Impeding Thermal Atomization Enables Synthesizing Fe₂N Cluster Liganded Single Fe‐Atom Catalyst for Highly Efficient Oxygen Reduction Reaction

Abstract

Anchoring a ligand, such as a functional group or a cluster, on the active metal center is an effective strategy for regulating the electronic structure of single‐atom catalysts (SACs). Herein, we present a nitridation‐induced‐clustering strategy to produce not only SAC Fe‐N₄ in N‐doped carbon nanorods, but also Fe₂N cluster as a ligand anchored on the active Fe site (Fe₂N_nc/Fe₁‐N‐C). Unlike the conventional iron atomization process, the reactive nitridation process can generate thermodynamically stable Fe₂N intermediates by nitriding the initially formed iron oxide, thereby impeding subsequent thermal atomization to fabricate Fe₂N_nc/Fe₁‐N‐C catalysts. Compared to the conventional SAC Fe₁‐N‐C with Fe‐N₄ active sites, the Fe₂N_nc/Fe₁‐N‐C nanorods are more active for oxygen reduction reaction (ORR), yielding a record high half‐wave potential of 0.957 V versus RHE in alkaline condition. The Fe₂N_nc/Fe₁‐N‐C nanorods can be utilized as air‐cathode catalysts for Zn‐air batteries with a charge‐discharge gap of only ∼0.658 V and outstanding cyclability up to 1000 h. Theoretical calculations show that the Fe₂N_nc ligands indeed modified the electronic structures of Fe‐N₄ sites, leading to a lower adsorption energy for the ORR intermediate OH* and facilitating the desorption of OH* and thus higher activity for ORR.

Fe₂N cluster liganded single‐atom Fe‐N‐C catalyst was made via impending the conventional thermal atomization process. The in situ generated Fe₂N clusters can optimate the electronic structures of Fe‐N₄ sites, thereby rendering Fe₂N_nc/Fe₁‐N‐C catalyst a record‐high oxygen reduction activity of 0.957 V versus RHE in alkali condition.

Introduction

Developing highly efficient and low‐cost electrocatalysts for the oxygen reduction reactions (ORRs) are highly desired for the deployment of advanced energy technologies, including fuel cells and metal‐air batteries.^{[ 1} , ² , ^{3 ]} Single‐atom catalysts (SACs) Fe‐N‐C (denoted as Fe₁‐N‐C) with atomic Fe‐N₄ active centers are widely viewed as promising candidates as efficient ORR catalysts to substitute the expensive and precious Pt/C catalysts, due in part to their desirable structural and functional properties, such as unique coordination micro‐environment and electronic structures.^{[ 4} , ⁵ , ^{6 ]} Nevertheless, to achieve high catalytic activity required for practical applications in energy devices, improving the state‐of‐the‐art Fe₁‐N‐C catalysts for ORRs is still required. To this end, under the guidance of Sabatier principle, optimal adsorption of the oxygen intermediates (e.g., O* and OH*) onto the active centers is needed for ORRs. It is known that the unfavorable electron distribution of the symmetric Fe‐N₄ configuration in conventional Fe₁‐N‐C SACs tends to strengthen adsorption of oxygen intermediates, compared to Pt/C, thereby leading to undesirable ORR performance.^{[ 7} , ⁸ , ⁹ , ^{10 ]} Therefore, rational modulation of the electronic structure of the Fe‐ N₄ active site for achieving higher ORR activities has received considerable attention over the past years.

Tailoring the electronic structures of classical Fe‐N₄ configuration by introducing axial coordination ligands and creating dual‐atomic sites has been proven to be an effective strategy to optimize the binding strength with oxygen containing intermediates during the ORRs.^{[ 11} , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ^{24 ]} Several recent studies have shown that strong electronic interactions between Fe‐N₄ and Fe containing clusters can result in favorable synergistic enhancement effect to improve the ORR activity of Fe₁‐N‐C SACs.^{[ 25} , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ^{30 ]} However, few studies were reported on using the multinuclear clusters as ligands for optimizing the ORR activities of Fe₁‐N‐C catalysts.^{[ 31 ]} For example, Shui et al. reported an advanced electrocatalyst with the coexistence of atomic Fe‐N₄ species and Fe clusters, which increased the activity and stability of ORR by increasing the intrinsic activity and reducing the energy barrier.^{[ 32 ]} Liu et al. demonstrated that the non‐ORR‐active NiFe‐LDH nanodots can lower the d‐band center of Fe‐N₄ sites and thus enhance the ORR activity of Fe₁‐N‐C SACs.^{[ 33 ]} In this regard, the synergistic effect of liganded clusters and single atoms not only enhances intrinsic ORR activity of SACs via the electron redistribution, but also optimizes the electron transfer at the active sites. Nevertheless, further in‐depth understanding the synergistic effect and developing new synthetical approaches to produce clusters liganded SACs are still highly demanded.

