Electrocatalytic Conversion of Nitrates and Nitrites to Ammonia

Abstract

Electrocatalytic reduction of waste nitrates (NO3−) enables the synthesis of ammonia (NH3) in a carbon neutral and decentralized manner. The present invention uses atomically dispersed transition metal-nitrogen-carbon (M-N—C) catalysts with varying metal centers to uniquely favor mono-nitrogen products (e.g., NH3) via the electrochemical nitrate reduction reaction (NO3RR).

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application claims the benefit of U.S. Provisional Appl. No. 63/324,537, filed Mar. 28, 2022, which is incorporated herein by reference.

STATEMENT OF GOVERNMENT INTEREST

This invention was made with Government support under Contract No. DE-NA0003525 awarded by the United States Department of Energy/National Nuclear Security Administration. The Government has certain rights in the invention.

STATEMENT REGARDING PRIOR DISCLOSURES BY THE INVENTOR OR A JOINT INVENTOR

The following disclosure is submitted under 35 U.S.C. 102(b)(1)(A): Eamonn Murphy, Yuanchao Liu, Ivana Matanovic, Shengyuan Guo, Peter Tieu, Ying Huang, Alvin Ly, Suparna Das, Iryna Zenyuk, Xiaoqing Pan, Erik Spoerke, and Plamen Atanassov, “Highly Durable and Selective Fe- and Mo-based Atomically Dispersed Electrocatalysts for Nitrate Reduction to Ammonia via Distinct and Synergized NO₂⁻ Pathways,” ACS Catalysis 12, 6651 (2022). The subject matter of this disclosure was conceived of or invented by the inventors named in this application.

FIELD OF THE INVENTION

The present invention relates to ammonia synthesis and, in particular, to the electrocatalytic conversion of nitrates and nitrites to ammonia.

BACKGROUND OF THE INVENTION

Recently, there has been a surge in global research efforts towards the electrochemical production of ammonia (NH₃). See R. Society, Ammonia: Zero-Carbon Fertiliser, Fuel and Energy Store, The Royal Society London, UK (2020); and Y. Ren et al., Energy Environ. Sci. 14(3), 1176 (2021). NH₃ is critical to produce N-based fertilizers and shows promise as an energy vector and carbon free liquid fuel. See R. Society, Ammonia: Zero-Carbon Fertiliser, Fuel and Energy Store, The Royal Society London, UK (2020); S. Giddey et al., Int. J. Hydrogen Energy 38(34), 14576 (2013); A. Grinberg Dana et al., Angew. Chemie—Int. Ed. 55(31), 8798 (2016); and C. Smith et al., Energy Environ. Sci. 13(2), 331 (2020). Currently, the near century-old Haber-Bosch (H-B) process is the only industrial NH₃ manufacturing approach. However, the H-B process operates at high temperatures and pressures and utilizes gray hydrogen. See A. Klerke et al., J. Mater. Chem. 18(20), 2304 (2008). As a result, the process consumes 1-2% of global energy and produces 1-2% of global CO₂ emissions. See C. Smith et al., Energy Environ. Sci. 13(2), 331 (2020); and J. Norskov and J. Chen, “Sustainable Ammonia Synthesis”, Report, DOE Roundtable 2016, 1-23. Electrochemical methods offer carbon-neutral pathways for NH₃ synthesis at ambient conditions and in a distributed manner. To date, the electrocatalytic nitrogen reduction reaction (eNRR) from dinitrogen gas is the most heavily researched. However, eNRR faces several challenges such as low solubility in aqueous electrolytes, competition from the more facile hydrogen evolution reaction (HER) and large thermodynamic activation barriers. See C. Tang and S. Z. Qiao, Chem. Soc. Rev. 48(12), 3166 (2019); D. R. MacFarlane et al., Joule 4(6), 1186 (2020); and B. H. R. Suryanto et al., Nat. Catal. 2(4), 290 (2019). As a result, the Faradaic efficiencies (FEs) and yields of NH₃ in aqueous systems remain low, even proving difficult for reliable analytical detection. See J. Choi et al., Nat. Commun. 11(1), 1 (2020); Y. Chen et al., Nat. Catal. 3(12), 1055 (2020); and S. Z. Andersen et al., Nature 570(7762), 504 (2019).

To circumvent these challenges, oxidized nitrogen species, such as nitrate (NO₃⁻) can be utilized as feedstock for efficient electrocatalytic NH₃ production. NO₃⁻ is the second most abundant form of nitrogen as an environmental pollutant present in industrial waste streams and water runoff at concentrations up to 2 M. See I. Katsounaros et al., J. Hazard. Mater. 171(1-3), 323 (2009); R. Chauhan and V. C. Srivastava, Chem. Eng. J. 386(3), 122065 (2020); and T. T. P. Nguyen et al., ECS Trans. 53(16), 41 (2013). Nitrate's appeal is owed to its large solubility in aqueous electrolytes and relatively low bond dissociation energy (204 kJ/mol), facilitating significantly more favorable kinetics compared to eNRR processes. See D. Xu et al., Front. Environ. Sci. Eng. 12(1), 1 (2018); W. T. Mook et al., Desalination 285, 1 (2012); and L. Su et al., Nano Lett. 19(8), 5423 (2019). The electrocatalytic nitrate reduction reaction (NO₃RR) has been studied extensively for denitrification and water purification. See D. Xu et al., Front. Environ. Sci. Eng. 12(1), 1 (2018); W. T. Mook et al., Desalination 285, 1 (2012); and H. Liu et al., ACS Catal. 2(3), 8431 (2021). Early NO₃RR research focused on selective reduction of NO₃⁻ to ½N₂ via a 5e⁻ transfer pathway (E⁰=1.25 V vs. RHE) and has been validated on a variety of bulk metals, heterogeneous catalysts (Ru, Rh, Pd, Ir, Pt, Cu, Ag and Au), and their alloys. See L. Su et al., Nano Lett. 19(8), 5423 (2019); J. Martinez et al., Appl. Catal. B Environ. 207, 42 (2017); M. T. De Groot et al., J. Electroanal. Chem. 562, 81 (2004); M. D'Arino et al., Appl. Catal. B Environ. 53(3), 161 (2004); A. Pintar et al., Appl. Catal. B Environ. 52(1), 49 (2004); X. Huo et al., Appl. Catal. B Environ. 211, 188 (2017); A. H. Pizarro et al., J. Environ. Chem. Eng. 3(4), 2777 (2015); M. Yanauchi et al., J. Am. Chem. Soc. 133, 1150 (2011); E. Perez-Gallent et al., Electrochim. Acta 227, 77 (2017); J. Yang et al., Chem. Commun. 50, 2148 (2014); and F. Calle-Vallejo et al., Phys. Chem. Chem. Phys. 15, 3196 (2013).

Recently, significant attention has been shifted to selectively reducing NO₃⁻ to NH₃ via an 8e⁻ transfer pathway (E⁰=0.82 V vs. RHE). The 8e⁻ transfer pathway is highly complex, involving many potential reaction intermediates (NO₂, NO₂⁻, NO, N₂O, N₂, NH₃, NH₂OH and N₂H₄). The initial 2e⁻ transfer reduction of adsorbed *NO₃⁻ to *NO₂⁻ is thought to be universal in the NO₃RR pathways and is regarded as the rate-limiting step. See Z. Wang et al., Catal. Sci. Technol. 11, 705 (2021); Y. Zeng et al., Small Methods 4(12), 1 (2020); and M. Duca and M. T. Koper, Energy Environ. Sci. 5, 9726 (2012). Further reduction to *NO is regarded as the selectivity determining step on extended lattice surfaces. From *NO, the reaction pathways diverge selectively towards NH₃ or N₂, depending on electrolyte conditions and intermediates' adsorption strength on catalyst surfaces. Several recent computation studies have examined the selectivity of the *NO to subsequent reaction products (N₂, NH₂OH, and NH₃) over a variety of transition metal surfaces. See C. A. Casey-Stevens et al., Appl. Surf. Sci. 552, 149063 (2021); H. J. Chun et al., ACS Catal. 7, 3869 (2017); and H. Wan et al., Angew. Chem. 133, 22137 (2021). By contrast, single-atom catalysts (SACs) provide an intrinsic advantage favoring NH₃ production over N₂ products. By creating isolated single-atom active sites, the possibility for two mono-nitrogen moieties can be activated on adjacent sites and form coupled N₂ products (N₂O, N₂, N₂H₄) is minimized, narrowing product selectivity largely to NH₃. See Z. Wu et al., Nat. Commun. 12(1), 1 (2021). Recent work reported the feasibility of Fe-based SACs for NO₃RR and some theoretical work indicated that single-atom active sites preferentially adsorb NO₃⁻ over H⁺, reducing competition from the HER. See Z. Wu et al., Nat. Commun. 12(1), 1 (2021); P. Li et al., Energy Environ. Sci. 14, 3522 (2021); H. Niu et al., Adv. Funct. Mater. 31(11), 2008533 (2020); J. Wu et al., Catal. Sci. Technol. 11, 7160 (2021); and J. Wu et al., J. Phys. Chem. Lett. 12, 3968 (2021). Although good progress has been made in the field of nitrate reduction for NH₃ production, developing well-established reaction mechanisms and electrocatalysts with long-term operational stability remains a significant challenge. See P. H. van Langevelde et al., Joule 5(2), 290 (2021).

SUMMARY OF THE INVENTION

The present invention is a method for the electrocatalytic conversion of nitrates and nitrites to ammonia, comprising providing an electrochemical cell, the electrochemical cell comprising a working electrode comprising an atomically dispersed transition metal-nitrogen-carbon (M-N—C) catalyst, a counter electrode, a hydrogen permeable membrane between the working and counter electrodes, and an electrolyte comprising an aqueous solution of nitrate or nitrite; and applying an electrode potential between the working electrode and the counter electrode sufficient to electrocatalytically reduce the nitrate or nitrite to ammonia at the working electrode. For example, the transition metal can comprise Fe, Mo, Cr, Mn, Co, Ni, Cu, Ru, Rh, Pd, W, La, or Ce, or a combination thereof. For example, the M-N—C catalyst can be a monometallic catalyst or a bimetallic catalyst, such as FeMo. The electrode potential can be more negative than −0.1 V.

