Atomically dispersed Fe/Zn synergy in sulfur-modified nitrogen-doped carbon for boosting oxygen reduction activity

Abstract

Single-atom catalysts (SACs) stabilized by multiple nitrogen-coordination architectures exhibit superior catalytic activity in pivotal electrocatalytic reactions, owing to their highly unsaturated coordination environments and robust metal-substrate interactions. Herein, atomically dispersed Fe and Zn species stabilized in specific Fe-N₄ and Zn-N₄ configurations without dimer formation were synthesized, confirmed by X-ray absorption near-edge structure analysis. Moderate S-doping strategically modulates the electronic structure of metal active sites, which advantageously regulates the adsorption/desorption characteristics of the atomic center towards the reaction intermediate. Electrochemical evaluations reveal remarkable oxygen reduction reaction (ORR) enhancement in the S-doped Fe₁Zn₁-NC catalyst (denoted as the Fe₁Zn₁-SNC-X series). The optimal catalyst Fe₁Zn₁-SNC-II demonstrates an exceptional onset potential of 0.999 V and half-wave potential of 0.871 V in 0.1 M KOH, surpassing the performance of Fe₁Zn₁-NC (S free). When assembled in zinc–air batteries (ZABs), the Fe₁Zn₁-SNC-II-ZAB outperforms the Pt/C-ZAB in both power density and cycling stability. This work provides fundamental insights into catalytic enhancement mechanisms through precisely tailoring the local coordination environment of M₁-N₄/M₂-N₄ moieties, establishing a paradigm for designing high-performance SACs by synergistic heteroatom engineering.

---

Fanpeng Meng

Fanpeng Meng received his BS degree and PhD degree from the Shandong University of Technology (2016) and Tiangong University (2021), respectively. Then, he worked as a Lecturer at the School of Chemistry and Chemical Engineering, Liaocheng University. His research focuses on the design and synthesis of advanced photocatalytic and electrocatalytic materials, with a primary goal of addressing key challenges in sustainable energy utilization and environmental remediation.

Introduction

Atomically dispersed metal–nitrogen–carbon materials (AD-MNCs) have emerged as a promising class of electrocatalysts, demonstrating exceptional performance in sustainable energy conversion technologies such as metal–air batteries, fuel cells, and electrochemical CO₂/N₂ fixation.^1–4 Their remarkable catalytic activity arises from the synergistic interplay between atomically dispersed metal centers in a highly unsaturated coordination environment and the conductive carbon support.^5–7 Beyond stabilizing metal centers through M-N₄ coordination, the surrounding nitrogen atoms serve bifunctions as electronic bridges facilitating charge transfer between metal centers and the carbon matrix, while simultaneously providing adaptive coordination sites responsive to reactive intermediates.^8,9 Therefore, the unique configuration establishes a tunable platform for catalytic optimization, where the electronic structure and catalytic behavior of metal centers can be precisely tailored through strategic manipulation of the local chemical environment.¹⁰

Nowadays, substantial advancements have been achieved in atomically resolved studies of AD-MNCs, particularly in elucidating active site architectures through cutting-edge characterization techniques, and the exploration of novel polynuclear metal centers to establish structure–activity correlations at atomic precision.^11–13 However, significant potential remains in precisely engineering the electronic properties of M-N_x moieties via localized heteroatom doping.^14,15 Emerging evidence reveals that the catalytic performance of AD-MNCs exhibits strong correlation with the atomic-scale modulation of M-N_x electronic structures, especially when heteroatoms (e.g., B, P, and S) are strategically incorporated into the nitrogen coordination environment.^16–20 However, achieving optimal heteroatom doping through rational introduction of heteroatom sources in precursor materials remains a significant challenge.²¹ Therefore, the rational regulation of heteroatom spatial distribution and chemical states in carbon materials through doping strategies holds profound implications for enhancing electrocatalytic performance in binary metal single-atom coordination systems.^22,23

In this work, the chemical reaction between formamide (serving as a N and C source) and thiourea (acting as a S source) achieves localized sulfur doping in atomically dispersed Fe-N₄ and Zn-N₄ coordination configurations. The spatial distribution of S dopants not directly bound to Fe/Zn was verified through fitting analysis of Fe K-edge and Zn K-edge X-ray absorption fine structure. Electrochemical measurements demonstrate that the S-doped bimetallic single-atom catalyst (denoted as Fe₁Zn₁-SNC) exhibits superior oxygen reduction reaction (ORR) performance in alkaline electrolyte, delivering an onset potential of ∼0.999 V vs. RHE and a half-wave potential of ∼0.871 V. Compared with its S-free Fe₁Zn₁-NC counterpart, the enhanced catalytic performance of Fe₁Zn₁-SNC can be attributed to the isolated exposure of dual metal single-atom sites (Fe-N₄ and Zn-N₄) that provide optimized adsorption/desorption energetics and the electronic structure modulation induced by S species in the carbon matrix. Additionally, the developed liquid Zn–air battery (ZAB) offers an open-circuit voltage of 1.58 V, a specific power of 228.0 mW cm⁻² and a specific capacity of 804.4 mAh g_Zn⁻¹, highlighting its promising potential for advanced electronics.

Results and discussion

The synthesis of the FeZn-SNC series is schematically depicted in Fig. 1a. Fe/Zn salts and thiourea were initially dissolved in formamide, where the strong coordination between metal ions and formamide derivatives ensured atomic-level dispersion. Thermal polycondensation at 180 °C induced a formamide-derived carbonization process, synergistically promoted by the uniformly distributed metal species.²⁴ This resulted in the in situ generation of S/N co-doped carbonaceous frameworks with optimized coordination environments.²⁵ A subsequent stabilization and carbonization step at 900 °C for 1 h under N₂ effectively immobilized the metal sites within the activated carbon matrix. The typical field-emission scanning electron microscopy (FE-SEM) images demonstrate that the catalyst exhibits a three-dimensionally interconnected framework constructed through the random assembly of irregular spherical architectures, with particle diameters spanning from 0.5 to 3 μm (Fig. 1b). The high-resolution transmission electron microscopy (HRTEM) images in Fig. 1c confirm the absence of metallic clusters or nanoparticles in Fe₁Zn₁-SNC-II, directly evidencing the atomic-level dispersion of metal species. Furthermore, the aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) image (Fig. 1d) reveals isolated metal atoms, appearing as white dots with high diffraction contrast uniformly dispersed against the black background of Fe₁Zn₁-SNC-II, indicating a homogeneous distribution of metal atoms on the carbon support. The EDS mapping spectrum (Fig. 1e and f) demonstrates the successful incorporation of C, N, O, S, Fe, and Zn components throughout Fe₁Zn₁-SNC-II, showing a uniform distribution. The Zn component acts as a spatial constraint during the heat treatment process, facilitating the dispersion and stabilization of iron atoms. Notably, it is significantly removed at 900 °C, as evidenced by the highly dispersed signal detected in the final catalyst.

