Metal-Based Electrocatalysts for Selective Electrochemical Nitrogen Reduction to Ammonia

Ammonia (NH3) plays a significant role in the manufacture of fertilizers, nitrogen-containing chemical production, and hydrogen storage. The electrochemical nitrogen reduction reaction (e-NRR) is an attractive prospect for achieving clean and sustainable NH3 production, under mild conditions driven by renewable energy. The sluggish cleavage of N≡N bonds and poor selectivity of e-NRR are the primary challenges for e-NRR, over the competitive hydrogen evolution reaction (HER). The rational design of e-NRR electrocatalysts is of vital significance and should be based on a thorough understanding of the structure–activity relationship and mechanism. Among the various explored e-NRR catalysts, metal-based electrocatalysts have drawn increasing attention due to their remarkable performances. This review highlighted the recent progress and developments in metal-based electrocatalysts for e-NRR. Different kinds of metal-based electrocatalysts used in NH3 synthesis (including noble-metal-based catalysts, non-noble-metal-based catalysts, and metal compound catalysts) were introduced. The theoretical screening and the experimental practice of rational metal-based electrocatalyst design with different strategies were systematically summarized. Additionally, the structure–function relationship to improve the NH3 yield was evaluated. Finally, current challenges and perspectives of this burgeoning area were provided. The objective of this review is to provide a comprehensive understanding of metal-based e-NRR electrocatalysts with a focus on enhancing their efficiency in the future.


Introduction
With the ever-increasing global population, pressing worldwide environmental concerns and the challenges in energy supply and conservation, the search for sustainable and eco-friendly energy pathways is strongly demanded to secure our energy future [1,2].Elemental nitrogen (N 2 ) is indispensable to human society and the planet's ecosystem.Although nitrogen is abundant in the atmosphere (~78% by volume), it cannot be directly utilized by humans.However, nitrogen fixation can convert atmospheric N 2 to ammonia (NH 3 ), which is a more active nitrogen-containing alternative than N 2 .Traditionally, NH 3 serves as a raw material for the synthesis of fertilizers to sustain the rising global population.NH 3 is also extensively applied to produce explosives, plastics, resins, pharmaceuticals, and many other chemical compounds for industrial use.Currently, NH 3 has received considerable attention as a promising carbon-free energy carrier, due to its high hydrogen content (17.65%) and energy density (4.3 kW h kg −1 ), as well as its easy storage in liquids for transportation (9-10 bar) [3,4].Compared to C-containing fuels, N-containing fuels do not emit CO 2 upon final decomposition.
Benefiting from one of the most significant scientific inventions in the early 20th century, the Haber-Bosch process (N 2 + 3H 2 2NH 3 , ∆ f H • = −45.940kJ mol −1 , ∆ f G • = −16.407kJ mol −1 ) is a huge leap towards the mass production of NH 3 [5,6].To activate the strong N≡N bonds (46.1 kJ mol −1 ), the Haber-Bosch process requires high temperature (300-500 • C) and high pressure (150-300 atm) with heterogeneous iron-based catalysts.Here, the H 2 required depends on the carbon-intensive steam reforming of methane, with the input of energy derived from fossil fuels [7].Consequently, NH 3 production accounts for 1-2% of the global energy consumption each year and over the 2% of world's natural gas, giving rise to 3% of energy-related CO 2 emissions.In this regard, alternative approaches of NH 3 synthesis need to be developed with relatively low energy consumption, low pollutant production and mild operating conditions.In 2016, the US Department of Energy (DOE) launched the Renewable Energy to Fuels through Utilization of Energy-dense Liquids (REFUEL) program, including NH 3 as the Carbon-Neutral Liquid Fuels (CNLFs).
