were synthesized by heating ZnO, N-containing ligands, and tris-1,10-phenanthroline iron(II) perchlorate (TPI) (as seen in Figure 4a) [37]. After pyrolization at 1050 °C, acid washing, and a second 950 °C heat treatment, a final catalyst with a uniformly dispersed iron complex can be formed. Even though there is a lack of porosity in some of the precursors, the pyrolyzed ZIF catalysts are still able to have great surface areas. Ahluwalia et al. took a similar approach using a similar Fe-ZIF precursor and pyrolyzing at 1100 °C [45]. Their synthesis yielded a stable Fe-N-C catalyst with atomically dispersed Fe sites.

As shown in Figure 5, another technique involving ZIFs and MOFs includes ball-milling. Zhan et al. used this approach to create bamboo-like carbon nanotubes (BLCNTs) with dispersed Fe-centered active sites (nitrogen-coordinated). Ball-milling NH₂-MIL-88B (iron source) and ZIF-8 (carbon source) precursors together destroys their crystal structure and uniformly mixes them. Zhan et al. controlled the proportion of precursors and utilized pyrolysis to control the morphology of the bamboo-like carbon nanotubes [46].

While MOFs, such as ZIFs, serve as sacrificial templates, they differ from traditional templates like silica. In silica-based templating, the silica is removed after pyrolysis, usually by acid etching, to generate porosity in the final material [47,48]. For example, Gianola et al. [47] highlighted how different silica removal methods, including acid-free approaches developed by Atanassov’s group, influence catalyst porosity and performance. MOFs, on the other hand, are not removed but instead undergo a transformation during thermal treatment, forming part of the carbon-based framework and contributing directly to the structure and functionality of the resulting catalyst [49].

Secondary Metals Present in Fe-N-C: To change/tune the electronic properties of Fe-N-C catalysts, secondary metals or molecules are introduced into the carbon matrix. G. Yang et al. explored atomically dispersed Mn-N in Fe-N-C [50], and Jin et al. explored porous rod-like Bi-O structures [51]. A quick overview of their synthesis methods is shown in Figure 6 and Figure 7. These will not be discussed in depth as they are similar to the synthesis methods discussed in the pyrolysis and metal–organic framework sections, with the addition of Mn and Bi precursors. In the case of Mn-N incorporation, Yang et al. reported that Mn atoms modulate the electronic environment around the Fe-N₄ active sites, resulting in a downshift of the d-band center of Fe. This downshift is described as electron delocalization from the Fe³⁺ centers, shifting their spin state from low spin (t_2g5 e_g0) to intermediate spin (t_2g4 e_g1), which enhances the overlap between Fe 3d orbitals and the π antibonding orbitals of O₂. Electronic tuning weakens the binding strength of oxygen intermediates, enhancing the overall ORR kinetics and improving catalytic activity and durability [50]. This type of electronic modulation is not unique to Mn; other secondary metals and Fe-N-C systems have also been shown to influence the spin state and d-band center of Fe, offering a versatile strategy to optimize catalytic performance [52].

3.2. Molecular Catalysts on Carbon Supports

Materials such as carbon matrices, carbon nanotubes, and graphene have also been studied independently for their application as cathode catalysts. Pristine graphene is considered a semiconductor with no bandgap due to its sp²-bonded carbon atoms, which limits application in energy storage and catalysis [53,54,55]. Through covalent bonding with graphene, atoms such as fluorine (F), nitrogen (N), and boron (B) can be substitutionally doped into graphene to disrupt the sp² network and cause sp³-defect regions [27]. This allows for the tuning of electronic properties and modification of surface chemistry. Fe-N-C complexes and macrocyclic compounds can be combined with carbon matrices, carbon nanotubes, and graphene to increase ORR activity and overall stability. Combining these materials in their pristine and modified (doped/exchanged) varieties has proven to be beneficial and favorable towards ORR activity and stability. Incorporating Fe-based macrocycles, like FePc, into conductive supports, such as graphene or CNTs, improves site dispersion and electron conductivity, leading to increases in onset potential by up to ~50 mV and 2–3× higher limiting current densities compared to unsupported counterparts [28,56,57,58]. In some cases, onset and half-wave potentials approach those of commercial Pt/C catalysts (e.g., ~0.92 V vs. RHE) [28], with enhanced stability, where over 90% of the initial activity is retained after 10,000 cycles in acidic media [59,60,61]. This approach also enables fine-tuning of the Fe coordination environment through ligand exchange or heteroatom doping of the carbon host, further enhancing ORR kinetics and promoting the desirable four-electron reduction pathway. First, doping of graphene and carbon nanotubes will be discussed before moving on to phthalocyanines and porphyrins that are commonly supported by these carbon-based structures as they prevent aggregation and can supply supplementary electron sites [28].