Fig. 1 (a) Schematic illustration of the synthesis process for Fe₁Zn₁-SNC-II. (b) SEM image, (c) HRTEM image, (d) AC-HAADF-STEM image, (e) EDS mapping images and (f) elemental overlaps of Fe₁Zn₁-SNC-II.

As shown in Fig. 2a, all XRD spectra exhibit two broadened peaks at ∼24.0° and 43.5°, corresponding to the (002) and (101) crystal planes of graphitic carbon, respectively.²⁶ Additionally, no detectable diffraction peaks associated with metal oxides or metallic species are observed in Fe₁Zn₁-SNC-II, definitively verifying the absence of aggregated Fe and Zn species in the catalyst. The Raman spectra (Fig. 2b) reveal characteristic D (∼1343 cm⁻¹) and G (∼1585 cm⁻¹) bands, which correspond to disordered lattice vibrations (sp³ defects) and in-plane sp²-hybridized carbon vibrations, respectively.²⁷ The calculated I_D/I_G intensity ratios for Fe₁Zn₁-NC, Fe₁Zn₁-SNC-I, Fe₁Zn₁-SNC-II, and Fe₁Zn₁-SNC-III progressively increase from 0.99 to 1.06, demonstrating that sulfur incorporation during high-temperature annealing leads to a decrease in the degree of graphitization. BET analysis reveals that Fe₁Zn₁-SNC-II exhibits the largest specific surface area of 1040.02 m² g⁻¹, followed by Fe₁Zn₁-NC (977.13 m² g⁻¹), Fe₁Zn₁-SNC-III (846.91 m² g⁻¹), and Fe₁Zn₁-SNC-I (637.55 m² g⁻¹). The pore size distribution analysis indicates that all carbon materials feature a hierarchical pore structure (Fig. S1†). The abundant pore systems and large specific surface areas provide essential conditions for the accessibility of ORR active sites.

Fig. 2 (a) XRD patterns, (b) Raman spectra, (c) XPS survey spectra, XPS fine spectra of (d) C 1s, (e) N 1s, (f) S 2p, (g) Fe 2p and (h) Zn 2p, and (i) ICP for Fe and Zn element contents of Fe₁Zn₁-NC, Fe₁Zn₁-SNC-I, Fe₁Zn₁-SNC-II and Fe₁Zn₁-SNC-III.

The chemical composition and valence states of the catalysts were systematically investigated by X-ray photoelectron spectroscopy (XPS). As clearly evidenced in Fig. 2c, the survey spectrum of catalysts confirms the coexistence of C, N, O, S, Fe, and Zn elements. Notably, the Fe signal remains barely detectable with attenuated peak intensity, suggesting that its concentration falls below the XPS detection limit, which aligns with the absence of Fe aggregates observed in XRD analysis. The high-resolution C 1s XPS spectrum (Fig. 2d) was deconvoluted into three characteristic peaks, corresponding to C–C/C=C bonds (284.8 eV), C=N bonds (286.0 eV), and C=O bonds (287.5 eV), respectively. The dominant contribution from C–C/C=C indicates the graphitic/sp² carbon framework, while the presence of C=N and C=O functional groups suggests surface oxidation and nitrogen-containing moieties, likely arising from the synthesis process or environmental interactions.²⁸ As shown in Fig. 2e, the high-resolution N 1s XPS spectrum reveals five nitrogen configurations: a peak at 398.5 eV, assigned to pyridinic N; a peak at 399.2 eV, ascribed to the metal-nitrogen (M-N_x) coordination; a peak at 399.8 eV, attributed to pyrrolic N; a peak at 401.2 eV, corresponding to graphitic N; and a broad peak centered at 403.0 eV, indicative of oxidized N species.²⁹ The high-resolution S 2p XPS spectrum (Fig. 2f) was fitted with three components: a doublet with S 2p_3/2 and S 2p_1/2 peaks centered at 163.3 eV and 164.5 eV, respectively, characteristic of thiophenic sulfur (C–S–C) in aromatic structures with a spin–orbit splitting of 1.2 eV and a broad peak at 167.7 eV assigned to oxidized sulfur species (SO_x⁻), which may originate from partial oxidation during synthesis or exposure to ambient conditions.³⁰

As shown in Fig. 2g, the high-resolution Fe 2p XPS spectrum was deconvoluted into four components, corresponding to the spin–orbit split doublets of Fe²⁺ (Fe²⁺ 2p_3/2 at 711.1 eV and Fe²⁺ 2p_1/2 at 723.0 eV) and Fe³⁺ (Fe³⁺ 2p_3/2 at 713.2 eV and Fe³⁺ 2p_1/2 at 725.8 eV) species. The coexistence of Fe²⁺ and Fe³⁺ states indicates that the single-atom Fe sites are in the oxidized valence state, suggesting stabilization by coordination with electronegative N atoms in the carbon matrix.³¹ The Zn 2p XPS spectrum (Fig. 2h) displays two well-resolved peaks at 1021.5 eV (Zn 2p_3/2) and 1044.5 eV (Zn 2p_1/2), corresponding to Zn²⁺ species, confirming the exclusive presence of Zn²⁺ in a single-atom configuration.³² The metal mass loadings of Fe and Zn in the single-atom catalysts were quantified by inductively coupled plasma optical emission spectroscopy (ICP-OES). As summarized in Fig. 2i and Table S1,† the Fe content increases from 0.93 wt% in Fe₁Zn₁-NC to 1.24–1.48 wt% in the sulfur-modified catalysts (Fe₁Zn₁-SNC-I to III), while the Zn content slightly decreases from 9.23 wt% (Fe₁Zn₁-NC) to 7.66–7.95 wt% (Fe₁Zn₁-SNC series). This trend suggests that sulfur incorporation enhances the anchoring capability for Fe atoms in the carbon substrate.

The local coordination environment and electronic structure of single-atom Fe sites were systematically probed through X-ray absorption fine spectroscopy (XAFS). Fe foil, Fe₂O₃ and Fe phthalocyanines (FePc) were also measured for comparison. As shown in Fig. 3a, high-resolution Fe K-edge X-ray absorption near-edge structure (XANES) spectroscopy reveals a distinct pre-edge feature at ∼7113.5 eV, characteristic system symmetry of a porphyrin-like square-planar structure with Fe-N₄ coordination.³³ Moreover, the absorption threshold of the Fe K-edge in Fe₁Zn₁-SNC-II is positioned intermediate between that of Fe foil and Fe₂O₃ and closer to that of FePc, which indicates the dominant oxidation state of Fe centers in Fe₁Zn₁-SNC-II. The extended X-ray absorption fine structure (EXAFS) spectra (Fig. 3b) in R-space analysis demonstrated the absence of Fe–Fe scattering paths at ∼2.20 Å of Fe₁Zn₁-SNC-II, confirming the atomic dispersion of Fe centers.³⁴ The dominant Fourier transform (FT) peak at 1.48 Å (phase-uncorrected) corresponds to first-shell Fe-N coordination with an average coordination number of 4.1 ± 0.2 (Table S2†). Wavelet transform (WT) analysis of the k³-weighted EXAFS oscillations further excluded the existence of metallic Fe clusters or nanoparticles (Fig. 3c).