The electrochemical nitrate reduction reaction (NO₃RR) not only promises an effective route to sustainable ammonia synthesis in a carbon-neutral and decentralized manner, but also offers potential impact to critical wastewater remediation. The bioinspired, atomically dispersed catalysts of the present invention have exception efficacy for NO₃RR in aqueous media via a catalytic cascade. Atomically dispersed catalysts have intrinsic selectivity towards mono-nitrogen species over their dinitrogen counterparts. As an example, a series of nitrogen-coordinated mono- and bimetallic, atomically dispersed, iron- and molybdenum-based electrocatalysts were evaluated for ammonia synthesis via the NO₃RR. The key role of the nitrite *NO₂/NO₂⁻ intermediates was identified both computationally and experimentally, wherein the Fe—N₄ sites and Mo—N₄/*O—Mo—N₄ sites carried distinct associative and dissociative adsorption of NO₃⁻ molecules, respectively. By integrating individual Fe and Mo sites on a single bimetallic catalyst, the unique reaction pathways were synergized, achieving a Faradaic efficiency of 94% toward ammonia. The utilization of catalytic cascades, synergizing distinct reaction pathways on heterogeneous single-atom sites, is largely unconstrained by linear scaling relations of reaction intermediates. The invention enables electrocatalysts for highly selective, efficient, and durable ammonia synthesis.

BRIEF DESCRIPTION OF THE DRAWINGS

The detailed description will refer to the following drawings, wherein like elements are referred to by like numbers.

FIG. 1 is a schematic illustration for an electrochemical cell for the electrocatalytic conversion of nitrates and nitrites to ammonia.

FIGS. 2A-2C illustrate distinct NO₃⁻ to NO₂⁻ pathways in the electrochemical nitrate reduction (NO₃RR) for NH₃ synthesis over Fe- and Mo-based M-N₄ sites. FIG. 2A illustrates associative adsorption of NO₃⁻ on Fe—N₄ sites as predicted by DFT calculations. FIG. 2B illustrates dissociative adsorption of NO₃⁻ into *O and NO₂⁻ on Mo—N₄ sites. FIG. 2C illustrates a proposed catalytic cascade for synergized NO₃⁻ reduction over Mo—N₄ and Fe—N₄ sites. FIGS. 2D-2F are graphs of the evolution of NH₃ and NO₂⁻ concentrations over a 24 h NO₃RR electrolysis catalyzed by: monometallic Fe—N—C catalyst (FIG. 2D), monometallic Mo—N—C catalyst (FIG. 2E), and bimetallic FeMo—N—C catalyst (FIG. 2F). Insets show top-down view of corresponding active sites. The electrolysis was performed at −0.45 V (vs. RHE) in 0.05M PBS+0.16M NO₃⁻ electrolyte. The errors were calculated from three independent electrolysis. Products were detected and quantified using an ultraviolet-visible (UV-Vis) spectrophotometer.

FIGS. 3A-3E show the electrocatalytic performance of NO₂⁻ and NO₃⁻ reduction for atomically dispersed Fe—N—C, Mo—N—C and FeMo—N—C catalysts. FIG. 3A is a linear sweep voltammetry (LSV) scan for the FeMo—N—C catalyst in 0.05M PBS and 0.05M PBS doped with either 0.01M NO₂⁻ or 0.16M NO₃⁻. FIGS. 3B-3C show the potential dependent NO₂RR (0.5 h) for the three catalysts at a NO₂⁻ concentration of 0.01 M, specifying: NH₃ Faradaic efficiency (FE) (FIG. 3B) and NH₃ yield rate (FIG. 3C). FIGS. 3D-3E show the potential dependent NO₃RR (4 h) for the three catalysts at a NO₃⁻ concentration of 0.16M. FIG. 3D is a graph of FE for NH₃ (solid bars) and NO₂⁻ (dashed bars) at varying potentials, with the dotted line indicating 100% FE. FIG. 3E is a graph of yield rates for NH₃. All errors were calculated from three independent electrolysis.

FIGS. 4A-4D show NO₃RR stability test and isotopic analysis for the FeMo—N—C catalyst. FIG. 4A shows 60 h electrolysis at −0.45 V (vs. RHE) in 0.05 M PBS+0.16 M NO₃⁻. Every 12 h, the electrolyte was sampled for NH₃ and NO₂⁻ quantification and then refreshed for the next cycle. FIGS. 4B-4D show ¹⁵NO₃RR electrolysis at −0.45 V with isotopically labeled ¹⁵NO₃⁻ (0.16 M). FIG. 4B shows ¹H-NMR spectra of the electrolyte at varying time intervals. The dashed line at the 6 h point indicates ¹⁵NH₃ detected in counter chamber due to the NH₃ crossing over the membrane. FIG. 4C is a graph of concentration of ¹⁵NH₃ as a function of the electrolysis time. FIG. 4D shows a comparison of the NH₃ FE and yield rate from ¹⁴NO₃⁻ and ¹⁵NO₃⁻ feeds, over 6 h of electrolysis.

FIGS. 5A-5G are related to physical characterization of the atomically dispersed catalysts. FIG. 5A is a XRD pattern of the mono- and bimetallic Fe—N—C, Mo—N—C and FeMo—N—C catalysts. FIG. 5B shows a low magnification aberration corrected high-angle annular dark-field (AC-HAADF) STEM image of the FeMo—N—C catalyst. FIG. 5C is an atomic resolution AC-HAADF STEM image, bright spots indicate atomically dispersed Fe and Mo sites. FIG. 5D is an AC-HAADF image and energy dispersive X-ray spectroscopy (EDS) mapping of the FeMo—N—C catalyst. FIG. 5E is a graph of N 1s X-ray photoelectron spectra (XPS) of the FeMo—N—C catalyst confirming the formation of M-N_x moieties. FIG. 5F is a graph of Fe 2p XPS indicating the presence of Fe—N_x sites. FIG. 5G is a graph of Mo 3d XPS indicating the presence of Mo—N_x sites.

FIG. 6A illustrates a DFT optimized structure of the Fe—N₄ center (side and top view). FIG. 6B illustrates a DFT optimized structure of the Mo—N₄ center. Only atoms belonging to one unit cell are shown (C—black, N—blue, Fe—gold, Mo—purple).

FIGS. 7A-7B show the Gibbs free energy landscape and intermediate geometries for the conversion of NO₃⁻ to *NO₂/NO₂⁻ on: Fe—N₄ sites (FIG. 7A), and Mo—N₄ sites (FIG. 7B). FIG. 7C shows the Gibbs free energy landscape and intermediate geometries for the cyclic conversion of NO₃⁻ to NO₂⁻ over oxygenated Mo—N₄ sites (*O—Mo). The associative adsorption pathways are marked in grey, while dissociative adsorption and dissociation steps are marked in red. A cathodic potential of 0 V and pH of 6.3 were used to estimate the Gibbs free energy levels.

FIGS. 8A-8E show the physical structure of the atomically dispersed transition-metal catalysts. FIG. 8A shows the selection of 3d, 4d, 5d and f metals synthesized via the sacrificial support method (SSM). FIG. 8B is a schematic of the nitrogen coordinated metal active site (M-N₄) on a prototype carbon matrix, illustrating in-plane and out-of-plane configurations. FIGS. 8C-8E show representative atomic resolution AC-HAADF STEM images for the Fe—N—C, Rh—N—C and La—N—C catalysts, respectively. High contrast points indicate atomically dispersed metal sites. Corresponding EDS mappings of the M-N—C are shown below the images.

FIGS. 9A-9L show the coordination environment of the single-atom metal centers. FIGS. 9A-9C show atomic resolution EELS point spectra of the N K-edge and corresponding metal-edge of Fe (L_3,2), Rh (M₃) and La (M_5,4), respectively, in the M-N—C catalysts, indicating nitrogen coordination of the single atom M-N_x site. All scale bars are 2 nm. FIGS. 9D-9F show FT EXAFS spectra of the Fe, Rh and La—N—C catalysts, respectively, with the corresponding metallic foil and metal oxide standards, demonstrating the presence of M-N_x sites. FIGS. 9G-9I show FT EXAFS of the Fe K-edge of Fe—N—C, Rh K-edge of Rh—N—C and La L₃-edge of La—N—C, respectively. FIGS. 9J-9L show N 1s XPS spectra for the Fe, Rh and La—N—C catalysts, respectively, confirming the presence of M-N_x moieties.

FIGS. 10A-10D show electrochemical activation, selectivity, and activity for the NO₃RR and NO₂RR. FIG. 10A is an example of a linear sweep voltammetry (LSV) curve for the Cr—N—C catalyst in electrolytes of 0.05M PBS (HER), 0.05M PBS+0.01 M KNO₂ (NO₂RR) and 0.05M PBS+0.16M KNO₃⁻ (NO₃RR). FIGS. 10B and 10C show representative LSV over the metal free N—C and a variety of 3d and 4d metals for the NO₃RR and NO₂RR, respectively. FIG. 10D is a gap analysis plot (GAP) for the electrochemical NO₃RR for all M-N—C catalysts in 0.05M PBS+0.16M KNO₃⁻ for 2 hours. The top section of the figure shows the Faradaic efficiency for NO₂⁻ (blue; top-down) and NH₃ (gray; bottom-up) as a function of the applied potentials between −0.2 V and −0.8 V vs. RHE. A horizontal line is set at 50% FE to guide the eye. The bottom section of the figure shows the corresponding yield rate (μmol h⁻¹ cm⁻²) for NO₂⁻ (blue; triangle) and NH₃ (gray; circle) as a function of the applied potentials. Error bars are determined from replicate trials at −0.40 V vs. RHE.

FIGS. 11A-11D show a mechanistic analysis of the NO₃RR via a NO₂⁻ intermediate Isotopic analysis of competing NO₃RR and NO₂RR reactions at −0.4 V vs. RHE. FIG. 11A is an experimental schematic for NO₃RR electrolysis in which small amounts of ¹⁵NO₂⁻ are doped in 0.16M ¹⁴NO₃⁻. FIG. 11B shows NMR spectra for the electrolysis with 10 ppm of ¹⁵NO₂⁻ doped in 0.16M ¹⁴NO₃⁻ (10,000 ppm), sampled at 2, 4, and 6 hours. FIG. 11C is a time course analysis of ¹⁴NH₃ and ¹⁵NH₃ concentration at 2, 4, and 6 hours for the NO₃RR with 10 ppm ¹⁵NO₂⁻ (left) and 100 ppm ¹⁵NO₂⁻ (right) in 0.16M ¹⁴NO₃⁻ (10,000 ppm). FIG. 11D shows electrochemical NO₂RR for all M-N—C catalysts in 0.05M PBS+0.01 M KNO₂⁻ for 0.5 hours. Faradaic efficiency for NH₃ as a function of the applied potential between −0.2 V and −0.8 V is shown in the top section. The bottom section shows the corresponding NH₃ yield rate (μmol h⁻¹ cm⁻²). Note, FEs over 100% are a result of electrolyte dilution to bring the concentration of NH₃ into the UV-vis detection calibrated range. FIGS. 11E and 11F show correlations between the NO₃RR Faradaic efficiency for NH₃ in FIG. 10E (y-axis) and the dividend of the NH₃ yield rate and Faradaic efficiency of the NO₂RR in FIG. 11D at −0.20 V (FIG. 11E) and −0.40 V (FIG. 11F) vs. RHE. The NO₂RR Yield_NH3/FE_NH3 is represented as the NO₂RR total current (NO₂RR, j_total). The Ru and Rh outliers in gray shades were excluded from the linear fit due to the dominant NO₃RR gaseous products as well as Co—N—C. FIG. 11G shows a summary of the goodness of fit, R values, for varying experimental NO₃RR—NO₂RR correlations.

