Breaking the electronic distribution symmetry at Fe–N₄ sites in iron phthalocyanines enhances CO₂ electrochemical reduction

Abstract

Iron phthalocyanines (FePc) feature a typical two-dimensional plane-symmetric structure and a symmetric electron distribution at the well-defined Fe–N₄ sites, resulting in low selectivity for CO₂ conversion to CO. Theoretical calculations reveal that the introduction of axially coordinated N atoms onto the Fe–N₄ motifs can break the electron density symmetry, facilitating electron transfer to CO₂. This enhances CO₂ adsorption and activation while reducing the binding energy of the CO intermediate. To validate these findings, a facile pyrolysis-free co-doping strategy is employed to fabricate the axial N-coordinated Fe–N₄ atomic configuration (Fe–N₅), identified as the active site. The synthesized Fe–N₅ structure exhibits excellent CO₂RR performance for CO production, achieving a selectivity of 96% and a turnover frequency of 5283 h⁻¹. This work provides a pyrolysis-free approach to optimize the local micro-environment of active sites for superior performance.

---

1 Introduction

The CO₂ reduction reaction (CO₂RR) is one of the keys to converting CO₂ into reusable energy and alleviating environmental issues.^1,2 Yet, the stability of CO₂ and the complex electron–proton transfer processes pose challenges.^3,4 Single transition metal-doped nitrogen–carbon materials have shown great potential as CO₂RR catalysts due to their high atom utilization, tunable coordination, and cost-effectiveness.⁵ Among them, iron-based single-atom catalysts (Fe-based SACs) show lower applied potentials for converting CO₂ to CO, demonstrating a significant cost advantage for syngas generation.^6,7 Iron macrocycle complexes like iron phthalocyanines and porphyrins, with well-defined Fe–N₄ configurations, offer consistent active centers and reaction environments, enhancing electrocatalytic performance.⁸ Additionally, the central iron site in the concave molecular geometry of Fe–N₄ SACs can bind axial ligands, further modulating the electronic structure to tune the adsorption of key intermediates and promote favourable reaction pathways.⁹ However, due to the inherent stability of the molecular geometry, the preparation of Fe-based SACs from such complexes remains difficult.

The commonly used pyrolysis method for Fe-based SAC preparation is generally carried out at high temperature and pressure, which damages the Fe–nitrogen configurations and causes the thermodynamically unstable Fe atoms to agglomerate as crystal particles according to their ratio in precursors. Therefore, the unpredictable Fe–nitrogen configurations lead to deviations in catalytic performance evaluation from the intrinsic activity of Fe-based SACs, and may result in the failure of iron atom immobilization under the applied potentials.^10,11 Although advanced characterization techniques like X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) can analyze the surface structure and nitrogen coordination number in Fe-nitrogen configurations of the ensemble Fe-based SACs,¹² straightforward and repeatable strategies are still needed to precisely manipulate the coordination environment of the Fe center at the atomic level. Instead of randomly creating Fe–nitrogen configurations in Fe-based SACs through pyrolysis, pyrolysis-free approaches anchor well-designed Fe-based SACs like iron phthalocyanine (FePc) on high-conductivity supports via intermolecular interactions. Therefore, well-designed Fe-based SACs prepared through a pyrolysis-free approach to regulate catalytic sites could overcome the problems of unpredictable Fe–nitrogen configurations and agglomeration of Fe atoms, thereby enhancing the understanding of their electrocatalytic mechanisms in CO₂RR. However, Fe-based SACs generally exhibit low selectivity in the CO₂RR, especially when FePc exhibits an Fe–N₄ configuration on the carbon matrix from the pyrolysis-free approach, compared to cobalt and nickel phthalocyanines.^6,7

According to the molecular geometry of Fe-based SACs, the typical two-dimensional and plane-symmetric structure leads to a symmetrical electron distribution around Fe–N₄ sites. This symmetry in the electronic structure of the Fe atom restricts CO₂ adsorption and activation, leading to strong binding of *CO intermediates, which can poison the active sites.^7,8,12–14 To improve the catalytic performance of Fe-based SACs, a novel strategy based on support modification was employed to disrupt the symmetry of the electronic structure. This strategy enables covalent grafting of molecular linkers with nitrogen-functional groups (e.g., pyridinic) onto a support, followed by metalation via axial ligation.¹⁵ The molecular linkers not only facilitate the distribution and immobilization of Fe-based SACs via strong intermolecular interaction but also induce the reconstruction of the electron structure at the Fe center. As reported in previous studies, coordinating Fe–N₄ of FePc with organic ligands rich in electron-donating groups (nitrogenous and oxygenic) can enhance the catalytic performance of the oxygen reduction reaction (ORR), which shares an analogous step with the hydrodeoxygenation in CO₂RR.^14,16–21 Pyrene groups can form strong π–π coordination between the FePc catalyst and supports,²² but they may impede direct electron transfer. Therefore, using an appropriate modification on the support can strengthen the covalent bond to the metal active center and promote inner-sphere electron transfer,²¹ disrupting the symmetric electronic density of Fe–N₄ sites.

In this study, a straightforward pyrolysis-free method was proposed to introduce native axial nitrogen to the Fe atom of Fe-based SACs. While the nitrogen-doped hierarchically mesoporous carbon support (NHPCs) is prepared through conventional thermal treatment, the subsequent assembly of FePc with axial nitrogen occurs without pyrolysis, maintaining the FePc molecular structure. For proof-of-concept, FePc with the typical Fe–N₄ configuration is employed to rivet the Fe center to NHPCs through a simple pyrolysis-free method, yielding the Fe–N₅/NHPCs catalyst. The strong interaction between FePc and NHPCs enabled effective axial N coordination to the Fe atom in FePc, forming the Fe–N₅ active site. XAS verified that each single Fe atom in Fe–N₅/NHPCs is bonded to five nitrogen atoms. DFT calculations indicated that the axial N coordination at the Fe–N₅ site significantly distorts the symmetry of Fe hybrid states, localizing electrons on the coordinated N atom, which promotes CO₂ adsorption and activation. The Fe–N₅/NHPCs catalyst achieved 96% selectivity for CO₂RR to CO, with a current density of 3.6 mA cm⁻² and a TOF of 5282 h⁻¹ at −0.7 V (vs. RHE) in a H-cell. Notably, it maintained stability for 20 hours at this current.