Recently, the thermal activation strategy has been adopted to fabricate carbon supported SACs, in which the metal oxide intermediates initially formed at low temperature would be redispersed with single metal atoms by forming covalent bonding between central metal and adjacent heteroatoms through thermal atomization.^{[ 6} , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ^{38 ]} Thus, one key point in generating cluster liganded SACs is to suppress the metal atomization process. In this work, using Fe₁‐N‐C as the model SACs, a nitridation‐induced‐clustering strategy was developed to simultaneously implant both Fe₂N clusters and atomic Fe‐N₄ species into N‐doped carbon matrix to produce unique Fe₂N cluster liganded Fe₁‐N‐C (denoted as Fe₂N_nc/Fe₁‐N‐C) nanorods. The experimental observations and theoretical computation reveal that the generation of more thermodynamically stable Fe₂N intermediates can impede the thermal atomization process, ultimately leading to the formation of Fe₂N_nc ligand decorated Fe₁‐N‐C structure. Compared to Fe₁‐N‐C nanorods with same Fe content, the Fe₂N_nc/Fe₁‐N‐C nanorods exhibit significantly enhanced ORR performance, ranking top in reported Fe₁‐N‐C based ORR catalysts in alkaline conditions. As a result, the Fe₂N_nc/Fe₁‐N‐C nanorods were further investigated as cathode catalysts for high‐performance Zn‐air batteries (ZABs) and anion exchange membrane fuel cells (AEMFCs). Theoretical calculations show that the decoration of Fe₂N_nc ligands can modify the electronic structures of atomic Fe‐N₄ sites and lower the energy barrier of OH* desorption, thereby enhancing the ORR performance.

Results and Discussion

Nitridation‐Induced Fabrication of Fe₂N_nc/Fe₁‐N‐C Nanorods

The schematic synthesis process of Fe₂N_nc/Fe₁‐N‐C nanorods is illustrated in Figures 1a and S1. In the first step, zeolitic imidazolate framework‐L (ZIF‐L) nanorods with length of ∼2 µm were fabricated by using a room‐temperature liquid‐phase method (as it is easy to scale up) (see Figures S2a and S2d). Subsequently, the thermodynamically unstable ZIF‐L phase can be converted into ZIF‐8 phase without obvious morphological alteration^{[ 39 ]} verified by the scanning electron microscope (SEM) images (Figure S2b). X‐ray diffraction (XRD) patterns in Figure S3 confirm that the characteristic peaks of ZIF‐L completely disappears and strong characteristic peaks of ZIF‐8 emerges after the ZIF‐phase‐conversion procedure. Transmission electron microscope (TEM) image (Figure S2e) verified the generation of obvious cavities in the ZIF‐8 nanorods. The as‐obtained ZIF‐8 nanorods were subsequently carbonized at 1100 °C for 2 h to fabricate N‐doped carbon nanorods (Figure S4). SEM and TEM images (Figures S2c and S2f) showed that the rod‐like shape was perfectly preserved after high‐temperature carbonization. During the carbonization process, the low boiling point Zn in ZIF‐8 will be reduced and evaporate out through a decarbonization reaction, thus generating defect‐rich N‐doped carbon matrix.^{[ 5} , ⁴⁰ , ^{41 ]} Subsequently, different amounts of anhydrous FeCl₃ were adsorbed on the N‐doped carbon nanorods, which are then subjected to thermal activation in inert argon and active 10% of NH₃ balanced in argon to obtain Fe₁‐N‐C and Fe₂N_nc/Fe₁‐N‐C nanorods, respectively. Note that the generation of nitride intermediates through nitridation treatment is the key step to obtain cluster liganded SACs (Figure 1a).

Figure 1.

Characterizations of Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods. a) Formation process of Fe₁‐N‐C and Fe₂N_nc/Fe₁‐N‐C catalysts in Ar and NH₃ atmosphere, respectively, b) SEM image, c) TEM image, d) HRTEM image (inset is the contrast intensity profile of (002) lattice fringes), e) HAADF‐STEM image of Fe₁‐N‐C nanorods, f, g) HAADF‐STEM image of Fe₂N_nc/Fe₁‐N‐C nanorods showing the co‐existence of both clusters (marked with red rectangles in (d and e) and isolated Fe atoms (marked with yellow circles in (e), h) HAADF‐STEM image and corresponding elemental mapping of C, N, and Fe.