In nature, biological N 2 fixation occurs through multiple proton-and electron-transfer steps relying on the partnership of reductase and nitrogenase enzymes in certain bacteria.Notably, nitrogenase enzymes operate under mild conditions with a significant energy input by the hydrolysis of adenosine triphosphate (ATP) molecules (N 2 + 6H + + nMg-ATP + 6e − (enzyme) → 2NH 3 + nMg-ADP + nPi) [8].Electrochemical catalytic reactions including the hydrogen evolution reaction (HER), oxygen evolution reaction (OER), oxygen reduction reaction (ORR), and carbon dioxide reduction reaction (CO 2 RR) have rapidly developed and achieved excellent results over the past few years [9][10][11][12][13][14].In addition, nitrate-containing wastewater streams could serve as a nitrogen source via the electrochemical reduction of nitrate into ammonia [15][16][17][18].Inspired by the biological nitrogen-fixation process, the emerging electrochemical nitrogen-reduction reaction (e-NRR) is promising for achieving NH 3 production, directly from N 2 and water under mild conditions with the assistance of renewable energy.
Given that steam reforming for hydrogen production accounts for approximately 75% of the energy consumption in the Haber-Bosch process (Figure 1a), the steam-reforming unit with an electrocatalysis unit is a highly effective strategy [19].Additionally, only two major blocks were observed in this electrocatalytic strategy (Figure 1b).If renewable electricity (from solar energy, wind, etc.) is available, its use in the electrocatalysis strategy will become nearly 100% renewable.In the NH 3 economy, electrochemical NH 3 synthesis and NH 3 -powered fuel cells are two critical technologies [20].The water and nitrogen in air are used as the only reactants to produce NH 3 .The generated NH 3 can be distributed to users, including farms, NH 3 /H 2 refuelling stations and residents.The NH 3 -based infrastructure provides a promising way to solve the challenges related to the spatiotemporal fluctuations and the mismatch between supply and demand of electricity.Typically, the detailed cathodic and anodic reactions for e-NRR can be expressed as shown below (Equations ( 1)-( 5)), under different pH aqueous electrolytes [21].
Cathodic reaction (acidic condition): N 2 +6H + +6e − → 2NH 3 (2) Anodic reaction (basic condition): Tremendous efforts have been devoted towards the development of e-NRR since 2016, aimed at promoting NH3 yield and Faradaic efficiency (FE).However, e-NRR activity is hindered by the poor selectivity of e-NRR and poor activity of current e-NRR electrocatalytic designs.The poor selectivity of e-NRR arises from the competing HER.And the poor catalytic activity is mainly due to the weak affinity of N2 to the catalyst surface, which hinders the activation of N2 and the corresponding e-NRR efficiency.Accordingly, it is necessary to explore suitable electrocatalysts to overcome these limitations and improve catalytic activity towards e-NRR.Recently, extensive research has focused on designing a suitable electrocatalyst for efficient e-NRR, including noble-metal-based materials, non-noble-metal-based materials, and metal-free materials [2,3,[22][23][24].
There are some reviews published elsewhere highlighting the progress of e-NRR electrocatalysts [3,[25][26][27][28].More specifically, Wen et al. [29] summarized the recent progress in low-dimensional nanomaterials with various structures and mentioned the relationship between this structure and e-NRR activities from both theoretical and experimental perspectives.Liu et al. [22] systematically outlined the latest development in novel electrocatalysts, including noble-metal-based catalysts, single-metal-atom catalysts, nonnoble metal and their compounds, as well as metal-free catalysts, with various strategies to enhance the e-NRR activities through surface control, defect engineering, and hybridization.From the view of defect engineering, structural manipulation, crystallographic tailoring, and interface regulation, Shi et al. [2] comprehensively summarized the recent development of heterogeneous e-NRR catalysts, together with the catalytic mechanisms, current issues, and critical challenges.
