Advanced approach for active and durable proton exchange membrane fuel cells: Coupling synergistic effects of MNC nanocomposites

2.1 PEMFCs and Acidic ORR Mechanism

PEMFCs convert the chemical energy of hydrogen and oxygen into electrical energy through spontaneous electrochemical reactions, namely the anodic hydrogen oxidation reaction (HOR) and cathodic ORR, both of which require catalysts to reduce their overpotentials (η) and establish efficient reaction pathways. PGMs supported on high-surface-area carbon have been commonly used for this purpose. The use of PGM catalysts, such as Pt NPs and Pt alloys, substantially accelerates the HOR and reduces its kinetic overpotential under the acidic conditions of PEMFCs.⁹ However, despite the use of PGM-based catalysts, the kinetic overpotential associated with the sluggish ORR remains a major contributor to performance losses.¹⁰ For this reason, the related research has focused on developing ORR electrocatalysts and investigating their reaction mechanisms to enhance the kinetics and overall performance of PEMFCs.

As a multi-electron-proton reaction, the ORR features complex elementary steps, which hinders the understanding of the intrinsic ORR mechanism and, hence, the design of PGM-based and PGM-free MNC electrocatalysts. The complete four-electron pathway (Equations 1, 2, and 3) is desirable, whereas the two-electron pathway (Equations 4 and 5) is not, as its final byproduct is the strongly oxidizing H₂O₂, which can destroy the polymer membrane and oxidize the carbon support to cause cell degradation.¹¹

The ORR properties (pathway, mechanism, activity and selectivity) are determined by the oxygen adsorption (Griffiths, Pauling, or Yeager) model and intermediate interaction/sorption energy or dissociation barrier of the catalyst surface.¹² ORR pathways are mainly classified into dissociative (Equation 2) and associative (Equation 3) ones. In the associative pathway, O₂ molecules are adsorbed according to the Pauling and Griffiths models to form MOO structures, with subsequent electron and proton addition affording *OOH. The addition of an extra electron and proton results in the formation of H₂O + *O or (in the case of the two-electron reaction) H₂O₂. In the case of the direct four-electron reaction, *O is reduced to *OH by the addition of a proton and electron, and the reduction of *OH by further proton and electron transfer affords H₂O as the final product. H₂O can also be produced via an indirect four-electron reaction, which involves the two-electron reduction of H₂O₂. In the case of the dissociative pathway, O₂ is adsorbed on the active sites in Yeager mode to form MOOM bridge-on-site structures. Subsequently, the OO bond is broken upon the addition of two protons and electrons to afford 2(M*OH), and the following addition of two protons and electrons affords H₂O (O₂ + 4H + 4e⁻ → 2H₂O). Compared with the associative pathway, the dissociative pathway features rapid one-step four-electron transfer and facilitates the ORR. Traditional close-packed metal surfaces, such as those of Pt and Pt alloy NPs, favor the dissociation of O₂ molecules under the action of two neighboring active metal sites and therefore result in high ORR efficiency and rate.¹²

2.2 Adsorption energy and electronic structures of active sites

The use of appropriate catalytic reaction descriptors allows one to effectively demonstrate activity, durability, and selectivity trends and establish catalyst design guidelines.¹³ In particular, the energy of oxygen atom and intermediate adsorption on the catalyst surface are used to evaluate catalyst efficiency.

Experimental and density functional theory (DFT) studies have shown that ORR activity is determined by specific energy changes corresponding to the protonation of oxygenated species in each step. The step with the highest increase in energy, forming high energy barriers and overpotentials, is the rate-determining step (RDS), governing the reaction rate in its entirety. To achieve high catalytic activity, one should decrease the energy barrier and overpotential of the RDS.¹⁴ The Sabatier principle explains the relationship between catalytic activity, catalytically active sites, and intermediate adsorption energy in heterogeneous catalytic reactions, such as the ORR.¹⁵ Overly high adsorption energy can hinder the removal of adsorbed oxygenated species occupying the active sites and thus interfere with subsequent steps, whereas overly low adsorption energy disfavor the adsorption of reactants, such as O₂ molecules, and lead to the formation of H₂O₂ by favoring two-electron reduction over four-electron reduction. The electronic structure descriptors of active sites, such as d-band center energy and spin state, are widely used to explain structure–ORR activity relationships. The variation in the adsorption energy of oxygenated species arises from changes in the electronic structure of catalytic sites due to those in their geometric structure. Therefore, by modifying the environment of catalytically active metal sites (e.g., through strain, ligands, defects, and heteroatom doping), one can alter their charge distribution and spin state and thus enhance activity.¹⁶

2.3 PGM-based ORR catalysts

Pt-based electrocatalysts, comprising Pt NPs loaded on porous carbon supports with high electrochemically active surface areas, exhibit outstanding stability and superior activity but are not widely used because of the scarcity and high cost of Pt.^{4, 5, 24} To meet the cost target of $35 kW⁻¹ required for the global commercialization of fuel cell systems, the US Department of Energy (DOE) has set some technical goals for PGM catalysts to be realized in 2025, aiming for a mass activity of 0.44 A mg_PGM⁻¹ and total PGM content of <0.1 g kW⁻¹ in PEMFCs, with the recommended electrode loading lying below 0.125 mg cm⁻².²⁵ Accordingly, the development of catalysts based on Pt and other PGMs primarily aims to diminish PGM content while boosting the intrinsic activity and durability of ORR active sites. Pt-based catalysts featuring modified morphologies, alloys, and intermetallic structures have been engineered to date. Pt-based alloy and intermetallic catalysts are particularly promising, as they allow one to substantially decrease Pt loading by incorporating other transition metals and thus increase mass activity.²⁶ These catalysts are also intrinsically more active than their Pt counterparts because of the complex metal–Pt interactions. The use of Pt alloys or intermetallic catalysts provides electronic and geometrical synergies and thus enables the alteration of critical reaction descriptors, such as the d-band center energy, adsorption energy, and reaction energy barrier.²⁷ By adjusting the types and contents of transition metals in the employed alloy, one can fine-tune the effects of strain and ligands on Pt active sites. The engineering of lattice mismatch–induced strain helps adjust the adsorption/desorption energy and thus influence catalytic performance. Compressive strain decreases interatomic spacings, broadening the d-band and weakening adsorbate–surface interactions, while tensile strain has the opposite effect. Theoretical research indicates that the strong binding of Pt to oxygenated intermediates makes compressive strain beneficial for enhancing the catalytic performance of Pt active sites.²⁸ The ligand effect occurs at the interfaces between atoms with different electronegativity and d-band centers, facilitating electronic charge transfer.²⁹ Ligand and strain effects are entwined during the alloying of Pt with other transition metals. Despite the favorable performance of Pt alloy and intermetallic catalysts, their stability is compromised by the corrosion of carbon supports and NP dissolution/agglomeration.²⁹ Further research is needed to address these challenges and fully exploit the potential of Pt-based catalysts for practical use in fuel cells and electrochemical devices.