SEM images (Figures 1b and S5) reveal that Fe₂N_nc/Fe₁‐N‐C nanorods are uniform with relative rough surface, which can promote the mass transfer of catalytic reactions. The low‐magnification TEM image (Figure 1c) of Fe₂N_nc/Fe₁‐N‐C nanorods verified that the rod‐like structure is well maintained, and many cavities can be found in nanorods. The high‐resolution TEM (HRTEM) image (Figure 1d) of Fe₂N_nc/Fe₁‐N‐C nanorods shows that the carbon matrix structure is partially graphitized due to the high‐temperature treatment. The lattice spacing of (002) plane is ∼0.389 nm, larger than the theoretical lattice spacing of 0.340 nm in graphite. The aberration‐corrected high‐angle annular dark‐field scanning transmission electron microscopy (AC‐HAADF‐STEM) of the samples obtained in both Ar and NH₃ atmosphere revealed that both atmospheres could implant well‐dispersed Fe atoms into the carbon matrix. The obvious difference is that all the Fe atoms are isolated from each other in Fe₁‐N‐C nanorods (Figure 1e), whereas the reactive nitriding treatment can bring about the coexistence of both clusters and single‐atoms in Fe₂N_nc/Fe₁‐N‐C nanorods. As shown in Figures 1f and S6, many bright areas around 2 nm (marked as red rectangles) belonging to the Fe‐containing clusters can be clearly observed. The high‐magnification AC‐HAADF‐STEM image (Figure 1g) also shows many isolated Fe single‐atoms (labeled as yellow circles) are distributed around the clusters to form single‐atom/cluster catalyst. XRD patterns of both Fe₁‐N‐C and Fe₂N_nc/Fe₁‐N‐C nanorods (Figure S7) have two broad diffraction peaks at 23.5° and 44°, which can be assigned to the (002) and (100) planes of graphite, respectively. An additional weak diffraction peak at ∼43° appears in Fe₂N_nc/Fe₁‐N‐C nanorods, indicating the generation of small Fe‐containing clusters. Further increasing the Fe content in Fe₂N_nc/Fe₁‐N‐C nanorods is used to accurately clarify the crystal phase of Fe containing aggregates. As shown in Figure S8, a supplemental peak at 40.6° belonging to Fe₂N phase (JCPDS card no. 50–0958) emerge, confirming the generation of Fe₂N clusters in Fe₁‐N‐C matrix after nitriding treatment. The above phase identification is consistent with the HAADF‐STEM results. The homogeneous distribution of C, N, and Fe elements in the elemental mapping images of Fe₂N_nc/Fe₁‐N‐C nanorods (Figure 1h) verify the Fe₂N clusters and Fe single‐atoms are uniformly distributed in the N‐doped carbon matrix. The results of inductively coupled plasma optical emission spectrometer (ICP‐OES) revealed that the loading mass of Fe was ∼0.95%, a middle‐level content in previously reported Fe₁‐N‐C catalysts (Table S1).

