Structural insight into [Fe–S₂–Mo] motif in electrochemical reduction of N₂ over Fe₁-supported molecular MoS₂

Abstract

The catalytic synthesis of NH₃ from the thermodynamically challenging N₂ reduction reaction under mild conditions is currently a significant problem for scientists. Accordingly, herein, we report the development of a nitrogenase-inspired inorganic-based chalcogenide system for the efficient electrochemical conversion of N₂ to NH₃, which is comprised of the basic structure of [Fe–S₂–Mo]. This material showed high activity of 8.7 mg_NH3 mg_Fe⁻¹ h⁻¹ (24 μg_NH3 cm⁻² h⁻¹) with an excellent faradaic efficiency of 27% for the conversion of N₂ to NH₃ in aqueous medium. It was demonstrated that the Fe₁ single atom on [Fe–S₂–Mo] under the optimal negative potential favors the reduction of N₂ to NH₃ over the competitive proton reduction to H₂. Operando X-ray absorption and simulations combined with theoretical DFT calculations provided the first and important insights on the particular electron-mediating and catalytic roles of the [Fe–S₂–Mo] motifs and Fe₁, respectively, on this two-dimensional (2D) molecular layer slab.

---

Introduction

NH₃ is a chemical that can be used as a fertilizer and carbon-free energy store. The industrial production of NH₃ from N₂ and hydrogen (H₂) via the Haber Bosch (HB) process is well developed, but simultaneously, it is very energy demanding and environmentally unfriendly.^1,2 The HB process is normally conducted at high pressure and high temperature (400–500 °C and 100–200 bar, respectively), which accounts for 1–2% of the global annual energy output.^3,4 This is due to the difficulty in this reaction route to dissociate the strong N≡N triple bond of N₂ for the production of ammonia.⁵ In addition, H₂ as a reactant for the HB process is predominately derived from fossil fuel, which is responsible for about 1% of the global greenhouse gas emission.² Thus, several new attempts have been developed to replace the HB process using renewable energies. For example, decentralized pilot plants have been built to convert solar, wind, and tidal power to H₂via renewable electricity for the synthesis of NH₃ (eHB) (see Fig. S1†).^6,7 Furthermore, as a potential new process, it is even more attractive to produce NH₃ directly from the electrochemical reaction of N₂ and H₂O under ambient conditions. However, this still has to be developed using more effective catalysts.⁸

For low-temperature N₂ catalytic fixation to NH₃, the associative mechanism most likely occurs through enzymatic, photo- or electro-chemical means.^9–11 For these processes, N₂ fixation through enzyme nitrogenase is the most efficient route to produce NH₃, which has also been adopted in nature. Thus, substantial efforts have been devoted to understanding and mimicking how the nitrogenase enzyme accomplishes the reduction of N₂ at ambient temperature and pressure.^12,13 Many homogeneous catalysts act as well-defined molecular systems to provide important mechanistic insights.^14–16 On the other hand, inorganic-based nitrogenase mimics can potentially accomplish N₂ fixation and convert it into NH₃ under ambient conditions with light or electricity input.^10,11,17,18 For example, heterogeneous catalysts in the form of transition metal chalcogenides, including Mo- and Fe-containing sulphide clusters, have been reported to catalyze the reduction of N₂ to NH₃.^17,18 However, these structures are not well-defined and cannot provide as much mechanistic guidance as that of the homogeneous catalysts.

In addition, a number of these solid electrocatalysts suffer from slow kinetics due to the low N₂ reduction. Also, H₂ from competitive proton (water) reduction occurs over the same active sites.⁸ It has been reported that proton reduction is thermodynamically more favorable than N₂ reduction under negative potentials (see Fig. S2†).¹⁹ The adsorption and reduction of H⁺ to surface H* are potential dependent and can be rate-limiting on a specific catalyst. If this competitive route is suppressed, then it may dramatically enhance the faradaic efficiency (η_FE) for N₂ reduction.^20,21 Therefore, the design of inherently active and selective electrocatalysts with a suitable surface for N₂ reduction relies on controlling the applied potential to attenuate or totally inhibit the H₂ evolution process, while enhancing the activation of N₂. The activity of transition metals for the synthesis of NH₃ has been rationalized in terms of the N₂ binding energy by Norskov and co-workers.²² Their results showed that transition metals with half-electron filled 3d orbitals, such as Ru, Os and Fe, have a relative lower adsorption energy (−55–10 kJ mol⁻¹ N₂) for N₂, which results in higher turnover frequencies for the synthesis of NH₃. As both a non-noble metal and the active ingredient of the nitrogenase enzyme,^23,24 Fe is a potential candidate for the electrochemical synthesis of NH₃.