In this review, we begin with the configurations and fundamentals of e-NRR under ambient conditions.Subsequently, the developed e-NRR metal-based catalysts based on noble-metal-based catalysts, non-noble-metal-based catalysts and metal compound catalysts were summarized, from experimental and theoretical perspectives, discussing the Tremendous efforts have been devoted towards the development of e-NRR since 2016, aimed at promoting NH 3 yield and Faradaic efficiency (FE).However, e-NRR activity is hindered by the poor selectivity of e-NRR and poor activity of current e-NRR electrocatalytic designs.The poor selectivity of e-NRR arises from the competing HER.And the poor catalytic activity is mainly due to the weak affinity of N 2 to the catalyst surface, which hinders the activation of N 2 and the corresponding e-NRR efficiency.Accordingly, it is necessary to explore suitable electrocatalysts to overcome these limitations and improve catalytic activity towards e-NRR.Recently, extensive research has focused on designing a suitable electrocatalyst for efficient e-NRR, including noble-metal-based materials, nonnoble-metal-based materials, and metal-free materials [2,3,[22][23][24].
There are some reviews published elsewhere highlighting the progress of e-NRR electrocatalysts [3,[25][26][27][28].More specifically, Wen et al. [29] summarized the recent progress in low-dimensional nanomaterials with various structures and mentioned the relationship between this structure and e-NRR activities from both theoretical and experimental perspectives.Liu et al. [22] systematically outlined the latest development in novel electrocatalysts, including noble-metal-based catalysts, single-metal-atom catalysts, non-noble metal and their compounds, as well as metal-free catalysts, with various strategies to enhance the e-NRR activities through surface control, defect engineering, and hybridization.From the view of defect engineering, structural manipulation, crystallographic tailoring, and interface regulation, Shi et al. [2] comprehensively summarized the recent development of heterogeneous e-NRR catalysts, together with the catalytic mechanisms, current issues, and critical challenges.
In this review, we begin with the configurations and fundamentals of e-NRR under ambient conditions.Subsequently, the developed e-NRR metal-based catalysts based on noble-metal-based catalysts, non-noble-metal-based catalysts and metal compound catalysts were summarized, from experimental and theoretical perspectives, discussing the structure-function relationship.Finally, current challenges and perspectives of this burgeoning area are provided.

Configurations of Electrochemical-Reactor for e-NRR
The electrochemical reactor is important for performing e-NRR.Generally, the configuration of such reactors can be divided into four categories, namely, back-to-back cell, protonexchange membrane (PEM)-type cell, single-chamber cell, and H-type cell (Figure 2) [30].
Nanomaterials 2023, 13, x FOR PEER REVIEW 4 of structure-function relationship.Finally, current challenges and perspectives of this bu geoning area are provided.

Configurations of Electrochemical-Reactor for e-NRR
The electrochemical reactor is important for performing e-NRR.Generally, the co figuration of such reactors can be divided into four categories, namely, back-to-back c proton-exchange membrane (PEM)-type cell, single-chamber cell, and H-type cell (Figu 2) [30].In the back-to-back cell (Figure 2a), two gas-diffusion electrodes (anode and cathod are usually separated by a dense membrane, and the anode and cathode are supplied w H2O and N2 gas, respectively.Among the various types of polymer membranes, p fluorosulfonic acid proton-exchange membranes such as Nafion membranes are wide used [31].Apart from the use of solid-state electrolytes, liquid electrolytes can also be u lized in back-to-back cells [7]. In the PEM-type cell (Figure 2b), the cathodic chamber is fed with nitrogen gas, a the synthesized NH3 gas is directly dissolved in the acidic electrolyte.Different from t back-to-back cell, the anodic chamber in the PEM-type cell is filled with liquid electroly where water is electrolyzed to supply protons for the cathodic reaction [32].In the bac In the back-to-back cell (Figure 2a), two gas-diffusion electrodes (anode and cathode) are usually separated by a dense membrane, and the anode and cathode are supplied with H 2 O and N 2 gas, respectively.Among the various types of polymer membranes, perfluorosulfonic acid proton-exchange membranes such as Nafion membranes are widely used [31].Apart from the use of solid-state electrolytes, liquid electrolytes can also be utilized in back-to-back cells [7].