In the deconvoluted Raman spectra of Fe₂N_nc/Fe₁‐N‐C nanorods (Figure S9), the D band at 1354.4 cm⁻¹ and the G band at 1599.1 cm⁻¹ represent the disordered carbon and sp² ‐type carbon, respectively. The I _D/I _G area ratio of Fe₂N_nc/Fe₁‐N‐C nanorods initially carbonized at 1100 °C is 2.98, while the ratio value is 3.47, 3.40, and 2.79 pyrolyzed at 900 °C, 1000 °C, and 1200 °C, respectively, indicating the highly disordered carbon matrix and a downward trend of carbon defects as the initial carbonization temperature rises. The specific surface area of Fe₂N_nc/Fe₁‐N‐C nanorods is high as ∼783.2 m² g⁻¹ and abundant pores with size below 3 nm can be observed in the pore distribution curve (Figure S10). The rich defects and high porosity of N‐doped carbon substrate can provide abundant sites for fully dispersing the Fe‐containing species, thereby enhancing the catalytic activity. The X‐ray photoelectron spectroscopy (XPS) survey spectrum confirmed the presence of Fe, N, and C in both Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods (Figure S11). The peaks in high‐resolution N1s spectra in both Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods can be mainly deconvoluted into pyridinic‐N, Fe‐N and graphitic‐N. Notably, compared to Fe₁‐N‐C, the overall N1s peaks of Fe₂N_nc/Fe₁‐N‐C (Figure S12) negatively shift to lower binding energies, suggesting the increase in electron density at N sites donated from the Fe atoms. This reflects the strong electron transfer between Fe and nitrogen atoms in Fe₂N_nc/Fe₁‐N‐C catalysts. The structural parameters and coordination geometry of both Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods at atomic level were further elucidated by X‐ray adsorption near‐edge spectroscopy (XANES). As shown by the normalized XANES in Figure 2a, the Fe K‐edge absorption energy of Fe₂N_nc/Fe₁‐N‐C is located between that of Fe and Fe₂O₃ and close to Fe₃O₄, indicating that Fe species in Fe₂N_nc/Fe₁‐N‐C have an oxidation state between +2 and +3. Besides, the Fe K‐edge absorption energy of Fe₂N_nc/Fe₁‐N‐C show a slightly negative shift compared to the Fe₁‐N‐C, suggesting that the valence state of Fe element decreases after reducing NH₃ treatment (inset in Figure 2a). The coordination microenvironment of Fe element was further analyzed by Fe K‐edge Fourier‐transformed extended X‐ray absorption fine structure (FT‐EXAFS). As shown in Figure 2b, a notable peak of ∼1.4 Å can be assigned to the Fe‐N bond in both Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods. In addition, an obvious Fe‐Fe coordination peak at ∼2.2 Å similar to that of Fe foil was detected in Fe₂N_nc/Fe₁‐N‐C nanorods, demonstrating the formation of Fe‐containing clusters. The slightly negative shift for Fe₂N_nc/Fe₁‐N‐C (compared to Fe₁‐N‐C) can be ascribed to the strong electronic interaction of Fe‐N and clusters. The EXAFS fitting analysis was conducted to quantitate the structural configuration of Fe atoms in Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C. As displayed in Figure 2c, these atomically dispersed Fe atoms in both Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods are isolated by neighboring four nitrogen atoms, thereby generating a typical Fe‐N₄ configuration. Moreover, a small portion of Fe₂N clusters containing Fe‐Fe coordination bonds can be also distinguish out in Fe₂N_nc/Fe₁‐N‐C nanorods distinguish out in Fe₂N_nc/Fe₁‐N‐C nanorods (Table S2). Furthermore, EXAFS wavelet transform (WT) oscillations were used to more clearly differentiate the coordination environment of Fe atoms, in which the k and R dependence of K‐edge absorption signals can be displayed simultaneously (Figure 2d). The contour intensity maximum of k space for Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C occurs at ∼5.6 Å⁻¹, suggesting the formation of Fe‐N bond, another maximum of contour feature is identified at ∼9.8 Å⁻¹ for Fe₂N_nc/Fe₁‐N‐C nanorods, undoubtedly assigned to Fe‐Fe bond. Moreover, the slight position difference in contour intensity maximum of Fe─N bonds further confirms the strong electronic interaction between Fe‐N₄ sites and Fe₂N clusters. Therefore, these structural characterizations demonstrate that Fe₂N_nc/Fe₁‐N‐C nanorods contain both Fe‐N₄ atomic sites and Fe₂N clusters while Fe₁‐N‐C nanorods only involve isolated Fe‐N₄ atomic sites.

Figure 2.

XAFS of Fe₂N_nc/Fe₁‐N‐C catalysts. a) Normalized Fe K‐edge XANES spectra (inset is the magnified pre‐edge region), b) Fourier‐transformed k ²‐weighted EXAFS spectra, c) The corresponding EXAFS R space fitting curves of Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods (inset is corresponding schematic model), and d) WT oscillations with Fe₂O₃ and Fe foil as references.

Clustering Mechanism of Fe₂N_nc/Fe₁‐N‐C Structures

During the thermal activation process, the active nitridation process plays a key role on the formation of iron single‐atom/cluster structures, which is contrary to the formation of carbon based single‐atom catalysts through conventional thermal atomization process.^{[ 6} , ^{37 ]} To unveil the formation mechanism of the Fe₂N_nc liganded Fe₁‐N‐C structures, the evolution process of Fe‐containing intermediate structure treated in both argon and NH₃ atmosphere was evaluated by ex situ XRD and TEM analysis. Obvious crystalline Fe₃O₄ phase would be generated after the thermal activation at 500 °C in both inert argon and reactive NH₃ atmosphere, which can be ascribed to the thermal decomposition of FeCl₃·6H₂O. And the Fe₃O₄ phase would be survived even heating to 700 °C. Further raising the thermal activation temperature to 800 °C and 900 °C, the diffraction peaks belonging to the Fe₃O₄ phase would gradually disappear in inert argon atmosphere (Figure S13), indicating the occurrence of thermal atomization process.^{[ 6} , ^{37 ]} On the contrary, the Fe₃O₄ would be nitride to generate Fe₂N in reactive NH₃ atmosphere at 800 °C. Interestingly, the Fe₂N phase can still survive at 900 °C, suggesting the atomization difficulty of Fe₂N phase in N doped carbon matrix (Figure 3a). HRTEM images of intermediate nanorods produced in active NH₃ atmosphere further confirmed the Fe₃O₄ nanoparticles generated before 700 °C (Figures 3b and S14) would be converted into Fe₂N clusters after 800 °C (Figures 3c and S15). It is anticipated that the formation of intermediate nitride clusters plays a key role to block the conventional atomization during the thermal activation process.