2.4 Atomically dispersed MNC catalysts

The need to reduce the cost and facilitate the large-scale application of PEMFC cathode catalysts has inspired extensive research on PGM-free and inexpensive transition metal–based catalysts.³⁰ Among the various PGM-free alternatives, MNC catalysts have drawn considerable attention because of their promising acidic ORR performance, comprising individual metal-atom active sites within porous nitrogen-doped carbon supports. These catalysts are prepared by high-temperature annealing, which results in the carbonization and nitrogen doping of the organic precursors to form a carbonaceous matrix with embedded metal atoms. Nitrogen atoms act as coordinating ligands for the embedded metal atoms, preventing their aggregation and promoting their atomic-scale dispersion within the carbon matrix. The synthesis method, precursor species, pyrolysis temperature, and metal precursor concentration strongly influence the composition, morphology, and performance of the resulting MNC catalysts.³¹ PGM-free MNC catalysts offer the benefits of high theoretical metal-atom utilization efficiencies and low production costs. Additionally, the electronic structures of MNC catalysts can be tailored to enhance catalytic activity and selectivity.³² The incorporation of heteroatoms, such as nitrogen, sulfur, or phosphorus, into the local coordination environment of metal active sites can modify their geometric and electronic structures and enhance their activity by optimizing the intermediate adsorption/desorption energy and changing the reaction pathway.^{33, 34} Among the MNC catalysts, those based on Fe and Co promote the acidic ORR. However, given that CoNC catalysts promote the undesired two-electron reduction because of weak oxygen adsorption and therefore suffer from the generation of H₂O₂, the related research has extensively focused on FeNC catalysts, which have a relatively low selectivity for H₂O₂ in the direct four-electron reaction and high ORR activity.³⁵ Various methods for refining the electronic structure of active sites and adjusting the catalyst morphology and pore structure have been used to obtain catalysts suitable for practical PEMFCs.³⁶ Despite the substantial progress in and potential of this field, the development and commercialization of atomically dispersed FeNC catalysts for PEMFCs remain challenging. For example, although these catalysts demonstrate performance comparable with that of commercial Pt/C in half-cell tests, their intrinsic activity remains lower than that of Pt/C. Additionally, FeN_x sites have been synthesized using low metal loadings to prevent the formation of agglomerated compounds, such as metals and oxides, which can affect the electronic structure of active sites. The low intrinsic activity and metal loading of FeNC catalysts necessitate the use of substantial catalyst loadings (>2 mg cm⁻²) to achieve high performance in practical PEMFC MEAs, which results in higher mass transport resistance and lower fuel cell performance compared with those observed for Pt/C.³⁷ Consequently, efforts are being made not only to enhance intrinsic activity by controlling the local coordination environment of active sites but also to increase the number of active sites participating in the reaction by developing synthesis methods compatible with high loadings (>5 wt%) and adjusting the pore structure to enhance the accessibility of active sites.³⁸ Moreover, FeNC catalysts suffer from low durability and rapid degradation in acidic environments because of carbon corrosion, demetallation, and nitrogen protonation.³⁹ The Fenton-like reaction between Fe³⁺ active sites and H₂O₂, a byproduct of the two-electron ORR pathway, generates reactive oxygen species (ROS), such as OH^˙, leading to the marked corrosion and oxidation of the carbon support and thus inducing carbon framework disintegration, pore structure alteration, and active site agglomeration/loss. Carbon support corrosion not only increases mass transport resistance but also reduces electron conductivity. Moreover, the collapse of the carbon support around the FeN_x active sites indirectly facilitates their demetallation. Under acidic ORR conditions, the coordinating nitrogen atoms of the FeN_x active sites are protonated to form NH bonds. This protonation favors FeN bond cleavage, leading to the detachment of the metal center from the nitrogen coordination sites and subsequent leaching during the ORR.⁴⁰ The durability of FeNC catalysts has been increased by developing various forms of FeN_x active sites with geometric and electronic structure adjustments and thus promoting the four-electron ORR and suppressing H₂O₂ generation. Additionally, the introduction of radical scavengers can be used to eliminate ROS.⁴¹

3.1 Classification of MNC nanocomposite catalysts

The interactions between single-atom active sites and diverse compounds (single atoms, ACs, NPs) in MNC nanocomposite catalysts exert synergistic effects on various electrochemical reactions.⁷ Recently, various MNC nanocomposites have been investigated and tested in the cathodes of PEMFC MEAs, demonstrating high performance (Tables 1 and 2). Based on the configurations of their metal atoms, MNC nanocomposite catalysts can be grouped into three categories. The first category encompasses dual-atom catalysts (DACs, M₁M₂NC), which feature two types of single-atom metal active sites and exhibit two active site arrangements (separated individual metal sites without bonding and bonded configurations defined by the bonding mode between the two metals (MM)). DACs exhibit apparent synergy through the interatomic modulation of the geometric and electronic structures of their single-atom active sites. The intermediate adsorption energy and reactant binding modes of the active sites can be optimized by modulating the electronic structure of DACs to favor the desired reaction pathway. Current research on DACs focuses on the use of ternary and high-entropy single-atom strategies to increase the number of active sites by incorporating additional metal species alongside dual-metal active sites.⁵⁸ Furthermore, this approach enables the fine-tuning of catalyst performance through heteroatom doping in the vicinity of active sites to enhance intrinsic activity and durability.⁵⁹

