Conductive metal-covalent organic frameworks as novel catalytic platforms for reduction of nitrate to ammonia†

Hao Huang and Kaiying Wang *
Department of Microsystems, University of South-Eastern Norway, Borre 3184, Norway. E-mail: Kaiying.Wang@usn.no

First published on 16th October 2023

---

---

Introduction

Ammonia (NH₃), as a vital constituent of all amino acids in living organisms, is the cornerstone in agriculture and aquaculture.¹ Its nitrogen-rich characteristics also make it important in the plastics, textile, and pharmaceutical industries.^2,3 Recently, there has been growing interest in the application of NH₃ in the field of renewable energy, driven by the pursuit of sustainable energy sources. With its high hydrogen content (17.7 wt%) and carbon-free composition, NH₃ serves as a promising hydrogen carrier with a high energy density of 4.3 kW h kg⁻¹.^4–6 Unlike hydrogen, liquid NH₃ can be readily obtained and stored under mild conditions (p ≥ 10 bar or T ≤ −33 °C).⁷ NH₃ is now recognized as a next-generation sustainable energy carrier.

Currently, the worldwide NH₃ production industry relies on the Haber–Bosch method, which involves the reaction between hydrogen and nitrogen under harsh conditions (high pressure, ≥200 atm and high temperature, ≥400 °C).^8,9 However, this method results in extreme energy consumption (≈38 GJ t_NH3⁻¹) and carbon emission (≈2.9 t_CO2 per t_NH3).¹⁰ In response to the growing demand for NH₃, there is a need to expand the production scale and improve technology. As an attractive alternative, electrochemical nitrogen reduction offers a strategy that uses nitrogen and water as sources for NH₃ synthesis under mild conditions.^11,12 However, direct dinitrogen electroreduction faces several challenges: dinitrogen has an ultrahigh energy barrier to break the inert NN bond of 941 kJ mol⁻¹ and it exhibits extremely low solubility in water (17.1 mg kg⁻¹_water).^13–15 Additionally, as a nonpolar molecule, dinitrogen shows weak absorption ability on the surface of electrodes.¹⁶ These three obstacles severely limit the utilization of direct dinitrogen electroreduction. As an alternative, nitrate (NO₃⁻) presents unique advantages to boost the electrochemical reduction strategy for NH₃ production: (1) low bond energy (204 kJ mol⁻¹); (2) large water solubility (383 g kg⁻¹_water for potassium nitrate at 25 °C); and (3) high chemical polarity for good absorption at active sites.^16–19 Furthermore, due to its long-term accumulation as a result of agricultural and industrial activities, NO₃⁻ acts as a noxious pollutant that can cause various diseases in human tissue.^20,21 Therefore, the reduction of NO₃⁻ to NH₃ (NRA, NO₃⁻ + 9H⁺ + 8e⁻ → NH₃ + 3H₂O, −0.12 V vs. SHE) not only enhances the efficiency of NH₃ production, but also greatly alleviates pollution.¹¹

As an emerging technology, various electrocatalysts have been exploited for the NRA reaction, including noble metal nanoparticles, transition metal nanostructures and alloys, oxides, phosphides, etc.^5,22,23 These electrocatalysts rely on inorganic metal-based nanomaterials, where the active sites for NRA are typically limited to the exposed metal atoms at the surface and/or the edge of the nanostructures.⁵ The internal metal atoms in these inorganic catalysts are often inert and difficult to utilize for the reduction of NO₃⁻ to NH₃. This natural limitation results in very low metal-atom utilization and catalytic efficiency of current NRA catalysts. Fortunately, owing to their abundant exposed metal sites and ordered in-plane porous structure, metal–organic frameworks (MOFs) offer a promising approach for maximizing the utilization of metal atoms (in principle, up to 100%) and enhancing the efficiency of electrocatalytic reactions, similar to single-atom catalysts.²⁴ However, the structural stability of metal linkages in MOFs is often fragile, leading to framework dissociation under applied potentials.^25,26 To address this challenge, the incorporation of conjugated covalent linkages into c-MOFs to form conductive metal-covalent frameworks (c-MCOFs) could be instrumental.²⁵ Although only a few studies have reported on the construction of electrocatalysts using c-MCOFs, most of them are based on porphyrin subgroups with very low metal content (∼1 at%).^27,28 Hexaazatriphenylene (HATN, Fig. S1, ESI†) is considered as a preferred alternative to overcome this limitation because its three bidentate tetraamine moieties can coordinate with most metal atoms, forming M₃·HATN with an ultra-high metal content of ∼14 at%.²⁹ Therefore, the construction of c-MCOFs based on M₃·HATN would pave the way for designing novel electrocatalysts toward the NRA reaction. The exploitation of c-MCOF structures is still in its early stage, especially in terms of c-MCOFs with high metal content for electrochemical NRA catalysts.