FIGS. 12A-12D show computational NO₃RR descriptors as calculated by using DFT with optB86b-vdW functional. FIG. 12A shows Gibbs free energies (Δ_rG) of the reaction for the first two electron transfer steps in the NO₃RR. Processes that generate NO₂⁻ in the bulk electrolyte are shown in blue. FIG. 12B is a Quadrant plot of the Δ_rG for associative adsorption [*+NO₃⁻→*NO₂] (Y₁-axis), dissociative adsorption [*+NO₃⁻→*O+NO₂⁻] (X-axis) and *NO₂ desorption [*NO₂→*+NO₂⁻] (Y₂-axis), forming sectors where certain reaction pathways are thermodynamically favored. The main quadrants were determined by the X-axis and Y₁-axis. Quadrants III and IV were further divided by the Y₂-axis into III.a, III.b, IV.a, and IV.b sub-sections. The three reaction coordinates determined two types of NO₃⁻ adsorptions and two types of NO₂⁻ evolutions as shown by the diagram above the figure. FIG. 12C shows the correlation between DFT-derived Δ_rG [*+NO₃⁻→*NO₂] and experiment-derived NO₃RR FE_NH3 at −0.2 V vs. RHE. FIG. 12D shows the correlation between DFT-derived Δ_rG [*+NO₂⁻→*NO₂] and experiment-derived NO₂RR FE_NH3 at −0.2 V vs. RHE. Oxygenated active sites (O-M) were included for the oxyphilic elements (Mo, La, Ce and W).

DETAILED DESCRIPTION OF THE INVENTION

In nature, the NO₃RR pathway is carried by enzymatic cascades, wherein the conversion of NO₃⁻ to NO₂⁻ and subsequent reduction of NO₂⁻ to NH₃ are catalyzed by the nitrate reductase Mo cofactor and nitrite reductase Fe cofactor, respectively. See R. D. Milton et al., Chempluschem 82(4), 513 (2017); and L. Y. Stein and M. G. Klotz, Curr. Biol. 26(3), R94 (2016). This indicates the unique role of different metal elements for nitrogen transformation. As described below, an embodiment of the present invention uses a highly efficient synthetic electrocatalytic cascade utilizing chemical dissociation of the NO₃⁻ ion to NO₂⁻ over single-atom Mo sites, followed by a kinetically fast 6e⁻ transfer reduction of NO₂ to NH₃ at 100% FE over single-atom Fe sites.

The biological enzymatic process indicates the unique role of different metal elements in nitrogen transformation. An embodiment of the invention uses atomically dispersed mono- and bimetallic Fe- and Mo-based electrocatalysts for the NO₃RR. As will be described below, distinct NO₃⁻ to *NO₂/NO₂⁻ intermediate pathways were identified both computationally and experimentally. The initial NO₃⁻ adsorption was found to be dissociative on Mo sites and associative on Fe sites, showing great biological similarity. By integrating both Mo and Fe sites on a bimetallic catalyst, the two reaction mechanisms were synergized, and the synthetic catalytic cascade achieved a NO₃RR FE for NH₃ of 94% and a yield of 18.0 μmol cm⁻² hr⁻¹ (153 μg_NH3 mg_cat⁻¹ h⁻¹). Moreover, durability studies showed a well-maintained FE above 90% over a 60 h electrolysis. These properties enable synergizing distinct reaction pathways on atomically dispersed electrocatalysts for highly selective and efficient NH₃ synthesis.

Synthesis of Fe, Mo and FeMo—N—C Catalysts

The transition metal-nitrogen-carbon (M-N—C) catalysts are based on transition metal (M) ions coordinated with nitrogen (N) atoms and embedded into a graphitic carbon (C) matrix. See T. Asset et al., Joule 4, 33 (2020); and K. Kumar et al., ACS Catal. 11, 484 (2021), which are incorporated herein by reference. Atomically dispersed mono- and bimetallic catalysts can be synthesized using the well-established sacrificial support method (SSM). See A. Serov et al., Adv. Energy Mater. 4(10), 201301735 (2014); A. Serov et al., Nano Energy 16, 293 (2015); T. Asset et al., ACS Catal. 9(9), 7688 (2019); and A. Serov et al., Electrochim. Acta 179, 154 (2015), which are incorporated herein by reference. The SSM class of M-N—C catalysts has been widely studied due to their good activity and stability for the oxygen reduction reaction (ORR), where the oxidative cathodic potentials (0.6-1 V) are even harsher for carbon corrosion as compared to the reductive potentials in NO₃RR (−0.8 to −0.2 V). See T. Asset et al., Joule 4, 33 (2020); and M. Primbs et al., Energy Environ. Sci. 13, 2480 (2020). The outstanding ORR performance for the SSM M-N—C was largely attributed to its high degree of graphitization and robust M-N_x sites, which also greatly contributed to the excellent stability of the FeMo—N—C for NO₃RR, as will be described below. See Y. Chen et al., Mater. Today 53, 58 (2020).

As examples of the invention, atomically dispersed Fe—N—C and Mo—Ni—C monometallic catalysts and FeMo—N—C bimetallic catalysts were synthesized using the sacrificial support method. Iron nitrate nonahydrate and ammonium molybdenum tetrahydrate were selected as the iron and molybdenum precursors, respectively, for the catalyst synthesis. First, a calculated amount of metal nitrate precursor (Fe—N—C precursor=0.6 g, Mo—N—C precursor=0.26 g) was combined with silica of varying surface areas, consisting of in-house Stöber spheres, LM-150 and OX-50 (0.5, 1.25, and 1.25 g, respectively). Nicarbazin (6.25 g) was added to the precursor mixture. For the molybdenum-containing catalysts, a calculated amount of urea was added to assist in reducing the molybdenum precursor. The amount of metallic precursor was calculated such that the number of metal atoms between the mono- and bimetallic catalysts was constant. Next, MilliQ water was added to the precursor ratio and sonicated for 30 min. The viscous mixture was then set at 45° C. under constant stirring until dry. The material was then ground and ball-milled for 1 h at 45 Hz. A first pyrolysis was performed at 650° C. for 45 min under a reductive 7% H₂-93% Ar atmosphere. The material was then etched 4 days in a 40 wt % hydrofluoric acid mixture (2:1) HF/H₂O. The material was then washed to a neutral pH before being dried and again ball-milled. Next the material underwent a second pyrolysis at 650° C. for 30 min under a 10% NH₃-90% Ar atmosphere, followed by a final ball-milling.

Electrochemical Measurements

The electrocatalytic conversion of nitrates and nitrites to ammonia can be accomplished in an electrochemical H-cell, as shown in FIG. 1. The electrochemical cell comprises a working electrode comprising an atomically dispersed transition metal-nitrogen-carbon (M-N—C) catalyst, a counter electrode, a hydrogen-permeable membrane between the working and counter electrodes, and an electrolyte comprising an aqueous solution of nitrate or nitrite. An electrode potential can be applied between the working electrode and the counter electrode to electrocatalytically reduce the nitrate or nitrite to ammonia at the working electrode.

The working electrode was a carbon paper cut to a geometric surface area of 0.45 cm². A catalyst ink was prepared by dispersing 5 mg of the M-N—C catalyst in 300 μL of MilliQ water, 670 μL of isopropanol and 30 μL of a 5 wt % Nafion solution and ultrasonicated for 1 hour. 100 μL of the catalyst ink was drop cast onto a carbon paper working electrode for a total catalyst loading of 0.5 mg/cm² and dried in a vacuum oven at 60° C. overnight. A standard three-electrode system was used with a reversible hydrogen electrode and a graphite rod as the reference and counter electrode, respectively. Electrochemical nitrate and nitrite reduction reactions were carried out in a customized glass electrochemical H-cell comprising working and counter electrode chambers separated by a Nafion 211 membrane at ambient conditions. The electrolyte used for all nitrate reduction experiments is 0.05 M phosphate buffer solution (PBS) and 0.16 M NO₃⁻ (KNO₃). Prior to an electrochemical test, N₂ gas was purged in both the working and counter chambers for 30 min at 80 sccm, which contain 30 mL and 25 mL of electrolyte, respectively. Electrochemical nitrite reduction experiments were performed analogously to nitrate reduction experiments, utilizing the same H-cell and three electrode system. The electrolyte was 0.05 M PBS and 0.01 M NO₂⁻ (KNO₂). Potentiostatic tests were performed at applied potentials of between −0.2 and −0.70 V vs RHE.

Distinct *NO₂/NO₂⁻ Intermediate Pathways Over Fe and Mo Active Sites

In biological NO₃⁻ reduction pathways, Mo-based active centers reduce NO₃⁻ to NO₂⁻ and Fe-based active centers further reduce NO₂⁻ to NH₃ (or N₂). To evaluate the catalytic activity of atomically dispersed Fe and Mo sites toward the NO₃RR to NH₃, a series of atomically dispersed M-N—C(M=Fe, Mo, and FeMo) electrocatalysts were employed.

Density Functional Theory (DFT) calculations and experimental NO₃⁻ reduction time-course studies predicted and confirmed two different reaction mechanisms over the M-N₄ sites. All the DFT calculations were performed using a generalized gradient approximation approach and projector augmented-wave pseudopotentials as implemented in the Vienna Ab initio Simulation Package. See P. E. Blöchl et al., Phys. Rev. B 49(23), 16223 (1994); D. Joubert, Phys. Rev. B 59(3), 1758 (1999); G. Kresse and J. Hafner, Phys. Rev. B 47(1), 558 (1993); G. Kresse and J. Hafner, Phys. Rev. B 49(20), 14251 (1994); and G. Kresse and J. Furthmüller, Comput. Mater. Sci. 6(1), 15 (1996). As shown in FIG. 2A, over exclusively Fe sites, associative NO₃⁻ adsorption is favorable with a Gibbs free adsorption energy calculated at −0.39 eV. Namely, Fe sites can stabilize the adsorbed NO₃⁻ regardless of the orientation NO₃⁻ binds to the active sites, through either one or two oxygen atoms. This determined the basis of a direct 8e⁻ transfer pathway from *NO₃ to NH₃, which was supported by its electrocatalytic time-course study as shown in FIG. 2D, wherein the NO₂ concentration observed was negligible and constant throughout the entire 24 h electrolysis, while the concentration of NH₃ steadily increased. The detected minute amount of NO₂⁻ ions might be from the low-probability of *NO₃ dissociation over the Fe sites, as described in the following section. In contrast, over exclusively Mo sites (FIG. 2B), DFT calculations predicted the reaction initiates via a highly favorable dissociative-adsorption pathway (Δ_rG=−4.2 eV), spontaneously breaking the O—N bond of the NO₃⁻ molecule to form a surface *O intermediate and a NO₂⁻ ion. The corresponding experimental time-course study supports this hypothesis (FIG. 2E), highlighting a significant NO₂⁻ evolution over NH₃⁻ through the entire 24 h electrolysis. In this case, the complete NO₃⁻ to NH₃ process follows a 2e⁻+6e⁻ transfer pathway.