In the PEM-type cell (Figure 2b), the cathodic chamber is fed with nitrogen gas, and the synthesized NH 3 gas is directly dissolved in the acidic electrolyte.Different from the back-to-back cell, the anodic chamber in the PEM-type cell is filled with liquid electrolyte, where water is electrolyzed to supply protons for the cathodic reaction [32].In the backto-back and PEM-type cells, no direct contact exists between the cathode and electrolyte.Consequently, they possess a favorable advantage for suppressing the HER by limiting the proton supply.However, these two types of cells are not ideal for e-NRR measurements.Several shortcomings including the complex preparation process and the effect from the use of an anion-exchange membrane lead to underestimated NH 3 production.
In the single-chamber cell (Figure 2c), the anode and cathode are placed in the same chamber without any separator.Nitrogen gas is continuously bubbled into the electrolyte and is subsequently reduced into NH 3 at the cathode under an applied potential.Simultaneously, OERs occur at the anode.One disadvantage of the single-chamber cell is that the NH 3 produced from the cathodic reaction may be oxidized at the anode, leading to inaccurate NH 3 determination.Given that the cathodic and anodic reactions are not separated, gases consisting of NH 3 , hydrogen, and oxygen are produced and discharged from the reactor together at the same time.The dominant HER could also suppress e-NRR at the same region of potential in an aqueous electrolyte-based cell.Consequently, the NH 3 production rate and FE are limited.
The H-type cell is the most widely studied reactor configuration.The anode and cathode chambers are separated by a membrane to prevent the mixing of products (Figure 2d-f).The working and reference electrodes are located on the same side, attributed to the accurate measurement of the applied potentials and significant decrease in resistance between the working and reference electrodes.In the cathode chamber, nitrogen gas is purged into the electrolyte and is reduced into NH 3 .The anodic reaction is mainly the oxidation of water molecules, also known as the OER.The anodic chamber was sealed without any gas purge in most reported work.Compared with the single-chamber cell, the products in each chamber can be separated in a double-chamber reactor, thereby preventing further oxidation and a mixture of gaseous products.Moreover, different electrolytes can be separated into two chambers so that the cathodic reactions are controlled independently with little influence from the anode.Accordingly, instrumental errors in measurements of e-NRR activity are minimized in the H-type cell compared with other reactors.However, the contribution of NH 3 dissolved in or leached from the ion-exchange membrane should be carefully handled.

Adsorption of Nitrogen onto the Catalyst Surface
The process of e-NRR includes N 2 adsorption onto the active sites, activation of N≡N bonds, and a final hydrogenation process.However, the high efficiency of e-NRR is hindered by the difficulty of N 2 adsorption and activation.Thus, a comprehensive understanding on the dominant reaction pathways is important to design highly efficient electrocatalysts.In the following subsection, an outline of the fundamental comprehension of e-NRR is discussed, with a focus on key steps and dominant reaction pathways.
The first step is the chemical adsorption of N 2 onto the catalyst surface.A large amount of N 2 adsorption sites can be provided by e-NRR catalysts with large surface areas.Accordingly, catalysts with relatively large specific surface areas are promising for the enhancement of e-NRR performance, such as porous-structure materials.The qualitative trends of catalytic activities on different metal surfaces have been summarized by Skúlason and co-workers through theoretical calculations [33].They assumed that the activation-energy scales with the free energy differed in each elementary step of e-NRR, for the range of flat and stepped transition metals.The key results of this study were illustrated as a volcano plot (Figure 3a), in which the theoretical limiting potentials (U) on different metal surfaces were plotted versus their adsorption energy of N atom (∆E N* ).This plot also showed the relatively limited region in white shading (N-binding), where binding N-adatoms were able to compete with the H-adatoms on the metal surface.With regard to minimizing the parallel HER process, Mo, Fe, Rh, and Ru on top of the volcano diagrams were bound to be the most active surfaces for e-NRR.Metals (Rh, Ru, Ir, Co, Ni, and Pt) on the right legs of the volcano plot were prone to adsorb H-adatoms instead of N-adatoms.In addition, more negative potentials were required for these metals to activate N 2 , resulting in HER that overwhelms e-NRR.Several flat metal surfaces of early transition metals such as Sc, Y, Ti, and Zr, on the left side of the volcano plots, tended to bind N-adatoms more strongly than H-adatoms.However, it remains unclear whether these early transition metals are effective e-NRR electrocatalysts, owing to easy oxidation.As a result, the authors encouraged experimental studies by using some of these metals.Beyond pure metal binding, some nanostructured catalysts with metal-nitrogen bonds have also been found capable of adsorbing N 2 , such as heteroatom-doped carbons (e.g., N/B-doped porous carbon) and metal/nonmetal nitrides (e.g., Mo 2 N, C 3 N 4 ) [34].Therefore, active sites on catalysts can be engineered to preferentially adsorb nitrogen species for e-NRR.