Herein, we propose a novel strategy for constructing NRA electrocatalysts using the c-MCOF structure with an M₃·HATN subgroup. Inspired by the active enzyme nitrate reductase found in denitrifying bacteria, we utilize Mo atoms to establish HATN-based c-MCOFs, known as Mo-HATN-COFs. The Mo-HATN-COFs are achieved via a facile coordination–condensation reaction, resulting in structures with an ultra-high Mo content of 12.5 at%, an ordered in-plane porous structure with a size of 1.2 nm and excellent metallicity for good conductivity of 1.5 S cm⁻². To further enhance the catalytic activity for NRA, a salt-template approach is employed to synthesize Mo-HATN-COFs with ultra-thin nanosheets (∼1.4 nm). As a result of our experiment, the Mo-HATN-COF nanosheets exhibit a remarkable ammonia yield rate of 8.52 mg h⁻¹ cm⁻² (0.50 mmol h⁻¹ cm⁻²) at −0.5 V vs. RHE. The highest faradaic efficiency (FE) achieved reaches an impressive value of 91.3%. Most importantly, unlike traditional inorganic catalytic structures, c-MCOFs with an M₃·HATN subgroup can serve as preferred models for building novel platforms to investigate the mechanism of metal-atom electrocatalysis. Our findings represent a significant contribution towards revealing the inherent mechanisms of and designing next-generation electrocatalysts for the NRA reaction. By pioneering this research, we hope to pave the way for future advancements in this field.

Experimental

Reagents and chemicals

1,2,4,5-Tetraaminobenzene tetrahydrochloride (≥95.0%), hexaketocyclohexane octahydrate (≥97.0%), anhydrous 1-methyl-2-pyrrolidinone (NMP ≥ 99.5%), anhydrous molybdenum(V) chloride (≥99.9%), anhydrous nickel(II) chloride (≥99.9%) and sodium chloride (≥99.0%) were obtained from Sigma-Aldrich. All reagents and solvents were used as received from commercial sources without further purification.

Preparation of bulk Mo-HATN-COFs

80 mL of NMP with 0.5 mL of sulfuric acid was purged with nitrogen for 30 min to remove the dissolved oxygen. Under nitrogen conditions, 1 g of 1,2,4,5-tetraaminobenzene tetrahydrochloride (3.52 mmol), 0.73 g of hexaketocyclohexane octahydrate (2.35 mmol) and 2.11 g of molybdenum chloride (7.76 mmol) were slowly added into the deoxygenated NMP and stirred for 1 h to ensure sufficient dissolution. The mixed solution was then heated to 180 °C for 8 h. After cooling to room temperature, 150 mL of deionized water was introduced to quench the reaction. The black Mo-HATN-COF sample was purified by dialyzing with deionized water and the final product obtained after freeze drying.

Preparation of Mo-HATN-COFs

0.12 g of 1,2,4,5-tetraaminobenzene tetrahydrochloride (0.42 mmol), 0.09 g of hexaketocyclohexane octahydrate (0.29 mmol) and 0.25 g of molybdenum chloride (0.92 mmol) were added into 5 mL of NMP with 0.04 mL of sulfuric acid. After 15 min of ultrasonication, the mixed solution was added into 20 g of NaCl and then stirred for 30 min for adequate adhesion on the surfaces of NaCl crystals. The mixture was transferred into a three-necked flask and heated to 180 °C for 12 h under nitrogen conditions. When it cooled down to room temperature, 50 mL of deionized water was added into the flask. The black Mo-HATN-COF nanosheets were purified by dialyzing with deionized water and the final product obtained after freeze drying.

Preparation of Ni-HATN-COFs

The Ni-HATN-COF nanosheets were prepared by a similar method. In this process, 0.12 g of nickel chloride (0.92 mmol) was used instead of 0.25 g of molybdenum chloride to prepare Ni-COF nanosheets.

Preparation of COFs without metal sites (HATN-COFs)

HATN-COF nanosheets were also obtained by a similar method, with no introduction of metal chloride into the reaction system.

Preparation of the M-HATN-COFs/carbon paper electrode

4 mg of M-HATN-COF (including Mo-HATN-COFs, Ni-HATN-COFs, HATN-COFs, and bulk Mo-HATN-COFs) powder and 20 μL of Nafion solution (5 wt%) were dispersed in a mixture of 1 mL of ultra-pure water and 1 mL of anhydrous alcohol by ultrasonic treatment for 1 h to form a homogeneous black ink. Then, 500 μL of the dispersion was evenly loaded on carbon paper with an area of 1 × 1 cm² and dried under a baking lamp for 1 h at room temperature.

The details of physical characterization, electrochemical measurements, and density functional theory (DFT) calculations can be found in the ESI.†

Results and discussion

The preparation and characterization of M-HATN-COFs

In this work, the HATN-COFs were synthesized by a one-step coordination–condensation reaction. In our opinion, the amino group in 1,2,4,5-tetraaminobenzene tetrahydrochloride (Fig. S2a†) firstly coordinates with Mo ions in NMP solvent to form a Mo-tetraaminobenzene molecule. Subsequently, as the temperature rises, the Mo-HATN structure is gradually formed via Schiff-base condensation between the carbonyl group in hexaketocyclohexane (Fig. S2b†) and the amino group in Mo-tetraaminobenzene (R′′CO + R′NH₂ → R′′CNR′ + H₂O); the prepared CN moiety acts as as a conjugated linkage.³⁰ Finally, the Mo-HATN-COFs (Fig. 1) are achieved via the ordered growth of Mo-HATN. Beyond traditional COFs, the Mo-HATN-COFs have abundant metal sites (Mo: 12.5 at%) with high unsaturation (Mo–N₂), providing enhanced activity for electrocatalytic reactions, including the NRA reaction. Not only that, the CN linkage and the Mo–N₂ moiety also contribute excellent conductivity to the Mo-HATN-COFs as metallic materials, which is very conducive to electron transfer in the electrocatalytic process. As a result, these factors provide Mo-HATN-COFs with very strong ability for electrocatalysis.