Fig. 3 (a) Normalized XANES spectra obtained at the Fe K-edge for Fe₁Zn₁-SNC-II and reference samples (Fe foil, Fe₂O₃, and FePc); (b) FT-EXAFS for Fe₁Zn₁-SNC-II and three Fe-relevant samples with their fitting curves. (c) WT-EXAFS of the Fe K-edge for Fe₁Zn₁-SNC-II. (d) Normalized XANES spectra obtained at the Zn K-edge for Fe₁Zn₁-SNC-II and reference samples (Zn foil, ZnO, and ZnPc); (e) FT-EXAFS for Fe₁Zn₁-SNC-II and three Zn-relevant samples with their fitting curves. (f) WT-EXAFS of the Zn K-edge for Fe₁Zn₁-SNC-II.

The electronic state and local coordination environment of Zn species in Fe₁Zn₁-SNC-II were further investigated through Zn XAFS. XANES analysis (Fig. 3d) reveals that the absorption edge energy of Zn in Fe₁Zn₁-SNC-II lies intermediate between that of Zn foil and the ZnO reference, with closer proximity to that of ZnPc, suggesting a formal oxidation state between Zn⁰ and Zn²⁺. The FT-EXAFS spectra exhibit a dominant peak at approximately 1.4 Å in the R-space, which corresponds to the first coordination shell of Zn-N/O scattering paths (Fig. 3e).³⁵ This characteristic agrees well with the reference spectrum of ZnPc containing four-coordinated Zn-N₄ moieties, while showing marked contrast to the Zn–Zn metallic bonding signature (∼2.2 Å) observed in Zn foil.³⁶

Quantitative analysis indicates an average N coordination number of 3.8 ± 0.3 with a Zn-N bond length of 2.01 ± 0.02 Å, suggesting a predominantly 4-coordinated environment around the atomically dispersed Zn centers provided by N atoms (Table S3†). Notably, WT analysis of the Zn K-edge EXAFS (Fig. 3f) demonstrates the absence of detectable Zn–Zn or Fe–Zn contributions within 2.6 Å which further excludes the existence of Zn clusters, Fe–Zn diatomic pairs, or metallic nanoparticles; this observation is also corroborated by the Fe K-edge analysis showing exclusive Fe-N coordination without Fe–Zn contributions, confirming the mononuclear and spatially isolated nature of both Fe and Zn atoms anchored on the nitrogen-doped carbon matrix.

The oxygen reduction reaction (ORR) electrocatalytic performance was first investigated by cyclic voltammetry (CV) measurements at a scan rate of 50 mV s⁻¹ in both O₂- and N₂-saturated 0.1 M KOH electrolytes. As illustrated in Fig. S2,† all catalyst-modified electrodes exhibited distinct oxygen reduction peaks in an O₂-saturated environment compared to the featureless capacitive responses observed under the N₂ atmosphere, confirming the occurrence of oxygen reduction processes. Moreover, Fe₁Zn₁-SNC-II demonstrated a well-defined cathodic peak centered at ∼0.87 V vs. RHE and displayed the most positive potential among all investigated samples. The linear sweep voltammogram (LSV) curves were evaluated using a rotating disk electrode (RDE) at rotational speeds of 1600 rpm. The sulfur-free Fe₁Zn₁-NC catalyst demonstrates promising ORR performance with an onset potential (E_onset) of 0.998 V vs. RHE and a half-wave potential (E_1/2) of 0.848 V (Fig. 4a). Remarkably, controlled sulfur incorporation induces distinct performance evolution: Fe₁Zn₁-SNC-I (3.0 mM thiourea) maintains a comparable E_onset of 0.999 V while achieving an enhanced E_1/2 of 0.853 V, whereas Fe₁Zn₁-SNC-II (4.5 mM thiourea) exhibits optimal activity with both an E_onset of 0.999 V and E_1/2 of 0.871 V exceeding those of commercial Pt/C (E_onset = 0.978 V, E_1/2 = 0.849 V). This reverses at higher sulfur loading (6.0 mM thiourea), as evidenced by Fe₁Zn₁-SNC-III showing a deteriorated E_onset of 0.968 V and E_1/2 of 0.836 V (Fig. 4b). The performance degradation at excessive sulfur content may originate from active-site coverage, as supported by subsequent electrochemical surface area analysis. To investigate the enhancement mechanism of the ORR, the site density (SD) of Fe₁Zn₁-NC and Fe₁Zn₁-SNC-II samples was determined (Fig. S3†). As shown in the calculation results summarized in Table S4,† the SD values of Fe₁Zn₁-NC and Fe₁Zn₁-SNC-II are 12.4 and 14.9 μmol g⁻¹, respectively, indicating that Fe₁Zn₁-SNC-II possesses a higher concentration of electrochemically accessible active Fe sites. This suggests that the incorporation of S contributes to activating the activity of the sites, which may be a key factor for the improved ORR performance.

Fig. 4 Electrochemical performances of catalysts in O₂-saturated 0.1 M KOH solution: (a) ORR LSV curves. (b) Histogram comparison of E_onset and E_1/2. (c) Tafel plots. (d) H₂O₂ yield and (e) electron transfer number (n) calculated using an RRDE. (f) Comparison of n from K–L plots of Fe₁Zn₁-NC, Pt/C and Fe₁Zn₁-SNC series. (g) Fitted EIS Nyquist plots. (h) Linear correlation curves of current density versus the scan rate with calculated C_dl values. (i) Methanol tolerance performance.

The kinetic characteristics of oxygen reduction were further elucidated through Tafel slopes derived from polarization curves.³⁷ As shown in Fig. 4c, Fe₁Zn₁-SNC-II demonstrates the lowest Tafel slope of 78.69 mV dec⁻¹, significantly outperforming Fe₁Zn₁-SNC-I (89.71 mV dec⁻¹), Fe₁Zn₁-SNC-III (86.59 mV dec⁻¹) and the sulfur-free control sample Fe₁Zn₁-NC (93.94 mV dec⁻¹) and even superior to commercial Pt/C (91.95 mV dec⁻¹). The reduced Tafel slope directly reflects enhanced electron transfer efficiency during the rate-determining step. Such kinetic superiority can be ascribed to the sulfur-induced modulation of metal active centers, which optimizes the adsorption/desorption energetics of intermediates and facilitates rapid proton-coupled electron transfer processes. The reaction pathway selectivity and catalytic efficiency were quantitatively evaluated through rotating ring-disk electrode (RRDE) measurements. Fe₁Zn₁-SNC-II exhibits exceptional 4e⁻ ORR characteristics, maintaining H₂O₂ yields below 5% across a broad potential range of 0.2–0.8 V vs. RHE (Fig. 4d and S4†). The calculated electron transfer number (n) reaches an average value of >3.90, significantly surpassing those of Fe₁Zn₁-NC, Fe₁Zn₁-SNC-I, Fe₁Zn₁-SNC-III and commercial Pt/C.