Doping of Graphene and Carbon Nanotubes: Dopants play a large part in the catalytic activity and electronic properties of carbon-based materials. Atomic-scale substitution can redistribute local strain within a catalyst particle, altering the energetic stability of surface configurations and potentially exposing more catalytically active sites. As shown by Husic et al., even a single dopant atom can trigger solid–solid transitions or shift structural preferences, leading to surface reconstructions that affect site accessibility and stability [62]. As mentioned in a previous section, sp³-defect regions can be generated in the sp² network of graphene by introducing substitutional dopants, such as nitrogen (N) and boron (B). Though other atoms such as phosphorus (P) and sulfur (S) can also be doped into the graphene matrix, they are not as similar in size to carbon as N and B atoms and would disrupt the matrix to a much higher degree and deform the planar nature of graphene [27]. With relatively low doping percentages of <3 at.%, the electronic, magnetic, and optical properties in graphene can be greatly influenced due to the polarization that is created by the differences in electronegativity. Nitrogen and boron are common graphene dopants due to their similarity in size and electronegativity to carbon and have been studied intensively for various applications. Nitrogen-doped graphene has been successfully synthesized using thermal annealing, chemical vapor deposition (CVD), solvothermal synthesis, supercritical reaction, arc discharge, plasma, nitrogen bombardment, and ball-milling processes at doping concentrations between 0.6 and 16 at.% [27,52,63,64,65].

As shown in Figure 8a, nitrogen-doped graphene is made of various types of nitrogen in the graphene matrix, including pyridinic (sp²-hybridized at edge sites), pyrrolic (sp³- hybridized), and graphitic (sp²-hybridized in the graphene matrix). Sun et al. mention that pyridinic and pyrrolic N allow for improved pseudo-capacitance and that graphitic N enhances conductivity (favoring electron transport during the charge/discharge process) [66]. The most common precursors for nitrogen-doped graphene include graphene or graphene oxide and a nitrogen-containing molecule such as NH₃. Boron-doped graphene for the application of ORR is slightly less common but does have a more energetically favorable formation energy of ~5.6 eV/atom (compared to 8.0 eV/atom for nitrogen-doping) [67]. Boron-doped graphene has been successfully synthesized using chemical vapor deposition (CVD), thermal annealing, thermal exfoliation, liquid process, arc discharge, and microwave plasma, and the Wurtz-type reductive coupling reaction processes dopant concentrations between 0.5 and 13.85 at.%. As shown in Figure 8b, two types of stable boron doping are possible in the boron-doped graphene. These include in-plane doping of boron (graphitic), which is sp²-hybridized, and the formation of a “boron silane”, which is where B atoms are located in a π-conjugated system [68]. Borane silane atoms have only been present in one synthesis method (CVD). The most common precursors for boron-doped graphene include graphene oxide and a boron-containing molecule, such as boron powder, BH₃-tetrahydrofuran (BH₃-THF), boron tribromide (BBr₃), and diborane (B₂H₆). The many available synthesis techniques make both N- and B-doped graphene a great candidate for the commercialization of fuel cells.

As shown in Figure 9, there are different types of carbon nanotube supports: single-walled, double-walled, and multi-walled nanotubes. The different types of carbon nanotubes behave differently from one another and provide support to other complexes deposited onto them. Most researchers purchase double- and multi-walled nanotubes and use them as received without any further purification step. However, in oxidative treatments, they can be subjected to 35% HNO₃ under reflux and heating conditions (100 °C) for 12 h. Single-walled carbon nanotubes can be produced using a laser ablation technique and can be purified to remove any amorphous carbon particles, which results in superficial oxidation [71].