2 Experimental section

2.1 Synthesis of NHPCs

NHPCs were synthesized by adopting the general procedures from previous work²¹ with modifications. Briefly, 4 mL of aqueous ammonia solution and 50 mL of deionized water were mixed in 20 mL ethanol under stirring for 30 minutes at 30 °C. Then, 2 mL of tetraethyl orthosilicate (TEOS) was added dropwise using a syringe, followed by vigorous stirring for another 30 minutes. Next, 1.2 g of polyethylene glycol (PEG-20000), 0.9 g of the triblock copolymer pluronic F127 (PEO₁₀₆PPO₇₀PEO₁₀₆) and 0.68 mL of formaldehyde were added into the above mixture, and stirred for 30 minutes, and then 0.945 g of melamine was added followed by 0.42 mL of formaldehyde. The mixture was stirred for 24 hours at 30 °C before being transferred to a 100 mL Teflon-lined hydrothermal autoclave and heated at 100 °C for 12 hours. The resulting samples were collected via centrifugation, washed three times with deionized water and ethanol individually, and then dried under vacuum at 70 °C. The dried powder was then heated at 800 °C for 2 hours at a heating rate of 2 °C per minute under flowing argon (Ar) gas and allowed to cool naturally to room temperature. Finally, the obtained black powder was etched with hydrofluoric acid (HF) for 48 hours to remove silicon and the obtained NHPC support was dried at 70 °C for 12 hours in a vacuum.

2.2 Synthesis of Fe–N₅/NHPCs

12 mg of FePc and 20 mg of NHPCs were separately dissolved in 20 mL of dimethylformamide (DMF). The two solutions were mixed together by ultrasonication for 1 hour. The FePc-DMF dispersed solution was then added dropwise to the NHPCs-DMF suspension while stirring vigorously. After stirring continuously for 24 hours at 30 °C, the composite samples were collected by centrifugation and washed with DMF repeatedly until the solution became colourless. The samples were then dried in a vacuum oven at 60 °C for 12 hours, resulting in the final product. It is worth noting that different amounts of FePc were employed to optimize the hybrid catalyst, specifically, 8 mg, 12 mg, and 16 mg. Among these, 12 mg was found to be the optimal amount for our catalyst, denoted as Fe–N₅/NHPCs. For comparison, FePc/NHPCs (a physical mixture of FePc and NHPCs) and Fe–N₅/NHPCs carbonized at 560 °C were also synthesized.

3 Results and discussion

3.1 Theoretical calculations

DFT calculations were performed to explore the state symmetry breaking around the Fe atom in FePc for the CO₂RR process. Optimized models and state distributions of FePc, FePc/NHPCs (without N coordination), and Fe–N₅/NHPCs (with axial N coordination) are shown in Fig. S1. FePc and FePc/NHPCs display symmetric charge distribution around the Fe atom, while Fe–N₅/NHPCs shows strong state localization on the axial N atom and asymmetric states around Fe–N₅ sites (Fig. 1a and S1). This electron localization on the axial N atom in Fe–N₅/NHPCs occurs because NHPCs transfer partial electrons to the axial N atom, as evidenced by the electron localization function (ELF) differences (Fig. S1). Although the weak van der Waals interaction between Fe–N₅ and NHPCs indicates that the charge and spin density of Fe–N₅ change little compared to the Fe–N₄ site in FePc (Fig. S2), the axial N coordination still accepts electron donation from NHPCs. This breaks the electronic state symmetry near the Fe–N₄ site and significantly alters its spin polarization, confirming that axial N coordination is crucial for state symmetry-breaking.^{12,14,23–25}

The interaction between the active site and adsorbed CO₂ was investigated through density of states (DOS), ELF, CO₂ activation energy, and Bader charge analysis. ELF maps revealed that axial N coordination promoted electron transfer from axial N to the Fe atom, forming the Fe_*CO₂ bond.¹³ This process contracted the state tightly, creating an apparent covalent interaction zone with an ELF value of about 0.2 (Fig. 1b and S1). Consequently, Fe–N₅/NHPCs exhibits relatively stronger covalent binding interaction for *CO₂ adsorption compared to FePc, enhancing the state symmetry disruption around the Fe atom of Fe–N₅/NHPCs, while the weaker interaction in FePc maintains the perfection of Fe atom states (Fig. 1b and S1). Additionally, the ELF map shows electron fluctuation in axial N (Fig. 1a, b, and S1). That is, as shown in Fig. 1a, the states were localized on the right side with an ELF value of about 0.9, and when CO₂ was adsorbed on Fe–N₅/HNPCs (Fig. 1b), the electron was localized on the opposite side with an ELF value close to 0.3. The opposite-direction electron fluctuation patterns indicate that the electrons were depleted in the axial N atom.²⁶ This implies that Fe–N₅/NHPCs experiences electron transfer from the Fe atom to *CO₂, depleting the localized states on the axial N atom. To investigate this further, the charge density differences for *CO₂ species on the Fe–N₅/NHPCs surface were analysed. The *CO₂ species bond more firmly on the Fe–N₅ sites compared Fe–N₄ sites (Fig. S1 and inset Fig. 1c), boosting CO₂ adsorption and activation, which is consistent with previous studies showing that the Fe–N₅ configuration enhances these processes through mechanisms like strong CO₂ adsorption,^9,21 electronic modulation,^14,27 synergistic ligand effect^9,18 and enhanced intermediate stabilization.^2,21,28Fig. 1c shows that the CO₂ adsorption energies on Fe–N₅/NHPCs and FePc are 0.79 and 1.48 eV, respectively. A more positive adsorption energy suggests a stronger repulsion force between the active site and CO₂ molecules.²⁹ This discrepancy in the surface binding environment indicates that the active site of FePc binds the CO₂ molecule too weakly. Fe–N₅/NHPCs is an ideal catalyst as it binds CO₂ moderately, following the Sabatier principle.³⁰ The charge transfer from Fe–N₅/NHPCs and FePc to the adsorbed CO₂ molecules is 1.23e and 1.11e, respectively (Fig. 1c). Projected density of states (PDOS) analysis reveals significant spin asymmetry in the d-orbitals of FePc (Fig. 1d). For Fe–N₅/NHPCs, reduced unpaired electron density in the d_xz, d_yz, and d_z² orbitals results in weaker *CO binding, which in turn enhances intrinsic activity (Fig. 1e). Additionally, the d-band centre of Fe–N₅/N shifts closer to the Fermi level than that of FePc (Fig. 1d and e), suggesting improved conductivity and electron transfer, consistent with the findings from Bader charge analysis. Overall, PDOS analysis confirms that the presence of a native ligand enables an extreme inner-sphere electron transfer mechanism, facilitating the full formation of covalent bonds between the 3d states of Fe and 2p states of N (axial N atom) while distorting the local state symmetry of Fe–N₄ sites.