The second category of MNC nanocomposite catalysts are those combining single-atom active sites in MNC with ACs and NPs. The strong electronic coupling between single atoms and other nanocomposite sites enhances the ORR kinetics by lowering the energy barrier for the reduction of oxygen molecules and oxygenated intermediates to afford an efficient reaction pathway.⁶⁰

Additionally, the synergistic effects of coupling can extend beyond catalytic activity. By adjusting the size of NPs and ACs, as well as the metal type and distance from single atoms, one can regulate the interaction between the single atoms and NPs/ACs. Moreover, the charge distribution alterations due to this interaction can impact the length and strength of MN bonds in MN_x active sites to inhibit demetallation, a major contributor to the low durability of MNC catalysts.⁶¹ Furthermore, the coupling of MNC catalysts with metal compounds capable of removing ROS, which can be generated via the incomplete MNC-promoted two-electron reduction and Fenton-like reactions, may prevent catalyst performance degradation resulting from carbon oxidation and demetallation.⁶²

As mentioned above, the scarcity and high cost of Pt hinder the commercialization of PEMFCs. Although atomically dispersed MNC catalysts are promising alternatives to Pt-based ones for the acidic ORR, the former are inferior to the latter in terms of activity and stability under practical PEMFC conditions. The drawbacks of both MNC and Pt-based catalysts can be addressed through the fabrication of MNC-supported Pt-based electrocatalysts. Thus, MNC-supported low-Pt catalysts are the third category of MNC nanocomposite catalysts. Similar to NP/AC-containing MNC nanocomposite catalysts, MNC-supported low-Pt catalysts prevent the dissolution of Pt and demetallation of single-atom sites through the strong electron–metal support interaction involving Pt-based catalysts and MN_x single-atom sites in MNC supports. Furthermore, this interaction enables a more delicate tuning of the adsorption energy of oxygenated species and intrinsic ORR activity of Pt and/or MN_x active sites. Both Pt-based catalysts and MN_x active sites demonstrate ORR activity under acidic conditions, which suggests that ORR performance can be enhanced by increasing the number of active sites and coupling nanocomposite and electronic structure modification under practical PEMFC conditions even at very low Pt loadings.⁵⁴ The design and synthesis of MNC nanocomposite catalysts, which exhibit high ORR performances while featuring low PGM contents, is expected to help meet the DOE's 2025 technical targets for the global commercialization of PEMFCs.

3.2 Dual-atom active sites with MNC catalysts

DACs can potentially overcome the limitations of single-atom catalysts (SACs) and are characterized by the presence of two adjacent metal atoms as dispersed active sites, thus offering potential for new catalytic mechanisms and enhanced structural tunability compared with SACs.⁶³

3.2.1 Bonded DACs

Bonded DACs feature a catalyst matrix comprising two distinct single-atom metal active sites chemically bonded to each other, enabling the creation of metal–metal interactions exhibiting electronic structure coupling effects. Huang et al. synthesized a dual-site FeMn catalyst featuring FeN₄MnN₃ moieties, showing that the formation of FeMn bonds induced a strain effect substantially optimizing the FeMnNC electronic structure.⁴² The electronic structure change of Fe induced by the incorporation of MnN₃ weakened the adsorption of oxygenated intermediates and thus facilitated the desorption of *OH. Moreover, the FeMn dual site cooperatively adsorbed *OOH to almost entirely suppress the two-electron ORR and achieve a low yield of H₂O₂ (Figure 1A,B). Considering the excellent ORR activity of FeMnNC and its low H₂O₂ yield, excellent performance was confirmed at the MEA level. Under H₂O₂ conditions, the power peak density of an FeMnNC-containing PEMFC reached 1.048 W cm⁻² at 2.9 A cm⁻², and the current density at 0.65 V reached 0.95 A cm⁻². Zhu et al. prepared FeMo sites coordinated by a nitrogen-doped carbon support (FeMoNC) using a general host–guest strategy.⁶⁴ Forming the FeMoN₆ active moiety induced a downward shift in the d-orbital center of Fe. Two different adjacent active site configurations resulted in different reaction mechanisms for the bridge-cis adsorption of oxygen molecules. This adsorption induced the elongation of the OO bond and promoted its cleavage, thus considerably increasing the ORR efficiency (Figure 1C). As the adsorption method changed, the OO bond of the FeMoN₆OH site became more stretched than that of the FeN₄ site, which favored the *OOH + H + e⁻ → *O + H₂O step (Figure 1D). As a result, the achieved ORR activity was comparable with that of commercial Pt/C catalysts in both half-cell and PEMFC MEA tests.

3.3 Coupling of NPs/ACs with MNC nanocomposite catalysts

3.3.1 NPs/ACs-modified MNC nanocomposite catalysts

The synthesis of MNC catalysts involves high-temperature annealing, which can reduce catalyst performance by inducing the aggregation of metal atoms and formation of inactive clusters and NPs to decrease the density of single-atom active sites.⁷¹ However, regarding the electronic structure of single-atom active sites, interactions with ACs or NPs result in synergistic activity and durability enhancement. Therefore, numerous works have focused on designing ORR catalysts by incorporating NPs/ACs without notably impacting the density of single-atom active sites. Despite these efforts, research has been predominantly conducted under limited alkaline conditions because of the instability and dissolution of NPs/ACs (which weakly interact with support materials) under acidic conditions.⁷² However, recent studies have tried to harness this synergy even under the acidic conditions present in PEMFCs.