Herein, we developed a structurally well-defined single-atom catalyst consisting of isolated Fe₁ anchored on exfoliated molecular-layered MoS₂ for the efficient N₂ reduction reaction (NRR) to NH₃ of 8.7 mg_NH3 mg_Fe⁻¹ h⁻¹ in water under an applied potential, which could also offer a high η_FE of 27% over H₂ evolution from water electrolysis. It is interesting to find that this single-atom Fe₁ catalyst possesses similar [Fe–S₂–Mo] motifs to the core-structure of the FeMo sulfur (S) clusters in the nitrogenase enzyme. This makes the single-atom Fe₁ the catalytic redox active centers, which combined with the electronic-mediating [Fe–S₂–Mo] units, boost the electrochemical reduction of N₂ in water. The electrochemical reduction of N₂ over the Fe₁ single-atom catalyst was investigated via operando synchrotron-radiation X-ray absorption fine structure (opXAFS), X-ray absorption near edge structure (XANES) spectroscopy and density functional theory (DFT) calculations. The mechanistic pathways and structure–activity relationships were deduced over this inorganic nitrogenase mimic [Fe–S₂–Mo], providing guiding principles for the NRR.

Results and discussion

Structure of Fe₁ single-atom on single-layer MoS₂

The MoS₂ matrix was firstly treated with n-butyllithium solution in hexane²⁵ to exfoliate bulk MoS₂ to form 2D mono-layered MoS₂. The X-ray diffraction (XRD, Fig. S3†) pattern and atomic force microscopy (AFM, Fig. S4†) image show that around 60% of the exfoliated MoS₂ is single molecular layers. Subsequently, single-atom Fe₁ was introduced on the three-sublayer S–Mo–S in trigonal prismatic 2-H structure mono-layered MoS₂via the hydrothermal method. No peaks corresponding to Fe-based aggregated species were detected in the XRD patterns and TEM images, demonstrating the high dispersion of the Fe₁ atoms.

The existence of dispersed Fe atoms on the basal plane of MoS₂ was clearly verified by high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), as shown in Fig. 1. Single-layer MoS₂ nanosheets with a 2H trigonal prismatic symmetry pattern can be clearly seen. For single-atom catalysis, the specific chemical environment of the atom is critical since its coordinated feature can significantly affect its catalytic behavior and performance. For most of the reported supported single-atom (active site) materials, although the atoms could be directly visualized using the recently developed HADDF-STEM technique, their atomic positions with respect to the support structures were not clear and well-defined; hence, obscuring the derivation of the important structure–activity relationships. In contrast, our single-atom Fe on single-molecular layered MoS₂ (sMoS₂) exhibited clear bonding environments. As shown in Fig. 1, isolated Fe₁ atoms are located at two types of preferred positions on the basal plane of sMoS₂, namely the Mo atop site and substituted S atom site. They were revealed by the brighter spots than the surrounding Mo or S₂ sites in the 2-H arrangement, typically as presented in the blue and pink squares, respectively. Further evidence was obtained from the DFT simulations (Fig. S5†), enlarged HAADF-STEM image, corresponding model and intensity profile analysis, as shown in Fig. 1b, c and f, respectively showing that the Fe₁ atom sits on the triangle S sites, which is directly on the top position of Mo as the atop site. Similarly, Fig. 1d, e and g show that the Fe atom is located on the S basal site of 2H-sMoS₂, where the intensity profile suggests that S is substituted by the Fe atom. It should be noted that most of the Fe₁ single atoms were found on the Mo atop sites, and occasionally on the S substitution sites.

Fig. 1 Morphology and structural characterization of Fe-sMoS₂. (a) HAADF-STEM image of Fe-sMoS₂ sheet, scale bar is 1 nm. Chemical environments of Fe₁ can be seen in the two enlarged square boxes, where Scan 1 (blue line) shows the Fe₁ atom on the Mo atop site and Scan 2 (pink line) shows the Fe₁ atom substituted on the S site. The red arrows indicate individual Fe₁ atoms on the Mo atop site. (b) HAADF-STEM scan, (c) corresponding DFT optimized model and (f) ADF intensity profile analysis of the Fe₁ atom on the Mo atop site. (d) HAADF-STEM scan, (e) corresponding DFT optimized model and (g) ADF intensity profile analysis of the Fe₁ atom as the substituted S site.

To obtain bonding information on the anchored Fe₁ atom, experimental XAFS (Fig. 2a) spectra were collected together with DFT simulations. Fig. 2a shows the Fourier transform spectrum of the Fe K-edge XAFS oscillations of the as-reduced Fe-sMoS₂ in comparison with the standard Fe foil. The absence of Fe–Fe interaction in the FT-XAFS spectra indicates the single-atom configuration of Fe₁. The peak at approximately 1.7 Å is mainly attributed to the Fe–S bonds at the Mo atop site. The simulation of the structure with the corresponding bonding distance is shown in the inset of Fig. 2a. Wavelet transformed analysis of XAFS (WT-XAFS, Fig. 2c) based on Morlet wavelets was conducted to differentiate the closely-related spatial interactions²⁶ of the Fe₁ atoms with their proximal atoms. As displayed in Fig. 2c, the Fe–Fe bonds in the Fe foil show an energy maximum in the range of 7–11 Å⁻¹, while that for Fe-sMoS₂ is in the range of 4–9 Å⁻¹. This again supports the fact that the Fe species are individually dispersed as single atoms, as shown by the HADDF-STEM image (Fig. 1), mainly at the Mo atop sites.