Fig. 1 (a and b) Electronic localization function of Fe–N₅/NHPCs and FePc with and without *CO₂ adsorption. (c) CO₂ adsorption energies and Bader charge transfer, (d and e) partial density of states, and (f) CO₂RR free energy diagrams of Fe–N₅/NHPCs and FePc.

Furthermore, the influence of electronic localization on CO₂RR processes was probed via comparative CO₂RR free energy profiles of Fe–N₅/NHPCs and FePc (Fig. 1f). For FePc, the initial CO₂ activation step (*CO₂ → *COOH) exhibits an extremely low free energy barrier of merely 0.02 eV; however, the subsequent CO desorption step is highly energetically unfavourable (1.86 eV). This severe discrepancy indicates significant product poisoning of FePc's active sites, which strongly suppresses catalytic turnover. By contrast, Fe–N₅/NHPCs features a moderate free energy barrier for *COOH formation (0.48 eV)—a value that enables efficient CO₂ activation while preventing excessive binding of the intermediate. The free energy pathway further reveals that Fe–N₅/NHPCs achieves optimized stabilization of reaction intermediates and a substantially lower barrier for CO desorption relative to FePc. These balanced energetic characteristics account for the superior CO₂RR performance of Fe–N₅/NHPCs: axial N coordination modulates the electronic environment of the Fe centre, thereby ensuring efficient CO₂ activation, stable intermediate binding, and facile CO release—three key requisites for high catalytic activity. These findings agree with the PDOS and charge density difference analyses of *CO₂ adsorption on Fe–N₅/NHPCs, suggesting that stable CO₂ adsorption and effective activation can facilitate the CO₂RR process. Based on these results, the axial N coordination in Fe–N₅/NHPCs is proposed to promote electron transfer from the Fe atom to *CO₂ and introduce state symmetry disruption around the Fe atom, enabling efficient CO₂ adsorption and activation, ultimately enhancing CO₂RR performance.

3.2 Morphology and structural characterization

Inspired by the DFT calculations, nitrogen-doped hierarchically mesoporous carbon was selected as the supporting material for FePc due to its unique properties. The mesopores enable coordination with the Fe–N₄ site of FePc via specific adsorption,³¹ while numerous negatively charged functional groups interact favourably with the positively charged Fe atom in FePc, generally through van der Waals–Debye forces initially. This interaction is stronger than the intermolecular aromatic ring stacking (π–π) interactions of FePc, facilitating an even distribution of FePc. A large surface area also promotes mass transport and suppresses FePc aggregation, enhancing catalytic performance. The Fe–N₅/NHPC catalyst was prepared by FePc deposition on NHPCs via a facile pyrolysis-free approach (Fig. 2a). NHPCs were synthesized through carbonization of the precursor at 800 °C followed by HF etching to remove silicon. Transmission electron microscopy (TEM) and scanning electron microscopy (SEM) characterization revealed the uniform dispersion of Fe atoms on NHPCs (Fig. 2b–d and S3), with the TEM image showing a hierarchical porous morphology (Fig. 2c) derived from silicon removal. High-resolution TEM (HR-TEM) images confirmed the absence of Fe nanoparticles (Fig. 2d and S3). To further verify the distribution of FePc in NHPCs, aberration-corrected high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) imaging, along with corresponding energy-dispersive X-ray spectroscopy (EDX) mapping, as displayed in Fig. 2e, demonstrated a uniform dispersion of C, N, O, and Fe atoms, confirming that FePc was uniformly immobilized on NHPCs (Fe, 0.71 at%, Table S1) and that Fe atoms in Fe–N₅/NHPCs were homogeneously distributed across the entire NHPC framework. Unlike other Fe-based CO₂RR catalysts that suffer pyrolysis and unavoidable Fe atom aggregation due to π–π interactions,⁷ Fe–N₅/NHPCs exhibited unique uniform dispersion of FePc molecules.

Fig. 2 (a) Schematic synthesis process, (b) SEM, (c) TEM, (d) HR-TEM, and (e) HAADF image, and corresponding EDX mapping for C, N, O and Fe atoms of the Fe–N₅/NHPCs catalyst. (f) FTIR spectra, (g) XRD pattern, (h) UV-vis spectra, and (i) Raman spectra of FePc, Fe–N₅/NHPCs and NHPCs.