By integrating both Mo and Fe sites in a bimetallic catalyst (FeMo—N—C), a synergized pathway was proposed to optimize the NH₃ production with a catalytic cascade, as shown in FIG. 2C. That is, NO₃⁻ was dissociated to NO₂⁻ on Mo sites, and NO₂⁻ can be subsequently associatively adsorbed over Fe sites and further reduced to NH₃. In the corresponding time-course study (FIG. 2F), the synergistic effect was confirmed by observing improved NH₃ yield and a largely suppressed NO₂⁻ concentration as compared to the Mo—N—C case, indicating the additional Fe sites readily converted the NO₂⁻ to NH₃.

Electrochemical Nitrite and Nitrate Reduction

By mimicking biological pathways and synergizing Mo and Fe sites, the activity of the bimetallic catalyst for the NO₃RR was significantly increased compared with its monometallic counterparts. To further explore the catalyst activity and reaction mechanism, detailed potential-dependent studies were performed for the nitrite reduction reaction (NO₂RR) and NO₃RR processes.

The cathodic current and reaction onset potential were first evaluated by linear sweep voltammetry (LSV) in a standard 0.05 M phosphate-buffered solution (PBS) electrolyte. Employing the bioinspired bimetallic FeMo—N—C as an example, the LSV curves in FIG. 3A show that the HER had a delayed onset potential below −0.5 V (vs. RHE), indicating the suppression of the parasitic reaction, a characteristic of atomically dispersed active sites for the NO₃RR. See Z.-Y. Wu et al., Nat. Commun. 12, 2870 (2021); P. Li et al., Energy Environ. Sci. 14, 3522 (2021); H. Niu et al., Adv. Funct. Mater. 31, 2008533 (2020); and Y. Wang and M. Shao, ACS. Catal. 12, 5407 (2022). This is also confirmed by DFT calculations which show that due to the larger Gibbs free energy of adsorption for NO₃⁻ and NO₂⁻ vs H adatom (−0.39 and −1.07 eV vs +0.52 eV for Fe—N₄ and −2.72 and −2.24 eV vs −0.18 eV on Mo—N₄ sites at pH=6.3), HER should be pushed to −0.6 and −1.1 V vs RHE on Fe- and Mo-containing sites. The addition of 0.01 M NO₂⁻ shifts the onset potential over Fe—N_x sites to 0.12 V, observed in both the Fe—N—C and FeMo—N—C catalysts, indicating the high activity of the Fe sites toward the NO₂RR at low overpotentials. Particularly, when compared with the Mo—N—C catalyst, the Fe-containing catalysts showed a steep NO₂RR current increase at potentials between 0 V and −0.2 V. Similarly, the addition of 0.16 M NO₃⁻ also positively shifted the onset potential to −0.20 V and increased current densities. Interestingly, the positive shift in the onset potential in NO₃⁻ to −0.09 V was most significant in the FeMo—N—C catalyst, likely due to the cascade pathway with Mo sites chemically converting NO₃⁻ to NO₂⁻ and Fe sites electrochemically reducing the NO₂⁻ to NH₃ at low overpotentials.

The 6e⁻ transfer reduction of NO₂⁻ ions or a *NOOH intermediate to NH₃ is a critical step in the NO₃RR and the key in synergizing the dissociative and associative reaction mechanisms as shown in FIG. 2C (2e⁻+6e⁻ transfer pathway). Therefore, the activity and selectivity of nitrite reduction were first studied, as shown in FIGS. 3B and 3C. The potential-dependent studies show that both the mono- and bimetallic electrocatalysts reduced NO₂⁻ to NH₃ at yields and FEs greater than those for nitrate reduction even under a 16-fold concentration of NO₃⁻ (FIGS. 3D and 3E), demonstrating the downstream 6e⁻ steps from NO₂⁻ to NH₃ are not rate limiting in the NO₃RR. This is also partially supported by previous experimental and theoretical works. See Z. Wang et al., Catal. Sci. Technol. 11, 705 (2021); Z. Wu et al., Nat. Commun. 12(1), 1 (2021); and G. F. Chen et al., Nat. Energy 5(8), 605 (2020). Additionally, it was revealed that the Fe sites were particularly active for the NO₂RR, in that its FE was consistently around 100% at all potentials and the yields were as high as 50.6 μmol cm⁻² hr⁻¹ (865 μg_NH3 mg_cat⁻¹ h⁻¹) at −0.7 V. While less active than Fe sites, monometallic Mo sites still maintained a FE above 60% below −0.35 V and reached a maximum yield of 18.3 μmol cm⁻² hr⁻¹ (177 μg_NH3 mg_cat⁻¹ h⁻¹) at −0.7 V. This also explains the moderate NH₃ evolution, compared to the dominant NO₂⁻ production of the Mo—N—C in the 24 h time-course study as shown in FIG. 2E. Interestingly, the difference in nitrite reduction efficacy between Fe and Mo sites mirrors the behavior seen in biological systems, where Mo-based reductases preferentially convert NO₃⁻ to NO₂⁻ and Fe-based reductases preferentially convert NO₂⁻ to NH₃.

Based on this understanding of the NO₂RR activities, the more complex electrochemical nitrate (NO₃⁻) reduction was investigated. FIG. 3D shows the FE for NH₃ and NO₂⁻, with the balance being undetected gas-phase products, likely H₂ or NO species since N₂ is unlikely to form on single-atom active sites. The corresponding NH₃ yields as a function of potential are shown in FIG. 3E.

In agreement with FIGS. 2D and 2E, the NH₃ and NO₂⁻ FEs confirmed distinct reaction mechanisms for Fe and Mo sites. NO₃RR over exclusively Mo sites resulted in a large portion of cathodic current going to the kinetically facile NO₃⁻ to NO₂⁻ reduction via the dissociative-adsorption mechanism, resulting in a FE_NO2− up to 33% (˜60% molar percent) at −0.45 V. Meanwhile, the FE_NH3 was 66% at −0.45 V with a corresponding yield of 5.6 μmol cm⁻² hr⁻¹ (48 μg_NH3 mg_cat⁻¹ h⁻¹). Although decreasing the electrode potential from −0.2 V to −0.7 V increased the NH₃ yield (FIG. 3E), its FEs failed to surpass 70% and a 1.5-fold FE_NO2− was observed at −0.7 V. In contrast, the NO₃RR over Fe sites demonstrate a highly selective and efficient reduction of NO₃⁻ to NH₃ (rather than NO₂) via a direct 8e⁻ pathway, resulting in a FE_NH3 up to 84% and a yield of 10.1 μmol cm⁻² hr⁻¹ (86 μg_NH3 mg_cat⁻¹ h⁻¹) at −0.45 V. Although the NO₃RR became slightly favorable to NO₂⁻ at more cathodic conditions, the FE_NH3 was well maintained above 80% and the NH₃ yield over Fe active sites was consistently higher than that of Mo active sites. Given Fe—N—C's excellent NO₂RR activity (FIGS. 3B and 3C), the first 2e⁻ transfer process is likely the rate-limiting step for the NO₃RR.

Therefore, by integrating heterogenous Mo sites, which readily reduce NO₃⁻ to NO₂⁻, and Fe sites, which are limited by NO₃⁻ to NO₂⁻ reduction, into a single bimetallic catalyst (FeMo—N—C), a synergistic catalyst system can be employed to optimize this multistage NO₃RR chemistry. As shown in FIG. 3D, the bimetallic FeMo—N—C catalyst showed an increased FE_NH3 up to 94% at −0.45 V, a value among the highest selectivity reported for NO₃⁻ reduction to NH₃ (the remaining 6% FE came from NO₂⁻). Additionally, the NH₃ yields were significantly improved to 18.0 μmol cm⁻² hr⁻¹ (153 μg_NH3 mg_cat⁻¹ h⁻¹), which was 1.8-fold greater than Fe—N—C and 3.5-fold greater than Mo—N—C at −0.45 V. As compared to the NO₂RR, where Fe—N—C showed the largest NH₃ yield over FeMo—N—C and Mo—N—C(FIG. 3C), the bimetallic catalyst showed a dominating NH₃ yield in the NO₃RR at all potentials (FIG. 3E), strongly supporting the synergy of the Mo sites (NO₃⁻ to NO₂⁻) and Fe sites (NO₂⁻ to NH₃) in the NO₃RR via a cascade reaction for optimized ammonia synthesis.

NO₃RR Stability and Reliability

A recent review by Koper and co-workers highlighted that long-term durability studies are an immediate need for the future development of ammonia synthesis via the NO₃RR. See P. H. van Langevelde et al., Joule 5(2), 290 (2021). Usually, durability was claimed from single or several cycles of short-term electrolysis (0.5-4 h). For example, titanium electrodes were reported to achieve a FE of 82% for the first 2 hours, after which the FE dropped to 40%, showing change of catalytic performance over long-term electrolysis. See J. M. McEnaney et al., ACS Sustain. Chem. Eng. 8(7), 2672 (2020). Therefore, durability studies were performed on the FeMo—N—C catalyst at −0.45 V for a total operation time of 60 h, broken into five 12 h segments, each with a refreshed electrolyte, as shown in FIG. 4A. After the first 12 h, the yield decreased from 13.8 to 9.3 μmol cm⁻² hr⁻¹ and then remained steady. However, the FE_NH3 was well maintained above 90% over the 60 h, demonstrating the robustness of the electrocatalyst for selective NO₃⁻ reduction to NH₃.

Isotopic analysis is essential to confirm the electrochemical transformation of N species to NH₃, especially with N-doped catalysts. A 6 h electrolysis was conducted at −0.45 V with a ¹⁵NO₃⁻ feed at the same concentration. FIG. 4B shows only the characteristic doublet of ¹⁵NH₃ in the ¹H-NMR spectrum, wherein the peak intensity steadily increased throughout the 6 h experiment, showing good linearity between the NH₃ concentration and reaction time, as shown in FIG. 4C. Consequently, the NH₃ production from the labeled ¹⁵NO₃⁻ and standard ¹⁴NO₃⁻ showed good consistency in terms of both FE_NH3 92% vs. 94% and yields 16.5 vs. 18.0 μmol cm⁻² hr⁻¹ (FIG. 4D), confirming the reliability of NO₃⁻ conversion to NH₃.