To further probe the catalytic kinetics for the ORR, the RDE measurements were conducted at rotational speeds ranging from 400 to 1600 rpm. The diffusion-corrected current densities exhibited progressive enhancement with increasing rotation rates, while the E_onset remained constant, indicating stable activation of catalytic sites at this applied voltage. The parallel Koutecky–Levich (K–L) plots at selected potentials (0.45–0.70 V vs. RHE) displayed excellent linearity (Fig. S5†), confirming first-order reaction kinetics with respect to the dissolved oxygen concentration. Based on the K–L equation slopes, the calculated electron transfer number (n) for Fe₁Zn₁-SNC-II averaged ∼4.0 across the potential window, decisively verifying a dominant 4e⁻ oxygen reduction pathway. Electrochemical impedance spectroscopy (EIS) was performed at 0.5 V (vs. RHE) with an amplitude of 5.0 mV across the frequency range of 0.1–10⁵ Hz. The Nyquist plot (Fig. 4g) exhibits a compressed semicircle in the high-frequency region corresponding to charge transfer resistance (R_ct), followed by a Warburg-type diffusion tail at low frequencies.³⁸ Through equivalent circuit modeling and fitting (Fig. S6 and Table S5†), Fe₁Zn₁-SNC-II demonstrates the smallest R_ct value of 53.52 Ω cm⁻² among the series (Fe₁Zn₁-NC: 59.72 Ω cm⁻², Fe₁Zn₁-SNC-I: 60.87 Ω cm⁻², and Fe₁Zn₁-SNC-III: 59.54 Ω cm⁻²), reflecting that the incorporation of moderate amounts of elemental sulfur enhances interfacial charge transfer kinetics.

The accessible catalytic sites were quantitatively evaluated through electrochemical surface area (ECSA) determination using cyclic voltammetry in a non-faradaic potential region (1.015–1.115 V vs. RHE). As shown in Fig. 4h, Fe₁Zn₁-SNC-II exhibited the largest double-layer capacitance (C_dl = 21.77 mF cm⁻²), corresponding to the highest electrochemical surface area (ECSA). Furthermore, the ECSA values for all sulfur-doped samples (Fe₁Zn₁-SNC-I/II/III) were substantially greater than that of Fe₁Zn₁-NC, highlighting a significant impact of sulfur doping on enhancing the accessibility of catalytically active centers by modulating the electronic structure.³⁹ The electrochemical stability and methanol tolerance of Fe₁Zn₁-SNC-II were evaluated through comparative studies with commercial Pt/C. Accelerated durability tests (ADTs) of 5000 continuous CV cycles revealed distinct degradation behavior between catalysts. After the ADT, Fe₁Zn₁-SNC-II exhibited a minimal shift of E_1/2 (∼14 mV), significantly less than the shift observed for commercial Pt/C (∼27 mV) (Fig. S7†). This shows that Fe₁Zn₁-SNC-II is much more stable than Pt/C during long-term operational applications. To investigate the mechanism underlying electrochemical cycling stability, the microstructural morphology and elemental valence states of Fe₁Zn₁-SNC-II were analyzed post-ORR ADTs (5000 cycles). EDS mapping (Fig. S8a†) confirmed the retention of a highly uniform elemental dispersion and absence of single-atom aggregation, indicative of structural preservation. Fig. S8b–e† demonstrate that the valence states of N, S, Fe, and Zn exhibit no significant changes after 5000 cycles in alkaline solution. These observations provide valuable insights into the origin of the catalyst's exceptional stability, highlighting its structural and compositional robustness under electrochemical operation. The methanol resistance assessment through LSV measurements was conducted with 1 M methanol introduction. Fe₁Zn₁-SNC-II maintained its original oxygen reduction trajectory with negligible current variation, while Pt/C manifested characteristic methanol oxidation currents between 0.6 and 0.8 V vs. RHE (Fig. 4i). This dichotomy originates from the distinct catalytic mechanisms: the metal-N₄ coordination sites in Fe₁Zn₁-SNC-II selectively catalyze the 4e⁻ ORR pathway without activating methanol oxidation, whereas Pt surfaces facilitate competitive methanol oxidation reactions through non-selective adsorption.

Besides the ORR activity, the OER activity of all samples was also measured in a 0.1 M KOH aqueous solution. Fe₁Zn₁-SNC-II exhibits a low overpotential (η₁₀) of ∼410 mV at a current density of 10 mA cm⁻², surpassing that of Fe₁Zn₁-SNC-I (η₁₀ = 430 mV), Fe₁Zn₁-SNC-III (η₁₀ = 520 mV), and commercial IrO₂ (η₁₀ = 480 mV) (Fig. S9a†). The Fe₁Zn₁-NC catalyst without S demonstrates even poorer OER activity, highlighting the critical role of S incorporation in promoting the transformation of oxygen-containing intermediates. Kinetic analysis via Tafel plots (Fig. S9b†) further reveals that Fe₁Zn₁-SNC-II possesses a Tafel slope of 93.96 mV dec⁻¹, significantly lower than that of IrO₂ (123.78 mV dec⁻¹) and its other counterparts, indicating faster OER kinetics.