Fourier transform infrared (FTIR), X-ray powder diffraction (XRD), ultraviolet-visible spectrophotometry (UV-vis), and Raman spectroscopy were employed to characterize the coordination interactions and structural changes of Fe–N5/NHPCs. In the FTIR spectrum of NHPCs (Fig. 2f), the strong broad bands at 1100–1630 cm⁻¹ and 3350–3500 cm⁻¹ correspond to carbon-containing nitrogen groups (CN)^30,31 and N–H/O–H stretching vibrations,³² respectively, consistent with the XPS elemental composition results (C, 82.82 atom% and N, 13.7 atom%, Table S1).³³ After FePc deposition, although the overall structure of NHPCs was maintained, several peaks in the Fe–N₅/NHPC spectrum shifted slightly compared to those of NHPCs and FePc. Specifically, the N–H/O–H absorption band in Fe–N₅/NHPCs becomes more intense and less broad, and exhibits a blue shift, indicating successful interaction between FePc and NHPCs via unsaturated nitrogen atoms.^4,38 The XRD patterns showed that the broad diffraction peak of Fe–N₅/NHPCs shifted slightly to a higher angle relative to that of NHPCs (Fig. 2g), confirming the strong FePc–NHPC interaction. The absence of distinct FePc characteristic peaks in the hybrid catalyst suggested homogeneous dispersion of FePc, while a high FePc loading led to its agglomeration, as evidenced by the emergence of characteristic peaks (Fig. S4a). UV-vis spectroscopy revealed a significant blue-shift of the FePc absorption band in Fe–N₅/NHPCs (Fig. 2h), ascribed to the extended π-electron delocalization in NHPCs,^14,32 indicating the uniform dispersion and altered environmental coordination of FePc on the NHPC surface. An intense absorption peak at 590 nm, corresponding to the ligand-to-metal charge transfer (LMCT) transitions,²⁰ further supported the coordination between FePc and NHPCs. Raman analysis exhibits that the I_D/I_G ratio of 1.19 on Fe–N₅/NHPCs was higher than the 0.99 observed on NHPCs (Fig. 2h), suggesting the presence of more defects or disordered structures in Fe–N₅/NHPCs. We deduce that this phenomenon is likely due to the electronic interaction between Fe and N atoms, which induces significant distortion.

XPS was utilized to explore the chemical composition and electron interactions of the Fe–N₅/NHPCs catalyst. The high-resolution Fe 2p spectra (Fig. 3a) show that the Fe 2p peak position in Fe–N₅/NHPCs shifts to a lower binding energy compared to that of FePc. This suggests that the interaction between NHPCs and Fe atoms modifies the electronic structure and alters the charge density of central Fe atoms.¹² The presence of an additional ligand increases the electron states around the Fe atom, leading to a less positive valence state than that in FePc.^13,34,35 The binding energy shifts towards a lower value, further implying that the electron transfer is from the NHPC substrate to Fe–N₅ sites via the axial N atom as shown in Fig. 1c. For the high-resolution N 1s spectra of NHPCs, deconvolution reveals four peaks at 404.7, 402.3, 400.3, and 398.2 eV, assigned to pyridinic N-oxide, graphitic N, pyrrolic N, and pyridinic N, respectively,^37–39 as displayed in Fig. 3b. Pyrrolic N is sp³ hybridized whereas graphitic and pyridinic N are sp² hybridized.^36,37 Previous studies suggest that pyrrolic N species, being more unsaturated, exhibit a stronger coordination affinity with central metal atoms than other N species.^36,37,39

Fig. 3 (a) High-resolution Fe 2p XPS spectra, (b) N 1s XPS spectra, (c) CO₂ adsorption isotherms measured at 301 K and (d) pore size distribution of Fe–N₅/NHPCs, NHPCs and FePc as indicated in each respective figure.

As depicted in Fig. 3b, upon compositing with FePc, the intensity of the pyrrolic N peak in NHPCs significantly decreases, while the pyridinic N peak becomes predominant. A new peak emerges at 399.7 eV in Fe–N₅/NHPCs, attributed to the formation of Fe–N_x(x≥4) bonds^19,40via coordination between pyrrolic N in NHPCs and FePc. The calculated pyrrolic N contents in NHPCs and Fe–N₅/NHPCs are 5.99 atom % and 4.19 atom%, respectively, consistent with the distinct N 1s high-resolution XPS spectra of FePc and Fe–N₅/NHPCs. This confirms the presence of chemical bonding between FePc and N species in NHPCs (Fig. S5a), in line with UV-vis results. The decrease in pyrrolic N content in Fe–N₅/NHPCs indicates coordination with FePc, whereas the increase in pyridinic N suggests that FePc does not coordinate with pyridinic N, likely because the lone pair of electrons in pyridinic N forms an active site for ORR upon metal-atom bonding.^35,36,40,41 Theoretical investigations of CO₂RR pathways on Fe–N₅/NHPCs with different N-coordination types (Fig. S5b) reveal that the Fe–N₅ site coordinated with pyrrolic N (Fe–N₅/NHPCs (N-5)) exhibits a lower potential barrier for converting adsorbed *CO₂ to *COOH and a significantly higher CO desorption energy (0.36 eV) compared to graphitic (Fe–N₅/NHPCs(NQ), −0.63 eV) and pyridinic (Fe–N5/NHPCs(N-6), −0.33 eV) coordination.^42,43 These results highlight the superior coordination ability of pyrrolic N for Fe–N₅/NHPCs in the CO₂RR.

To evaluate CO₂ adsorption on different Fe-N_x configurations, the CO₂ adsorption–desorption performances were measured on Fe Fe–N₅/NHPCs and FePc at room temperature (Fig. 3c and S6a). Fe–N₅/NHPCs exhibits a more robust CO₂ adsorption response than FePc. The nitrogen adsorption–desorption experiments revealed the surface area and porosity characteristics of NHPCs and Fe–N₅/NHPCs, as shown in Fig. 3d and S6b. The rapid nitrogen uptake at low relative pressures (P/P₀ < 0.1) and type IV isotherm curves indicate the presence of mesopores. Based on the pore size distributions in Fig. 3d, S6c and d, NHPCs have hierarchically mesoporous structures with a peak at ∼10 nm and a pore size range of 2–60 nm, matching well with the TEM results. Upon FePc immobilization, the pore size distribution peak shifts to ∼5.9 nm. The presence of numerous mesopores facilitates the reaction intermediates transport and electrolyte flow during CO₂RR.⁴⁴ The specific surface areas of NHPCs and Fe–N₅/NHPCs are 371.6 and 416.7 m² g⁻¹, respectively (Table S2). Compared to NHPCs and FePc, Fe–N₅/NHPCs exhibit higher a surface area and larger pore volumes, which are expected to expose more active sites and facilitate mass transport during electrochemical reactions.^44–46