Moreover, the adsorbed amount of N 2 near the active sites possibly affects e-NRR activity.The solubility of N 2 gas in water-based electrolytes is about 0.66 mmol/L.Suryanto et al. [35] reported that the use of hydrophobic fluorinated aprotic electrolyte effectively enhanced N 2 solubility, which could significantly improve the FE of e-NRR.Their results also indicated that the availability of protons was effectively limited and thus suppressed the competing HER.Therefore, the increase in N 2 solubility near the active sites may be a promising strategy to achieve high e-NRR performance.
N-adatoms.In addition, more negative potentials were required for these metals to activate N2, resulting in HER that overwhelms e-NRR.Several flat metal surfaces of early transition metals such as Sc, Y, Ti, and Zr, on the left side of the volcano plots, tended to bind N-adatoms more strongly than H-adatoms.However, it remains unclear whether these early transition metals are effective e-NRR electrocatalysts, owing to easy oxidation.As a result, the authors encouraged experimental studies by using some of these metals.Beyond pure metal binding, some nanostructured catalysts with metal-nitrogen bonds have also been found capable of adsorbing N2, such as heteroatom-doped carbons (e.g., N/B-doped porous carbon) and metal/nonmetal nitrides (e.g., Mo2N, C3N4) [34].Therefore, active sites on catalysts can be engineered to preferentially adsorb nitrogen species for e-NRR.
Moreover, the adsorbed amount of N2 near the active sites possibly affects e-NRR activity.The solubility of N2 gas in water-based electrolytes is about 0.66 mmol/L.Suryanto et al. [35] reported that the use of hydrophobic fluorinated aprotic electrolyte effectively enhanced N2 solubility, which could significantly improve the FE of e-NRR.Their results also indicated that the availability of protons was effectively limited and thus suppressed the competing HER.Therefore, the increase in N2 solubility near the active sites may be a promising strategy to achieve high e-NRR performance.

Catalytic Activation of N 2
Due to the high dissociation energy of N≡N bonds (941 kJ mol −1 ) and first-bond cleavage energy (410 kJ mol −1 ), N 2 is kinetically inert under mild reaction conditions.Activation of N 2 occurs following the chemical adsorption of N 2 , and is usually considered as one of the rate-limiting steps in e-NRR.The change in the electron density and electron density distribution of the N 2 molecule can trigger its activation, during adsorption on the catalyst surface.In Figure 3b, an N atom has five valence electrons outside its nucleus, arranged in the configuration 2s 2 2p 3 [36].After bonding, the hybridization of the atomic orbitals is divided into four bonding orbitals and four anti-bonding orbitals.Furthermore, there is a large gap (10.82 eV) between the highest occupied molecular orbital (HOMO) with the lowest unoccupied molecular orbital (LUMO), as well as high ionization energy (15.58 eV) that blocks electron transfer [34].Consequently, N 2 activation is extremely difficult to achieve under ambient conditions.The first strategy to accelerate N 2 activation is the improvement of the electron donation and back-donation effect between catalysts and the adsorbed N 2 .Appropriate strong binding on metals, positively charged carbons or vacancies may promote electron transferring from the electrocatalyst matrix to the dinitrogen molecule, therefore accelerating the triple bond activation.Oxygen vacancies in oxides, nitrogen vacancies in nitrides, and surface defects on metals have been further investigated.In another way, it was reported that N 2 could be activated using the lithium (Li)-mediation assistant method [38].Li can react directly with N 2 and dissociate to form Li 3 N. Subsequently, NH 3 was transformed after the hydrolysis of Li 3 N.However, the procedures of this strategy are relatively complicated.