Figure 3.

Nitridation‐induced‐clustering mechanism. a) Ex situ XRD patterns of Fe₂N_nc/Fe₁‐N‐C intermediate nanorods obtained at 500 °C, 700 °C, 800 °C, and 900 °C. HRTEM images of b) Fe₃O₄ and c) Fe₂N nanoparticles obtained at 700 °C and 800 °C, respectively. The transition processes of Fe₃O₄ cluster (d) and Fe₂N cluster (e) to single‐atom (IS: initial states; TS: transition states; FS: final states. The grey, blue, red, and orange balls indicate C, N, O, and Fe, respectively). f) Energy profiles based on CI‐NEB calculations.

To prove this conjecture, DFT calculations of thermal atomization process of Fe₃O₄ and Fe₂N clusters on defect‐rich N‐C were conducted. As shown in Figure 3d,e, the formation process of Fe₁‐N‐C is described as follows: i) The pre‐generated bulk Fe₂N or Fe₃O₄ can thermally emit iron single‐atoms; ii) the iron atoms are subsequent trapped by the defects in N‐doped carbon substrate to form Fe₁‐N‐C. The Fe single‐atom emitting energy (ΔE_e ) from (002) and (211) surfaces of Fe₂N intermediate are 5.18 and 5.85 eV, respectively. In contrast, the ΔE_e for (220) and (331) surfaces of Fe₃O₄ are 4.87 and 1.40 eV, respectively (see Figure S16). These indicates that Fe atom is much easier to escape from Fe₃O_4, especially from (331) surface, to supply Fe atoms to form the isolated Fe‐N₄ species. Moreover, diffusion activation energy for Fe single‐atom emitting and subsequent trapping were calculated by using climbing image nudged elastic band (CI‐NEB) method, as shown in Figure 3f. Although both are exothermic processes (Fe₂N: ‐2.57 eV; Fe₃O₄: ‐2.92 eV), the diffusion energy barrier of Fe atom emitting from Fe₂N is almost twice larger than that of Fe₃O₄ (Fe₂N: 0.43 eV; Fe₃O₄: 0.26 eV). According to the DFT results, Fe₂N not only hinders the formation of single Fe atom, but also defers the migration to form Fe₁‐N‐C due to the stronger Fe‐ N bond.^{[ 42 ]} Therefore, the formation of Fe₂N_nc liganded Fe₁‐N‐C structures can be ascribed as impeded thermal atomization of Fe₂N intermediates. Interestingly, Ni and Co based nitride cluster‐liganded SACs (Figures S17 and S18) can be also fabricated, demonstrating the generality of the nitridation‐induced‐clustering strategy.

Electrochemical Performance of Fe₂N_nc/Fe₁‐N‐C Catalysts

The ORR performance of Fe₂N_nc/Fe₁‐N‐C nanorods was evaluated in 0.1 M KOH electrolyte using rotating disk electrode (RDE) and rotating ring disk electrode (RRDE) technologies with commercial 20% Pt/C and conventional Fe₁‐N‐C nanorods as the comparison samples. According to the linear sweep voltammetry (LSV) curves (Figure 4a), the Fe₂N_nc/Fe₁‐N‐C nanorods exhibit a very positive onset potential (E _onset) of 1.105 V versus RHE and an ultrahigh half‐wave potential (E _1/2) of 0.957 V versus RHE. While the E _1/2 of Fe₁‐N‐C nanorods is only conventional 0.905 V versus RHE, indicating the presence of the liganded Fe₂N cluster can boost the ORR activities of Fe₁‐N‐C catalysts. Meanwhile, the Fe₂N_nc/Fe₁‐N‐C nanorods give rise to an ultrahigh kinetic current density (J _k) of up to 93.4 and 224.3 mA cm⁻² at 0.9  and 0.85 V, respectively, much higher than that of Fe₁‐N‐C and commercial Pt/C (Figure S19). The Fe₂N_nc/Fe₁‐N‐C nanorods manifest the smallest Tafel slope of 54.2 mV dec⁻¹ than Fe₁‐N‐C (59.6 mV dec⁻¹) and Pt/C (69.3 mV dec⁻¹), revealing their outstanding ORR kinetics (Figure 4b). Notably, the ORR performance of Fe₂N_nc/Fe₁‐N‐C nanorods rank among the top in advanced Fe₁‐N‐C based catalysts (Figure 4c and Table S1).^{[ 22} , ²³ , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ^{48 ]} The E _1/2 value of Fe₂N_nc/Fe₁‐N‐C nanorods in acid is ∼0.77 V versus RHE, showing a comparable activity for Fe‐N‐C catalysts in acid (Figure S20). The carbonization temperature of N‐doped carbon carrier, reflecting the graphitization degree and defect density, also affects the ORR activities. The Fe₂N_nc/Fe₁‐N‐C nanorods carbonized at 1100 °C (Figure S21) exhibit the best ORR activity when compared to other carbonization temperatures, this is 900 °C (E _1/2: 0.926 V vs. RHE), 1000 °C (E _1/2: 0.931 V vs. RHE), and 1200 °C (E _1/2: 0.926 V vs. RHE). In addition, the nitriding temperature plays a key role on the ORR performance (Figure S22). We note that the Fe₂N_nc@Fe₁‐N‐C nanorods with 5 mg FeCl₃ deliver the optimal ORR activity (Figure S23).