To elucidate the electronic structure and local coordination environment of Fe atoms in Fe–N₅/NHPCs, Fe K-edge XAS was performed. The X-ray absorption near-edge structure (XANES) spectra of Fe–N₅/NHPCs, FePc, Fe₂O₃, and Fe foil mirror the trends observed in their XPS spectra (Fig. 4a). Relative to Fe foil, the pre-edge peak in Fe–N₅/NHPCs and FePc (Fig. 4a, region 1), arises from a 1s to 4p_z shakedown transition characteristic of square-planar D_4h local symmetry.^45,46 The intensified pre-edge peak in Fe–N₅/NHPCs, compared to FePc, indicates distorted D_4h symmetry due to pentacoordinate Fe–N₅ motifs.^46,47 Additionally, the negative shift of near-edge and white line features in Fe–N₅/NHPCs (Fig. 4a, region 2) confirms a lower Fe oxidation state (+1.62 vs. +2 in FePc), consistent with XPS results.^48,49 These electronic changes arise from NHPCs–FePc interactions and coordination geometry modifications.^23,37

Fig. 4 (a) XANES of Fe₂O₃, Fe foil, FePc, and Fe–N₅/NHPCs. (b) FT-EXAFS of Fe₂O₃, Fe foil, FePc, and Fe–N₅/NHPCs in R space, and (c) EXAFS fitting curves of Fe–N₅/NHPCs. (d) WT-EXAFS of Fe–N₅/NHPCs and control samples at the Fe K-edge.

In Fig. 4b and S7a, the Fe K-edge Fourier-transformed (FT) k³-weighted extended X-ray absorption fine structure (EXAFS) of Fe–N₅/NHPCs and control samples reveal predominant peaks at 1.50 (Fe–N₄ in FePc) and 1.41 Å (Fe–N₅), respectively. The shorter bond in Fe–N₅/NHPCs indicates distorted D_4h symmetry. The absence of a ca. 2.2 Å peak (Fe–Fe scattering) confirms the atomic dispersion of Fe sites, consistent with HR-TEM and XRD results (Fig. 2d and g). EXAFS fitting (Table S3) yields a five-coordinate first shell (Fe–N₅), supported by excellent agreement between experimental and simulated curves (Fig. 4c). The wavelet transform (WT)-EXAFS analysis (Fig. 4d) further validates these findings. Collectively, these results confirm the successful interconnection of FePc and NHPCs via axial N coordination, inducing electron localization at Fe–N₅ sites and distorting the local symmetry around Fe–N₄ motifs.

3.3 Electrochemical CO₂RR of the Fe–N₅/NHPCs catalyst

Building on the Fe–N₅/NHPCs structure, its CO₂RR activity was evaluated in CO₂-saturated 0.5 M KHCO₃. Linear sweep voltammetry (LSV, Fig. 5a) reveals that Fe–N₅/NHPC exhibits superior current density compared to NHPCs and FePc. This enhancement is attributable to the larger electrochemically active surface area (Fig. S8a) and efficient electron transfer enabled by FePc–NHPC interactions, consistent with prior reports.^29,40 Furthermore, the peak current density increases linearly with the scan rate (Fig. S8a), indicating that the peak originates from a CO₂ mass transfer-controlled electrolysis process.^50–52 Comparative LSV curves in CO₂⁻vs. Ar-saturated electrolytes (Fig. 5b) demonstrate a substantial CO₂RR current increase under CO₂, underscoring the dominant role of Fe–N₅ sites in CO₂ activation.

Fig. 5 CO₂RR electrocatalytic performance in 0.5 M KHCO₃ electrolytes saturated with CO₂. (a) LSV curves of NHPCs, FePc, and Fe–N₅/NHPCs. (b) LSV curves of Fe–N₅/NHPCs in Ar- and CO₂-saturated KHCO₃ (dotted and continuous lines in Ar and CO₂, respectively). (c) FE_CO, and (d) FE_H2 of NHPCs, FePc, and Fe–N₅/NHPCs against applied potentials. (e) Stability test of Fe–N₅/NHPCs at −0.7 V. (f) Partial current density of CO for FePc and Fe–N₅/NHPCs at different applied potentials.

To examine the selectivity of the Fe–N₅/NHPCs catalyst, electrolysis tests were performed in a gastight H-cell (Fig. S8–S9).⁵³ The Fe–N₅/NHPCs outperformed FePc across all potentials, achieving a maximum CO faradaic efficiency (FE_CO) of 96% at −0.7 V (vs. 20.5% for FePc) with >90% FE_CO from −0.6–−0.8 V (Fig. 5c). The FE_H2 of Fe–N₅/NHPCs is suppressed compared to those of FePc and NHPCs (Fig. 5d), reflecting uniform FePc dispersion. Notably, the Fe–N₅/NHPCs presents sixfold higher current density (3.6 mA cm⁻², normalized by the geometrical area) than FePc (0.6 mA cm⁻²) at −0.7 V. While the Fe–N₅/NHPCs catalyst has not yet matched the current density or operational durability of state-of-the-art CO₂RR systems, it delivers valuable insights into axial ligand coordination as a tuneable strategy for engineering single-atom active sites under mild synthesis conditions. This work thereby establishes a promising direction for the rational design of next-generation CO₂RR catalysts (Table S4). A stability test of the Fe–N₅/NHPCs shows no obvious decay in the current density or FE_CO (>90%) over 20 h (Fig. 5e), with minimal structural changes (Fig. S10), confirming the robust Fe–N₅ sites.¹⁴ To further assess the catalytic performance, Fe–N₅/NHPCs were tested at varying current densities in a gas-diffusion electrode (flow cell) setup. At a low current density of 2 mA cm⁻², the faradaic efficiency for CO (FE_CO) reached 97%; however, FE_CO decreased progressively as the current density increased (Fig. S11a). This trend implies that under high current density conditions (e.g., 100 mA cm⁻²), the Fe–N₅ active sites may undergo partial structural transformation—either converting to Fe–N₄-type configurations (e.g., pristine FePc) or losing their coordination environment entirely—thereby compromising CO selectivity. Post-reaction characterization after 20 h of operation revealed only partial ligand degradation, with no evidence of complete structural collapse of the catalyst (Fig. S11b), indicating a moderate level of operational stability. The NHPCs' hierarchical porosity (BET analysis) enhances active site accessibility and mass transport. Control experiments confirm negligible CO production from bare NHPCs (FE_CO <20%), Ar-saturated electrolyte, or carbon cloth alone, establishing Fe–N₅/NHPCs and CO₂ as essential for CO generation (Fig. 5c). Isotope labelling (¹³CO₂) confirms ¹³CO formation (m/z = 29, Fig. S12), verifying CO₂ as the carbon source. FE_CO increases with FePc loading up to an optimal point; however, excess FePc reduces selectivity due to aggregation, which blocks active sites and impedes electron transfer.⁵⁴