Fig. 2 Bonding environment of Fe₁ atom in Fe-sMoS₂. (a) Fourier transform Fe K-edge XAFS spectra of Fe-sMoS₂ with reference to Fe foil. Inset shows the DFT model for Fe₁ atom at Mo atop site with the peak matching to that expected from the corresponding Fe–S distance and absence of Fe–Fe in both models. (b) Fe K-edge XANES spectra of atomically dispersed Fe-sMoS₂. Fe foil, hydrated FeCl₂, and FeCl₃ were used as references. (c) Wavelet transformation for the k²-weighted Fe K-edge XAFS signals of Fe-sMoS₂ and Fe foil based on Morlet wavelets with optimum resolutions at the first and higher coordination shells. The intensity reflects the content of scattering signals. Intensity decreases in order of red, yellow, green, and blue.

The bonding environment of the Fe₁ atom at the Mo atop site was simulated by DFT, and the result is shown in Fig. S5.† Interestingly, the structure in the model of Fe₁ at the Mo atop site has almost the same inorganic motif of [Fe–S₂–Mo] with that of the core structure of FeMoco, the primary cofactor of nitrogenase, giving equivalent bond lengths and geometries of the coordinated Fe₁ shells (see Table S1†).^24,27 In addition to the similar bonding environment of Fe, according to the XANES analysis, the absorption edge is clearly located between Fe^II and Fe^III, indicating that the oxidation state of the Fe species in Fe-sMoS₂ is also close to that of the working state of FeMoco,²⁸ as shown in Fig. 2b. Of particular interest is the characteristic peak below the absorption edge of Fe-sMoS₂. It is well known that this pre-edge feature is due to the 1s → 3d orbital forbidden transition, which would be excluded by dipole selection rules for a symmetry site.²⁹ The observed pre-edge peak matches with the characterized isolated Fe₁ on s-MoS₂.

Electrochemical N₂ reduction

The material was then tested for electrochemical N₂ conversion to NH₃ in water under ambient conditions. The catalyst was deposited on carbon paper as a cathode under a flowing stream of N₂ feed gas. It has been noted in the literature in this field that carefully designed blanks must be employed to confirm the nitrogen reduction reaction activity of any material. For example, it has been reported that contaminants such as NO_x may also participate in the synthesis of ammonia.^30,31 NaClO₂ is known to be one of the most efficient chemicals for NO_x oxidation due to its strong oxidation power.^32,33 Thus, to remove the interference of NO_x, two traps filled with 0.2 M NaClO₂ solution and 1 mM H₂SO₄ solution were used to purify the feed gas before it was flowed into the three-electrode single cell. The ability to remove NO_x is evidenced in Fig. S6.† In a previously established N₂ purification protocol, gas cleaning of the filters following by acid trapping were employed to remove NO_x.³⁴ The two methods were compared, and the results were within an acceptable deviation of 8.3%. Prior to each test, blank measurements in the absence of N₂ and catalyst were conducted.^35–38 The obtained reaction assay was measured by two independent methods, namely the indophenol blue method (Fig. S7†) and ammonia selective electrode.³⁹ The detail mechanism of the ammonia selective electrode is presented in the ESI.† The calibration curves are shown in Fig. S8 and S9,† respectively. The results from the two methods were in a good agreement. We firstly optimized the over-potential required for the maximum production of NH₃ over Fe-sMoS₂ in the range of −0.05 V to −1.00 V (versus the reversible H₂ electrode (RHE)). As shown in Fig. S10,† the current density increased as the applied potential increased, and was more stable at low potential. The highest rate of NH₃ production was 24.5 μg_NH3 cm⁻² h⁻¹ (8.7 mg_NH3 mg_Fe⁻¹ h⁻¹) at −0.10 V (versus RHE, see Fig. 3a) with a maximum η_FE of ca. 27.0% (Fig. 3b). To the best of our knowledge, this electrocatalytic performance is among the best reported results for the electrochemical synthesis of NH₃ using non-noble Fe-based systems in the literature (Table S2†). However, both the rate and η_FE decreased dramatically when the applied potential was beyond −0.20 V (Fig. 3b). The Tafel plot in Fig. S11† was used to determine the rate-determining step for the H₂ evolution reaction and oxygen reduction reaction. Mechanistically, three principal steps can participate in the conversion of 2H⁺ to H₂, namely the Volmer, Heyrovsky and Tafel steps. If the Volmer process is the rate-determining step (rds), a slope of ∼120 mV per decade will likely be obtained. In contrast, a rate-determining Heyrovsky or Tafel step gives characteristic slopes between 30–40 mV per decade.⁴⁰ Our Tafel slope for Fe-sMoS₂ in argon gas was measured to be ∼156 mV per decade at a low over-potential range (Fig. S11†), which suggests the rds is the Volmer step, involving the initial highly activated adsorption and reduction of H⁺ to produce active surface H* (H⁺ + e⁻ + * = *H). On the other hand, the corresponding Tafel slope for Fe-sMoS₂ for N₂ reduction to ammonia was measured to be ∼121 mV per decade, which is substantially lower that of the proton reduction. Therefore, to reduce the extent of H₂ production on this surface during N₂ reduction, it appears to be necessary to apply an optimal potential.