NPs–modified MNC nanocomposite catalysts

The encapsulation of NPs by MNC catalysts can not only change the geometric structure of the latter, for example, promote the formation of concave surface structures, but also stabilize NPs under acidic conditions. Furthermore, the hybrids of metal compounds and MNC catalysts exhibit reaction mechanisms different from those observed for MNC catalysts alone. Cheng et al. utilized metalorganic gaseous doping based on two-step pyrolysis to create Co_NP and single-atom (SA) CoN₄ composite sites and effectively increase the total CoN₄ site density (Figure 2A).⁴⁷ The d-(Co_NP/Co_SANC) composite catalyst, featuring Co_NP encapsulated by a Co_SANC shell, exhibited a certain degree of local curvature and distortion at the outer layers. Theoretical studies suggested that both the Co_NPCoN₄ composite and deformed CoN₄ sites induced the side-on adsorption of O₂ molecules and decreased the energy barrier for the dissociative pathway (Figure 2B,C). Hence, the Co_NPCoN₄ sites inhibited H₂O₂ formation and efficiently promoted the direct four-electron pathway to improve ORR activity and catalyst durability. As a result, d-(Co_NP/Co_SANC) exhibited an outstanding PEMFC peak power density of 1.207 W cm⁻². Pan et al. employed an in-situ doping–adsorption phosphatization strategy to realize a well-defined coupling of atomically dispersed FeN₄ and FeP sites in Fe_SA-Fe₂P NPs anchored on N, P codoped carbon frame (NPCF) catalysts (Fe_SA-Fe₂P_NP/NPCF).⁷³ DFT calculations further demonstrated that the strongly electron-donating doped P atoms increased the energy of Fe 3d orbitals and facilitated O₂ adsorption. The reaction mechanism in acidic media was investigated through x-ray absorption spectroscopy (XAS), in-situ Raman spectroscopy, and attenuated total reflection surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS). The FeP site promoted the formation of a PFeOH* intermediate, the OH group of which could be cleaved to form spillover H species that combined with an N₄FeOH* intermediate, leading to rapid ORR kinetics (Figure 2D).

Radical-scavenging NPs–modified MNC nanocomposite catalysts

When radical scavengers are not within the diffusion distance, catalyst surface oxidation may occur before they are completely consumed because of the high reactivity of ^˙OH and H₂O₂. Hence, radical scavenging sites adjacent to Fe single atoms are necessary for instant ROS scavenging. Cheng et al. confirmed the effectiveness of CeO₂ NPs anchored in locations adjacent to FeN₄ sites (FeNC/Sca^ad-CeO2) for scavenging ROS and chemically decomposing H₂O₂ (Figure 2E).⁷⁴ The strategy of instant radical elimination through the adjacency of the CeO₂ NPs to the FeN₄ sites reduced the survival time of radicals and localized the area of radical-induced damage. Compared with that prepared with FeNC, the fuel cell prepared with FeNC/Sca^ad-CeO2 exhibited a smaller power density decay after 30 000 cycles, as determined by the US DOE–relevant AST (69% vs. 28%, respectively) (Figure 2F,G). Luo et al. designed a similar catalyst structure combining radical-scavenging NPs and single-atom sites.⁷⁵ The authors developed a sol–gel method to synthesize a new aerogel with a TiN-interpenetrated FeNC structure for the ORR in fuel cells. The cell with FeNC/TiN exhibited a peak power density of 0.88 W cm⁻² and voltage loss of 17 mV at 0.80 A m⁻² after 10 000 V cycles during AST. This high performance was attributed to the high radical-scavenging efficiency of TiN and its strong adsorptive interaction with H₂O₂ and free radicals.

ACs–modified MNC nanocomposite catalysts

Coupling with adjacent acid-stable small ACs modulates the electronic structure of single-atom active sites. Chen et al. synthesized a structure featuring Fe single atoms with adjacent Fe-atom clusters by annealing a mixture of Fe-doped zeolitic imidazolate frameworks (ZIFs) and graphitic carbon nitride (gC₃N₄) at a high temperature (900°C).⁴⁸ Few-Fe-atom clusters markedly impacted the spin state of the adjacent Fe single-atom sites and enhanced ORR activity and fuel cell performance. Specifically, the cluster-free Fe_SA@NC catalyst primarily featured a high-spin e_g²t_2g³ Fe(III) configuration with a relatively low splitting energy. However, the introduction of Fe ACs increased the splitting energy and decreased the electron filling in the e_g orbital of the Fe single-atom sites in Fe_SA/AC@NC, resulting in electron rearrangement and an e_g¹t_2g⁵ Fe(II) configuration (Figure 3A,B). This modification of the single-Fe-atom spin state resulted in an optimized energy of oxygenated intermediate sorption. The MEA fabricated with Fe_SA/AC@NC exhibited a peak power density of 0.80 W cm⁻² under H₂O₂ conditions. Xue et al. developed a novel approach involving the construction of Fe_n ACs and satellite single-atom FeN₄ active sites.⁴⁹ The high activity of FeN₄ (S1) sites was achieved through an α-Fe₂O₃ atomization process assisted by the pyrrolic N of polyvinylpyrrolidone. The Fe_n clusters effectively stabilized the FeN₄ (S1) active center, preventing oxidation and inhibiting the protonation of N atoms in the FeN₄ moiety (Figure 3C). This process alleviated the Fenton reaction and demetallation, resulting in enhanced catalyst durability and activity. The H₂O₂ fuel cell with Fe_n@FeN_pyrrC delivered a maximum power density of 0.80 W cm⁻² and demonstrated a higher durability during 10 000 V cycles than the cell with FeN_pyrrC. Wan et al. fabricated a similarly structured FeNC catalyst with Fe clusters or NPs around FeN₄ active sites for the ORR in PEMFCs, demonstrating that the ORR activity of these (Fe_SA/Fe_AC2DNPC) catalysts was superior to that of other FeNC catalysts (Figure 3D).⁵⁰ In particular, the Fe_SA/Fe_AC2DNPC catalysts were characterized by an enhanced intrinsic activity of Fe single-atom active sites and long-term durability. According to DFT calculations, the Fe clusters induced the formation of OH ligands at the FeN₄ sites; hence, the electronic structure modulation of FeN₄ active sites decreased the ORR energy barrier and increased intrinsic activity (Figure 3E,F). A PEMFC MEA with Fe_SA/Fe_AC2DNPC showed a high maximum peak power density of 0.8 W cm⁻² under H₂O₂ conditions (Figure 3G). Furthermore, a decay of only 0.11% after a 150 h chronoamperometry (0.5 V) stability test was observed for Fe_SA/Fe_AC2DNPC, whereas a higher value of 50% was observed for Fe_SA-2DNPC after 100 h (Figure 3H). Fe clusters exerted a pinning effect and promoted incoherent vibrations between Fe_AC and FeN₄, which led to FeN bond shortening and thereby suppressed demetallation and enabled long-term stability.