To investigate the impact of S introduction on the oxygen reduction activity of single atom Fe and Zn sites, first-principles theoretical calculations based on density functional theory (DFT) were performed using the Vienna Ab initio Simulation Package (VASP). Fig. 5a and b display the total projected density of states (PDOS) plots for two models, namely FeN₄-ZnN₄-SNC and FeN₄-ZnN₄-NC. A conspicuous disparity in the electronic structures at the Fermi level is evident upon S introduction. This significant alteration in the electronic structure near the Fermi level strongly suggests that the incorporation of S has a profound impact on the catalytic performance, potentially modifying the adsorption and activation processes of reactant molecules involved in the ORR process. Analysis of the d-band centers of Zn reveals that, whether or not S is doped, the d-band centers of Zn are located far from the Fermi level, at −6.47 eV and −6.49 eV, respectively, which does not meet the typical conditions for active centers (Fig. 5c).⁴⁰ In contrast, with S doping, the d-band center of Fe shifts to −0.43 eV, significantly closer to the Fermi level compared to −1.67 eV in the absence of S doping (Fig. 5d). This shift suggests an upward movement of the Fe d-band center, enhancing the intermediate adsorption strength and facilitating oxygen activation. Furthermore, Bader charge analysis was conducted to gain deeper insights into the electronic structure modifications induced by S doping. The results reveal that S doping triggers charge transfer at the active sites, indicating a significant alteration in the local electronic environment. As shown in Fig. S10,† upon the introduction of S, the Bader valence state of Fe decreases from +1.10 to +0.91, suggesting an accumulation of electrons at the central site. This electron redistribution likely enhances the interaction between the active site and reaction intermediates, potentially affecting the catalytic performance. In contrast, the Bader valence state of Zn remains almost unchanged, shifting marginally from +1.17 to +1.16, which implies that S doping has a negligible impact on the electronic structure of Zn sites compared to Fe sites. According to the 4e⁻ transfer mechanism proposed by Nørskov,⁴¹ the Gibbs free energies of the intermediates “*OH, *O, and *OOH” adsorbed on the active sites of the catalyst surface are crucial determinants of the reaction barriers (Fig. 5e). These intermediates play pivotal roles in the ORR process, and their adsorption energies and corresponding Gibbs free energy (ΔG) changes govern the overall reaction kinetics. In Fig. 5f, the FeN₄-ZnN₄-SNC model exhibits a lower theoretical overpotential (η = 0.42) compared to the FeN₄-ZnN₄-NC model (η = 0.55). This significant reduction in overpotential can be primarily attributed to the decreased adsorption energy of OH on the active sites of the FeN₄-ZnN₄-SNC catalyst. A lower adsorption energy of OH facilitates the desorption of the reaction intermediate, thereby accelerating the ORR kinetics and reducing the energy barrier required for the reaction to proceed. These findings underscore the critical role of sulfur doping in optimizing the ORR performance by modulating the adsorption behavior of key reaction intermediates.

Fig. 5 Total DOS of (a) FeN₄-ZnN₄-SNC and (b) FeN₄-ZnN₄-NC. (c) DOS of Zn sites in FeN₄-ZnN₄-SNC and FeN₄-ZnN₄-NC with the d-band center. (d) DOS of Fe sites in FeN₄-ZnN₄-SNC and FeN₄-ZnN₄-NC with the d-band center. (e) Schematic diagrams of the ORR pathways. (f) Calculated adsorption energies for the ORR intermediates on the surface of FeN₄-ZnN₄-NC and FeN₄-ZnN₄-SNC.

As schematically depicted in Fig. 6a, a standard zinc–air battery (ZAB) configuration comprising a zinc anode, alkaline electrolyte, and a triphasic air cathode integrated with the Fe₁Zn₁-SNC-II or Pt/C catalyst-coated gas diffusion layer was used. The open-circuit voltage (OCV) measurements of ZABs revealed significant differences in performance between ZABs assembled with Fe₁Zn₁-SNC-II and Pt/C catalysts (Fig. 6b). Specifically, the Fe₁Zn₁-SNC-II-based zinc–air battery (Fe₁Zn₁-SNC-II-ZAB) achieved a remarkably high OCV of 1.58 V, surpassing the 1.49 V recorded for the Pt/C-ZAB, with both systems demonstrating exceptional voltage stability over a continuous one-hour measurement period, highlighting the Fe₁Zn₁-SNC-II's superior ability to maintain stable interfacial charge equilibria at the triple-phase interface.⁴² As shown in Fig. 6c, the Fe₁Zn₁-SNC-II-ZAB obtained a peak power density of 228.0 mW cm⁻², significantly exceeding the 185.1 mW cm⁻² attained by the Pt/C-ZAB. The rate capability assessment of ZABs was measured at incremental current densities of 5, 10, 20, and 30 mA cm⁻² (Fig. 6d). The Fe₁Zn₁-SNC-II-ZAB demonstrated superior voltage retention with polarization voltages, outperforming the Pt/C-ZAB. The specific capacity of the Fe₁Zn₁-SNC-II-ZAB was calculated to be 804.4 mAh g_Zn⁻¹ based on the mass loss of metallic Zn, demonstrating a 5.5% enhancement compared to the Pt/C-ZAB (762.1 mAh g_Zn⁻¹) (Fig. 6e). During galvanostatic discharge at 10 mA cm⁻², the Fe₁Zn₁-SNC-II-ZAB maintained consistently higher voltage plateaus than the Pt/C-ZAB throughout the discharge process, which is consistent with the rate capability. The Fe₁Zn₁-SNC-II-ZAB demonstrated exceptional operational stability in practical energy delivery scenarios, as evidenced by its ability to continuously power a commercial LED light panel (rated voltage = 1.2 V) under ambient conditions (Fig. 6f). Practical assessment of the FeZn-SNC-ZAB demonstrated remarkable stability in charge–discharge voltage, retaining over 98% of its initial voltage after 600 hours (1800 cycles) with minimal degradation (Fig. 6g). In contrast, the Pt/C-ZAB air cathode displayed a noticeable voltage decay after only 50 hours (150 cycles), demonstrating the practical potential of the Fe₁Zn₁-SNC-II-ZAB as a viable and cost-effective alternative to precious metal catalysts in electrochemical catalysis.

Fig. 6 (a) Scheme of liquid ZABs using Fe₁Zn₁-SNC-II as the cathode catalyst layer. (b) Open-circuit voltage tests of ZABs; the inset graphic shows voltage values of the Fe₁Zn₁-SNC-II-ZAB measured with a multimeter. (c) Discharge polarization plots and power density curves of the Fe₁Zn₁-SNC-II-ZAB and Pt/C-ZAB. (d) Discharge rate performance and (e) galvanostatic discharge curves. (f) LED light panel (rated voltage = 1.2 V) using the Fe₁Zn₁-SNC-II-ZAB. (g) Charge–discharge cycling performance of the Zn–air batteries, with a zoomed-in view of voltage profiles.

A comparative analysis was conducted with recently reported non-noble metal catalysts, as summarized in Table S6.† The results highlight that our catalyst outperforms its counterparts in key metrics, including OCV, power density, and specific capacity, showcasing its remarkable efficiency and competitiveness in electrochemical applications. To further elucidate the influence of S-incorporated samples on battery device performance, Fe₁Zn₁-NC-ZAB performance was evaluated. As shown in Fig. S11,† the OCV of the Fe₁Zn₁-NC-ZAB was measured at 1.53 V, significantly lower than that of the Fe₁Zn₁-SNC-II-ZAB. This disparity in OCV indicates a more favorable electrochemical reaction initiation in the S-doped system. Moreover, the specific capacity of the Fe₁Zn₁-NC-ZAB was 773.2 mAh g_Zn⁻¹, exhibiting a substantial reduction compared to the Fe₁Zn₁-SNC-II-ZAB, highlighting the pivotal role of sulfur doping in enhancing energy storage capacity. Additionally, the Fe₁Zn₁-NC-ZAB exhibited lower voltage profiles across varying current densities, and its inferior power density and charge–discharge cycling stability further corroborated the advantages of the sulfur-doped catalyst. These results collectively demonstrate that sulfur doping significantly improves the overall performance of ZABs, rendering Fe₁Zn₁-SNC-II a promising candidate for advanced battery applications.