Fig. 3 N₂ reduction in aqueous solution and relationship between activity and structure. (a) N₂ reduction activity to NH₃ over Fe-sMoS₂ under applied potentials in the range of −0.05 V to −1.00 V. Activity was evaluated at least 3 times under the same conditions to generate the measurement errors for the ammonia production rate (R_NH3) and faradaic efficiency (η_FE). (b) Linear sweep voltammetry from 0.10 V to −0.50 V versus RHE over Fe-sMoS₂ under Ar and N₂. N₂ reduction efficiency for NH₃ production at different applied potentials was extrapolated from the linear sweep voltammetry curves. The NH₃ yield is expressed as μg cm⁻² h⁻¹ (μg_NH3 per centimeter square of electrode per hour). (c) Calculated amount of electrons consumed for the nitrogen reduction reaction (NRR) and hydrogen evolution reaction (HER) at −0.1 V in 1 h over [M–S₂–Mo] (M represents metal as shown in x-axis) dwelling in single-layered MoS₂ assuming no heat was generated from the current. (d) Electron paramagnetic resonance spectra of over different thickness MoS₂ samples and transition metals.

It is well known that exposed lattice vacancies can act as active sites for the activation of H₂ and N₂.^41–43 Consequently, different concentrations of S vacancies in different-layered MoS₂ samples using electron paramagnetic resonance were obtained (EPR, Fig. 3d). As can be seen in Fig. 3d, the peak intensity due to S vacancies (unpaired electrons at g = 2.00 detected infer the formation of S vacancies) increased with S decrease in the thickness of the MoS₂ layers (S vacancies were created during exfoliation²⁵), which correlates well with their electrochemical performances at a potential of −0.1 V (versus RHE, Fig. 3d and S12†). The activity for both N₂ reduction and H₂ evolution appeared to be greatly promoted when single-layered MoS₂ was used. Notably, the activity for NH₃ yield apparently increased with a reduction in the thickness of the MoS₂ slab. However, the overall η_FE for N₂ reduction to NH₃ by the single-layered MoS₂ was significantly lower than that of few-layered MoS₂ and bulk MoS₂. This implies that S vacancies promote a greater degree of H₂ evolution than N₂ reduction due to the more favorable thermodynamics in the former case. The addition of a transition metal causes an obvious decrease in the EPR signal, presumably because the transition metal dopant can occupy the S vacancies of 2H-MoS₂, as shown by the HAADF-STEM analysis (Fig. 1c). Fe-sMoS₂ exerts strong magnetic perturbation due to the presence of paramagnetic Fe, which accounts for the perturbed zig-zag oscillation of the background ESR signal. Fig. S13† shows a comparison of the activities and η_FE for N₂ reduction to NH₃ over different metal-doped sMoS₂ such as Au and Ru with the previously reported values.^44,45 The presence of trace Li⁺ during the preparation of the molecular layer of MoS₂ may facilitate the activity and η_FE since Li⁺ has been reported to play a vital role in the NRR.⁴⁶ However, the result from Fig. S13† indicates that the metal doping affects much more than the residual Li⁺. Polarization due to protruded transition metal atoms on the thin MoS₂ surface suggested by L. Zhang and co-workers may play a role in their activity.⁴⁷ However, we believe that the intrinsic atomic arrangements of Fe-sMoS₂, which has the core structure of nitrogenase, can give the best activity and η_FE. In fact, among the Haber–Bosch catalysts and biological enzymes, Fe is well-known to bind N and H competitively to give ammonia compared to other metals. This is further supported by the high electron consumption for the nitrogen reduction reaction over Fe-sMoS₂, as shown in Fig. 3c.

The electrochemical performance for N₂ fixation to NH₃ on Fe-sMoS₂ was also studied by linear sweep voltammetry (LSV). As shown in Fig. 3b, a clearly higher cathodic current density can be observed in the sweeping potential range of −50 mV to −300 mV versus RHE when the electrolyte was purged with N₂ instead of Ar. In addition, the η_FE for N₂ reduction by dividing the current density in Ar is very close to that in the synthesis of NH₃ (maximum of 27%), verifying that N₂ is activated and converted to NH₃ by the Fe-sMoS₂ catalyst. Isotopic labeling using 98% ¹⁵N-enriched N₂ gas was carried out to prove the derivation of NH₃. Controlled experiments in the absence of ¹⁵N₂, catalyst, and applied potential were firstly conducted, and no clear ammonia signal was observed in the proton NMR spectra, as shown in Fig. S14.† In contrast, a doublet in the region near 7.0 ppm was found for the test over Fe-sMoS₂ at −0.10 V with a flow of ¹⁵N₂. The quantitative results (Fig. S15†) indicated that the product rate is around 22 μg cm⁻² h⁻¹, which is consistent with the result using ¹⁴N₂. These results show that both the catalyst and the applied potential are necessary for nitrogen fixation. Thus, based on the result from LSV, nitrogen fixation occurs at a potential in the range of 0 to −0.5 V. Subsequently, liquid chromatography-mass spectrometry (LC-MS) analysis was conducted, which identified two major species containing indophenol derivatives from natural ¹⁴N and enriched ¹⁵N (see Fig. S16†).^17,48 The fragments containing ¹⁵N have a much higher area ratio at 199/198 m/z (mass/charge ratio) compared to that of the control fragments containing ¹⁴N. The isotopically labeled ¹⁵N₂ authenticated that the NH₃ synthesized originated from N₂ reduction. These results gave sufficient proof that N₂ can be fixed to NH₃ over Fe-sMoS₂. We conducted a 10 h chronoamperometry test, which demonstrated that the activity and η_FE slightly changed, as shown in Fig. S17.†