Heteronuclear NPs/ACs–modified MNC nanocomposite catalysts

The hybridization of two different metal-based ACs with MNC nanocomposites can result in more pronounced synergistic effects and different electronic structures of single-atom sites than the introduction of homonuclear ACs. Liang et al. introduced ultrasmall copper nanoclusters into porous N-doped carbon containing isolated FeN₄ sites (Fe_SACu_NC/NC).⁵¹ The Fe_SACu_NC/NC nanocomposite was synthesized using a two-step vacuum gas diffusion method. Initially, Fe(acac)₃ was in-situ encapsulated in ZIF-8, and the resulting material (Fe(acac)₃@ZIF-8) was subjected to copper precursor (Cu(acac)₂) deposition followed by high-temperature annealing. The thus synthesized Fe_SACu_NC/NC composite exhibited a synergistically enhanced intrinsic ORR activity and durability at all tested pH values. According to DFT calculations, when Cu nanoclusters and Fe single atoms were within a certain distance from each other, the former donated electrons to the latter and thus regulated their spin state. The modified electronic structure of Fe single atoms accelerated electron transfer in the RDS of the ORR under acidic, neutral, and alkaline conditions. The Fe_SACu_NC/NC catalyst was concluded to hold promise for practical electrochemical applications, delivering a peak power density of 0.971 W cm⁻² in a PEMFC operated under H₂O₂ conditions. Han et al. fabricated a Co₄@/Fe₁@NC catalyst through a preconstraining strategy using Co₄ molecular clusters and Fe(acac)₃-implanted ZIF-8 (Figure 4A).⁵² This synthesis ensured the controllable preparation of atomically dispersed Co₄/Fe₁ and Fe₄/Fe₁ polymetallic centers. The combination of Co₄ or Fe₄ ACs with Fe SACs substantially optimized the free energy of the oxygen-containing intermediate states of the FeN₄ configuration (Figure 4B), leading to improved intrinsic activity and PEMFC performance (peak power density = 0.84 W cm⁻²) (Figure 4C).

Precious metal NPs/ACs with MNC nanocomposite catalysts

Pt-based catalysts have long been regarded as ideal for the ORR. However, non-Pt precious metals, including Ru, Ir, Rh, Pd, and Au, have electronic structures and d-band center energy different from those of Pt. These differences result in varying adsorption/desorption energy of oxygenated intermediates and inferior ORR activity compared with those of Pt-based catalysts. Similar to transition metal ACs and NPs, precious metal clusters exhibit synergistic effects when coupled with MNC catalysts. Wei et al. synthesized hybrid catalysts comprising single Fe atoms and Pd clusters.⁵³ Although the Pd clusters individually exhibited lower ORR activity than FeNC, the coupling of FeNC with Pd clusters in the FeNC/Pd_NC hybrid markedly enhanced the ORR activity of Fe active sites, resulting in a half-wave potential of 0.87 V in the half-cell test. This enhancement was attributed to the interaction between Pd clusters and single Fe atoms, which induced a transition from low-spin Fe in FeNC to medium-spin Fe in FeNC/Pd_NC (Figure 4D). Generally, FeNC catalysts favor the indirect four-electron associative pathway because of the high energy barrier of O₂ dissociation. In the medium-spin state induced by the Pd NC introduction, Fe(II) adsorbed O₂ in a side-on fashion, thereby favoring the direct four-electron dissociative pathway over the indirect four-electron associative pathway (Figure 4E). The dissociative pathway of FeNC/Pd_NC was confirmed by the detection of different oxygenated intermediates on the catalyst surface (Figure 4F). Furthermore, the partially occupied d_z2 orbitals of single Fe atoms alleviated the strong adsorption of OH*, consequently decreasing the ORR overpotential. The enhanced ORR stability was attributed to the strong metal–support interactions, electronic interactions between Fe and Pd, and low production of H₂O₂, which helped avoid the agglomeration and demetallation of metal sites. In the PEMFC MEA test, FeNC/Pd_NC delivered a peak power density of 0.92 W cm⁻².

Concerning the relationship between the durability of FeNC catalysts and interactions with precious metal NCs, Gao et al. elucidated the durability enhancement mechanism and monitored Fe dissolution using online ICP-MS and DFT calculations.⁷⁶ Metal NPs/ACs were shown to act as electron donors (Figure 4G), increasing the electron density at FeN₄ sites and therefore strengthening the FeN coordination bonds, inhibiting the electrochemical dissolution of Fe, and enhancing catalyst stability (Figure 4H). This mechanism remained consistent across various metal particle types, forms, and contents. Consequently, a stable particle/cluster-modified Fe-based acidic ORR catalyst capable of reliably operating for up to 430 h in direct methanol fuel cells was identified (Figure 4I).

3.3.2 Low-PGM catalysts with MNC supports

With the increasing commercialization of PEMFCs, research has focused on reducing Pt usage and enhancing the intrinsic activity of Pt-based catalysts. However, decreasing the Pt loading of catalysts increases the mass transport resistance of reactants at low voltages and thus hinders efficient oxygen reduction. This problem can be mitigated by hybridizing PGMs with SACs.⁷⁷ The utilization of MNC as supports for PGM NPs addresses the individual weaknesses of these materials through various synergistic effects when applied to practical PEMFCs. The MN_x sites in MNC supports provide additional ORR active sites and enhance the efficiency of Pt utilization. The strong interaction with MNC supports increases the activity and stability of Pt-based catalysts. Additionally, given that PGMs and single-atom active sites have different energy barriers for each step of the four-electron ORR, an efficient and complete four-electron ORR is achieved through a relay or cascade reaction pathway. Furthermore, this synergistic reaction pathway helps prevent carbon oxidation in the MNC support and polymer membrane degradation by hindering H₂O₂ generation.