The Hydrogenation Pathway of N 2 to NH 3
Generally speaking, the current reaction mechanisms of the e-NRR on the heterogeneous catalysts are mainly divided into two kinds: the dissociative pathway (Figure 3c), and the associative pathway (Figure 3d,e) [37].In the dissociative pathway, the triple bonds of an adsorbed nitrogen are firstly cleaved before the hydrogenation reaction, and two independent N atoms subsequently go through a catalytic hydrogenation reaction [39].The traditional Haber-Bosch process for industrial NH 3 production mainly follows this mechanism, in which extraordinarily high energy input is required, whereas the process of e-NRR tends to undergo the associative pathway under ambient conditions [40].In this case, according to the different types of hydrogenation sequences, the associative pathway can be carried out in three possible pathways: the distal, alternating, and enzymatic pathways [30,36].It is assumed that only one N atom is fixed to the active site, namely, end-on adsorption [41].In the distal pathway, the N-atom distant from the adsorption site is preferentially hydrogenated continuously.After the release of the first NH 3 , the other N atom bound to the catalyst surface begins to form the second NH 3 , through the hydrogenation process.In contrast, the alternating pathway is to hydrogenate two N atoms in turn, with two NH 3 molecules generated simultaneously.Instead of an end-on adsorption mode, the enzymatic pathway exhibits side-on adsorption, in which two N-atoms are both adsorbed on the active sites.Additionally, the hydrogenation process involved in the enzymatic path is similar to the alternating pathway.The reduction of nitrogen to NH 3 undergoes these possible mechanisms, resulting in different intermediates, such as diazene (N 2 H 2 ), NH 3 , and hydrazine (N 2 H 4 ).
However, apart from the difficulty of nitrogen activation, the e-NRR in aqueous solution is limited by the competition of HER [42].Several processes at the electrode-electrolyte interface occur concurrently, involving the diffusion and adsorption of reactant species, transfer of electrons and protons, as well as desorption of species, where e-NRR and HER share some reaction species for basically electro-hydrogenation reactions [43,44].Moreover, the standard equilibrium potential of HER (E 0 = 0 V, vs. RHE) is similar to that of e-NRR (E 0 = 0.092 V, vs. RHE), but HER has much faster reaction kinetics [36].As a result, e-NRR typically suffers from a low reaction rate and low selectivity (FE) for NH 3 production.The competition between HER and e-NRR can be controlled by optimizing the electrolyte and potential, the local availability of protons and N 2 molecules near the catalyst, and revealing the relationships between structure and activity for rational catalyst design.

Advances in Metal Catalysts Design for e-NRR
In recent years, significant efforts have been devoted to design and fabricate an efficient electrocatalyst for NH 3 production [45].In view of the different compositions and characteristics, the various e-NRR metal catalysts can be classified into metal-based materials and metal compound materials.The recent progress of reported e-NRR metal catalysts is summarized and discussed, with a particular emphasis on their e-NRR performance and catalytic reaction.Noble metal catalysts have been proved as promising electrocatalysts in plenty of electrochemical reactions (such as HER, OER, and ORR), due to their marvelous conductivity, active polycrystalline surfaces, and appropriate adsorption of various reactants.Recently, noble metal catalysts such as Pt, Au, Ag, Ru, and Rh, have been explored for e-NRR to NH 3 synthesis.