Iron Phthalocyanine on Carbon Nanotubes: As shown previously in Figure 9, carbon nanotubes can either be single-walled, double-walled, or multi-walled. However, catalyst preparation tends to be the same no matter the type of carbon nanotube (CNT). Morozan et al. were able to synthesize their catalysts by ultrasonicating a 1:2 w/w ratio of iron phthalocyanine, which was dissolved in THF and CNT powder, and then, the THF was evaporated off. Their proposed chemical representation of a hybrid catalyst for another phthalocyanine they studied is shown in Figure 10a [73]. Govan et al. took a slightly different approach by introducing eight cyano-ligand groups into their phthalocyanine, as shown in Figure 10b, in an attempt to tune their catalyst [74]. Their synthesis also included a physical deposition with sonication to ensure homogeneous dispersion in solution.

Iron Phthalocyanine on Reduced Graphene: Jiang et al. were able to synthesize iron phthalocyanine supported on chemically reduced graphene by first synthesizing graphene sheets through the partial reduction of graphene oxide, following the modified Hummers’ method. Then, by having both iron phthalocyanine and graphene dispersed in N,N-dimethylformamide (DMF), they were able to incorporate iron phthalocyanine into the graphene by stirring for 6 h and sonicating [28]. This catalyst and doped-graphene variants were explored computationally by Helsel et al. and can be seen in Figure 11 [75,76,77].

Iron Polyporphyrin: Polyporphyrins can offer a high density of iron macrocyclic centers that are nitrogen-coordinated. Achieved successfully by Yuan et al., the overall synthesis and polymerization are displayed in Figure 12. After these steps were taken to synthesize the catalyst (PFeTTPP), it was thermally activated by being pyrolyzed from 600 to 1000 °C [29]. The simulated stacking of their catalyst is shown in Figure 12b.

4. Performance Evaluation of Non-PGM Catalysts

To be able to experimentally characterize a catalyst for fuel cell applications, electrochemical measurements and/or fuel cell tests must be conducted. Papers that involve purely computational work use a parameter called overpotential to quantify their catalyst’s ORR activity. Research that involves mainly experimental work mostly utilizes a potentiostat, electrode rotator, and 3/4-electrode electrochemical cell to quantify parameters such as onset and half-wave potentials. If equipped with an available fuel cell, some researchers utilize actual fuel cell tests to obtain a better picture of the stability and practicality in actual applications.

4.1. Electrochemical Characterization Techniques

Electrochemical testing for fuel cell catalysts typically employs a three-electrode configuration using a rotating disk electrode (RDE) or rotating ring–disk electrode (RRDE), with the working electrode used to evaluate the catalyst. The working electrode is often glassy carbon, while RRDE setups include a surrounding ring, commonly platinum. However, for platinum-group metal-free (PGM-free) catalyst studies, platinum should be avoided as a counter electrode due to the risk of contamination through Pt dissolution and redeposition. Instead, graphite rods are preferred because they are chemically inert under typical electrochemical conditions, minimizing the risk of introducing unwanted species that might affect the catalyst’s performance [78]. A reference electrode provides a stable potential (e.g., Ag/AgCl or saturated calomel), and reported values are commonly converted to the reversible hydrogen electrode (RHE) scale for consistency [79]. The three/four electrodes are arranged and then submerged in an acidic (<7 pH) O2-saturated electrolyte solution, which is typically 0.1 M HClO₄ for Fe-N-C catalysts. Fe-N-C catalysts are susceptible to poisoning by hydrochloric acid (HCl) as chloride ions strongly adsorb onto the Fe-N_x active sites, blocking oxygen access and significantly reducing ORR activity [80]. The studied mechanism, ORR, involves reducing oxygen. An O₂-saturated electrolyte solution allows for the reaction to occur without much resistance when certain voltages are applied.