Chong et al. developed a hybrid nanocomposite catalyst with an ultralow Pt loading (2–3 wt%) consisting of PtCo alloy NPs supported on CoNC, achieving a mass activity of 1.77 A mg_Pt⁻¹ at 0.9 V.⁵⁵ Similarly, Huang et al. fabricated CoNCsupported hybrid nanocomposite catalysts, revealing that the PtCo@CoNC, NCNT, and rGO hybrid (PtCo@CoNC/NTG) substantially outperformed commercial Pt/C because of its multiple (e.g., PtCo, CoN, and CN) active sites.⁷⁸ DFT calculations suggested that H₂O₂ generated at the CoNC sites migrated to the adjacent PtCo sites for the further reduction to water. This synergistic pathway enhanced catalytic efficiency. As a result, PtCo@CoNC/NTG demonstrated a high mass activity (1.52 A mg_Pt⁻¹ at 0.9 V) and durability (1.3% decay after 30 000 V cycles). Xiao et al. synthesized a hybrid nanocomposite ORR catalyst with an ultralow Pt loading (1.7 wt%) and an FeNC support (Figure 5A).⁵⁶ This hybrid catalyst, featuring Pt and Fe single-atom sites and PtFe alloy NPs, achieved a mass activity of 0.77 A mg_Pt⁻¹ at 0.9 V and high durability, retaining 97% of its activity after 100 000 V cycles (Figure 5B). Multiple active sites (PtN₁C₃, FeN₁C₃, and PtFe@Pt) induced synergistic ORR activity despite the low Pt loading, as confirmed by DFT simulations (Figure 5C). The enhanced durability of the hybrid catalyst was attributed to the reduced H₂O₂ formation and resulting mitigation of membrane and ionomer degradation (Figure 5D). Yin et al. prepared FeN₄ active sites electronically coupled with PtFe alloys using vapor deposition, thus obtaining a hybrid catalyst with an ultralow Pt loading (0.64 wt%).⁵⁷ X-ray photoelectron spectroscopy and Fourier-transform extended x-ray absorption fine structure analyses indicated that the oxidation state of Fe in PtFe–FeNC was minimally lower than that in FeNC, and the construction of the PtFe bond suggested strong electronic coupling between the Fe single-atom sites and PtFe alloy (Figure 5E,F). This interaction shifted the Pt d-band center of PtFe NPs, hindering OH adsorption on the Pt surface and enhancing ORR activity. A PtFeFeNC-based PEMFC MEA cathode demonstrated a peak power density of 1.087 W cm⁻² at a Pt loading of 0.012 mg_Pt cm⁻² under H₂O₂ conditions (Figure 5G). After 70 000 potential cycles, the PtFeFeNC catalyst experienced no notable performance degradation (Figure 5H). Zeng et al. explored the synergistic effect between single-atom sites and Pt-based catalysts applied in PEMFCs for heavy-duty vehicles (HDVs), which require high power density and long-term durability.⁸ The authors experimentally and theoretically demonstrated the effects of coupling between Pt and single metal (Ni, Co, Mn)-rich carbon, ZIF-derived carbon, and Ketjenblack as supports, revealing that the integration of Mn_SANC and Pt enhanced interactions and weakened OH adsorption (Figure 5I–K). Among the single metal–rich carbons, Mn_SANC was an ideal support for improving PEMFC activity. Pt and ordered L1₂Pt₃Co NPs on the Mn_SANC support were designed as cathode ORR catalysts, featuring initial mass activity of 0.63 and 0.91 A mg_Pt⁻¹, respectively. After 30 000 AST cycles, the mass activity remained higher than the DOE target (0.44 A mg_Pt⁻¹) under the light duty–vehicle condition (Figure 5L). Additionally, L1₂Pt₃Co@Mn_SA–NC exhibited a minor power density loss and retained 82% of its current density at 0.7 V after 90 000 AST cycles under HDV conditions (at a higher Pt loading of 0.2 mg_Pt cm⁻² and backpressure of 250 kPa_abs) (Figure 5M,N). This catalyst showed minimal particle coarsening after AST, probably because of the strong chemical coupling between the Mn_SANC support and PGM NPs and the physical hosting effect of the curved surface and porous structure of Mn_SANC support. Wei et al. fabricated a high-performance ORR catalyst (Pt_NPMn_SA/C) composed of rich Pt_NPMn_SA pairs.⁷⁹ The interaction between Pt NPs and Mn_SA enhanced OO bond cleavage and the dissociative ORR pathway, resulting in lower H₂O₂ production through electronic polarization. The synergistic interactions between Pt and Mn_SA weakened the PtO interaction, thus accelerating intermediate desorption. Pt_NPMn_SA/C exhibited a high ORR activity, achieving a peak power density of 1.214 W cm⁻² in the MEA cell test. Regarding durability, the low H₂O₂ and radical production, combined with strengthened structural stability, resulted in only a 5% decline in mass activity after 80 000 AST cycles.