Figure 4.

Electrocatalytic performance of Fe₂N_nc/Fe₁‐N‐C nanorods. a) LSV curves and b) corresponding Tafel plots of Fe₂N_nc/Fe₁‐N‐C, Fe₁‐N‐C and Pt/C, c) comparison of ORR performance with advanced Fe‐N‐C based electrocatalysts in alkaline conditions. d) chronoamperometric stability test of Fe₂N_nc/Fe₁‐N‐C and Pt/C. e) Rate curves at different current density and f) curves of discharge polarization and the corresponding power density of Fe₂N_nc/Fe₁‐N‐C and Pt/C, g) charge‐discharge cycling curves of rechargeable ZABs using Fe₂N_nc/Fe₁‐N‐C + RuO₂ and Pt/C + RuO₂ mixtures as cathode catalysts. h) AEMFC performance obtained by using Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C as cathode catalyst with a loading mass of 2 mg cm⁻².

Determined by Koutecky‐Levich (K‐L) equation based on LSV curves at different rotating speeds ranging from 400 to 2500 rpm, the average electron transfer number (n) is very close to 4.0, demonstrating the typical four‐electron conversion of O₂ to H₂O catalyzed by the Fe₂N_nc/Fe₁‐N‐C nanorods (Figure S24). The n value and H₂O₂ production of Fe₂N_nc/Fe₁‐N‐C and Pt/C were determined by the RRDE test (Figure S25). The yield of H₂O₂ is less than 1% and the n of Fe₂N_nc/Fe₁‐N‐C fluctuates between 3.97 and 4.0 in the potential window between 0.4  and 0.8 V. Additionally, the Fe₂N_nc/Fe₁‐N‐C nanorods demonstrate robust stability with only 0.4% current decay after 20 hours of chrono‐current testing (Figure 4d) and high methanol tolerance (Figure S26), all of which surpass those of the Pt/C catalyst. The XPS and TEM results (Figures S27 and S28) after stability test further confirm the robust structural stability of Fe₂N_nc/Fe₁‐N‐C nanorods.

Considering the outstanding ORR performance, disposable ZABs are assembled using Fe₂N_nc/Fe₁‐N‐C nanorods as air‐cathode catalyst. The open circuit voltage (OCV) of Fe₂N_nc/Fe₁‐N‐C based ZAB is high up to 1.564 V (Figure S29). Impressively, the discharge voltage of Fe₂N_nc/Fe₁‐N‐C based battery is continuously higher than that of Pt/C at different current densities ranging from 2 to 50 mA cm⁻², demonstrating its outstanding discharge performance arising from the remarkable ORR activity (Figure 4e). Fe₂N_nc/Fe₁‐N‐C based battery delivers a maximum power density of up to 202.0 mW cm⁻², higher than the commercial Pt/C (170.4 mW cm⁻², Figure 4f). The specific capacity of Fe₂N_nc/Fe₁‐N‐C is higher than that of Pt/C based electrode, which is up to 812.9 mAh g⁻¹ at a discharge current density of 20 mA cm⁻² (Figure S30). No obvious discharge voltage drop can be observed after 150 h continuous discharging at a high current density of 20 mA cm⁻² (Figure S31). Moreover, mixture of Fe₂N_nc/Fe₁‐N‐C and commercial RuO₂ was used as air‐cathode catalysts to assemble rechargeable ZABs, in which RuO₂ was served as oxygen evolution catalyst during the charging process. As shown in Figure 4g, the battery demonstrated a narrow charge‐discharge gap of only∼0.658 V, corresponding to a round‐trip efficiency of up to 65%. Impressively, only ∼10 mV decay of charge‐discharge gap can be observed even after 1000 h of cycling (1500 cycles), demonstrating the remarkable long‐term catalytic stability of Fe₂N_nc/Fe₁‐N‐C nanorods. On the contrary, Pt/C||RuO₂ based ZABs exhibited unsatisfied cycling stability in terms of severe degradation even after 400 h of cycling. As another practical application of this promising ORR catalyst, AEMFCs were also assembled using Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C nanorods as the cathode catalysts. The preparation method of the membrane electrode and the AEMFC test parameters were expressed in the Experimental Section. As shown in Figure 4h, the Fe₂N_nc/Fe₁‐N‐C based AEMFC exhibited a higher peak power density of 524.1 mW cm⁻² at 1.38 A cm⁻², surpassing that of bare Fe₁‐N‐C (355.1 mW cm⁻² at 0.91 A cm⁻²).