Cyclic voltammetry (CV) is one of the most common electrochemical techniques in studying the reduction and oxidation processes of a particular electrode, catalyst, or species. In essence, cyclic voltammetry is where a range of potentials is swept in forward and reverse directions. CV reveals the cathodic peak potential and the half-wave potential, which can be calculated by taking the middle point between the cathodic and anodic peaks [81]. Linear sweep voltammetry (LSV) is another common electrochemical technique used where the potentials have a single forward sweep. Taking the intersection of two tangents on the non-faradaic zone and faradic zone in an LSV allows for the determination of the onset potential (E_onset) [82]. In order to study the kinetics and mechanisms involved in a glass carbon electrode, a method such as Koutecky–Levich analysis must be used. The analysis includes running multiple LSV sweeps at different rotation rates. This analysis reveals the number of electrons (n) involved in the reaction mechanism, or which ORR pathway the deposited catalyst takes [83]. Another key metric used to understand the activity of a catalyst is the Tafel slope. Tafel slopes help elucidate the kinetics of a catalyst by aiding the determination of the rate-limiting step, understanding its mass transport limitations, and the efficiency of the catalyst [84]. Due to the nature of this review, Tafel slopes will not be discussed, and reaction mechanisms can be deduced via the total number of electrons transferred from the Koutecky–Levich analysis. Half-wave potential alone does not provide a complete picture of ORR performance. Catalyst loading and mass activity (e.g., A mg⁻¹ catalyst or A mg⁻¹ Fe) are also critical metrics, especially when comparing selectivity and minimizing the effect of artifacts, such as peroxide re-reduction in high-loading films. Therefore, where possible, mass activity and loading information are highlighted in this review to provide more accurate benchmarking.

When utilizing a full membrane electrode assembly (MEA) with a prepared catalyst, the peak power density, along with the volumetric and gravimetric current densities, are considered key performance metrics. These values can be measured from the current-voltage polarization curves and power density tests. The values determined from the fuel cell test are more realistic than purely the electrochemical measurements; however, the cost, complexity, and time associated with operating a fuel cell, even in a lab setting, are more expensive than the general electrochemical cell setup used for preliminary catalyst screening [85].

Generally, to test the stability and durability of the studied catalyst on a lab scale, chronoamperometry is used over the course of a few hours to a few days by holding the electrode at a typical working potential for fuel cell operation, and then, current retention is measured over this time frame. It is generally directly compared to a control commercial Pt/C catalyst and can yield promising information regarding a catalyst’s stability [34].

4.2. Comparison of ORR Activity

Table 1. Comparison of loading, onset potential (E_onset), half-wave potential (E_1/2), and number of electrons (n) for various PGM-free catalysts. Loadings denoted with ^ are estimated from the given drop-casted catalyst ink weights/volumes and estimated working electrode areas. Entries in the number of electrons column with * denote that n and %H₂O₂ formed was listed instead if present.

Environment	Catalyst	Loading (mgcat cm−2)	Electrolyte	Eonset (V)	E1/2 (V)	n	Ref.
Acidic	NGr	^ 0.71	0.1 M HClO4	0.84	0.58	* (13.4% H2O2)	[86]
	B, N-Gr	^ 0.71	0.1 M HClO4	0.86	0.61	* (1.8% H2O2)
	Fe, Mn/N-C	0.1	0.1 M HClO4	*	0.804	3.91	[50]
	Fe SAs/N-C	0.25	0.1 M HClO4	0.95	0.798	*	[36]
	PFeTTPP-700	0.4	0.1 M HClO4	0.93	0.73	3.96	[29]
	Fe-N-C-10/1-950	*	0.1 M HClO4	0.96	0.78	~4	[46]
	Zn(mlm)2TPIP	0.4	0.1 M HClO4	0.902	0.76	3.9	[37]
	Zn(elm)2TPIP	0.4	0.1 M HClO4	0.914	0.78	3.8
	Fe/Bi-RNC	^ 1.4	0.5 M H2SO4	0.899	0.97	3.93	[51]
	PmPDA-FeNx/C (950 °C)	0.6	0.1 M H2SO4	0.94	0.82	* (<1% H2O2)	[33]
Alkaline	Fe SAs/N-C	0.25	0.1 M KOH	1.02	0.91	3.91	[36]
	Fe, Mn/N-C	0.1	0.1 M KOH	*	0.928	*	[50]
	FePc/a-MWCNTs	0.185	0.1 M NaOH	0.917	0.817	3.81	[73]
	OCNFePc-CNT	^ 0.2	0.1 M NaOH	0.95	*	3.8	[74]
	FeNCNs-800	0.36	0.1 M KOH	0.985	0.89	3.9	[35]
	Fe-NC SAC	0.6	0.1 M KOH	0.98	0.9	* (<3.5% H2O2)	[44]
	g-FePc	^ 0.113	0.1 M KOH	0.98	0.88	3.96	[28]