To understand the frameworks, the structure of Mo-HATN-COFs was simulated by density functional theory (DFT) calculations. After geometry optimization, the Mo-HATN-COF is a two-dimensional (2D) covalent framework, constituted by a periodic hexagonal ring of six Mo₃·HATN subgroups connected through conjugated CN linkages. As with P6/mmm symmetry, the in-plane lattice length of HATN-COFs is about 16.54 Å (a = b; α: 90°, β: 90°, γ: 120°) with an ordered in-plane porous structure of ∼1.2 nm. Meanwhile, experimental data are also used to comprehend the frameworks (Fig. S3†). The X-ray powder diffractometer (XRD) pattern of Mo-HATN-COFs shows a prominent peak at 26.5°, which belongs to the (002) plane, well matched with the simulated XRD result and previous works.³¹ This result demonstrates that the Mo-HATN-COFs are successfully prepared through the coordination–condensation approach. In addition, the Ni-HATN-MCOF and HATN-COF slabs also show hexagonal chemical structures with similar lattice lengths of 16.48 Å and 16.68 Å, respectively (Table S1†). The introduction of metal atoms can reduce the lattice length from 16.68 Å to 16.48 Å, which indicates that the metal coordination reaction can make the HATN-COF structure more compact and may bring about a higher degree of conjugation for better conductivity. To confirm this, partial density of states (PDOS) calculations were carried out for investigating the electronic band structure of various M-HATN-MCOF slabs. As a result, the HATN-COF slab exhibits semiconductor properties with a band-gap of 1.73 eV (Fig. 1b). In contrast, the Mo-HATN-COFs and Ni-HATN slabs show metallicity without any band-gap (Fig. 1c and d). The PDOS result of the Mo-HATN-COF slab reveals that the electron distribution near the Femi level mainly originates from the electrons of the Mo atoms. Meanwhile, a similar result is observed with the Ni-HATN-MCOF slab (Fig. S4 and S5†). Comparison of the d orbitals shows that the d-center of Mo-HATN-COFs is located at −3.08 eV, which is more negative than that of Ni-HATN-COFs (−2.67 eV). This result demonstrates that the Mo sites are more inert for H⁺ absorption (HER process) than the Ni sites, which means that the Mo sites may be advantageous to the NRA reaction (Fig. S6†). Four-point probe technology reveals that the compressed pellet of Mo-HATN-COFs has great electrical conductivity of 1.5 S cm⁻², much larger than that of the HATN-COFs without Mo sites (0.06 S cm⁻²). To directly prove the enhanced conductivity of Mo-HATN-COFs, we employed EIS tests on Mo-HATN-COF and HATN-COF samples (Fig. S7†). The R_ct value of HATN-COFs is 521.9 Ω, which is much lower than that of non-conjugated COFs. After introducing Mo atoms into the HATN-COF structure, the R_ct value drops to 117.2 Ω, indicating that the incorporation of Mo atoms can enhance the conductivity of the HATN-COF structure.

For fully tapping into the potential for NRA electrocatalysis, an ultrathin structure is also incorporated into the M-HATN-COFs. A salt-template serves in a directed strategy for achieving ultrathin M-HATN-COFs nanosheets, providing an effective route toward the growth of various few-layered inorganic nanostructures.^32–34 As a result of its excellent stability, flat surface and low cost, a sodium chloride crystal can be an inert planar substrate to guide the growth of most 2D materials.^35–37 In this work, we first propose that a sodium chloride crystal (Fig. S8†) serves as an agent to orientate construction of ultrathin COF nanosheets (Fig. 2a). The morphology of various M-HATN-COFs is confirmed by scanning electron microscopy (SEM). Without the NaCl salt-template, Mo-HATN-COFs show severely reunited layered structures as bulk materials (Fig. 2b). In contrast, the SEM image of template-directed Mo-HATN-COFs shows that they have an obvious ultrathin layered morphology with generous wrinkles (Fig. 2c). The interaction between these nanosheets results in not only an extensive flower-like structure, but also a large-scale hierarchical porous structure for Mo-HATN-COFs, where the size distribution of pores ranges from 5 to 100 nm (Fig. S9†). As far as we know, the creation of this hierarchical structure is a very efficient strategy for rapid mass transfer, improving the interaction between the catalytic substrate and active centers. In this work, the hierarchical mesoporous structure can rapidly transfer nitrate from the electrolyte to the Mo sites and releases ammonia once the reduction is completed, which is beneficial for accelerating the process of adsorption and desorption in the electrochemical reduction reaction. To investigate the porous structure, N₂ adsorption/desorption testing was employed. The isotherm of Mo-HATN-COF nanosheets indicates a large surface area of 445.6 m² g⁻¹ and two types of size distribution: (1) a wide peak at about 18.7 nm, which can be attributed to the interaction with hierarchical pores; and (2) a sharp peak at 1.3 nm attributed to the ordered in-plane porous structure, the same as that for the DFT slab (Fig. S10†). In contrast, the Mo-HATN-COFs without a salt-template show a blocky-shaped nanostructure, which consists of agglomerated layered nanosheets. This comparison indicates the effective function of the salt-template-directed strategy for engineering the 2D structure of HATN-COFs. Similarly, the SEM images show that the Ni-HATN-MCOF and HATN-COF samples also have ultrathin-layered structures with hierarchical pores, indicating the universality of the salt-template-directed strategy for 2D COF materials (Fig. 2d and S11†).