ORR Enhancement Mechanism of Fe₂N_nc/Fe₁‐N‐C Catalysts

The remarkable ORR performance of Fe₂N_nc/Fe₁‐N‐C nanorods can be ascribed to syngenetic enhancement effect of liganded Fe₂N clusters and Fe single‐atoms. To gain deeper insight into the underlying synergistic effect of Fe₂N clusters and Fe₁‐N‐C from orbital scale, density functional theory (DFT) computations were performed. Considering the different distance between Fe₂N cluster and surrounding Fe‐N₄ species, three possible models (Figure S32) were considered and compared, including axial Fe₂N_nc/Fe₁‐N‐C, non‐axial Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C. To evaluate the four‐electron ORR catalytic activities of these models, the optimized configurations of OOH*, O* and OH* on axial Fe₂N_nc/Fe₁‐N‐C, non‐axial Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C were first established (Figure S33). And the corresponding Gibbs free energy profiles of four‐electron pathway for each surface at alkaline conditions were investigated (Figure 5a). The potential determining step (PDS) for all considered systems were determined to be the last electron transferred step, in which OH* is desorbed from the surface to spare the active site for the following reaction. Due to excessively strong adsorption of OH* intermediate, the Fe₁‐N‐C exhibits the lowest theoretical onset potential of 0.415 V. In comparison, introduction of liganded Fe₂N clusters can significantly weaken the bonding of OH* and thus increase the onset potential from 0.415  to 0.616 V for axial configuration, and to 0.464 V for non‐axial configuration. Although the theoretical onset potential obtained by CHE model would be underestimated, the increment of onset potential induced by introduction of Fe₂N clusters are well consistent with experimental results. In addition, the introduction of Fe₂N clusters can also promote the dissociation of OOH* (Table S3). Importantly, compared with axial configuration and non‐axial configuration, the former outperforms the latter. This suggests that direct inductive effects of Fe₂N_nc ligands on Fe₁‐N‐C can further modulate the adsorption of intermediate with an optimal bonding strength. These results indicate the synergistic effects of the liganded Fe₂N cluster and Fe₁‐N‐C can enhance the ORR catalytic activity, which agrees well with the experimental results.

Figure 5.

DFT calculations. a) Energy profiles of ORR at U = 1.23 V. b) The relationship between d_d and ΔG _OH*. c) Schematic diagram of the hybridization of d_z2−p_z, d_xz−p_x, and d_yz−p_y. Calculated PCOHP of OH* adsorbed on axial Fe₂N_nc/Fe₁‐N‐C (d), non‐axial Fe₂N_nc/Fe₁‐N‐C (e) and bare Fe₁‐N‐C (f) (The dashed lines indicate the Fermi level). Side view of differential charge density for OH* adsorbed on Fe₂N_nc/Fe₁‐N‐C (g), non‐axial Fe₂N_nc/Fe₁‐N‐C (h), and bare Fe₁‐N‐C (i) (The orange, grey, blue, red, and white balls indicate Fe, N, O, and H, respectively. The yellow and blue colors represent charge accumulation and depletion).