Conclusions

In conclusion, we have successfully developed a reliable strategy to enhance the ORR performance of atomically dispersed Fe and Zn active sites through optimized sulfur local doping. Comprehensive characterization studies using XRD, XPS, AC-HAADF-STEM, and XAFS techniques conclusively demonstrate the atomic dispersion of both Fe-N₄ and Zn-N₄ moieties in Fe₁Zn₁-SNC-II, along with homogeneous sulfur incorporation within the carbon matrix. This dual-site engineering approach effectively creates favorable adsorption/desorption kinetics for oxygen intermediates, resulting in synergistic enhancement of intrinsic activity and active site accessibility. Electrochemical measurements reveal significantly enhanced ORR activity of Fe₁Zn₁-SNC-II, achieving an onset potential of 0.999 V and a half-wave potential of 0.871 V in 0.1 M KOH, surpassing those of commercial Pt/C by 21 mV and 22 mV, respectively. Remarkably, the Fe₁Zn₁-SNC-II-ZAB demonstrates superior performance compared to the Pt/C-ZAB, particularly in terms of power density and discharge stability. This work provides fundamental insights into the rational design of dual-atom catalysts through heteroatom doping engineering, offering promising potential for advanced energy conversion applications such as fuel cells and metal–air batteries.

Experimental section

Chemicals and reagents

Formamide (CH₃NO, AR, 99%) was obtained from Damao Chemical Reagent Factory. Ferric nitrate nonahydrate (Fe(NO₃)₃·9H₂O, AR, 98.5%) was purchased from Macklin Company, while zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O, AR, 99%) was supplied by Sinopharm Chemical Reagent Co., Ltd. Thiourea was procured from Tianjin Kemiou Chemical Reagent Co., Ltd. A commercial Pt/C catalyst (20 wt% Pt) was purchased from Shanghai Hesen Electric Co., Ltd. All reagents were used as received without further purification.

Preparation of Fe₁Zn₁-NC and Fe₁Zn₁-SNC-X (X = I, II, III) materials

Zn(NO₃)₂·6H₂O (33.6 mM, 300 mg), Fe(NO₃)₃·9H₂O (0.8 mM, 10 mg), and thiourea (either 3.0 mM, 6.9 mg; 4.5 mM, 10.3 mg or 6.0 mM, 13.7 mg) were incorporated into formamide (30 mL) in a 50 mL reaction kettle under continuous stirring. The resulting solution was then placed in an oven and heated at 180 °C for 8 hours. The precursor products were collected by centrifugation, thoroughly purified three times with a mixture of water and ethanol, and subsequently dried under vacuum at 60 °C for 12 hours.

The obtained precursor was finely ground using an agate mortar, and 100 mg of the powder was placed in a porcelain boat. Subsequently, the sample was annealed in a tube furnace under a N₂ atmosphere with the following thermal program: heating to 900 °C at 5 °C min⁻¹ and holding for 1 h. The final products obtained with 3.0, 4.5, and 6.0 mM thiourea additions were denoted as Fe₁Zn₁-SNC-I, Fe₁Zn₁-SNC-II, and Fe₁Zn₁-SNC-III, respectively.

For the Fe₁Zn₁-NC control sample, the synthesis followed an identical protocol to Fe₁Zn₁-SNC-X except for omitting thiourea.

Material characterization

The crystallographic structures were characterized by powder X-ray diffraction (XRD, Rigaku SmartLab 9) with Cu Kα radiation (λ = 1.5418 Å) at 40 kV. Chemical states and elemental composition were analyzed via X-ray photoelectron spectroscopy (XPS, Thermo Scientific ESCALAB 250Xi) with a monochromatic Al Kα source. Raman spectra were acquired on a Renishaw InVia reflex confocal Raman microscope using a 532 nm excitation laser. Morphological features were examined by scanning electron microscopy (SEM, Thermo Fisher Scientific Apreo 2S, acceleration voltage: 40 kV) and transmission electron microscopy (TEM, FEI Talos F200x operated at 200 kV) equipped with energy-dispersive X-ray spectroscopy (EDS). The atomic dispersion of metal species was verified by high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) using an aberration-corrected STEM (Titan 80–300, Thermo Fisher Scientific) operated at 300 kV accelerating voltage. Quantitative elemental composition was determined by inductively coupled plasma optical emission spectrometry (ICP-OES, Agilent 5110) with calibration curves established using certified standard solutions. Fe/Zn K-edge X-ray absorption fine structure (XAFS) measurements were conducted at the BL14W beamline of the Shanghai Synchrotron Radiation Facility (SSRF, 3.5 GeV storage ring) using a double-crystal Si (111) monochromator. Powder samples were uniformly loaded into aluminum holders with Kapton tape encapsulation to prevent atmospheric oxidation.

Electrochemical evaluations

Catalyst inks were prepared by ultrasonically dispersing 5.0 mg of catalyst in a homogeneous mixture consisting of 495 μL of ethanol, 495 μL of deionized water, and 10 μL of Nafion solution (5 wt%). All electrochemical measurements were conducted using a CHI 760C workstation (Chenhua Instruments, Shanghai) equipped with an ALS RRDE-3A rotating electrode system. A standard three-electrode system was employed with the following configuration: A glassy carbon rotating disk electrode (GC-RDE, 4 mm diameter) modified with catalyst ink (catalyst loading: 0.4 mg cm⁻²) was the working electrode; Ag/AgCl (3 M KCl) and a Pt wire (d = 0.5 mm, L = 23 cm) were used as the reference electrode and counter electrode, respectively. All potentials were converted to the reversible hydrogen electrode (RHE) using the Nernst equation:⁴³

E_RHE = E_Ag/AgCl^θ + 0.059 pH + E_Ag/AgCl without iR correction.