In transmission electron microscopy (TEM) images, similar to the results of SEM, these M-HATN-COF nanosheets have a smooth silk-like layered structure (Fig. 3a–c) and the Mo-HATN-COF nanosheet thickness is about 1.4 nm (Fig. 3d), which is the same as those of the Ni-HATN-MCOF and HATN-COF nanosheets (Fig. S12 and S13†).³⁸ The high-resolution TEM images show that the M-HATN-COF nanosheets exhibit good crystallinity (Fig. 3e and f). Thus, the results of morphological characterization illustrate that the salt-template-directed strategy can serve as a highly feasible method to control the growth of 2D structures on COFs. To inspect the elemental composition, energy dispersive spectroscopy (EDS) mapping was carried out on the HATN-COF nanosheets. The mapping image reveals that the Mo and N elements are present and uniformly distributed in the Mo-HATN-COF nanosheets (Fig. 3g).

Inductively coupled plasma (ICP-OES) technology was employed to analyze the actual Mo content. The ICP-OES result shows that the Mo content in the Mo-HATN-COF is 7.9 at% (40.08 wt%), which is lower than its theoretical value (12.5 at%; 52.90 wt%). In our opinion, this discrepancy in Mo-HATN-COFs may be attributed to insufficient coordination between Mo ions and 1,2,4,5-tetraaminobenzene tetrahydrochloride in the synthesis process. To further confirm the structure of Mo-HATN-COFs, XPS analysis was carried out. The XPS analysis confirms the presence of Mo, N, and C elements in Mo-HATN-COFs. The Mo XPS spectra of Mo-HATN-COFs are consistent with previously reported results (Fig. S14a†), showing two peaks at 230.9 and 234.1 eV, which are ascribed to the Mo⁵⁺ 3d_5/2 and 3d_3/2, respectively.³⁹ The peak at 237.3 eV indicates the presence of an Mo⁶⁺ structure, possibly resulting from mild oxidation during the synthetic process. The XPS spectra of N 1s (Fig. S14b†) exhibit two characteristic peaks at 398.3 and 401.1 eV, attributed to NC and N–H, respectively, indicating the existence of CN linkages.⁴⁰ Moreover, the peak observed at 394.9 eV indicates the formation of an Mo–N moiety.⁴¹ For C 1s, the spectrum of the Mo-HATN-COF is deconvoluted into three carbon species of C–C (284.9 eV), CN (286.3 eV) and C–O (288.7 eV), confirming the existence of C-based frameworks (Fig. S14c†). Thus, all experimental data and analyses manifest the achievement of M-HATN-COF nanosheets in this work.

The electrocatalytic property for NRA

To study the NRA properties of various c-MCOF nanosheets, a typical three-electrode system is employed by using modified carbon paper as the working electrode (Fig. S15†). The SEM images of the modified carbon papers reveal the structural state of Mo-HATN-COF, Ni-HATN-MCOF and HATN-COF nanosheets on the surface of a carbon substrate. After freeze drying, most HATN-COFs nanosheets stand vertically on the substrate and intersect with each other to form the hierarchical mesoporous structure (Fig. S16–18†), which indicates that the various c-MCOFs maintain their best state for NRA electrocatalysis. In this work, various electrochemical technologies were employed to estimate the catalytic activity of these c-MCOFs in 0.1 M Na₂SO₄ (supporting electrolyte) with 0.1 M NaNO₃ (nitrate source). Firstly, the linear sweep voltammogram (LSV) curves show that the current densities of these c-MCOF samples are much larger than those without nitrate in the electrolyte, indicating that the NRA reaction has occurred on the surfaces of these modified substrates (Fig. 4a). The NRA behavior of Mo-HATN-COF nanosheets is initiated at −0.31 V vs. RHE and the maximum difference of catalytic current density (NRA catalytic current density) of 60.7 mA cm⁻² occurs at −0.5 V (vs. RHE), inferring that the Mo-HATN-COFs may reach their best activity at −0.5 V for NRA. In contrast, the bulk Mo-HATN-COFs show a lower NRA catalytic current of 23.2 mA cm⁻², indicating that the strategy of achieving ultrathin structures can improve the electrochemical activity of MCOF catalysts. The HATN-COF electrodes exhibit partial NRA catalytic current significantly lower than that of Mo-HTN-MCOFs, which is due to the lack of highly active metal sites for NRA. However, under higher applied potential, the hydrogen evolution reaction (HER) as a competitive reaction dominates the electrode reaction due to its lower electron-transfer number. On the electrode of Ni-HATN-MCOFs, this phenomenon is more obvious, although the catalytic current density is larger than that of HATN-COFs, which is mostly dedicated to the HER (Fig. S19†). This result implies that the NRA activity of Ni sites may be lower than that of Mo sites, because of the higher catalytic activity for the HER.