To understand the origin of the outstanding electrocatalytic performance of the synthesized catalyst, Nyquist plots were analyzed, revealing lower impedance on Fe–N₅/NHPCs than that on FePc (Fig. S13a), indicating reduced charge transfer resistance and faster electron transfer during CO₂RR, consistent with DFT predictions (Fig. 1d). Tafel slopes of 233 mV dec⁻¹ (Fe–N₅/NHPCs) and 429 mV dec⁻¹ (FePc) (Fig. S13b) suggest that the former exhibits faster reaction kinetics and better *CO₂⁻ stabilization, with the first electron transfer being the rate-limiting step for FePc. Double-layer capacitance measurements (Fig. S8a) show that Fe–N₅/NHPCs has a higher C_dl (5.44 vs. 2.62 mF cm⁻²), indicating greater electrochemically active surface area (ECSA), in line with BET results. ECSA-normalized partial currents confirm higher intrinsic activity for Fe–N₅/NHPCs across all potentials (Fig. S8b). Turnover frequency (TOF) calculations further highlight its superior performance, with Fe–N₅/NHPCs achieving 5283 h⁻¹ at −0.7 V compared to 334 h⁻¹ for FePc.

To further investigate the structure–activity relationship, Fe–N₅/NHPC was pyrolyzed at 560 °C for 2 h under Ar (its decomposition temperature¹⁴), yielding a maximum FE_CO of 53.1% at −0.6 V, substantially lower than that of the unpyrolyzed sample (>90%, Fig. S8c), indicating that structural degradation of Fe–N₅/NHPCs is responsible for the FE_CO attenuation. XRD analysis of Fe–N₅/NHPCs revealed that heating the material to 560 °C induced a ∼4.2° shift of the main XRD peaks toward lower 2θ values (Fig. S8f), indicative of a notable increase in interplanar spacing. This lattice expansion is attributed to the onset of FePc thermal decomposition at 560 °C,¹⁴ which triggers distortion of the macrocyclic structure of FePc. Such structural distortion relaxes the catalyst lattice and enlarges the d-spacing surrounding the Fe–N₅ coordination sites. These observations clearly demonstrate that pyrolysis at this temperature disrupts the Fe–N₅ coordination configuration and reduces the accessibility of isolated Fe–N₅ active sites—ultimately resulting in decreased electrocatalytic CO₂RR performance, as evidenced by the activity data in Fig. S8c. Furthermore, physical mixtures of FePc and NHPCs (FePc/NHPCs, Fig. S1 and S8d) and FePc-supported porous carbon (HPCs, prepared without N sources) were also evaluated. For HPCs-Air and HPCs-Ar (carbonized in air or Ar, respectively), the major product was H₂, with FE_CO maxima of 6.5% (−0.5 V) and 17.6% (−0.6 V), respectively (Fig. S9a–d). The UV-vis spectra showed that FePc in HPCs exhibits a red-shifted absorption peak (755 nm) versus Fe–N₅/NHPCs (660 nm, Fig. S9c), reflecting distinct coordination environments (axial O in HPCs vs. N in NHPCs). Oxygen's strong electron-withdrawing nature likely forms FePc–O complexes or oxidizes Fe centers,¹⁴ whereas axial N coordination in Fe–N₅/NHPCs induces symmetry-breaking around Fe, enhancing CO₂ adsorption/activation and driving superior CO₂RR performance.

4 Conclusions

In summary, a pyrolysis-free co-doping strategy is reported for synthesizing axial N-coordinated Fe–N₅ atomic configurations. The introduction of axial N coordination breaks the symmetry around Fe–N₄ sites in FePc, inducing electronic localization at Fe–N₅ sites that enhances CO₂ adsorption and activation, thereby boosting the catalytic performance of Fe–N₅/NHPCs. XPS and XAS analyses reveal a negative energy shift of the Fe atom, indicating a lower oxidation state (+1.62) in Fe–N₅/NHPCs compared to FePc (+2.0), attributed to electron transfer from axial N ligands to the Fe center. Structural characterization confirms single Fe atoms coordinated with five N atoms in a distorted D_4h local symmetry. As a result, the Fe–N₅/NHPC catalyst exhibits superior CO₂RR selectivity (FE_CO = 96%), activity (3.6 mA cm⁻² at −0.7 V), and stability (20 h), alongside a low Tafel slope (233 mV dec⁻¹) and high TOF (5283 h⁻¹). For proof of concept, these findings provide a novel strategy to tailor local coordination environments of Fe and other metal centres for enhanced CO₂RR and inspire the design of metal–N₅ structures beyond Fe-SACs.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data for this article are available at pan.baidu.com at https://pan.baidu.com/s/1jmu7HFEU6KKO-V91nVJKaQ?pwd=JJMA.