Molecular activation and reduction of N₂

Operando Fe K-edge opXAFS and opXANES are sensitive techniques to monitor the chemical environment of Fe atoms, which were performed in this study at the B18 Beamline, Diamond Light Source, UK to study the structural dynamics involving the single Fe atoms upon the competitive adsorption and activation of N₂ with a proton from water over Fe-sMoS₂. Fig. 4a and c show the Fourier transform (FT) opEXAFS spectra and the corresponding Fe K-edge opXANES spectra under different experimental conditions. Particularly, the peak relative to the Fe–S bond attributed to the Mo atop site (see Fig. 3a) under open-circuit voltage is compared in N₂, Ar/H₂O, and N₂/H₂O, and at −0.1 V (versus RHE) in N₂/H₂O. It is clear that the FT intensity near the Fe–S bonds at the Mo atop site clearly changes in this region to the different treatments. After switching to highly acidic electrolyte solution purged with Ar, the intensity of the Fe–S peak attributed to the Fe atop site exhibited the lowest value, which indicates the lowest coordination number for this Fe species. Based on our DFT simulation, the Fe–N bonding interactions of these molecule-absorbed Fe-sMoS₂ species are at around 1.8 Å (Fig. S18†), which is comparable in distance with the Fe–S interactions. Thus, it was anticipated that competitive replacement of this absorbed N₂ species from Fe by H⁺ would cause a reduction in the intensity of this peak. Interestingly, upon switching the gas stream back to N₂, we noted that the peak intensity increased to a higher value, resuming the higher contribution from the Fe–N scattering. These results clearly show that N₂ and H⁺ can be competitively activated by the Fe₁ atoms on MoS₂, as reflected by the atop Fe–S peak modulations. During the typical conditions for electrochemical N₂ reduction to NH₃ at the previously optimized −0.1 V versus RHE, the intensity of opXAFS was significantly enhanced, which implies that the dynamic N₂ adsorption to NH₃ was greatly strengthened, corresponding to the same phenomenon observed in Fig. 3a. This also indicates that the electron from the applied potential is biased at the Fe₁ atom and could be used for competitive N₂ activation to NH₃. In addition, the Fourier transform spectra showed that there was a slight but significant shift in the Fe–S interaction of the atop site during N₂ activation, as shown in Fig. 4a. The main peak of Fe–S shifted to a shorter length (∼Δ 0.04 Å) upon switching to Ar flushing and returned to the original position when the N₂ flow was resumed. Clearly, the electron back-donation of Fe orbitals from the adsorbed N₂ can attenuate its bonding with the S ligands, accounting for the longer Fe–S interaction. Apparently, applying a negative over-potential for N₂ over H⁺ in the dynamic synthesis of ammonia places the peak position between these two values.

Fig. 4 Operando X-ray absorption spectroscopy and N₂ activation process. (a) Fourier transform magnitudes of the experimental Fe K-edge srXAFS spectra of Fe-sMoS₂ under open-circuit voltage bias in N₂, Ar/H₂O, and N₂/H₂O, and at −0.1 V (versus RHE) in N₂/H₂O. H₂O represents electrolyte solution containing 0.1 M hydrogen chloride. (b) Structural evolution of the active site in electrochemical NH₃ synthesis with N₂ absorption and applied potential. (1) Before N₂ adsorption, the Fe–S bond is 2.14 Å in length. (2) After the adsorption of N₂ on the Fe₁ atom, the Fe–S bond is extended to 2.20 Å in length. (3) Low unoccupied molecular orbital of Fe₁ Mo atop site. Green net represents positively charged orbital and orange net represents negatively charged orbital. After applying a potential, the electron will transfer to the Fe₁ atom. Blue, yellow, brown, green, and red balls are Mo, S, Fe, N, and H atoms, respectively. (c) Normalized operando Fe K-edge XANES spectra for Fe-sMoS₂ under open-circuit voltage bias in Ar/H₂O, and N₂/H₂O, and at −0.1 V (versus RHE) in N₂/H₂O. (d) Simulated Fe K-edge XANES spectra for Fe-sMoS₂, N₂-adsorbed Fe-sMoS₂, and N₂-adsorbed Fe-sMoS₂ with electron-rich Fe. Blue, yellow, brown, cyan, and red balls are N, S, Fe, and Mo atoms, respectively.