The NH₃ yields of various c-MCOF samples were analyzed using chromogenic methods on the electrolyte after electrocatalysis (Fig. S20 and 21†). On comparing with absorbance calibration curves, it was observed that the Mo-HATN-COF nanosheets exhibit an NH₃ yield of 8.52 mg h⁻¹ cm⁻² (0.50 mmol h⁻¹ cm⁻²), dramatically superior to that of the bulk Mo-HATN-MCOFs (2.37 mg h⁻¹ cm⁻²) at −0.5 V vs. RHE (Fig. 4b). At the applied potential of −0.4 V, the TOF (turnover frequency) value of the Mo-HATN-COF, calculated from the LSV curves, is 0.714 NH₃ per s per site, indicating that Mo-HATN-COFs exhibit the best intrinsic activity for ammonia production. This shows that engineering the 2D structure is an important factor for regulating the electrocatalytic behaviors of COFs. In this work, the ultrathin structure improves the NRA activity of HATN-COFs due to the increased exposure to metal sites. To investigate the source of electrocatalytic activity, the NRA behaviors of Ni-HATN-MCOF and HATN-COF nanosheets were also evaluated. The Ni-HATN-MCOF and HATN-COF samples demonstrate lower NH₃ yields of 1.92 and 0.49 mg h⁻¹ cm⁻², respectively, revealing that the metal atoms are the highly active sites for NRA electrocatalysis and the activity order is Mo sites > Ni sites > N sites. The NH₃ yield on Mo-HATN-COFs firstly increases from −0.2 to −0.5 V (vs. RHE); subsequently, the yield decreases with the continued increase of applied potentials (Fig. 4c). The FE is enhanced with an increase in applied potential (from −0.2 to −0.4 V vs. RHE) and reaches the maximum value of 91.3%; beyond −0.40 V (vs. RHE), the FE obviously decreases because the competitive reaction (HER) dominates at a large applied potential and the Mo sites begin to adsorb protons instead of nitrates. The activity of Mo-HATN-COFs is compared with other MOF-, COF- and Ni-based catalysts (Table S2†). It can be concluded that the Mo-HATN-COFs in this work show better electrocatalytic properties than the other MOF-, COF- and Ni-based catalysts. The bulk Mo-HATN-COF sample exhibits the best FE of 77.8% at −0.4 V (vs. RHE) (Fig. S22†). The Ni-HATN-MCOF nanosheets reach the best FE of 31.6% at −0.30 V (vs. RHE) (Fig. S23†), even lower than that of HATN-COFs (57.1% at −0.70 vs. RHE, Fig. S24†). The lower FE of Ni-HATN-MCOFs may be attributed to the better HER activity of Ni sites. Thus, the strategy of simulating enzymes is feasible; the HATN-COFs exhibit the optimal electrocatalytic activity for the NRA reaction. The selectivity for NH₃ is another factor to estimate the activity of NRA catalysts, because of the nitrite by-product (NO₃⁻ + 2H⁺ + 2e⁻ → NO₂⁻ + H₂O, 0.93 vs. SHE). In this work, the Mo-HATN-COF nanosheets demonstrate good NH₃ selectivity of ∼95.8% in the NRA reaction, indicating their high efficiency for NH₃ production (Fig. 4d). For the Ni-HATN-MCOF nanosheets, the selectivity is ∼92.7%, higher than that of HATN-MCOFs (80.6%). This result indicates that the metal sites (Mo or Ni) exhibit better catalytic activity for reduction of NO₃⁻ to NH₃, but some reduction reactions stop at the NO₂⁻ step on nanosheets with the non-metal sites (N sites). Finally, to investigate the effect of nitrate concentration on catalytic activity, nitrates of various concentrations (0.01, 0.02, 0.05, 0.1 and 0.5 M) were used as nitrogen sources in the electrolyte (Fig. 4e). The yield curve shows that the NH₃ production becomes faster as the concentration increases. The FE slightly increases as the concentration is increased until 0.1 M NO₃⁻. However, at higher concentrations, the FE decreases because the produced ammonia cannot leave the active sites in time.²⁰ All electrochemical results demonstrate that HATN-COF nanosheets have the optimal catalytic activity for NRA, which originates from three factors: (1) highly active Mo sites for NRA; (2) the fully conjugated framework for electron transfer; and (3) the hierarchical and in-plane porous structure for mass transfer. Moreover, the M-HATN-MCOF nanosheets also act as ideal platforms to investigate the catalytic activity of metal atoms. In this work, we find that the activity of the Mo atom is better than that of the Ni atom and the non-metal atom (nitrogen). The scope of the MCOF platform can be extended to most metal species due to the very strong coordinating ability of the bidentate tetraamine in the HATN subgroup.