Supplementary information is available. See DOI: https://doi.org/10.1039/d5ta04508h.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (No. 2022YFB3803600), the National Natural Science Foundation of China (No. 52472205 and 12275199), the Fundamental Research Funds for the Central Universities (No. CCNU25JC002), the Hubei Provincial Natural Science Foundation of China (No. 2025EHA032). We extend our gratitude to the team at the XAFCA beamline at the Singapore Synchrotron Light Source (SSLS) for their assistance with XAS measurements.

References

1. Y. R. Bao, J. Y. Xiao, Y. K. Huang, Y. Z. Li, S. Y. Yao, M. Qiu, X. X. Yang, L. C. Lei, Z. J. Li, Y. Hou, G. Wu and B. Yang, Angew. Chem., Int. Ed., 2024, 63, e20240630.
2. J. Xiao, J. J. Masana, H. Dong, H. Zhang, H. L. Yang, M. Qiu and Y. Yu, J. Mater. Chem. A, 2023, 11, 24359–24370.
3. J. J. Masana, B. W. Peng, Z. Y. Shuai, M. Qiu and Y. Yu, J. Mater. Chem. A, 2022, 10, 1086–1104.
4. J. Xiao, J. J. Masana, M. Qiu and Y. Yu, Mater. Today Phys., 2024, 48, 101565.
5. J. F. Sun, Q. Q. Xu, J. L. Qi, D. Zhou, H. Y. Zhu and J. Z. Yin, ACS Sustainable Chem. Eng., 2020, 8, 14630–14656.
6. Z. Zhang, J. P. Xiao, X. J. Chen, S. Yu, L. Yu, R. Si, Y. Wang, S. H. Wang, X. G. Meng, Y. Wang, Z. Q. Tian and D. H. Deng, Angew. Chem., Int. Ed., 2018, 57, 16339–16342.
7. N. Corbin, J. Zeng, K. Williams and K. Manthiram, Nano Res., 2019, 12, 2093–2125.
8. Q. L. Liu, Y. F. Wang, Z. Z. Hu and Z. Q. Zhang, RSC Adv., 2021, 11, 3079–3095.
9. L. J. Zhang, N. Jin, Y. B. Yang, X. Y. Miao, H. Wang, J. Luo and L. L. Han, Nano-Micro Lett., 2023, 15, 228.
10. Z. X. Song, L. Zhang, K. Doyle-Davis, X. Z. Fu, J. L. Luo and X. L. Sun, Adv. Energy Mater., 2020, 10, 2001561.
11. J. F. Sun, Q. Q. Xu, J. L. Qi, D. Zhou, H. Y. Zhu and J. Z. Yin, ACS Sustainable Chem. Eng., 2020, 8, 14630–14656.
12. W. Xia, Z. F. Hou, J. Tang, J. J. Li, W. Chaikittisilp, Y. N. Kim, K. Muraoka, H. J. Zhang, J. P. He, B. X. Han and Y. Yamauchi, Nano Energy, 2022, 94, 106868.
13. Y. Pan, R. Lin, Y. J. Chen, S. J. Liu, W. Zhu, X. Cao, W. X. Chen, K. L. Wu, W. C. Cheong, Y. Wang, L. R. Zheng, J. Luo, Y. Lin, Y. Q. Liu, C. G. Liu, J. Li, Q. Lu, X. Chen, D. S. Wang, Q. Peng, C. Chen and Y. D. Li, J. Am. Chem. Soc., 2018, 140, 4218–4221.
14. K. J. Chen, K. Liu, P. D. An, H. J. W. Li, Y. Y. Lin, J. H. Hu, C. K. Jia, J. W. Fu, H. M. Li, H. Liu, Z. Lin, W. Z. Li, J. H. Li, Y. R. Lu, T. S. Chan, N. Zhang and M. Liu, Nat. Commun., 2020, 11, 4173.
15. H. Aga, A. Aramata and Y. Hisaeda, J. Electroanal. Chem., 1997, 437, 111–118.
16. R. Cao, R. Thapa, H. Kim, X. Xu, M. G. Kim, Q. Li, N. Park, M. L. Liu and J. Cho, Nat. Commun., 2013, 4, 2076.
17. M. A. C. de Oliveira, V. C. A. Ficca, R. Gokhale, C. Santoro, B. Mecheri, A. D'Epifanio, S. Licoccia and P. Atanassov, J. Solid State Electrochem., 2021, 25, 93–104.
18. Y. M. Zhao, P. C. Zhang, C. Xu, X. Y. Zhou, L. M. Liao, P. J. Wei, E. Liu, H. Q. Chen, Q. G. He and J. G. Liu, ACS Appl. Mater. Interfaces, 2020, 12, 17334–17342.
19. V. N. Nemykin, A. A. Purchel, A. D. Spaeth and M. V. Barybin, Inorg. Chem., 2013, 52, 11004–11012.
20. Z. B. Han, X. Han, X. M. Zhao, J. T. Yu and H. Xu, J. Hazard. Mater., 2016, 320, 27–35.
21. J. J. Masana, J. Xiao, H. Zhang, X. Y. Lu, M. Qiu and Y. Yu, Appl. Catal., B, 2023, 323, 122199.
22. A. Maurin and M. Robert, J. Am. Chem. Soc., 2016, 138, 2492–2495.
23. M. Mukherjee, M. Samanta, P. Banerjee, K. K. Chattopadhyay and G. P. Das, Electrochim. Acta, 2019, 296, 528–534.
24. Y. Wei, G. C. Xing, K. Liu, G. Li, P. Dang, S. S. Liang, M. Liu, Z. Y. Cheng, D. Jin and J. Lin, Light: Sci. Appl., 2019, 8, 1–9.