The processes for N₂ activation were then investigated by DFT calculations (Fig. S19†). Fig. 4b(1) shows that the DFT-optimised Fe–S bond of the initial Fe₁ atom at the Mo atop site is 2.14 Å. After the absorption of N₂, the bond is extended by absorbed N₂ to 2.20 Å (Fig. 4c(2)). The increment in the bond length (∼0.06 Å) is close to the observed value (∼0.04 Å), as measured by opXAFS. The electron ground state of Fe-sMoS₂ was simulated in the form of the highest occupied molecular orbital (HOMO). As shown in Fig. 4c(3), Fe₁ is relatively positively charged under N₂, which allows the external electrons to occupy it under HOMO excitation. The experimental result from opXAFS also confirmed that the external electrons from the applied potential will be accommodated at the Fe₁ atom, as above. Therefore, the electron can then be used for the activation and reduction of N₂ to NH₃ on [Fe–S₂–Mo].

Fig. 4c shows the corresponding normalized operando Fe K-edge opXANES spectra in Ar. The shoulder signal of the main absorption edge at ∼7118 eV is due to the 1s → 4p transition. The introduction of N₂ caused a shift to the right, showing an electron withdrawing effect from the Fe species to N₂, indicating the electronic effects on the Fe₁ atom via N₂ adsorption. The shift to a higher oxidative state is due to the effective delocalization of the unpaired electron in the 3d orbitals of Fe and the spontaneous charge transfer from Fe to both the N₂ 2p orbital and proton 1s orbital. During electrochemical N₂ reduction, the Fe K-edge of Fe-sMoS₂ shifted back to a lower shift value, indicating the recovery of the electronic state of the orbitals of Fe₁ due to the injection of external electrons. We further monitored this process using XANES simulations for Fe-sMoS₂ under different conditions (Fig. 4d). As shown, the N₂ adsorption on the Fe₁ atom significantly shifted the edge of 1s → 4p transition, which returned to a lower energy value after applying one electron to the Fe atom. The simulations confirmed the trend of the effect of N₂ activation and potential applied. Similar phenomena of opXAFS and opXANES were observed for the molecular activation and reduction of CO₂.⁴⁹

Thus, based on these operando studies, the Fe₁ single atom on Fe-sMoS₂ serves as the active site for the electrochemical fixation of N₂ to NH₃. During the adsorption and electrochemical reduction of N₂, [Fe–S₂–Mo] responds to the tension and contraction of the Fe–S bond by changing the bond length and Fe electronic state.

To demonstrate the electron-mediating and catalytic roles of the [Fe–S₂–Mo] unit in promoting the synthesis of ammonia, we compared the electronic structure of N₂–Fe-sMoS₂ and N₂–Fe₃S₄. As shown in Fig. 5, the Bader charge of adsorbed N₂ is −0.29 electrons (−0.01–0.28) over Fe-sMoS₂. This value is much lower than that (0.23–0.43 = −0.20 electrons) over Fe₃S₄, indicating that more electrons are donated from the Fe₁ site to the antibonding orbital of the adsorbed N₂ on Fe-sMoS₂ with [Fe–S₂–Mo] units than Fe₃S₄ without Mo. Consequently, the activation of N₂ is promoted with a longer N–N bond length from 1.10 Å to 1.15 Å (see Table S3†). Meanwhile, the bond length of Fe–N is shortened within the unit of [Fe–S₂–Mo]. In addition, the average Bader charge of the S atom in the [Fe–S₂–Mo] unit is also more negative than that without the nitrogenase-mimic structure (−0.61 vs. −0.54 electrons), indicating that the removal of a proton from the competitive active site of the Fe atom is easier for a higher efficiency of nitrogen reduction over the nitrogenase mimic Fe-sMoS₂ under the same potential.

Fig. 5 Electronic structure of N₂ activation over nitrogenase-mimic Fe-sMoS₂ and Fe₃S₄. The presented data is the Bader charge of the corresponding atoms in units of electrons.

Conclusions

In summary, a new inorganic-based electrocatalyst with Fe₁ on a 2D single-layer MoS₂ slab was described. The structure contained dispersed Fe atoms on nitrogenase-like [Fe–S₂–Mo] motifs, which showed superior electrochemical activity and η_FE for electrochemical N₂ fixation to NH₃ over proton reduction in water under the application of the optimal potential at 0.1 V. Operando Fe K-edge srXAFS, XANES and DFT calculations indicated that N₂ can be adsorbed and reduced at the catalytic Fe₁ site on the essential electron-mediating [Fe–S₂–Mo] motifs. To activate the N₂ molecule, the strain of the Fe–S bonds and redox states of the Fe₁ atom will adapt to accelerate the absorption and reduction processes. This work not only demonstrated that single-atom heterogeneous catalysis accelerates the electrochemical reduction of N₂, but also offers unique insight into the synergistic active site with electronic and structural transitions during N₂ fixation over the nitrogenase mimic [Fe–S₂–Mo] structure.

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

The support of this project from the IUK-EPSRC of UK (DGE 102000) is gratefully acknowledged. The authors wish to thank Diamond Light Source (Diamond, UK) for accessing STEM and XAS facilities (B18; SP20856-1). LL also acknowledges the use of the Computing Facilities of Wuhan University of Science and Technology in the completion of the theoretical part of this work.