To further investigate the electrocatalytic property for NRA, several control experiments were conducted. Firstly, in the absence of nitrate, no NH₃ was detected in the electrolyte (Fig. S25†). Only trace amounts of NH₃ were produced during the electroreduction process under open circuit conditions. This indicates that the presence of a nitrogen source and an appropriate applied potential are necessary conditions for NH₃ production. Secondly, isotopic labeling experiments were performed to verify the origin of NH₃ (Fig. 4f).^42,43 The ¹H NMR result shows three signals assigned to the ¹⁴N atom in the electrolyte containing normal ¹⁴NO₃⁻. In contrast, the NMR spectrum exhibits two ¹⁵N signals when labeled Na¹⁵NO₃ is used. This clearly illustrates that the nitrogen element in as-prepared NH₃ originates solely from the nitrate in the electrolyte. Moreover, the stability of these electrocatalysts determines their practical utility. When an applied potential of −0.5 V vs. RHE was used, the current density of Mo-HATN-COFs showed only a slight loss of 13.4% after 10 hours of electrocatalytic NRA (Fig. 5a). Furthermore, the NH₃ yield rates of Mo-HATN-COFs remained high, exceeding 8.31 mg h⁻¹ cm⁻² with the FEs of at least 83.6% (Fig. 5b) in 5 consecutive electrochemical tests. SEM and XRD techniques were used to investigate the effect of the electrochemical process on the structure. The SEM image reveals a thin-layered morphology with a porous structure, which has not changed at all. This also indicates that the Mo-HATN-COFs maintain stable chemical structures in long-term electrochemical reduction reactions (Fig. 5c). More importantly, the XRD result provides direct evidence to prove the structural stability of Mo-HATN-COFs. After 10 h of electrocatalysis, the XRD pattern of Mo-HATN-COFs shows an obvious diffraction peak at 26.4° (Fig. S26†), which is similar to that of the as-prepared Mo-HATN-COF sample (26.5°). Thus, these results demonstrate excellent structural stability during the long-term electrochemical NRA process. These findings indicate the good stability of Mo-HTAN-MCOFs toward NRA. All above-mentioned results collectively highlight that M-HATN-COF nanosheets serve as exceptional catalytic models for efficient NRA electrocatalysis.

The investigation of the catalytic mechanism of NRA

More importantly, M-HATN-COFs featuring single-metal sites serve as suitable models for revealing the mechanism of NRA. To investigate the effect of single-metal sites, we conducted DFT calculations to analyze the free energy of the NRA process across various M-HATN-COFs (Fig. S27–29†).⁴⁴ In this work, the unsaturated single-metal sites act as catalytic centers for NO₃⁻ absorption. The free energy profiles show that the Mo sites exhibit the most favourable NO₃⁻ absorption, with an adsorption energy of −0.53 eV, lower than that of the Ni sites (0.24 eV) and N sites (−0.19 eV). The lower adsorption energy indicates that the Mo sites play a crucial role in nitrate absorption and in the subsequent electrocatalytic process (Fig. 6a). During the NRA catalytic process, the hydrogenation of *NOH (*NOH → *NH₂OH) serves as the potential-determining step (PDS) at Ni sites with an energy barrier of 1.30 eV, which is larger than that at Mo sites. In the Mo-HATN-COF slab, the PDS involves the hydrogenation of *NO (*NO → *NOH), and its barrier is only 0.88 eV, making it superior to that of the HATN-COF slab (2.61 eV). These results demonstrate that the Mo sites exhibit the best activity for NRA catalysis, with the activity order of Mo site > Ni site > non-metal site (N), aligning well with the electrochemical experimental results. Moreover, in the *NO slabs, the bond length of NO in Mo-HATN-COFs is 1.665 Å (Fig. 6b), larger than that in Ni-HATN-COFs (1.273 Å, Fig. 6c). The larger bond length indicates a weaker interaction between N atoms and O atoms in NO, which facilitates the further hydrogenation of the N atom. Furthermore, the charge difference densities of various M-HATN-COF slabs with NO are also studied in this work (Fig. 6d, e and S30†). It is evident that the electrons of NO are more inclined towards Mo sites rather than Ni sites and N sites. This signifies that the Mo sites have stronger interactions with NO, leading to enhanced catalytic activity for NRA. Consequently, the DFT results indicate that the Mo atom holds promise as an active center for constructing NRA catalysts, and the M-HATN-COF is an effective platform for studying the catalytic mechanism of single-metal atoms.

Conclusions

By utilizing the M₃·HATN as a subgroup, we have designed novel conductive metal-covalent organic framework (M-HATN-COF) nanosheet, which were achieved through a salt-template-directed coordination–condensation reaction. In this process, metal atoms directly coordinate with the abundant bidentate tetraamine in the HATN subgroup (M at% ≈ 12.5%), resulting in single-atom metal sites with high unsaturation. Both DFT and four-probe results demonstrate that the incorporation of M–N₂ moieties introduces metallic conductivity to the HATN-COF structure through π–d conjugation. Regarding the NRA reaction, the Mo-HATN-COFs display excellent electrocatalytic activity, showing high ammonia yield rate (8.52 mg h⁻¹ cm⁻²) and FE (91.3%). In particular, the M-HATN-COFs serve as suitable platforms for studying the catalytic mechanism of metal atoms, in contrast to using conventional inorganic catalytic structures. Inspired by these findings, both theoretical and experimental results demonstrate a significant outcome: the activity of the Mo site surpasses that of the Ni site and the non-metal site (Mo site > Ni site > N site). The M-HATN-COFs presented in this work offer a new strategy for designing next-generation catalysts, thereby paving the way for NRA application in industry.