25. Y. S. Zhou, F. L. Che, M. Liu, C. Q. Zou, Z. Q. Liang, P. De Luna, H. F. Yuan, J. Li, Z. Q. Wang, H. P. Xie, H. M. Li, P. N. Chen, E. Bladt, R. Quintero-Bermudez, T. K. Sham, S. Bals, J. Hofkens, D. Sinton, G. Chen and E. H. Sargent, Nat. Chem., 2018, 10, 974–980.
26. T. Lu and Q. X. Chen, Theor. Chem. Acc., 2020, 139, 1–12.
27. X. Li, G. L. Chai, X. Xu, J. J. Liu, Z. Zhong, A. H. Cao, Z. J. Tao, W. F. You and L. T. Kang, Carbon, 2020, 167, 658–667.
28. H. C. Zhang, P. X. Cui, D. H. Xie, Y. J. Wang, P. Wang and G. P. Sheng, Adv. Sci., 2023, 10, 2205681.
29. S. Wjihi, F. Aouaini, A. Erto, M. Balsamo and A. Ben Lamine, Chem. Eng. J., 2021, 406, 126669.
30. S. L. Hu and W. X. Li, Science, 2021, 374, 1360–1365.
31. W. Ju, A. Bagger, G. P. Hao, A. S. Varela, I. Sinev, V. Bon, B. Roldan Cuenya, S. Kaskel, J. Rossmeisl and P. Strasser, Nat. Commun., 2017, 8, 1–9.
32. S. Matsumoto, K. K. Chattopadhyay, M. Inagaki and T. Ohsaka, Electrochim. Acta, 2017, 245, 123–132.
33. Q. Guo, Y. Xie, X. Wang, S. Lv, T. Hou and X. Liu, Chem. Phys. Lett., 2003, 380, 84–87.
34. Y. Y. Jiang, Y. Z. Lu, X. Y. Lv, D. X. Han, Q. X. Zhang, L. Niu and W. Chen, ACS Catal., 2013, 3, 1263–1271.
35. M. Zhu, J. Chen, R. Guo, J. Xu, X. Fang and Y.-F. Han, Appl. Catal., B, 2019, 251, 112–118.
36. C. He, Y. Zhang, Y. F. Zhang, L. Zhao, L. P. Yuan, J. N. Zhang, J. M. Ma and J. S. Hu, Angew. Chem., Int. Ed., 2020, 59, 4914–4919.
37. A. Ejaz and S. Jeon, Int. J. Hydrogen Energy, 2018, 43, 5690–5702.
38. H. Jiang, L. Liu, K. Zhao, Z. Liu, X. S. Zhang and S. Z. Hu, Electrochim. Acta, 2020, 337, 135758.
39. P. Zhou, L. Jiang, F. Wang, K. J. Deng, K. L. Lv and Z. H. Zhang, Sci. Adv., 2017, 3, e1601945.
40. P. Lazar, R. Mach and M. Otyepka, J. Phys. Chem. C, 2019, 123, 10695–10702.
41. H. H. Chen, X. Guo, X. D. Kong, Y. L. Xing, Y. Liu, B. L. Yu, Q. X. Li, Z. G. Geng, R. Si and J. Zeng, Green Chem., 2020, 22, 7529–7536.
42. P. H. Matter, L. Zhang and U. S. Ozkan, J. Catal., 2006, 239, 83–96.
43. P. H. Matter and U. S. Ozkan, Catal. Lett., 2006, 109, 115–123.
44. H. Jiang, J. X. Gu, X. S. Zheng, M. Liu, X. Q. Qiu, L. B. Wang, W. Z. Li, Z. F. Chen, X. B. Ji and J. Li, Energy Environ. Sci., 2019, 12, 322–333.
45. X. C. Qiao, H. L. Peng, C. H. You, F. F. Liu, R. P. Zheng, D. W. Xu, X. H. Li and S. J. Liao, J. Power Sources, 2015, 288, 253–260.
46. H. Jiang, Y. Q. Wang, J. Y. Hao, Y. S. Liu, W. Z. Li and J. Li, Carbon, 2017, 122, 64–73.
47. G. Rossi, F. d'Acapito, L. Amidani, F. Boscherini and M. Pedio, Phys. Chem. Chem. Phys., 2016, 18, 23686–23694.
48. A. Kumar, V. Q. Bui, J. Lee, L. L. Wang, A. R. Jadhav, X. H. Liu, X. D. Shao, Y. Liu, J. M. Yu, Y. Hwang, H. T. D. Bui, S. Ajmal, M. G. Kim, S. G. Kim, G. S. Park, Y. Kawazoe and H. Lee, Nat. Commun., 2021, 12, 6766.
49. H. B. Yang, S. F. Hung, S. Liu, K. D. Yuan, S. Miao, L. P. Zhang, X. Huang, H. Y. Wang, W. Z. Cai, R. Chen, J. J. Gao, X. F. Yang, W. Chen, Y. Q. Huang, H. M. Chen, C. M. Li, T. Zhang and B. Liu, Nat. Energy, 2018, 3, 140–147.
50. B. X. Zhang, J. L. Zhang, J. B. Shi, D. X. Tan, L. F. Liu, F. Y. Zhang, C. Lu, Z. Z. Su, X. N. Tan, X. Y. Cheng, B. X. Han, L. R. Zheng and J. Zhang, Nat. Commun., 2019, 10, 2980.
51. J. Q. Guan, Z. Y. Duan, F. X. Zhang, S. D. Kelly, R. Si, M. Dupuis, Q. G. Huang, J. Q. Chen, C. H. Tang and C. Li, Nat. Catal., 2018, 1, 870–877.
52. J. Q. Feng, H. S. Gao, L. R. Zheng, Z. P. Chen, S. J. Zeng, C. Y. Jiang, H. F. Dong, L. C. Liu, S. J. Zhang and X. P. Zhang, Nat. Commun., 2020, 11, 4341.
53. W. Zhang, C. Huang, Q. Xiao, L. Yu, L. Shuai, P. An, J. Zhang, M. Qiu, Z. Ren and Y. Yu, J. Am. Chem. Soc., 2020, 142, 11417–11427.
54. S. Gong, W. Wang, X. Xiao, J. Liu, C. Wu and X. Lv, Nano Energy, 2021, 84, 105904.

---

Footnote

† These authors contributed equally to this work.

---