Physical Structure of the Fe, Mo, and FeMo—N—C SACs

Inductively coupled plasma mass spectrometry (ICP-MS) results showed a metal loading of 0.73 wt % for Fe—N—C and 1.36 wt % for Mo—N—C, while the bi-metallic FeMo—N—C preserved 0.38 wt % Fe and 0.77 wt % Mo. Raman spectroscopy showed that the three electrocatalysts had a similar degree of graphitization. The resulting carbon matrix exhibited a hierarchical porous structure with leading mesoporosity around 20 nm and 100 nm, as shown by nitrogen physisorption results and scanning electron microscopy (SEM) images. The N₂ sorption isotherm also confirmed a large amount of micropores that contributed to a Barrett-Emmett-Teller (BET) surface area of ca. 600 m²/g for all catalysts. The hierarchical pore structure was favorable for promoting mass transport of the electrolyte to the active sites. X-ray diffraction (XRD) patterns for the mono- and bi-metallic catalysts (FIG. 5A) show only two peaks, characteristic of the (002) and (100) graphitic planes. The absence of any Fe or Mo crystalline peaks supports the formation of only atomically dispersed metal sites. The aberration-corrected high-angle annular dark-field (AC-HAADF) image for the FeMo—N—C catalyst shown in FIG. 5B again shows the well-defined porous structure and the absence of any metallic nanoparticles. The high magnification AC-HAADF image in FIG. 5C reveals distinct bright spots, indicating the abundance of atomically dispersed Fe and Mo atoms. Energy dispersive X-ray spectroscopy (EDS) maps (FIG. 5D) reveal a homogenous distribution of C, N, Mo, and Fe elements. Fe 2p X-ray photoelectron spectroscopy (XPS) spectra (FIG. 5E) reveals the presence of N-coordinated Fe—N_x moieties (708.5 eV). Mo 3d XPS spectra (FIG. 5F) similarly indicates the formation of N-coordinated Mo—N_x moieties (229 eV). N 1s XPS spectra (FIG. 5G) show the presence of several N moieties, assigned to be pyridinic (398 eV), pyrrolic (401 eV), quaternary (401.5 eV), and graphitic (403 eV). See A. Serov et al., Adv. Energy Mater. 4(10), 201301735 (2014); and K. Artyushkova et al., J. Phys. Chem. C 121(5), 2836 (2017). Furthermore, a strong peak for M-N_x (399 eV) moieties is also observed, indicating the formation of N-coordinated metal active sites.

Associative and Dissociative Reaction Mechanisms

DFT was further used to study the mechanism of NO₃⁻ conversion to NO₂⁻ and *NO₂ on the M-N₄ centers of the Fe—N—C and Mo—N—C catalysts, illustrated in FIGS. 6A and 6B. Theoretically, the initial 2e⁻ transfer of the NO₃RR, reducing NO₃⁻ to NO₂⁻/*NO₂, is complex and encompasses distinct dissociative- and associative-adsorption mechanisms. The subsequent 6e⁻ transfer process of NO₂⁻/*NO₂ to NH₃ has been shown through NO₂RR experiments (FIGS. 3A-3C) be kinetically fast and not involved in the rate determining steps. Other DFT-based computational work also support this finding, showing the facile reduction of *NO₂ to NH₃ on single-atom catalysts. See H. Niu et al., Adv. Funct. Mater. 31(11), 2008533 (2020).

Both associative and dissociative adsorption of NO₃⁻ molecules were considered as the initial step of the NO₃RR (FIGS. 2A and 2B). Specifically, it was found that the Fe—N₄ sites can stabilize adsorbed NO₃⁻ molecule regardless of molecular conformation, either through one or two O atoms. As shown in FIG. 7A, a Gibbs free energy of −0.39 eV was achieved when two O atoms were coordinated by the Fe site, making associative-adsorption a favorable pathway. The following protonation step from *NO₃ to *HNO₃ is slightly endergonic, +0.37 eV, followed by an exergonic, −1.62 eV, step for *NO₂ intermediate formation. Further protonation of *NO₂ to *HNO₂ (+0.19 eV) was more favorable than the desorption of NO₂* into the bulk (+1.07 eV), facilitating direct 8e⁻ pathway for the NO₃RR to NH₃ process.

In contrast, the Mo—N₄ sites showed a very strong oxygen affinity, such that when the NO₃⁻ molecule was coordinated on the Mo site through a single O-atom, the N—O bond was chemically cleaved (−4.2 eV), leading to an oxygen atom adsorbed (*O) on Mo site and a desorbed NO₂⁻ ion, as shown in FIG. 2B. Although the *NO₃ intermediate could be stabilized when two O atoms were coordinated to the Mo site (−2.72 eV), the subsequent dissociative-adsorption pathway was thermodynamically more favorable by an energy downhill of −1.48 eV as depicted *O+NO₂⁻ in FIG. 7B. Furthermore, during the optimization of NO₃⁻ on the Mo site, the N—O bond spontaneously dissociated when coordinated with one O atom, indicating a small or zero kinetic barrier for the dissociative-adsorption pathway. However, the regeneration of Mo site through the reduction of the *O to H₂O was not realistic, given the Gibbs free energy penalties of +1.39 eV and +2.24 eV in the sequential electron-transfer steps. Although the associative-adsorption pathway could be an option, it is expected to have low selectivity over the *O+NO₂⁻ step and faces a large energy barrier of +1.35 eV for the protonation of the *NO₂ surface intermediate. Therefore, the bare Mo sites inevitably would end being populated by *O species, denoted as the *O—Mo site.

Interestingly, further investigation showed that due to the strong O affinity of the Mo active site, high coordination number, and orientation of its 4d-orbitals, the oxygenated *O—Mo site can simultaneously bind an additional NO₃⁻ molecule (FIG. 7C) or NO₂⁻ molecule. For example, with the coordination of a secondary NO₃⁻ molecule, the strong *O—Mo interaction was split between the *O and *NO₃, substantially reducing the energy barrier to +0.12 eV for the *O protonation (*O+*NO₃ to *OH+*NO₃ in FIG. 7C) as compared to +1.39 eV on the monocoordinated *O—Mo site (*O+NO₂⁻ to *OH+NO₂⁻ in FIG. 7B). Similarly, the energy barrier for the subsequent protonation of the *OH intermediate was also reduced from +2.24 eV to +0.81 eV (FIGS. 7B and 7C) but remained large enough that it is unlikely to be made energetically downhill by applying an electrochemical potential. Then, the remaining *NO₃ was reoriented on the Mo site and subsequently dissociated, releasing a NO₂⁻ molecule and regenerating the *O—Mo site (gray pathway). It was highlighted that the more energetically favorable pathway (red pathway) occurs through the dissociation of the adsorbed *NO₃ on *O—Mo site, releasing a NO₂⁻ molecule into the bulk and thus leaving two O atoms on the Mo site, as shown by the *OO+NO₂⁻ step in FIG. 7C. Due to the binding energy being shared between the two O atoms, the subsequent reduction of one *O intermediate to H₂O was more energetically facile, resulting in an energy barrier of only +0.22 eV. After the release of H₂O via two protonation steps, the *O—Mo site was regenerated and ready to coordinate and dissociate another NO₃⁻, restarting the catalytic cycle for NO₂⁻ generation. This mechanism agrees well with the continuously increasing NO₂⁻ concentration observed throughout the entire 24 h electrolysis, as shown in FIG. 2E.

In the calculations, the dissociative-adsorption pathway was also found to be thermodynamically plausible over Fe sites, as shown by FIG. 7A (red pathway). Furthermore, the active site regeneration from *O—Fe is only slightly endergonic at 0 V and can become exergonic at reductive potentials. Fe sites are less oxophilic than Mo sites and are widely recognized as active nonplatinum group metal (PGM) oxygen reduction catalysts. See U. Tylus et al., J. Phys. Chem. C 118(17), 8999 (2014); T. Asset and P. Atanassov, Joule 4(1), 33 (2020); and J. Li et al., Nat. Catal. 4(1), 10 (2021). However, during the experimental time-course study, it was observed that the associative-adsorption, 8e⁻ transfer pathway generating NH₃ was the dominant path, with only a negligible and constant NO₂⁻ concentration being observed. The dominance of the associative-adsorption pathway could be due to the smaller kinetic barrier for the protonation of *NO₃ (*NO₃ to *HNO₃). Additionally, it must be emphasized that the free energy diagrams were calculated at the experimental pH=6.3. As the dissociation step is not pH dependent, while the associative step is pH dependent, any local increase in H⁺ concentration will make the associative step more favorable. Furthermore, the solvent effect by (H₃O)(H₂O)_n⁺ species could also additionally downshift the energy level of *HNO₃.

As described above, the experimental results gave strong evidence of the excellent NO₂⁻ reduction capability of Fe—N—C catalyst, good NO₂⁻ generation capability of Mo—N—C catalysts, and synergistic effect of bimetallic FeMo—N—C catalysts (FIG. 3D). It should be noted that when comparing the energy landscapes in FIG. 7A and FIG. 7C, the first 2-electron transfer step (NO₃⁻ to *NO₂) on Fe—N—C showed a maximum energy barrier of +0.37 eV, while the potential determining step on *O—Mo sites showed an energy barrier of +0.22 eV for the completed catalytic cycle of NO₂⁻ generation. Therefore, the Mo sites are more efficient in the first 2e⁻ transfer step in NO₃RR, producing NO₂⁻ molecules that can be reduced by Fe sites. This provides strong support for the bimetallic FeMo—N—C's synergistic effect on the increased ammonia yield rate (FIG. 3E).

Summary of Fe, Mo and FeMo—N—C Catalysts

In summary, taking inspiration from the cascade mechanism of biological nitrate/nitrite reductase enzymes, distinct NO₃RR mechanisms were identified for single-atom Mo—N₄ and Fe—N₄ active sites, where Mo sites preferentially followed a dissociative-adsorption pathway, spontaneously breaking the N—O bond, releasing NO₂⁻ into the bulk and creating the *O—Mo active site for the continuous catalytic cycle of NO₂⁻ generation. Then, the NO₂⁻ molecules could be re-adsorbed and further reduced to NH₃, following a 2e⁻+6e⁻ process. Over Fe sites, an associated-adsorption and direct 8e⁻ transfer pathway was favorable in which NO₃⁻ adsorbs and was directly reduced to NH₃. Additionally, independent NO₂RR electrolysis revealed the ability of the Fe sites to reduce NO₂⁻ to NH₃ at 100% FE_NH3 and high yields. DFT results further revealed that the Mo—N—C catalyst was more efficient in the first 2e⁻ transfer step than its Fe—N—C counterpart, which enables the synergistic effect of the bimetallic FeMo—N—C catalysts in terms of both NH₃ selectivity and yield rate. Specifically, the cascade NO₃RR synergized the favorable dissociation of NO₃⁻ to NO₂⁻ over Mo sites with the fast kinetics of Fe sites in reducing NO₂⁻ to NH₃. As a result, selective NO₃RR to NH₃ at a high FE_NH3 of 94% was achieved at −0.45 V, outperforming either of the Fe—N—C or Mo—N—C mono-metallic catalysts. Additionally, long-term durability tests over 60 hours demonstrated the robustness of the FeMo—N—C catalyst, maintaining a FE_NH3 over 90%. Therefore, heterogeneous single-atom sites utilizing distinct reaction pathways can be synergized on a single electrocatalyst to achieve highly selective and efficient NO₃⁻ reduction to NH₃.