shows a table of the catalysts discussed in the previous synthesis method section, along with a few key additions that were not discussed. The catalyst loadings are either reported directly from the reference or were estimated from the given drop-casted catalyst ink weights/volumes and estimated working electrode areas (denoted by ^ in Table 1). In the case of Fe-N-C-10/1-950 (BLCNTs) [46], the drop-casted volume was not specified; therefore, the loading could not be calculated from the given data.

It is important to note that while the focus of this review is on Fe-N-C catalysts for PEMFC cathodes operating in acidic media, studies conducted on alkaline electrolytes, such as KOH, have contributed significantly to the understanding of ORR mechanisms and the development of active site structures. Therefore, select examples from alkaline studies are included for comparison and insight. Mass activity (i_m, A mg⁻¹) should be interpreted alongside catalyst loading (mg cm⁻²) as both directly influence observed current density and selectivity, especially in systems like Fe-N-C, where thick films may mask peroxide production through re-reduction. However, mass activity values were not included in the comparison table, as they were not consistently reported across the reviewed studies and could not be reliably calculated in the absence of complete loading and kinetic current data.

Regardless of the preparation method, Table 1 shows that the main dominating pathway for Fe-N-C catalysts is the direct four-electron pathway, as most electron transfer numbers are around four, and there is minimal hydrogen peroxide (H₂O₂) formation in the catalysts, for which electron transfer numbers were not calculated. With 13.4% H₂O₂ formation, NGr (nitrogen-doped graphene) by itself seems to have a lower electron transfer number than other graphene. This could be due to nitrogen having one more valence electron than the carbon atoms in the graphene. Looking at carbon networks or materials that have iron and nitrogen built into them in a single catalyst, these materials have very promising half-wave and onset potentials. A high onset potential is indicative of a low overpotential required for actual fuel cell operation. In catalysts, such as Z. Yang et al.’s Fe SAs/N-C, where 3.5 wt.% single-iron-atom active sites are dispersed in a porous nitrogen-doped carbon matrix, the onset and half-wave potentials are exceptionally favorable, with the onset potential being over 1 V vs. RHE in alkaline media [36]. This could potentially mean that the space isolation agent they used to create the single-iron-atom active sites (3.5 wt.%) within the carbon network has a larger effect than the increased metal loading found by the cascade anchoring strategy employed by L. Zhao et al. [44]. H. Jin et al. and Z. Yang et al. both modified their Fe-N-C catalysts with a secondary metal (Mn and Bi) to increase ORR activity [36,51]. Though the addition of the secondary metal improved the performance of both catalysts, neither catalyst performed better than the previously mentioned Fe SAs/N-C. There should be further studies conducted to investigate whether manganese or bismuth could be incorporated into the Fe SAs/N-C matrix to improve overpotential. Iron phthalocyanine on reduced graphene, studied by Y. Jiang et al., performed very well, even better than porous polyporphyrin with a high surface area [29]. Iron phthalocyanine is supported by Van der Waals dispersion forces, and it is known that the combination of iron phthalocyanine, supported by the reduced graphene substrate, allows for extra supplementary electron sites to be formed. Additionally, π- π interactions aid in preventing the aggregation of iron phthalocyanine, which aids in conductivity [28]. Experimental research should be conducted to tune the graphene substrate, like substrate doping (with N and B atoms), as mentioned previously, to potentially increase the overall ORR activity of the catalyst. The carbon nanotube (CNT) catalysts discussed in this review either have iron phthalocyanine physically adsorbed on the CNTs through Van der Waals dispersion forces or iron highly dispersed throughout the CNTs. A. Morozan et al. discussed the use of iron phthalocyanine deposited on single-, double-, and multi-walled carbon nanotubes, in which the acid-treated multi-walled CNTs displayed the best performance [73]. When looking at Govan et al.’s similar experiment with iron phthalocyanine that has eight hydrogens substituted with cyano groups deposited on single-walled CNTs, this material offers a higher onset potential than even iron phthalocyanine on multi-walled CNTs (the single-walled CNTs performed worse) [74]. This increase in ORR activity could potentially mean that the extra cyano groups on the ends of the phthalocyanine are aiding the charge transfer to make the reaction easier to complete. Overall, it seems that catalysts that had single iron atoms or Fe-N₄ -coordinated molecules performed better than just uniformly dispersing Fe in a carbon network, even at much higher loading percentages.