The electronic properties, including population analysis, projected density of states (PDOS), projected crystal orbital Hamilton population (PCOHP),^{[ 49 ]} and charge transfer analysis were conducted or calculated to further show the mechanism of liganded Fe₂N clusters on enhanced ORR activity of Fe₁‐N‐C catalysts. As the desorption of OH* is the PDS for all the considered systems, we focus on this step in the following discussions. In this work, active center Fe is coordinated four nitrogen atoms to form square‐planar Fe‐N₄ configuration. Accordingly, Fe 3d orbital can split into d_x2−y2, d_z2, doubly degenerated d_xz and d_yz, and d_xy energy levels. The interaction between active center Fe and OH* is dominated by d_z2, d_xz, and d_yz of Fe and p orbitals of O in OH*, in which the Fe d_z ² orbital and O p _z orbitals hybridize to form σ bond, while the Fe d_xz/d_yz orbital hybridizes with O p _x/p _y to form π bond (Figure 5c). These orbital hybridizations dictate the adsorption strength of OH*, and ultimately affect the ORR activity. After the introduction of Fe₂N_nc ligand, a significant charge redistribution between Fe₂N_nc ligand and Fe₁‐N₄‐C can be observed (Figure S34). The d_z2 orbital of Fe in Fe₁‐N₄‐C hybridizes with d_z2 orbital of Fe atom and p _z orbital of N atom in Fe₂N_nc ligand, leading to a marked electron delocalization of d_z2 orbital of active site Fe in Fe₁‐N₄‐C (Figures S35–S38). This electron delocalization appears to weaken the Fe‐OH* adsorption.^{[ 50} , ⁵¹ , ^{52 ]} In addition, we calculated the energy profiles and electronic properties of common iron nitride (Fe₄N) and iron carbides (Fe₂C and Fe₃C) by using DFT method for comparison (see Figures S39–S42 and Table S4).

Based on above analysis, we introduced a descriptor d_d to quantitively describe the delocalization extent of 3d_z2 orbital. As shown in Figure 5b, d_d exhibits a linear relationship between the adsorption strength of OH^* (ΔG _OH*). The underlying mechanism of this linear relationship can be explained by that the delocalization of Fe 3d_z ² orbital would deteriorate the overlapping with O p _z orbitals, and make the bonding and antibonding orbitals upshift (Figure 5c). This explanation can be supported by COHP analysis, which is widely used to quantitatively investigate the bonding and anti‐bonding properties. As shown in Figure 5d–f, anti‐bonding of d_z2− p _z of spin up and all the anti‐bonding of spin down are located above the Ferm level, indicating that the orbitals mentioned above are half‐filled, while the rest are full‐filled. Clearly, both the bonding and anti‐bonding orbitals of axial Fe₂N_nc/Fe₁‐N‐C become more dispersed owing to delocalization of d orbital. And this dispersion of the bonding and anti‐bonding orbitals leads to more positive integrated COHP (ICOHP) values of Fe‐O bond (Table S5), especially hybridization of d_z2− p _z of both spin states (axial Fe₂N_nc/Fe₁‐N‐C, −0.46 and −0.41; non‐axial Fe₂N_nc/Fe₁‐N‐C, −0.54 and −0.65; Fe₁‐N‐C, −0.57 and −0.67). This up‐shift of ICOHP suggests a weaker bonding affinity of OH*, and thus boosting the ORR activity. Bonding properties can be further clarified by the structural properties and charge analysis. From the Figure 5g–i and Mulliken charge analysis, the O atom of adsorbed OH* accepts less electron of 0.78 e with longer Fe‐O bond of 1.846 Å for axial Fe₂N_nc/Fe₁‐N‐C. On the contrary, the charge transfer and the bond length for non‐axial Fe₂N_nc/Fe₁‐N‐C and Fe₁‐N‐C are larger and shorter (0.80 e and 1.822 Å; 0.83 e, 1.821 Å, respectively). Therefore, our DFT calculations show that synergistic effect between liganded Fe₂N cluster and Fe₁‐N‐C can break the localization of Fe‐3d orbitals, and then fine‐tune the OH* affinity, thereby boosting the intrinsic catalytic performance of Fe active site.

Conclusion

In summary, we developed a nitridation‐induced‐clustering method to produce Fe₂N clusters liganded Fe₁‐N‐C catalysts to boost sluggish ORRs. Using ZIF‐derived N‐doped carbon nanorods as substrate, Fe₂N_nc/Fe₁‐N‐C nanorods can be fabricated through a facile nitridation treatment. Comprehensive ex situ experiments and DFT calculations revealed that the generation of Fe₂N intermediates would retard the conventional thermal atomization process, thus resulting in the co‐existence of Fe₂N clusters and single‐atom FeN₄ species. The synergistic effect between liganded Fe₂N clusters and Fe₁‐N‐C can effectively modulate the localization of Fe‐3d orbitals to optimize the OH* affinity, thus finally boosts the intrinsic ORR performance of Fe active site. As a result, the as‐obtained Fe₂N_nc/Fe₁‐N‐C nanorods exhibited superb ORR performance with a record‐high E _1/2 of 0.957 V versus RHE, which holds great potential for application in ZABs and AEMFCs. This work brings a new avenue on the preparation of highly active cluster liganded single‐atom catalysts through decoupling the thermal atomization process.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Ma F.‐X., Liang X., Liu Z.‐H., Chen Y., Liu Z.‐Q., Zhang W., Zhen L., Zeng X. C., Xu C.‐Y., Angew. Chem. Int. Ed.. 2025, 64, e202504935. 10.1002/anie.202504935

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.