Other Transition Metal M-N—C Catalysts

Other atomically dispersed transition metal and rare earth element M-N—C catalysts (M=Cr, Mn, Fe, Co, Ni, Cu, Mo, Ru, Rh, Pd, W, La, and Ce) can also be used for the electrocatalytic reactions. In order to characterize the utility of these catalysts for the electrocatalytic conversion of nitrates and nitrites, electrochemical descriptors were derived experimentally through the reaction onset potential (HER, NO₃RR and NO₂RR), the NO₃RR selectivity to both NO₂⁻ and NH₃, and the NO₂RR selectivity to NH₃. Isotopically doped ¹⁵NO₂⁻ in the NO₃RR elucidated the complex production/consumption mechanism of the NO₂⁻ intermediate. Density functional theory (DFT) was employed to evaluate the Gibbs free energy for the NO₃RR and NO₂RR following either the dissociative adsorption or associative adsorption pathway. By relating the experimentally determined and computationally determined activity descriptors, strong correlations were observed for the NO₂RR (R=0.72) the NO₃RR (R=0.73) selectivity to NH₃. This work bridges the gap between computation and experiment for the NO₃⁻ and NO₂⁻ reduction reactions over atomically dispersed M-N—C catalysts by providing a powerful set of activity descriptors that correlate strongly with the experimentally observed activity. These descriptors can be utilized to guide future atomically dispersed M-N_x catalyst development for highly active and efficient NO₃⁻ reduction to NH₃.

Synthesis of Transition Metal M-N—C Catalysts

Atomically dispersed M-N—C catalysts were synthesized through the well-established sacrificial support method (SSM). See A. Serov et al., Nano Energy 16, 293 (2015); M. M. Hossen et al., J. Power Sources 375, 214 (2018); and A. Serov et al., Electrochim. Acta 179, 154 (2015). As described above, SSM has been extensively applied for atomically dispersed Fe—N—C catalysts for the ORR and CO₂RR. See also T. Asset et al., ACS Catal. 9(9), 7668 (2019); and T. Asset and P. Atanassov, Joule 4(1), 33 (2020). Extending beyond Fe—N—C to a variety of 3d-, 4d-, 5d- and f-block metals, a set of atomically dispersed M-N—C catalysts (M=Cr, Mn, Fe, Co, Ni, Cu, Mo, Ru, Rh, Pd, La, Ce, and W), as shown in FIG. 8A, were synthesized as described below. The metallic centers ideally have a first coordination shell of 4-nitrogen atoms and are coordinated in either an in-plane or out-of-plane configuration as shown in FIG. 8B. With the synthesis conditions being adjusted to maintain the atomically dispersed nature for each metal element, the resulting catalysts showed consistent porosity, degree of graphitization and a well-maintained metal loading at 0.5-1.5 wt %.

Synthesis of Mn, Fe, Co, Ni, Cu, Mo and W—N—C catalysts. First a slurry of a carbon-nitrogen containing precursor, Nicarbazin (6.25 g), a silica sacrificial support comprising LM-150 fumed silica (1.25 g), OX-50 (1.25 g) and Stöber spheres (0.5 g), and the corresponding metal salt precursor (manganese(II) nitrate tetrahydrate, Mn=0.266 g; iron(III) nitrate nonahydrate, Fe=0.60 g; cobalt(II) nitrate hexahydrate, Co=0.272 g; nickel(II) nitrate hexahydrate, Ni=0.271 g; copper(II) nitrate hemi pentahydrate, Cu=0.345 g; ammonium molybdate tetrahydrate, Mo=0.262 g; or ammonium paratungstate, W=0.095 g), in 50 mL of MilliQ water was created. Next, the slurry was sonicated for 30 minutes before being dried overnight at 45° C., under constant stirring. The mixture was then further dried in an oven at 45° C. for 24 hours. The resulting powder was then ball milled at 45 Hz for 1 hour. The catalyst powder was then pyrolyzed at 975° C. (with a ramp rate of 15° C./min) under a reductive H₂/Ar (7%/93%) atmosphere for 45 min. The pyrolyzed powder was then ball milled a second time at 45 Hz for 1 hour. The silicate template was then etched in a hydrofluoric acid (15M) solution for 96 hours. The catalyst was then recovered by filtration and washed to neutral pH, followed by drying at 60° C. overnight. The catalyst underwent a second pyrolysis in a reductive NH₃/N₂ (10%/90%) atmosphere at 950° C. for 30 min (with a ramp rate of 20° C./min). The catalyst was then ball milled a final time at 45 Hz for 1 hour.

Synthesis of Cr—N—C catalyst. The synthesis is identical to the previous procedure, with only the pyrolysis temperatures being reduced to 650° C. in both the first and second pyrolysis, to maintain an atomic dispersion of the Cr metal. The metal salt precursor loading of chromium(III) acetylacetonate was Cr=0.519 g.

Synthesis of Ru—N—C catalyst. The synthesis is identical to the previous procedure, with the pyrolysis temperatures being reduced to 650° C. and an inert argon pyrolysis atmosphere, to maintain an atomic dispersion of the Ru metal. The metal salt precursor loading of ruthenium(III) nitrosylnitrate was Ru=0.235 g.

Synthesis of Rh and Pd—N—C catalysts. The synthesis is identical to the previous procedure, with the high pyrolysis temperatures of 975° C. and 950° C. for the first and second pyrolysis, respectively; however the pyrolysis is performed in an inert argon atmosphere, to maintain an atomic dispersion of the Rh and Pd metals. The metal salt precursor loading of rhodium(III) nitrate hydrate was Rh=0.215 g, and for palladium(II) nitrate dihydrate was Pd=0.198 g.

Synthesis of the La and Ce—N—C catalysts. The synthesis is identical to the previous procedure, with the pyrolysis temperatures being reduced to 650° C. under an inert argon pyrolysis atmosphere. Furthermore, the etching of the silica template was performed in an alkaline 4M NaOH environment at 80° C. for 96 hours. If etched using a hydrofluoric acid environment, the La and Ce readily form LaF₃ and CeF₃ nanoparticles. The metal salt precursor loading of lanthanum(III) nitrate hexahydrate was La=0.060 g, and for cerium(III) nitrate hexahydrate was Ce=0.061 g.

Physical Structure of Single-Atom Metal Centers

X-ray diffraction (XRD) shows only the characteristic (002) and (100) peaks for carbon, confirming the absence of bulk metallic phases. FIGS. 8C-8E show representative aberration corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) images for the Fe, Rh, and La—N—C catalysts, where atomically dispersed metal sites are clearly observed through the high contrast points. Energy dispersive X-ray spectroscopy (EDS) maps confirm the homogenous distribution of carbon, nitrogen, and the relevant metal throughout the catalyst, as shown in FIGS. 8C-8E. The physical characterization Fe—, Rh—, and La—N—C are shown as a 3d, 4d, and f-metal representative for the set of M-N—C catalysts herein.

The electron energy loss spectroscopy (EELS) in FIGS. 9A-9C shows the EELS point spectra of single atoms of Fe, Rh, and La, respectively. The EELS point spectra show the co-existence of the N K-edge and corresponding metal-edge (Fe-L_3,2, Rh-M₃, La-M_5,4), supporting the bond formation between the single-metal-atom and supporting nitrogen atoms (M-N_x). X-ray absorption spectroscopy (XAS) was used to further investigate the chemical state and coordination environment of the catalysts, including X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS). The absorption intensity in the rising edges of the Fe K-edge (7,112 eV), Rh K-edge (23,219.9 eV) and La L₃-edge (5,890.6 eV) as compared to their reference foil or oxide counterparts indicate nonzero oxidation states of ca. Fe²⁺, Rh³⁺, and La³⁺, in agreement with the Fe 2p, Rh 3d and La 3d XPS analysis. Similarly for other M-N—C catalysts, the chemical state is assessed by XPS. Furthermore, the EELS valence state analysis for Fe—N—C and La—N—C further supports the metal oxidate state of Fe (Fe²⁺/Fe³⁺) and La (La³⁺). However, the Rh K-edge EELS signal is too weak to perform quantitative analysis. EXAFS provides structural information of the metal centers inner coordination spheres, confirming the presence of shorter M-N bonds rather than longer metallic M-M bonds. Fourier transformed (FT) EXAFS spectra for the Fe K-edge, Rh K-edge and La L₃-edge are shown in FIGS. 9D-9F. In the Fe—N—C and Rh—N—C spectra, a single sharp peak was observed at a bond distance of ca. 1.5 Å (phase uncorrected) for both, characteristic of the Fe—N and Rh—N coordination in the first shell. While for La—N—C, an f-metal with a complex coordination environment (shown to sit out of plane during the modelling of the active site, FIG. 8B), a peak was observed at higher distances between 1-2.5 Å (phase uncorrected). A similar peak can also be observed in the FT-EXAFS of amorphous La₂O₃. Therefore, in the La—N—C sample, a strongly disordered local environment around the single atom site including the formation of small oxide-clusters can be assumed. For Ce—N—C, a more complex peak split feature was observed indicating the complex nature of f-metal M-N—C catalysts. To deconvolute minor intensities at larger bond distances, wavelet transforms of the EXAFS oscillations, providing high resolution information in both k-space and R-space are shown in FIGS. 9G-9I. For Fe—N—C and Rh—N—C, in addition to the high intensity Fe—N, Rh—N and La—N peak, a low intensity peak is observed at ca. 3.5 Å, 3.2 Å and 3.7 Å (at low k-space values), respectively, due to Fe—C, Rh—C and La—C interactions in the second coordination shell.

X-ray photoelectron spectroscopy (XPS) was used to further examine the metal-nitrogen coordination and other nitrogen moieties present. The representative N 1s spectra for the Fe—, Rh—, and La—N—C catalysts in FIGS. 9J-9L, respectively, reveals the presence of metal-nitrogen moieties (M-N_x). N 1s XPS spectra for the remaining M-N—C catalysts all demonstrate the formation of M-N_x moieties. The deconvoluted high resolution N 1s XPS spectra showed a variation in the N-content and N-moiety percentage (e.g., pyridinic, pyrrolic and graphitic nitrogen).