Although Fe-N-C catalysts often exhibit excellent ORR activity in alkaline media, their performance typically declines under acidic conditions due to harsher operating environments and proton-rich electrolytes, which lead to lower onset and half-wave potentials. Protonation of pyridinic-N species and carbon support corrosion are the main factors in this activity decline [87]. In acidic media, the commercial Pt/C catalyst has an approximate onset potential of 0.9 V and a half-wave potential of 0.8–0.85 V vs. SHE [88]. Though no Fe-N-C catalyst has yet surpassed Pt/C in terms of activity and stability due to the harsh operating conditions of PEMFCs, the gap in performance for this class of catalysts has narrowed significantly in the last few years. The facile synthesis methods described in this review have enabled greater active site dispersion and the modulation of electronic properties through nitrogen coordination and secondary metal co-doping, thereby enhancing the structural and catalytic performance of Fe-N-C catalysts.

While most Fe-N-C studies focus on half-cell measurements, such as RDE or RRDE, to evaluate intrinsic ORR activity, a smaller number extend testing to full MEA fuel cell configurations. As summarized in

Table 2. A comparison of the H₂/O₂ fuel cell test performance of the studied Fe-N-C catalysts in various media. Peak power densities and volumetric current densities denoted with ^ are estimated from polarization powder density plots from the referenced publications. The current densities marked with ^# denote that the volumetric current density was not given, just the normal current density.

Catalyst	Loading (mgcat cm−2)	Volumetric Current Density (A cm−3)	Peak Power Density (mW cm−2)	Ref.
PFeTTPP-700	4	20.2 @ 0.8 V	730 @ 0.4 V	[29]
Zn(mlm)2TPIP	2.2	67 @ 0.8 V	620 @ 0.43 V	[37]
Zn(elm)2TPIP	2.2	88.1 @ 0.8 V	500 @ 0.34 V
Fe SAs/N-C	1.5	# (^ 1.5 A cm−2 @ 0.89 V)	750 @ ^ 0.5 V	[36]
Fe-N-C-10/1-950	4	# (0.85 A cm−2 @ 0.7 V)	770 @ ^ 0.48 V	[46]

, studies on catalysts, such as PFeTTPP-700, Zn(mlm)₂TPIP, Zn(elm)₂TPIP, Fe SAs/N-C, and Fe-N-C-10/1-950, report polarization curves and peak power densities under H₂-O₂ conditions, with loadings ranging from 1.5 to 4.0 mg_cat cm⁻². These studies also report or enable the estimation of volumetric current densities as a more device-relevant metric that normalizes currents by both geometric area and catalyst layer thickness. For instance, the Fe-N-C-10/1-950 catalyst achieves an estimated volumetric current density of ~0.85 A cm⁻³ at 0.7 V and a peak power density of 770 mW cm⁻² at approximately 0.48 V. While some values are directly reported and others are estimated from polarization plots, this comparison highlights that several Fe-N-C catalysts now approach the performance range of commercial Pt-based systems, particularly when evaluated by space-constrained metrics relevant to practical device integration. For context, leading commercial catalysts, such as Pt/C and Pt-Co/C, have achieved peak power densities of up to 0.84 W cm⁻² at 0.67 V (LANL, using 0.1 mg_Pt cm⁻² loading under H₂/air, 80 °C, 100% RH, 150 kPa) and 0.94–1.0 W cm⁻² with Pt-Co alloy catalysts under similar conditions, as reported by General Motors and other groups. These values reflect the current benchmarks for DOE target-aligned PEMFC performance using precious-metal-based electrodes [89]. While Fe-N-C catalysts still fall short of these absolute performance levels, the gap has narrowed significantly, particularly in terms of volumetric current density and power output per unit loading. These results suggest that with continued improvements in site accessibility, conductivity, and catalyst layer architecture, non-precious metal catalysts are becoming increasingly viable for practical PEMFC applications.