4.1 Real-time monitoring of active site dynamics via in-situ/operando XAS

The accurate analysis of the reaction mechanism and changes in the geometric and electronic configurations of the active sites is important for identifying the origin of activity enhancement in nanocomposite catalysts. XAS is a versatile method for studying fuel cell catalysts, measuring the x-ray absorption of a material across a range of photon energy. The obtained spectra comprise two key regions, namely x-ray absorption near edge structure (XANES) and extended x-ray absorption fine structure (EXAFS). Single and multiple scattering approaches are used to analyze oscillations in EXAFS spectra and determine the local geometric structure, including the coordination numbers (CNs) and interatomic distances around the active-site atoms of catalysts.⁸⁰ In the XANES region, core electrons transition to unoccupied states, which results in strong absorption and edges sensitive to the oxidation state of the probed element. Moreover, the type of ligand or local coordination environment of active sites influences the shape of XANES spectra and position of the absorption edge. During the ORR, the electronic and atomic structures of active sites may change. In-situ XAS enables the monitoring of active-site structural changes in real environments and thus contributes to the understanding of catalytic mechanisms and design of robust electrocatalysts with high activity and optimized selectivity. Generally, operando XAS tests are conducted in fluorescence mode using a three-electrode system where a gas diffusion electrode serves as the cathode, while the air working electrode is composed of a catalyst layer, microporous layer, and carbon paper (Figure 6A).⁸¹ Luo et al. investigated the geometric and electronic structures of the active sites and interactions between the two metals in a bimetallic MNC catalyst synthesized by doping a secondary metal (Sn, Co) into FeNC.⁶⁸ Secondary metal–doped FeNC showed a higher ORR activity than FeNC. Operando XAS analyses of Fe, Co, and Sn K-edges were conducted in N₂-saturated electrolytes to investigated structural and reactivity effects. The change in the threshold energy of Fe K-edge XANES spectra in the 0.2–0.9 V_RHE region indicated a Fe(III)/Fe(II) redox transition and the structural modification of the FeN_x site (Figure 6B). The Fe K-edge XANES spectra of FeSnNC and FeNC showed no difference at 0.9 and 0.6 V_RHE (Figure 6C), although a minimal shift to higher energy was observed at 0.4 and 0.2 V_RHE (Figure 6D). This indicated that FeSnNC had a higher average Fe oxidation state than FeNC in the low-potential region. The same behavior was observed for Co, suggesting that Fe sites in secondary metal–doped FeNC had a higher oxidation state than those in FeNC. The Sn and Co K-edge XANES spectra showed no dependence on potential, consistent with results obtained for SnNC and CoNC (Figure 6E,F). After additional NH₃ treatment, the catalysts exhibited higher ORR activity and Fe K-edge oxidation states than FeSnNC and FeCoNC. This indicated that the SnN_x and CoN_x sites did not undergo changes in structure and oxidation state within the ORR potential range and implied that changes in the actual oxidation state of Fe were related to the variations in ORR performance. Consequently, the enhanced intrinsic catalytic activity was related to the higher mean oxidation state of FeN₄ sites in the bimetallic catalysts (Figure 6G). As reported in another paper, the combination of Fe₂P NPs with single-atom FeN₄ sites can result in the formation of atomically coupled FeN₄/FeP active centers (CACs) and thus facilitate synergistic ORR electrocatalysis.⁷³ This synergistic effect arises from the combined catalytic activity of FeP and FeN bonds, resulting in improved ORR activity, particularly in acidic environments. Operando XAS revealed changes in the electronic and atomic structures of CACs during the ORR in O₂-saturated 0.5 M H₂SO₄ and 0.1 M KOH. In 0.5 M H₂SO₄, an upward shift in the Fe K-edge was observed upon a potential increase from 0.4 to 0.8 V in the ORR region, indicating an increase in the oxidation state of Fe centers due to O or OH adsorption (Figure 6H). The results of Fourier transform EXAFS spectrum fitting demonstrated that the changes in the FeP bond and CN had a larger contribution to the increased ORR activity than those in the FeN bond with increasing potential. However, in the non-ORR region at 0.9 V, the CN of FeP remained unchanged (Figure 6I,J). In 0.1 M KOH, in the ORR region at 0.6 and 0.9 V, the operando Fe K-edge XANES spectra showed minimal changes, while the corresponding first-derivative Δμ-XANES spectra exhibited a minimal shift to higher energy as the potential increased from 0.6 to 0.9 V. This shift suggested an increase in the oxidation state of Fe, indicating the adsorption of O or OH on the Fe centers. However, the CN of FeP sites remained nearly unchanged, indicating that these sites were not involved in the alkaline ORR. In conclusion, regardless of electrolyte pH, the FeN₄ sites in FeN₄/FeP CACs act as the primary active centers. Structural analysis via in-situ XAS revealed changes in FeN and FeP bonding and CN. Under acidic conditions, both FeN₄ and FeP were involved in the ORR, with FeP facilitating H species spillover to the N₄FeOH* intermediate, which indicated a novel reaction mechanism for rapid ORR kinetics (Figure 2D). In contrast, under alkaline conditions, FeP remained unchanged, and only FeN₄ experienced an electronic structure change for the optimal adsorption of oxygenated intermediates, which led to high ORR performance.

4.2 In-situ/operando Raman and infrared spectroscopy for ORR mechanism understanding during ORR process

The detection and monitoring of intermediates and products during reactions are crucial for mechanistic studies, particularly for processes involving various oxygenated intermediates, such as the ORR. One of the most effective on-site characterization methods for monitoring oxygenated species is vibrational spectroscopy, including electrochemical Raman and infrared spectroscopy.⁸² In situ/operando Raman spectroscopy enables the monitoring of real-time changes on catalyst surfaces and identification of intermediates during the ORR.

Wei et al. demonstrated that Pd_NC incorporation resulted in a reaction pathway different from that observed for pristine FeNC, as revealed by in-situ Raman spectroscopy, leading to improved ORR performance.⁵³ When voltage was applied, OO stretching peaks corresponding to OOH^* (around 1260 cm⁻¹) and HOOH_ad (around 1396 cm⁻¹) appeared in the in-situ ATR-IR spectra of FeNC (Figure 4F). However, in the case of Pd_NC/FeNC, the intensity of both peaks was markedly reduced, indicating a reduction in the association pathway and the dominance of the dissociation pathway. The dissociation pathway (direct four-electron process) featured a lower OO bond cleavage barrier and faster O₂ dissociation than the association pathway (indirect four-electron process), thereby enhancing ORR kinetics. Consequently, the ORR reaction pathway in Pd_NC/FeNC was altered, which prevented H₂O₂ formation and improved both activity and durability. Liang et al. determined the RDS in various media using operando Raman spectroscopy.⁵¹ In acidic media, a band attributed to OH intermediates (1085 cm⁻¹) was observed for both Fe_SACu_NC/NC and Fe_SA/NC (Figure 7A,B). However, no Raman signal was detected in neutral and alkaline media, which indicated that oxygen-containing intermediates were not involved in the RDS under these conditions. Consequently, in acidic media, the RDS involved the desorption of *OH (*OH + e⁻ + H⁺ → H₂O + *), while in neutral and alkaline media, the RDS was considered to involve oxygen adsorption and electron transfer (* + O₂ + e⁻ → *O₂⁻) (Figure 7C,D).