Notes and references

1. J. G. Chen, R. M. Crooks, L. C. Seefeldt, K. L. Bren, R. M. Bullock, M. Y. Darensbourg, P. L. Holland, B. Hoffman, M. J. Janik, A. K. Jones and M. G. Kanatzidis, Science, 2018, 360, eaar6611.
2. E. Worrell, L. Price, M. Neelis, C. Galitsky and Z. Nan, World best practice energy intensity values for selected industrial sectors, Tech. Rep., LBNL-62806 Ernest Orlando Lawrence Berkeley National Laboratory, 2008.
3. D. Tilman, K. G. Cassman, P. A. Matson, R. Naylor and S. Polasky, Nature, 2002, 418, 671–677.
4. H. Wang, L. Wang, Q. Wang, S. Ye, W. Sun, Y. Shao, Z. Jiang, Q. Qiao, Y. Zhu, P. Song and D. Li, Angew. Chem., Int. Ed., 2018, 57, 12360–12364.
5. H. Li, J. Shang, Z. Ai and L. Zhang, J. Am. Chem. Soc., 2015, 137, 6393–6399.
6. L. Ye, R. Nayak-Luke, R. Banares-Alcantara and E. Tsang, Chem, 2017, 3, 712–714.
7. J. Zheng, F. Liao, S. Wu, G. Jones, T. Y. Chen, J. Fellowes, T. Sudmeier, I. J. McPherson, I. Wilkinson and S. C. E. Tsang, Angew. Chem., Int. Ed., 2019, 58, 17335–17341.
8. S. Chen, S. Perathoner, C. Ampelli, C. Mebrahtu, D. Su and G. Centi, Angew. Chem., Int. Ed., 2017, 56, 2699–2703.
9. R. D. Milton, R. Cai, S. Sahin, S. Abdellaoui, B. Alkotaini, D. N. Leech and S. D. Minteer, J. Am. Chem. Soc., 2017, 139, 9044–9052.
10. H. Hirakawa, M. Hashimoto, Y. Shiraishi and T. Hirai, J. Am. Chem. Soc., 2017, 139, 10929–10936.
11. H. Tao, C. Choi, L. X. Ding, J. Zheng, Z. Han, M. Jia, Q. Fan, Y. Gao, H. Wang, A. W. Robertson, S. Hong, Y. Jung, S. Liu and Z. Sun, Chem, 2019, 5, 204–214.
12. E. J. Vicente and D. Dean, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 3009–3011.
13. B. M. Hoffman, D. Lukoyanov, Z.-Y. Yang, D. R. Dean and L. C. Seefeldt, Chem. Rev., 2014, 114, 4041–4062.
14. T. Kandemir, M. E. Schuster, A. Senyshyn, M. Behrens and R. Schlögl, Angew. Chem., Int. Ed., 2013, 52, 12723–12726.
15. M. J. Chalkley, T. J. Del Castillo, B. D. Matson, J. P. Roddy and J. C. Peters, ACS Cent. Sci., 2017, 3, 217–223.
16. J. S. Anderson, J. Rittle and J. C. Peters, Nature, 2013, 501, 84–87.
17. A. Banerjee, B. D. Yuhas, E. A. Margulies, Y. Zhang, Y. Shim, M. R. Wasielewski and M. G. Kanatzidis, J. Am. Chem. Soc., 2015, 137, 2030–2034.
18. J. Liu, M. S. Kelley, W. Wu, A. Banerjee, A. P. Douvalis, J. Wu, Y. Zhang, G. C. Schatz and M. G. Kanatzidis, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 5530–5535.
19. B. M. Lindley, A. M. Appel, K. Krogh-Jespersen, J. M. Mayer and A. J. Miller, ACS Energy Lett., 2016, 1, 698–704.
20. D. R. Cummins, U. Martinez, A. Sherehiy, R. Kappera, A. Martinez-Garcia, R. K. Schulze, J. Jasinski, J. Zhang, R. K. Gupta and J. Lou, Nat. Commun., 2016, 7, 11857.
21. J. H. Montoya, C. Tsai, A. Vojvodic and J. K. Nørskov, ChemSusChem, 2015, 8, 2180–2186.
22. C. J. Jacobsen, S. Dahl, B. S. Clausen, S. Bahn, A. Logadottir and J. K. Nørskov, J. Am. Chem. Soc., 2001, 123, 8404–8405.
23. K. M. Lancaster, M. Roemelt, P. Ettenhuber, Y. Hu, M. W. Ribbe, F. Neese, U. Bergmann and S. DeBeer, Science, 2011, 334, 974–977.
24. J. Chen, J. Christiansen, N. Campobasso, J. T. Bolin, R. C. Tittsworth, B. J. Hales, J. J. Rehr and S. P. Cramer, Angew. Chem., Int. Ed., 1993, 32, 1592–1594.