4.3. Future Perspectives and Challenges

Recent advances in Fe-N-C catalysts for hydrogen fuel cells have led to significant progress in multiple areas. These include high oxygen reduction reaction (ORR) activity approaching that of Pt/C (E_1/2 = ~0.8 V), high power density in MEAs (e.g., >0.7 A cm⁻² at 0.7 V), and site densities exceeding 10¹⁹ sites gram⁻¹ [90]. The development of atomically dispersed Fe-N_x active sites and the use of 2D or porous supports, such as g-FePc and BLCNTs, have further improved active site utilization [28,46]. Additionally, volumetric current density has emerged as an increasingly relevant performance metric, reflecting improvements in catalyst layer architecture and practical device integration.

Despite these achievements, several critical challenges remain. The key issues include limited long-term stability under fuel cell operating conditions, acid leaching of Fe sites, and lower performance under H₂/air compared to H₂/O₂ conditions. The ability to accurately characterize which Fe-N_x sites are electrochemically active and how accessible they are in a working electrode remains a major limitation to rational catalyst design. Trade-offs also persist between maximizing site density and ensuring sufficient mass transport, particularly in thicker catalyst layers. In addition, kinetic limitations in O-O bond cleavage and the initial electron transfer step in the ORR continue to hinder performance relative to Pt-based benchmarks. Among these, stability remains the most significant barrier to practical deployment. While a few recent studies, such as Fe-N-C-10/1-950 retaining 70% of its current density after 24 h in an MEA, have shown encouraging durability, most Fe-N-C catalysts still undergo substantial degradation over time [46]. The causes are multifaceted: acid-induced demetalation of Fe-N_x sites, carbon corrosion at high potentials, and mechanical or chemical collapse of the porous structure during prolonged operation. The harsh acidic environment and voltage cycling during fuel cell start-up/shutdown further accelerate these degradation pathways. Moreover, Fe-N_x sites may leach or transform into inactive species, especially in the presence of H₂O₂ intermediates.

Efforts to address these issues, such as stronger Fe anchoring, sacrificial scavengers for reactive oxygen species, and graphitized supports, have shown partial success, but durability still lags far behind that of commercial Pt/C. Moving forward, research should place greater emphasis on in situ diagnostics to track catalyst evolution under working conditions and standardized durability testing protocols that enable fair comparison across studies. Ultimately, achieving a commercially viable performance will require integrated advances in both materials design and electrode engineering, including attention to ionomer distribution, water management, and catalyst layer morphology. An overview of the key challenges and progress made in Fe-N-C catalysts can be found in Figure 13.

5. Conclusions

Fe-N-C catalysts have emerged as the most promising non-precious alternatives to Pt/C for the oxygen reduction reaction in proton exchange membrane fuel cells. While still inferior to the commercial Pt/C and Pt-Co/C catalysts, recent advances in synthesis strategies have allowed for significant activity improvements in the durability and activity of Fe-N-C catalysts in acidic conditions. Pyrolysis, metal–organic framework (MOF)-derived structures, wet impregnation, and cascade anchoring have enabled improved site dispersion, enhanced nitrogen coordination, and the fine-tuning of electronic properties through heteroatom co-doping. Allowing for the increase in active site density and overall catalyst stability, Fe-N-C catalysts are now closer to practical use in fuel cells. Long-term stability and scalable production of these catalysts remain a key challenge in non-PGM fuel cell research. Continued investigation into the relationship between catalyst structure, electronic properties, and performance is necessary to fully bridge the gap to commercial viability. Achieving commercially viable stability will require coordinated progress in both molecular design and electrode-level engineering. Strategies such as graphitized supports, stronger Fe-N bonding, and protective architectures show promise but remain insufficient under realistic MEA operating conditions. Future work should prioritize operando diagnostics and standardized durability protocols to better understand degradation pathways and guide rational improvements.