Cyclic voltammetry (CV) and linear sweep voltammetry (LSV) were conducted in O₂-saturated 0.1 M KOH. The scan rates for CV are 100 mV s⁻¹ (for steady-state polarization) and 50 mV s⁻¹ (for data collection) within 0.2–1.2 V vs. RHE. For LSV, a 10 mV s⁻¹ scan rate with electrode rotation rates from 400 to 1600 rpm was used. The electron transfer number is calculated using the following Koutecky–Levich (K–L) equation:⁴⁴

where J represents the experimentally measured current density, while J_K and J_L denote the kinetic current density and diffusion-limiting current density, respectively. ω stands for the electrode rotating rate. F corresponds to the Faraday constant (96 485 C mol⁻¹). The oxygen-related parameters include , the bulk concentration of dissolved oxygen in 0.10 M KOH electrolyte (1.2 × 10⁻⁶ mol cm⁻³), and D_O2, the oxygen diffusion coefficient in the same medium (1.9 × 10⁻⁵ cm² s⁻¹). The kinematic viscosity of the electrolyte (ν) was maintained at 0.01 cm² s⁻¹ throughout the measurements.

The H₂O₂ yield and electron transfer number (n) were quantified using rotating ring-disk electrode (RRDE) voltammetry through the following relationships:

where I_D corresponds to the faradaic current measured at the disk electrode, I_R represents the ring electrode current, and N denotes the pre-calibrated collection efficiency of the Pt ring (N = 0.44).

DFT calculations

First-principles theoretical calculations based on density functional theory (DFT) were performed using the Vienna Ab initio Simulation Package (VASP). In these calculations, the Perdew–Burke–Ernzerhof (PBE) functional of the generalized gradient approximation (GGA) was adopted to describe the exchange–correlation functional, and the interactions between ionic cores and valence electrons were described by the projector-augmented wave (PAW) method. A plane-wave cutoff energy of 500 eV was set to ensure the accuracy of the calculations, and the Brillouin zone integration was sampled with a Monkhorst–Pack k-point mesh of 3 × 3 × 1. The vacuum layer was set at 15 Å to avoid interactions between the periodically repeated structures in the z-direction. The convergence criteria of structure optimization were chosen as the maximum force on each atom of less than 0.02 eV Å⁻¹ with an energy change of less than 1 × 10⁻⁵ eV.

Electrochemical assembly and performance evaluation of Zn–air batteries

The liquid Zn–air battery was configured with a sandpaper-polished zinc plate (99% purity) anode pretreated by acid activation, a catalyst-coated gas diffusion layer cathode, and a 6 M KOH electrolyte. Homogeneous catalyst inks were prepared by dispersing 4.6 mg Fe₁Zn₁-SNC-II or 20 wt% Pt/C (2.3 mg) in 800 μL ethanol containing 10 μL Nafion (5 wt%). After 30 min of ultrasonication, the resulting colloidal suspension was airbrushed onto carbon paper substrates achieving a Fe₁Zn₁-SNC-II loading density of 2.0 mg cm⁻² or Pt/C (1.0 mg cm⁻²). The electrodes were thermally stabilized at 60 °C for 3 h in a vacuum. The performance was evaluated using a LAND-CT2001A test station under ambient conditions. The current densities for variable current discharge are set at 5, 10, 20, 30 mA cm⁻², respectively. The polarization curves were measured with a scan rate of 10 mV s⁻¹. The long-term cycling test was carried out with a period of 10-min charge and 10-min discharge in the electrolyte dissolved with 0.2 M ZnO.

Data availability

All data are available within the article (and the ESI†).

Author contributions

Z. Li and H. Li conceived the project and designed experiments. Z. Li, F. Meng and H. Li performed data analysis and revised the manuscript. T. Wang and W. Kang conducted the experiments. T. Wang, Z. Li, R. Li, K. Qu and L. Wang jointly analyzed the data. Z. Li and H. Li wrote and revised the manuscript. Z. Li, F. Meng and H. Li obtained the funding sources. All authors have approved the final version of the manuscript.

Conflicts of interest

The authors declare that they have no conflict of interest, nor do they have a competing financial interest.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (No. 22305109), the Natural Science Foundations of Shandong Province (No. ZR2023QB069, ZR2019MB064 and ZR2024QB046), the Young Talent of Lifting Engineering for Science and Technology in Shandong (No. SDAST2024QTA033) and the Research Projects of Liaocheng University (No. 318052271 and 318050021).