Author contributions

Hao Huang: c-MCOF design, preparation, characterization, data analysis, and writing the first draft. Kaiying Wang: data analysis, supervision, and critical reviewing.

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

This work was supported by the MSCA-IF-2020-Individual Fellowships (Grant No. 101024758) and the Norwegian Micro- and Nano-Fabrication Facility (NorFab, No. 245963/F50). We thank Dr Miao Yu for support in the ¹H NMR and electrochemical analyses.

References

1. Y. T. Wang, H. J. Li, W. Zhou, X. Zhang, B. Zhang and Y. F. Yu, Angew. Chem., Int. Ed., 2022, 61, e202202604 CrossRef CAS PubMed .
2. V. Rosca, M. Duca, M. T. de Groot and M. T. M. Koper, Chem. Rev., 2009, 109, 2209–2244 CrossRef CAS PubMed .
3. W. H. Guo, K. X. Zhang, Z. B. Liang, R. Q. Zou and Q. Xu, Chem. Soc. Rev., 2019, 48, 5658–5716 RSC .
4. C. He, Z. Y. Wu, L. Zhao, M. Ming, Y. Zhang, Y. P. Yi and J. S. Hu, ACS Catal., 2019, 9, 7311–7317 CrossRef CAS .
5. J. Li, Y. Zhang, C. Liu, L. R. Zheng, E. Petit, K. Qi, Y. Zhang, H. L. Wu, W. S. Wang, A. Tiberj, X. C. Wang, M. Chhowalla, L. Lajaunie, R. H. Yu and D. Voiry, Adv. Funct. Mater., 2022, 32, 2108316 CrossRef CAS .
6. M. D. Fryzuk, Nature, 2004, 427, 498–499 CrossRef CAS PubMed .
7. D. R. MacFarlane, P. V. Cherepanov, J. Choi, B. H. R. Suryanto, R. Y. Hodgetts, J. M. Bakker, F. M. F. Vallana and A. N. Simonov, Joule, 2020, 4, 1186–1205 CrossRef CAS .
8. J. Wang, L. Yu, L. Hu, G. Chen, H. L. Xin and X. F. Feng, Nat. Commun., 2018, 9, 1795 CrossRef PubMed .
9. D. Qi, F. Lv, T. Wei, M. Jin, G. Meng, S. Zhang, Q. Liu, W. Liu, D. Ma, M. S. Hamdy, J. Luo and X. Liu, Nano Res. Energy, 2022, 1, e9120022 CrossRef .
10. J. Lim, C. A. Fernandez, S. W. Lee and M. C. Hatzell, ACS Energy Lett., 2021, 6, 3676–3685 CrossRef CAS .
11. Y. T. Wang, C. H. Wang, M. Y. Li, Y. F. Yu and B. Zhang, Chem. Soc. Rev., 2021, 50, 6720–6733 RSC .
12. J. Chen, X. He, D. L. Zhao, J. Li, S. J. Sun, Y. S. Luo, D. D. Zheng, T. S. Li, Q. Liu, L. S. Xie, Y. X. Lin, A. M. Asiri and X. P. Sun, Green Chem., 2022, 24, 7913–7917 RSC .
13. P. H. van Langevelde, I. Katsounaros and M. T. M. Koper, Joule, 2021, 5, 290–294 CrossRef .
14. L. Q. Li, C. Tang, X. Y. Cui, Y. Zheng, X. S. Wang, H. L. Xu, S. Zhang, T. Shao, K. Davey and S. Z. Qiao, Angew. Chem., Int. Ed., 2021, 60, 14131–14137 CrossRef CAS PubMed .
15. J. Liang, Q. Liu, A. A. Alshehri and X. Sun, Nano Res. Energy, 2022, 1, e9120010 CrossRef .
16. J. Long, S. M. Chen, Y. L. Zhang, C. X. Guo, X. Y. Fu, D. H. Deng and J. P. Xiao, Angew. Chem., Int. Ed., 2020, 59, 9711–9718 CrossRef CAS PubMed .
17. R. Daiyan, T. Tran-Phu, P. Kumar, K. Iputera, Z. Z. Tong, J. Leverett, M. H. A. Khan, A. A. Esmailpour, A. Jalili, M. Lim, A. Tricoli, R. S. Liu, X. Y. Lu, E. Lovell and R. Amal, Energy Environ. Sci., 2021, 14, 3588–3598 RSC .
18. A. Stirling, I. Papai, J. Mink and D. R. Salahub, J. Chem. Phys., 1994, 100, 2910–2923 CrossRef CAS .
19. Y. Y. Wang, L. L. Zhang, Y. J. Niu, D. Fang, J. Wang, Q. X. Su and C. Wang, Green Chem., 2021, 23, 7594–7608 RSC .
20. M. Yu, H. Huang, J. Hu, S. Wang, J. P. Li and D. S. Wang, J. Mater. Chem. A, 2022, 10, 23990–23997 RSC .
21. J. Liang, B. Deng, Q. Liu, G. L. Wen, Q. Liu, T. S. Li, Y. L. Luo, A. A. Alshehri, K. A. Alzahrani, D. W. Ma and X. P. Sun, Green Chem., 2021, 23, 5487–5493 RSC .