25. G. Liu, A. W. Robertson, M. M. Li, W. C. Kuo, M. T. Darby, M. H. Muhieddine, Y. C. Lin, K. Suenaga, M. Stamatakis, J. H. Warner and S. C. E. Tsang, Nat. Chem., 2017, 9, 810–816.
26. H. Fei, J. Dong, M. J. Arellano-Jimenez, G. Ye, N. D. Kim, E. L. Samuel, Z. Peng, Z. Zhu, F. Qin, J. Bao and M. J. Yacaman, Nat. Commun., 2015, 6, 8668.
27. R. Bjornsson, F. Neese, R. R. Schrock, O. Einsle and S. DeBeer, J. Biol. Inorg Chem., 2015, 20, 447–460.
28. R. Bjornsson, F. A. Lima, T. Spatzal, T. Weyhermüller, P. Glatzel, E. Bill, O. Einsle, F. Neese and S. DeBeer, Chem. Sci., 2014, 5, 3096–3103.
29. A. Kropf, B. Bunker, M. Eisner, S. Moss, L. Zecca, A. Stroppolo and P. Crippa, Biophys. J., 1998, 75, 3135–3142.
30. J. Jakob, J. K. Norskov and I. Chorkendorff, ACS Energy Lett., 2019, 4, 2986–2988.
31. J. Long, S. Chen, Y. Zhang, C. Guo, X. Fu, D. Deng and J. Xiao, Angew. Chem., Int. Ed., 2020, 59, 9711–9718.
32. A. Pourmohammadbagher, E. Jamshidi, H. Ale-Ebrahim and S. Dabir, Ind. Eng. Chem. Res., 2011, 50, 8278–8284.
33. C. Brogren, H. T. Karlsson and I. Bjerle, Chem. Eng. Technol., 1998, 21, 61–70.
34. B. R. Deshwal, D. S. Jin, S. H. Lee, S. H. Moon, J. H. Jung and H. K. Lee, J. Hazard. Mater., 2008, 150, 649–655.
35. J. Choi, H. L. Du, C. K. Nguyen, B. H. Suryanto, A. N. Simonov and D. R. MacFarlane, ACS Energy Lett., 2020, 5, 2095–2097.
36. L. F. Greenlee, J. N. Renner and S. L. Foster, ACS Catal., 2018, 8, 7820–7827.
37. S. Z. Andersen, V. Colic, S. Yang, J. A. Schwalbe, A. C. Nielander, J. M. McEnaney, K. Enemark-Rasmussen, J. G. Baker, A. R. Singh, B. A. Rohr, M. J. Statt, S. J. Blair, S. Mezzavilla, J. Kibsgaard, P. C. K. Vesborg, M. Cargnello, S. F. Bent, T. F. Jaramillo, I. E. L. Stephens, J. K. Norskov and I. Chorkendorff, Nature, 2019, 570, 504–508.
38. C. Tang and S. Z. Qiao, Chem. Soc. Rev., 2019, 48, 3166–3180.
39. Y. Ma, T. Yang, H. Zou, W. Zang, Z. Kou, L. Mao, Y. Feng, L. Shen, S. J. Pennycook, L. Duan, X. Li and J. Wang, Adv. Mater., 2020, 2002177.
40. T. Shinagawa, A. T. Garcia-Esparza and K. Takanabe, Sci. Rep., 2015, 5, 13801.
41. T. Wu, Z. Xing, S. Mou, C. Li, Y. Qiao, Q. Liu, X. Zhu, Y. Luo, X. Shi and Y. Zhang, Angew. Chem., Int. Ed., 2019, 58, 18449–18453.
42. C. Lv, Y. Qian, C. Yan, Y. Ding, Y. Liu, G. Chen and G. Yu, Angew. Chem., Int. Ed., 2018, 57, 10246–10250.
43. H. Li, C. Tsai, A. L. Koh, L. Cai, A. W. Contryman, A. H. Fragapane, J. Zhao, H. S. Han, H. C. Manoharan and F. Abild-Pedersen, Nat. Mater., 2016, 15, 48–53.
44. S. J. Li, D. Bao, M. M. Shi, B. R. Wulan, J. M. Yan and Q. Jiang, Adv. Mater., 2017, 29, 1606550.
45. B. H. Suryanto, D. Wang, L. M. Azofra, M. Harb, L. Cavallo, R. Jalili, D. R. G. Mitchell, M. Chatti and D. R. MacFarlane, ACS Energy Lett., 2018, 4, 430–435.
46. G.-F. Chen, X. Cao, S. Wu, X. Zeng, L.-X. Ding, M. Zhu and H. Wang, J. Am. Chem. Soc., 2017, 139, 9771–9774.
47. J. Li, S. Chen, F. Quan, G. Zhan, F. Jia, Z. Ai and L. Zhang, Chem, 2020, 6, 808–810.
48. Y. Zhao, Y. Zhao, R. Shi, B. Wang, G. I. Waterhouse, L. Z. Wu, C. H. Tung and T. Zhang, Adv. Mater., 2019, 31, 1806482.
49. H. B. Yang, S.-F. Hung, S. Liu, K. Yuan, S. Miao, L. Zhang, X. Huang, H.-Y. Wang, W. Cai and R. Chen, Nat. Energy, 2018, 3, 140–147.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0sc04575f

---