In-situ Fourier transform infrared spectroscopy enables the real-time tracking of groups formed on the catalyst surface during the ORR, with spectral frequency or intensity corresponding to the types or amounts of homogeneous materials, respectively.⁸⁴ Wei et al. reported that the introduction of Mn_SA facilitated the construction of Pt_NPMn_SA pairs and provided a dissociative pathway that enhanced OO bond cleavage because of a stronger polarization effect.⁷⁹ In-situ ATR- Fourier transform infrared spectroscopy was used to analyze the intermediates formed during the ORR, offering more definitive evidence. The intensity of the *OOH peak of Pt_NP/C markedly increased with the decreasing potential (Figure 7E), which indicated that Pt_NP/C favored the associative pathway to generate *OOH, inevitably producing H₂O₂ as a byproduct. In contrast, Pt_NPMn_SA/C showed the opposite trend, which strongly suggested that the associative pathway was suppressed, and the dissociative pathway was more dominant. By assembling the Pt_NPMn_SA pair with an optimal electronic structure, the pathway was switched in Pt_NPMn_SA/C, preventing the formation of H₂O₂ (Figure 7F).

Compared with other ATR or external reflection techniques, ATR-SEIRAS is more sensitive to the interfacial region. Wang et al. developed a catalyst comprising stable Fe(pyrrolic N)₄ active sites with axial Fe₄C ACs and investigated the related catalytic mechanism.⁸³ The Fe₄C clusters induced the delocalization of the Fe d-orbitals, enhancing the charge distribution toward the N bonds of the Fe(pyrrolic N)₄ sites, which not only optimized the activity of the Fe single-atom sites but also stabilized them. In-situ ATR-SEIRAS was used to elucidate the reaction mechanism and oxygenated intermediates formed on cyanFeNC (with Fe₄C) and FeNC. The two peaks at 1225 and 1380 cm⁻¹ in the ORR potential region were ascribed to the OO stretching mode of surface-adsorbed superoxide (OOH_ad) and the OOH bonding mode of surface-adsorbed H₂O₂ (HOOH_ad), respectively (Figure 7G). The *OOH peak shift for cyanFeNC exceeded that observed for FeNC. The shift of the *OOH peak suggested a weak interaction with intermediates rather than stable chemical bonds, thereby confirming the associative mechanism. Additionally, the *HOOH peak position and intensity changes indicated a strong interaction between the catalysts and H₂O₂. The negligible change in the *HOOH peak position and notable change in the peak intensity were observed for both cyanFeNC and FeNC (Figure 7H). The result obtained for the *HOOH peak indicated that Fe-based clusters and particles on the catalyst surface further adsorbed H₂O₂, which explained the low H₂O₂ yield.

4.3 In-situ/operando analysis for elucidating the demetallation mechanism of MNC catalysts

The long-term stability of electrocatalysts during electrochemical redox reactions is a major challenge. In the case of MNC-based catalysts, the demetallation of single-atom active sites and corrosion of carbon supports are the principal degradation factors. However, the corresponding degradation mechanisms have not been sufficiently precisely analyzed. ICP-MS can detect elements dissolved in the electrolyte down to ppt (part per trillion, 10⁻¹²) levels. Thus, by performing online ICP-MS measurements while altering electrochemical reaction parameters such as time, temperature, and potential one can analyze the demetallation or dissolution behavior of catalysts.

Choi et al. investigated the in-situ demetallation mechanism of FeNC catalysts using online SFC/ICP-MS.⁸⁵ In a 0.1 M HClO₄ solution at room temperature, the real-time ICP-MS signal for the Fe mass released per mass of FeNC catalyst was observed below 0.8 V_RHE, with the release rate peaking at 0.4 V_RHE, indicating that leaching began at relatively low voltages (Figure 8A). Online DEMS analysis combined with SFC was carried out under chronoamperometric conditions, revealing the cause of the oxidation current. The increasing DEMS signals for CO₂ and CO suggested that the substantial demetallation of FeN_x sites is due to the corrosion of carbon producing CO₂ and leads to substantial activity degradation (Figure 8B). Gao et al. monitored Fe dissolution and confirmed inhibition behavior induced by the interaction between metal-cluster Fe single-atom sites.⁷⁶ Fe dissolution from FeNC catalysts and FeNC catalysts with hybrid catalyst precious-metal (Au, Pt) clusters was quantified using online ICP-MS at different potentials (Figure 8C). The combined results of online ICP-MS and XANES spectroscopy showed that the amount of Fe dissolution depended on the Fe oxidation state (Figure 4G,H). This trend indicated e higher integrated crystal orbital hamilton population (ICOHP) values strengthen the FeN bond, resulting in reduced Fe dissolution (Figure 8D). Notably, Pt exhibited a greater inhibition potential than Au. The reduction in Fe dissolution due to FeN bond strengthening exhibited a similar trend across various metal types (Figure 8E).

Bae et al. argued that one should consider the presence of Pt single atoms instead of Pt NPs with FeNC and the effects of Pt single-atom sites on the enhanced stability of FeNC catalysts.⁷⁰ After synthesizing Fe_0.5NCPt with PtN₄ sites, the authors assessed its stability using an online ICP-MS instrument connected to a gas-diffusion electrode–type electrochemical flow cell at 0.6 V_RHE and 353 K under O₂ flow. Although its initial ORR activity was close to that of Fe_0.5NC, Fe_0.5NCPt exhibited a considerably better activity retention during a 2 h potentiostatic hold. The Fe loss of Fe_0.5NC–Pt was about three times less than that observed for Fe_0.5NC under Ar flow, while no discernible Pt dissolution was observed. The stability diagram of Fe_0.5NCPt showed an interrelation of current density (j) loss with a decrease in turnover frequency, suggesting that the enhanced stability resulted from the Pt-induced stabilization of FeN₄ sites. This enhanced stability was confirmed in a PEMFC operation test, with Fe_0.5NCPt exhibiting high stability during 50 h of operation. The postmortem characterization of the Fe_0.5NCPt cathode showed no discernible changes in Pt species distribution, ruling out the formation of highly active Pt nanoclusters or particles. The fundamental origin of enhanced stability was identified as the isolated Pt–induced stabilization of catalytic FeN₄ sites against demetallation. This proof-of-concept study validated the various synthetic approaches used to avoid Fe demetallation, a critical degradation path during PEMFC operation.