Notes and references

1. X. Wang, M. Yu and X. Feng, eScience, 2023, 3, 100141.
2. S. Zhang, M. Chen, X. Zhao, J. Cai, W. Yan, J. C. Yen, S. Chen, Y. Yu and J. Zhang, Electrochem. Energy Rev., 2021, 4, 336–381.
3. K. Zeng, X. Zheng, C. Li, J. Yan, J.-H. Tian, C. Jin, P. Strasser and R. Yang, Adv. Funct. Mater., 2020, 30, 2000503.
4. Y.-C. Ting, C.-C. Cheng, F.-Y. Yen, G.-R. Li, S.-I. Chang, C.-H. Lee, H.-Y. T. Chen and S.-Y. Lu, EnergyChem, 2024, 6, 100134.
5. H. Liu, L. Jiang, Y. Liu, B. Lu, L. Li, Y. Tang, Y. Sun and J. Zhou, Appl. Catal., B, 2025, 362, 124705.
6. B. Singh, M. B. Gawande, A. D. Kute, R. S. Varma, P. Fornasiero, P. McNeice, R. V. Jagadeesh, M. Beller and R. Zbořil, Chem. Rev., 2021, 121, 13620–13697.
7. Y. Zhou, R. Lu, X. Tao, Z. Qiu, G. Chen, J. Yang, Y. Zhao, X. Feng and K. Müllen, J. Am. Chem. Soc., 2023, 145, 3647–3655.
8. G. Qin, S. Sun, X. Zhang, Z. Han, Y. Li, G. Han, Y. Li and S. Zhu, Energy Storage Mater., 2024, 69, 103421.
9. K. Wu, S. Li, C. Hu, G. Wen, X. Zeng, M. Wang, J. Wang, M. Chu, H. Shang, M. Ye, W. Wang, F. Wu and S. Peng, Appl. Catal., B, 2024, 357, 124288.
10. L. Xie, W. Zhou, Z. Qu, X. Meng, Y. Huang, X. Zhang, C. Yang, J. Li, J. Li, F. Sun, J. Gao and G. Zhao, J. Mater. Chem. A, 2025, 13, 1788–1795.
11. H. Huang, D. Yu, F. Hu, S.-C. Huang, J. Song, H.-Y. Chen, L. L. Li and S. Peng, Angew. Chem., Int. Ed., 2022, 61, e202116068.
12. K.-M. Zhao, D.-X. Wu, W.-K. Wu, J.-B. Nie, F.-S. Geng, G. Li, H.-Y. Shi, S.-C. Huang, H. Huang, J. Zhang, Z.-Y. Zhou, Y.-C. Wang and S.-G. Sun, Nat. Catal., 2025, 8, 422–435.
13. S. Yin, Y.-N. Yan, L. Chen, N. Cheng, X. Cheng, R. Huang, H. Huang, B. Zhang, Y.-X. Jiang and S.-G. Sun, ACS Nano, 2024, 18, 551–559.
14. C. Shao, J. Hua, Q. Li, Y. Xia, L. Sun, L. Wang and B. Li, Nano Energy, 2024, 126, 109668.
15. X. Xu, H. Wu, Y. Yan, Y. Zheng, Y. Yan, S. Qiu, T. Liang, C. Deng, Y. Yao, J. Zou and M. Liu, Adv. Energy Mater., 2025, 2405236.
16. Q. Ji, L. Bi, J. Zhang, H. Cao and X. S. Zhao, Energy Environ. Sci., 2020, 13, 1408–1428.
17. Q. Li, J. Zhao, P. Li, Z. Xu, J. Feng, B. Chen and R. Liu, ACS Nano, 2024, 18, 21975–21984.
18. L. Zong, K. Fan, P. Li, F. Lu, B. Li and L. Wang, Adv. Energy Mater., 2023, 13, 2203611.
19. Z. Li, Y. Yan, M. Liu, Z. Qu, Y. Yue, T. Mao, S. Zhao, M. Liu and Z. Lin, Proc. Natl. Acad. Sci. U. S. A., 2023, 120, e2218261120.
20. Y. Fu, Q. Zhao, Q. Wei, C. R. Bowen, W.-Y. Wong and W. Yang, Mater. Sci. Eng., R, 2024, 160, 100822.
21. R. K. Singh, J. C. Douglin, V. Kumar, P. Tereshchuk, P. G. Santori, E. B. Ferreira, G. Jerkiewicz, P. J. Ferreira, A. Natan, F. Jaouen and D. R. Dekel, Appl. Catal., B, 2024, 357, 124319.
22. D. Deng, W. Zhang, J. Qian, Y. Chen, C. Pu, H. Li, H. Li and L. Xu, Nano Energy, 2025, 134, 110579.
23. P. Sabhapathy, P. Raghunath, A. Sabbah, I. Shown, K. S. Bayikadi, R.-K. Xie, V. Krishnamoorthy, M.-C. Lin, K.-H. Chen and L.-C. Chen, Small, 2023, 19, 2303598.
24. F. Zheng, Y. Guo, T. Wang, X. Meng, K. Qu, L. Wang, R. Li, Z. Li, W. Kang and H. Li, Chem. Eng. J., 2024, 498, 155487.
25. X. Yan, Y. Tu, H. Yuan, Y. Xia, Y. Jiang, S. Zhu, C. Li, H. Tang, P. Du and M. Lei, Adv. Compos. Hybrid Mater., 2023, 6, 71.
26. X. Yang, R. Takada, X. Li, K. Narimatsu, K. Miyake, Y. Uchida and N. Nishiyama, J. Mater. Chem. A, 2025, 13, 13884–13897.
27. Y. Wang, Z. Li, P. Zhang, Y. Pan, Y. Zhang, Q. Cai, S. R. P. Silva, J. Liu, G. Zhang, X. Sun and Z. Yan, Nano Energy, 2021, 87, 106147.
28. Z. Li, S. Ji, C. Wang, H. Liu, L. Leng, L. Du, J. Gao, M. Qiao, J. H. Horton and Y. Wang, Adv. Mater., 2023, 35, 2300905.
29. C. Cheng, C. Zhi, Z.-T. Sun, Y.-Q. Ming, T.-T. Xiang, Q.-C. Zhu, Z.-R. Wu, B. Li, Y. Li, C. Jin, Y. Cao and J. Yang, Rare Met., 2024, 43, 1524–1536.
30. C. Li, Y. Zhang, M. Yuan, Y. Liu, H. Lan, Z. Li, K. Liu and L. Wang, Chem. Eng. J., 2023, 471, 144515.
31. T. Zhou, Y. Guan, C. He, L. Zhang, X. Sun, Z. Song, Q. Zhang, C. He, X. Jiang, Z. Luo, W. Xing and X. Ren, Carbon Energy, 2024, 6, e477.
32. M. Ma, Z. Pei, Y. Peng, R. Hua, Y. Ma, W. Wang, X. Xu, X. Zhang and H. Zheng, J. Mater. Chem. A, 2025, 13, 7529–7538.
33. Y. Xiong, H. Li, C. Liu, L. Zheng, C. Liu, J.-O. Wang, S. Liu, Y. Han, L. Gu, J. Qian and D. Wang, Adv. Mater., 2022, 34, 2110653.
34. C. Liu, R. Yang, J. Wang, B. Liu, X. Chang, P. Feng, X. Zhang, L. Zhong, X. Zhao, L. Niu, S. Gan, Y. Xi, M. Huang and H. Wang, Angew. Chem., Int. Ed., 2025, e202501266.
35. Q. Jin, C. Wang, Y. Guo, Y. Xiao, X. Tan, J. Chen, W. He, Y. Li, H. Cui and C. Wang, Adv. Sci., 2023, 10, 2302152.
36. H. Li, S. Di, P. Niu, S. Wang, J. Wang and L. Li, Energy Environ. Sci., 2022, 15, 1601–1610.
37. L. Yin, S. Zhang, M. Sun, S. Wang, B. Huang and Y. Du, Adv. Mater., 2023, 35, 2302485.
38. X. Guo, M. Duan, J. Zhang, B. Xi, M. Li, R. Yin, X. Zheng, Y. Liu, F. Cao, X. An and S. Xiong, Adv. Funct. Mater., 2022, 32, 2209397.
39. P. Zhang, H.-C. Chen, H. Zhu, K. Chen, T. Li, Y. Zhao, J. Li, R. Hu, S. Huang, W. Zhu, Y. Liu and Y. Pan, Nat. Commun., 2024, 15, 2062.
40. L. Xie, W. Zhou, Z. Qu, Y. Huang, L. Li, C. Yang, J. Li, X. Meng, F. Sun, J. Gao and G. Zhao, Nanoscale Horiz., 2025, 10, 322–335.
41. J. A. Zamora Zeledón, M. B. Stevens, G. T. K. K. Gunasooriya, A. Gallo, A. T. Landers, M. E. Kreider, C. Hahn, J. K. Nørskov and T. F. Jaramillo, Nat. Commun., 2021, 12, 620.
42. L. Yan, B. Xie, C. Yang, Y. Wang, J. Ning, Y. Zhong and Y. Hu, Adv. Energy Mater., 2023, 13, 2204245.
43. Y.-C. Ting, C.-C. Cheng, S.-H. Lin, T.-Y. Lin, P.-W. Chen, F.-Y. Yen, S.-I. Chang, C.-H. Lee, H.-Y. T. Chen and S.-Y. Lu, Energy Storage Mater., 2024, 67, 103286.
44. Z. Li, Z. Tian, H. Cheng, T. Wang, W. Zhang, Y. Lu, Y. Lai and G. He, Energy Storage Mater., 2023, 59, 102764.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta03390j

---