In summary, 13 atomically dispersed M-N—C catalysts were synthesized using the SSM. By combining AC-STEM, EELS, XAS and XPS, the atomically dispersed nature of each metal site was comprehensively visualized and confirmed to be in a M-N_x coordination. Critically, the well-deciphered coordination environment of the metal center allows elucidation of the intrinsic activity of the different M-N_x moieties towards the electrocatalytic transformation of reactive N-species.

Electrochemical Performance

The NO₃RR is somewhat universal in that even metal-free nitrogen-doped-carbon (N—C) showed limited but observable activity towards the generation of NH₃ and NO₂⁻ under an isotopically labeled ¹⁵NO₃⁻ feed at neutral pH. With the addition of nitrogen-coordinated metal sites Cr—N—C as an example, the FE_NH3 increased to 91.2%. The introduction of Fe—N_x sites boosted the FE_NH3 to 99% at −0.4 (V vs. RHE) and the corresponding Yield_NH3 increased from 1.3 to 10.0 μmol h⁻¹ cm⁻².

It should be noted that, for the NO₃RR activity, as compared to neutral media, the current density and NH₃ yield rate could be significantly enhanced under alkaline conditions, while maintaining a high FE_NH3. Given Fe—N—C as an example, when using a 1M KOH electrolyte (pH=14), the NH₃ partial current density increased from 2 mA cm⁻² (11.4 μmol h⁻¹ cm⁻²) to 35 mA cm⁻² (156.5 μmol h⁻¹ cm⁻²) at −0.4 V and further to 175 mA cm⁻² (750.4 μmol h⁻¹ cm⁻²) at −0.6 V, both with a FE_NH3 above 95%. Other work on alkaline NO₃RR also found similar trends. See Y. Wang et al., J. Am. Chem. Soc. 142(12), 5702 (2020); P. Li et al., Energy Environ. Sci. 14(6), 3522 (2021); and J. Li et al., J. Am. Chem. Soc. 142(15), 7036 (2020). However, a neutral pH (pH 6.3) was used herein in order to evaluate the intrinsic activity of the M-N_x sites toward the electrocatalytic NO₃RR and NO₂RR.

Regardless of the NO₃RR pathways, the surface intermediates *NO₂ and *NOOH are inevitable before the reaction pathway diverges to its several possible products. See Y. Wang and M. Shao, ACS Catal. 12(9), 5407 (2022); H. Niu et al., Adv. Funct. Mater. 2008533(3), 1 (2020); and S. Wang et al., Nano Energy, 107517 (2022). The NO₂⁻ molecule is also the first desorbable reaction intermediate, playing a key role in the NO₃RR via a cascade pathway. See E. Murphy et al., ACS Catal. 12(11), 6651 (2022); and H. Liu et al., ACS Catal. 2(3), 8431 (2021). For the present set of M-N—C catalysts, the NO₂RR displayed an earlier activation than the NO₃RR by linear sweep voltammetry (LSV), as shown in FIG. 10A, indicating that the 6e⁻ transfer from NO₂⁻/*NOOH to NH₃, in the NO₃RR was more facile than the initial 2e⁻ transfer from *NO₃⁻ to *NO₂⁻/*NOOH, which is the rate limiting step and thus the focus of the computational analysis. Representative LSV for the metal free N—C and a variety of 3d and 4d metals demonstrates the range of NO₃RR (FIG. 10B) and NO₂RR (FIG. 10C) activities, with Cu—N_x being the most active NO₃RR site and Fe—N_x being the most active NO₂RR site. As a unique nitrate-to-nitrite electrocatalyst, Mo—N—C was even more inert than the metal free N—C. See E. Murphy et al., ACS Catal. 12(11), 6651 (2022). A full analysis of the reaction onset potentials clearly visualizes the range of onset potentials for the diverse M-N_x sites.

To investigate the NO₃RR activity and selectivity towards NH₃ and NO₂⁻, a series of 2-hour potential holds were performed at potentials between −0.2 V and −0.8 V. As shown in FIG. 10D, a gap analysis plot (GAP) was constructed to visualize the nitrogen conversion landscape. Collectively, as the cathodic potential decreased, the yield rate of NH₃ and NO₂⁻ increased for all catalysts, but the selectivity varied. Specifically, both Fe—N—C and Cr—N—C showed the highest NH₃ selectivity (FE_NH3 92%-100%) at −0.4 V and −0.6 V. In contrast, Mo—N—C showed comparable, consistent, and somewhat potential-independent selectivity for NO₂⁻ (40%-50% FE_NO2−) and NH₃ (50%-55% FE_NH3), as expected given that Mo—N_x sites have been shown to be as an active nitrate-to-nitrite converter via dissociative adsorption (chemical process). See E. Murphy et al., ACS Catal. 12(11), 6651 (2022). Similarly, the Mn—, Pd—, W—, La—, and Ce—N—C showed moderate NO₃⁻ and NH₃ selectivity at high overpotentials (˜70% FE_NH3 vs. ˜30% FE_NO2−) and the FE for nitrogen conversion (FE_NH3+FE_NO2−) was maintained at 100%. Other early transition metals (Co—, Ni— and Cu—N—C) showed irregular potential dependence on selectivity, wherein Co—N—C showed a volcano trend for FE_NH3 but consistent FE_NO2− and the nitrogen conversion was far below 100% FE through the entire potential range. The NO₃RR electrolysis results (FE_NH3/NO2− and Yield_NH3/NO2−) were separated based on the applied potential, where it was clearly observed that Cr—N—C was leading the other catalysts between −0.2 V and −0.6 V, being exceptional at −0.2 V (9.33 μmol h⁻¹ cm⁻² and 94% FE_NH3). However, Ni—N—C and Cu—N—C showed a strong potential dependence on Yield_NH3, where Cu—N—C outperformed Cr—N—C at −0.8 V (61 vs. 59 μmol h⁻¹ cm⁻²). Particularly, Cu—N—C maintained 100% FE for NH₃+NO₂ (FE_NH3=72%) at −0.8 V, indicating that no current is wasted on the parasitic HER. Two distinct exceptions were Ru—N—C and Rh—N—C, albeit their early activation indicated by LSV, both of which showed inferior NH₃ and NO₂⁻ activity (FIG. 10D), displaying unique and significant gaps between the NH₃ and NO₂⁻ FE bars. This gap in FE is likely due to insoluble, undetected gas phase products, likely from the HER, as the isolated active sites present in atomically dispersed catalysts are unfavorable for the coupling of N—N bonds for N₂.

For the role of potential nitrite (as by-product or intermediate), previous reports usually considered the NO₃RR as a direct 8e⁻ transfer pathway with certain irreversible NO₂⁻ desorption or leaching. See Z. Wu et al., Nat. Commun. 12(1), 2870 (2021); P. Li et al., Energy Environ. Sci. 14(6), 3522 (2021); and G. F. Chen et al., Nat. Energy 5(8), 605 (2020). Here, doping of isotopic ¹⁵NO₂⁻ in the NO₃RR, schematically shown in FIG. 11A, revealed that trace amounts (e.g., 1 ppm) of ¹⁵NO₂⁻ could be easily reduced to ¹⁵NH₃ even under a concentrated ¹⁴NO₃⁻ environment (10,000 ppm), which applied to both M-N_x sites and metal-free N—C sites. FIG. 11B shows the NMR spectra for a time course electrolysis over Fe—N—C, wherein even at a concentration ratio of 1000:1 (¹⁴NO₃⁻:¹⁵NO₂⁻), the ¹⁵NO₂⁻ competed for an active site, yielding comparable ¹⁵NH₃. A similar trend was observed for most metals with the exception of Co— and Ni—N—C, where significantly more isotopic ¹⁵NH₃ was generated than standard ¹⁴NH₃, in support of FIG. 10D, suggesting their poor activity towards NO₃⁻ but high activity towards NO₂⁻. As the concentration ratio is reduced to 100:1 (¹⁴NO₃⁻:¹⁵NO₂⁻), the ¹⁵NO₂⁻ outcompeted ¹⁴NO₃⁻, yielding significantly more ¹⁵NH₃ than ¹⁴NH₃ over the 6-hour electrolysis (FIG. 11C). This indicates that for the nitrogenous products in the NO₃RR, the potential participation of bulk NO₂⁻ or locally channeled NO₂⁻ significantly complicates the mechanism of NH₃ production (e.g., 2e⁻+6e⁻ vs. 8e⁻). See G. F. Chen et al., Nat. Energy 5(8), 605 (2020); and I. Wheeldon et al., Nat. Chem. 8(4), 299 (2016). Therefore, unambiguously identifying whether the underlying reaction mechanism is a direct 8e⁻ pathway, a (rapid) 2e⁻+6e⁻ pathway with a bulk/channeled NO₂⁻ intermediate, or a combination thereof, is of great significance for the design and optimization of NO₃RR systems and even more complex nitrate-involving reactions. See J. Li et al., Nat. Rev. Chem. 6(5), 303 (2022); and Y. Wu et al., Nat. Sustain. 4(8), 725 (2021).

FIG. 11D shows the NO₂RR electrolysis for all M-N—C catalyst under a 0.01 M NO₂⁻ feed, a concentration mimicking the bulk NO₂⁻ concentrations in the NO₃RR electrolysis. Obviously, both the NH₃ selectivity and activity of the NO₂RR were significantly higher than that of the NO₃RR for all catalysts. Specifically, Fe—, Co—, and La—N—C showed 100% FE_NH3 over the whole potential range and Cr—, Ni—, Cu—, Pd—, and W—N—C showed increasing FE_NH3 as cathodic potentials decreased reaching 100% at −0.8 V. Again, Mo—N—C showed a unique and potential-independent FE_NH3 around 75%. Similar to the NO₃RR, the NO₂RR for Ru—N—C and Rh—N—C showed a distinct decreasing trend on the FE_NH3 and consistent Yield_NH3 over the entire potential range, likely being outcompeted by the HER as the cathodic potential decreased.

To examine the relationship between the NO₂RR and NO₃RR, FIGS. 11E and 11F correlate the activity of NO₂RR (FE_NH3/Yield_NH3) with the selectivity of NO₃RR (FE_NH3), revealing a linear relationship at −0.20 V and −0.40 V. This linear relationship suggests a major contribution from the bulk or locally channeled NO₂⁻ towards NH₃ (2e⁻+6e⁻ pathway). Meanwhile, the correlation between the FE_NO2− in the NO₃RR and the [FE_NH3/Yield_NH3] in NO₂RR has a poor linear fit, highlighting the complex reaction mechanism. In contrast, the poor correlations between the NO₃RR FE_NO2− and NO₂RR Yield_NH3 (FIG. 11G) confirmed that in the NO₃RR, the bulk NO₂⁻ species were in a complex production-consumption process rather than an irreversibly desorbed final by-product.