22. Y. C. Zeng, C. Priest, G. F. Wang and G. Wu, Small Methods, 2020, 4, 2000672 CrossRef CAS .
23. X. F. Wan, W. W. Guo, X. Dong, H. H. Wu, X. F. Sun, M. G. Chu, S. T. Han, J. X. Zhai, W. Xia, S. Q. Jia, M. Y. He and B. X. Han, Green Chem., 2022, 24, 1090–1095 RSC .
24. J. W. Li, P. Liu, J. Y. Yan, H. Huang and W. B. Song, Adv. Sci., 2023, 10, 2206165 CrossRef CAS PubMed .
25. J. Dong, X. Han, Y. Liu, H. Li and Y. Cui, Angew. Chem., Int. Ed., 2020, 59, 13722–13733 CrossRef CAS PubMed .
26. L. Zhang, R. Bu, X. Y. Liu, P. F. Mu and E. Q. Gao, Green Chem., 2021, 23, 7620–7629 RSC .
27. C. Z. Zhu, S. F. Fu, Q. R. Shi, D. Du and Y. H. Lin, Angew. Chem., Int. Ed., 2017, 56, 13944–13960 CrossRef CAS PubMed .
28. W. H. Lai, Z. C. Miao, Y. X. Wang, J. Z. Wang and S. L. Chou, Adv. Energy Mater., 2019, 9, 1900722 CrossRef CAS .
29. H. Huang, Y. Zhao, Y. M. Bai, F. M. Li, Y. Zhang and Y. Chen, Adv. Sci., 2020, 7, 2000012 CrossRef CAS PubMed .
30. L. Wang, X. S. Zheng, L. Chen, Y. J. Xiong and H. X. Xu, Angew. Chem., Int. Ed., 2018, 57, 3454–3458 CrossRef CAS PubMed .
31. Z. H. Li, J. Tan, X. D. Zhu, S. J. Xie, H. Y. Fang, M. X. Ye and J. F. Shen, Energy Storage Mater., 2022, 51, 294–305 CrossRef .
32. H. Huang, F. M. Li, Q. Xue, Y. Zhang, S. W. Yin and Y. Chen, Small, 2019, 15, 1903500 CrossRef CAS PubMed .
33. X. Xiao, H. Wang, P. Urbankowski and Y. Gogotsi, Chem. Soc. Rev., 2018, 47, 8744–8765 RSC .
34. X. Xiao, H. B. Song, S. Z. Lin, Y. Zhou, X. J. Zhan, Z. M. Hu, Q. Zhang, J. Y. Sun, B. Yang, T. Q. Li, L. Y. Jiao, J. Zhou, J. Tang and Y. Gogotsi, Nat. Commun., 2016, 7, 11296 CrossRef CAS PubMed .
35. C. H. Li, H. X. Liu and Z. Y. Yu, Appl. Catal., B, 2019, 241, 95–103 CrossRef CAS .
36. J. Xiong, W. W. Cai, W. J. Shi, X. L. Zhang, J. Li, Z. H. Yang, L. G. Feng and H. S. Cheng, J. Mater. Chem. A, 2017, 5, 24193–24198 RSC .
37. X. Xiao, H. M. Yu, H. Y. Jin, M. H. Wu, Y. S. Fang, J. Y. Sun, Z. M. Hu, T. Q. Li, J. B. Wu, L. Huang, Y. Gogotsi and J. Zhou, ACS Nano, 2017, 11, 2180–2186 CrossRef CAS PubMed .
38. J. Bai, H. Huang, F. M. Li, Y. Zhao, P. Chen, P. J. Jin, S. N. Li, H. C. Yao, J. H. Zeng and Y. Chen, J. Mater. Chem. A, 2019, 7, 21149–21156 RSC .
39. Z. Hao, X. Li, Y. Tian, T. Ding, G. Yang, Q. Ma, N. Tsubaki and X. Li, Catalysts, 2021, 11, 230 CrossRef CAS .
40. D. O. Oluwole, S. L. Manoto, R. Malabi, C. Maphanga, S. Ombinda-Lemboumba, P. Mthunzi-Kufa and T. Nyokong, Dyes Pigm., 2018, 150, 139–150 CrossRef CAS .
41. Z. Guo, Y. Zhong, Z. Xuan, C. Mao, F. Dua and G. Li, RSC Adv., 2015, 5, 62624–62629 RSC .
42. 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, 574, E5 CrossRef CAS PubMed .
43. C. Y. Hu, X. Chen, J. B. Jin, Y. Han, S. M. Chen, H. X. Ju, J. Cai, Y. R. Qiu, C. Gao, C. M. Wang, Z. M. Qi, R. Long, L. Song, Z. Liu and Y. J. Xiong, J. Am. Chem. Soc., 2019, 141, 7807–7814 CrossRef CAS PubMed .
44. R. Zhang, L. Jiao, W. J. Yang, G. Wan and H. L. Jiang, J. Mater. Chem. A, 2019, 7, 26371–26377 RSC .

---

Footnote

† Electronic supplementary information (ESI) available: Details of physical characterization, electrochemical measurements, and density functional theory calculations. See DOI: https://doi.org/10.1039/d3gc01914d

This journal is © The Royal Society of Chemistry 2023