Computational Descriptors for the NO₃RR and NO₂RR

As discussed above, the first 2e⁻ transfer is the rate limiting step in the NO₃RR, therefore DFT was used to study the energetics of key intermediates in the conversion of NO₃⁻-to-NO₂⁻ on various catalytic active sites. This included the *NO₃ surface intermediate formed by oxidative associative adsorption of NO₃⁻, the *O surface intermediate formed by neutral dissociative adsorption of NO₃⁻, and the *NO₂ surface species formed in the reductive adsorption of NO₃⁻; these result in four thermodynamic reaction descriptors, as shown in FIG. 12A. See E. Murphy et al., ACS Catal. 12(11), 6651 (2022).

FIG. 12A shows that the Mo—N₄, La—N₄, Ce—N₄ and W—N₄ sites could strongly adsorb NO₃⁻ (Δ_rG˜4 eV) in a dissociative manner, forming an *O surface species and a free NO₂ molecule. For these oxyphilic metals, upon exposure to NO₃⁻ molecules, the single-atom sites are oxygenated with a fifth ligand and the new active site (O-M-N₄) can still coordinate a NO₃⁻/NO₂⁻ molecule for further reduction. See E. Murphy et al., ACS Catal. 12(11), 6651 (2022). The O-M-N₄ sites had a weaker interaction with NO₃⁻ but could stabilize the *NO₃/*NO₂ surface intermediates. For the Cr—N₄, Mn—N₄, Fe—N₄ and Ru—N₄ sites, associative and dissociative adsorption of NO₃⁻ is likely a competitive process that also depends on the cell potential. Namely, while the associative adsorption of NO₃⁻ is potential dependent, dissociative adsorption is not as no electrons are involved in this process. Co—N₄, Ni—N₄, Cu—N₄, Pd—N₄, and La—N₄ sites adsorb NO₃⁻ associatively and produce NH₃ via a *NO₂ intermediate. Additionally, the DFT descriptors suggest limited NO₃RR activity over Rh—N4 sites, being only favorable via the reductive adsorption of NO₃⁻, while in other computational based works, Rh—N₄ sites suffer some competition between *H and *NO₃, in agreement with the poor NO₃RR activity observed in the experimental observations. See S. Wang et al., Nano Energy 100, 107517 (2022); and Y. Wang and M. Shao, ACS Catal. 5407 (2022).

The quadrant plot in FIG. 12B graphically shows the correlations between the DFT-derived descriptors, mapping out the competitive NO₃⁻ adsorption and sources for NO₂⁻ evolution. Specifically, the far ends of Quadrant I, Quadrant II and Quadrant IV.b represent unfavored NO₃⁻ adsorption, exclusive NO₂⁻ evolution and unfavored NO₂⁻ evolution, respectively. These three blank sectors explain the above-mentioned universal NO₃RR activity for all M-N—C and metal free N—C catalysts, as well as the variant but non-dominant FE_NO2− in the NO₃RR. The diagonal of Quadrant III (x=y) indicates competitive associative and dissociative adsorption of NO₃⁻. The largest NO₂⁻ producing catalyst, Mo—N—C(O—Mo—N₄ and Mo—N₄ sites), is located at the diagonal, indicating comparable associative and dissociative adsorption. Several metal centers (e.g., Mn, Fe, Ru, La and Ce) fall below the diagonal in Quadrant III.b and upper Quadrant IV.b, suggesting a favored associative NO₃⁻ adsorption for high NH₃ selectivity in NO₃RR electrolysis (FIG. 11E). Quadrant IV.b (Co, Rh, O—Ce and O—La) marks the region where a solely direct 8e⁻ pathway to NH₃ is feasible, with formation of the *NO₂ intermediate being favored, while both descriptors for the generation/desorption of NO₂⁻ are unfavored. Quadrant IV.a (Ni, Cu and Pd, pyridinic-N and graphitic-N) shows a weak stabilization of the *NO₂ intermediate, while simultaneously showing activity for the desorption of *NO₂ for NO₂⁻ evolution. It should be noted that Mo—N—C is the only metal center favoring NO₂⁻ evolution through both the dissociative NO₃⁻ adsorption and *NO₂ desorption paths, explaining the unique selectivity of Mo—N—C in the NO₃RR (FIG. 11E).

To evaluate the practical relevance of these computational descriptors, a set of correlations were developed between the DFT-derived free energies (Δ_rG) in FIG. 12A and the electrocatalytic performance in FIG. 10D and FIG. 11D. Specifically, FIG. 12C shows the correlation between the adsorption energy of NO₃⁻ (Δ_rG [*+NO₃⁻→*NO₂]) and the NO₃RR NH₃ selectivity (FE_NH3), wherein a linear correlation (R=0.73) is observed at −0.2 V. This result highlights the importance of a stable *NO₂ intermediate for high NH₃ selectivity in both the 8e⁻ pathway or the 2e⁻+6e⁻ pathway (which requires the re-adsorption of NO₂⁻). It should be noted that Co, Ru, Rh were excluded as the outliers due to their dominant gaseous products in NO₃RR electrolysis (FIG. 10D), while Co is indicated as an outlier due to its unique NO₃RR performance showing strong activity towards the reduction of NO₂⁻, while showing poor activity towards NO₃⁻, as evidenced by the ¹⁵NO₂⁻ doping experiments, being a unique metal center showing a dominating concentration of ¹⁵NH₃ over ¹⁴NH₃ (at a 16,000 ppm ¹⁴NO₃⁻ and 10 ppm ¹⁵NO₂⁻). Additionally, the involvement of the oxo-form for the highly oxyphilic elements (O—Mo, O—W, O—La, O—Ce) shifts them back to the trend as compared to their bare M-N₄ counterparts. These results confirmed the distinct active sites/reaction mechanism of early transition 4d and 5d metal-based M-N—C catalysts in neutral/alkaline environments and is attributed to the oxophilicity and large coordination number of these metals that allow simultaneous coordination of multiple intermediates. See K. A. Moltved et al., J. Phys. Chem. C 123(30), 18432 (2019). Similarly, FIG. 12D shows a comparable correlation (R=0.72) between the adsorption energy of NO₂⁻ (Δ_rG [*+NO₂⁻→*NO₂]) and the NO₂RR ammonia selectivity (FE_NH3), indicating that the stabilization of the *NO₂ intermediate also plays a key role in the NO₂RR as well as the downstream 6e⁻ transfer in the NO₃RR.

However, the NO₃RR FE_NO2− showed minimum correlations with either Δ_rG [*NO₂→*+NO₂⁻ ] or Δ_rG [*+NO₃⁻→*O+NO₂⁻]. This agrees well with the above-mentioned poor correlation between the NO₃RR FE_NO2− and NO₂RR Yield_NH3 (FIG. 11G), wherein the bulk NO₂⁻ species in the NO₃RR were in an active but and complex production-consumption process, making it difficult to deconvolute the NO₂⁻ production and consumption rates from its net yield.

Summary of Transition Metal M-N—C Catalysts

In summary, a rich set of atomically dispersed 3d-, 4d-, 5d- and f-block M-N—C catalysts with a well-established M-N_x coordination was synthesized. The gap analysis plot revealed diverse NO₃RR performance, wherein Cr—N—C and Fe—N—C were the most selective catalysts, achieving near 100% FE_NH3, while Mn—, Cu— and Mo—N—C were highly selective for NO₂⁻. For the NO₂RR, several elements including Cr—, Fe—, Co—, Ni—, Cu— and La—N—C achieved a FE_NH3 of 100% at high overpotentials, with Fe and Co—N—C showing 100% FE_NH3 over the entire potential range. Isotopically doped ¹⁵NO₂⁻ in concentrated ¹⁴ NO₃⁻ demonstrated the ease at which minute concentrations of NO₂⁻ in the bulk electrolyte can preferentially reduce to NH₃, convoluting the possibility of a direct 8e⁻ pathway or a 2e⁻+6e⁻ cascade pathway with NO₂⁻ as transient intermediate for the NO₃RR. The correlation between experimental NO₃RR ammonia selectivity and experimental NO₂RR activity suggested a universal contribution of the NO₂⁻ intermediate in NO₃RR (2e⁻+6e⁻). The DFT-derived thermodynamic descriptors theoretically explain the electrocatalytic selectivity for each metal center. Furthermore, these computational descriptors showed strong correlations with the experimental performances, such that a simple computationally evaluated descriptor can be utilized to estimate the activity of M-N—C catalysts for the NO₃RR and NO₂RR. While these computational-experimental activity descriptors provide strong correlations for the NO₃RR and NO₂RR activity, these correlations are limited when the H⁺ adsorption is a competing factor leading to the HER over NO₃RR (Ru— and Rh—N—C) and when there is poor NO₃⁻ activation in contrast with significant NO₂⁻ activation (Co—N—C). This work enables the design of tandem NO₃RR systems and nitrate-containing systems composed of either multi-metallic M-N—C catalysts or extended catalytic surfaces supported by synergistic M-N—C supports.

The present invention has been described as electrocatalytic conversion of nitrates and nitrites to ammonia. It will be understood that the above description is merely illustrative of the applications of the principles of the present invention, the scope of which is to be determined by the claims viewed in light of the specification. Other variants and modifications of the invention will be apparent to those of skill in the art.

Claims

1. A method for the electrocatalytic conversion of nitrates and nitrites to ammonia, comprising:

providing an electrochemical H-cell, comprising a working electrode comprising an atomically dispersed transition metal-nitrogen-carbon (M-N—C) catalyst, a counter electrode, a hydrogen-permeable membrane between the working electrode and the counter electrode, and an electrolyte comprising an aqueous solution of nitrate and/or nitrite; and

applying a cathodic potential between the working electrode and the counter electrode sufficient to electrocatalytically reduce the nitrate and/or nitrite to ammonia at the working electrode.

2. The method of claim 1, wherein the transition metal comprises Fe, Mo, Cr, Mn, Co, Ni, Cu, Ru, Rh, Pd, W, La, or Ce, or a combination thereof.

3. The method of claim 1, wherein the M-N—C catalyst comprises a monometallic catalyst.

4. The method of claim 1, wherein the M-N—C catalyst comprises a bimetallic catalyst.

5. The method of claim 1, wherein the nitrate comprises potassium nitrate.

6. The method of claim 1, wherein the nitrite comprises potassium nitrite.

7. The method of claim 1, wherein the electrolyte comprises a concentration of greater than 0.01 M nitrate or nitrite.

8. The method of claim 1, wherein the electrolyte is alkaline.

9. The method of claim 1, wherein the electrolyte is neutral or acidic.

10. The method of claim 1, wherein the cathodic potential is more negative than −0.1 V.