Au electrocatalysts have been studied as the most promising noble catalysts for e-NRR [46], by controlling the morphology-dependent effect and metal-support synergetic effect.The former involves the creation of additional active sites by controlling morphology, crystal facets orientation, and crystallinity.As shown in Figure 4, Yan et al.
[47] synthesized tetra-hexahedral Au nanorods (Au THH NRs) as heterogeneous electrocatalysts and characterized e-NRR activity in an N 2 -saturated 0.10 M KOH solution.The measured angle implied that the bevels on the THH Au NR were high-index (730) planes, composed of the (210) and (310) sub-facets (Figure 4a,b).A large number of active sites can be provided to capture and activate N 2 , due to the exposed high-index (210) and (310) facets.The Au THH NRs endowed a highest NH 3 production rate of 1.648 µg h −1 cm −2 and maximum FE of 4.02% at −0.20 V vs. reversible hydrogen electrode (RHE).The density functional theory (DFT) calculations predicted that the e-NRR process preferably follows the alternating pathway, with the rate-determining step of N 2 dissociation for both Au (210) and Au (310) (Figure 4c).Wang et al. [48] reported a rapid approach for the fabrication of flower-like Au microstructures (Au flowers).In a comparative experiment, the e-NRR performance of Au flowers (NH 3 rate: 25.57µg h −1 mg cat. −1 , FE: 6.05%) outperformed the Au sphere counterpart.It indicated that Au flowers (particle size: ~900 nm) with highly dendritic structures could provide abundant electrocatalytically active sites, and therefore, promote e-NRR activity.The use of hollow gold nanocages (Au HNCs) as an effective electrocatalyst was also evaluated for e-NRR in 0.50 M LiClO 4 [49].The highest FE of Au HNCs (30.2%) was achieved at −0.40 V vs. RHE, while the maximum NH 3 production (3.90 µg h −1 cm −2 ) was obtained at −0.50 V vs. RHE.In contrast experiments, the e-NRR activity of Au HNCs was much better than other Au nanoparticles of various shapes (i.e., Au nanorods, Au nanospheres, and Au nanocubes) in the same conditions, resulting from the increased surface area and confinement effects.
Unlike the morphology-dependent effect, the metal-support synergetic effect contributes to improving the intrinsic activity of active centers.Therefore, the combination of the metal with support as composite may present a new route for reducing the usage of noble metal.For example, Zhao et al. [50] reported a nano-gold catalyst supported on a boron organic polymer (Au/M-BOP) as electrocatalyst for electrochemical reduction from N 2 to NH 3 .Yan and co-workers [51] studied the Au/TiO 2 catalyst as a heterogeneous catalyst for e-NRR, synthesized using Au sub-nanoclusters (~0.50 nm) embedded in commercial TiO 2 support.Unexpectedly, the obtained Au/TiO 2 endowed the e-NRR with a high yield (NH 3 : 21.4 µg h −1 mg −1 , FE: 8.11%) at −0.20 V vs. RHE.Moreover, it should be noted that the apparent catalytic activity decreased after tuning particle size of Au species dispersed on TiO 2 ranging from nanometer down to sub-nanometer sizes.This work also indicated that the isolated precious metal onto oxide supports provided a well-defined system.The proposed pathway for the NH 3 synthesis using Au/TiO 2 catalyst was shown in Figure 4d, displaying a distal hydriding pathway.
boron organic polymer (Au/M-BOP) as electrocatalyst for electrochemical reduction from N2 to NH3.Yan and co-workers [51] studied the Au/TiO2 catalyst as a heterogeneous catalyst for e-NRR, synthesized using Au sub-nanoclusters (~0.50 nm) embedded in commercial TiO2 support.Unexpectedly, the obtained Au/TiO2 endowed the e-NRR with a high yield (NH3: 21.4 µg h −1 mg −1 , FE: 8.11%) at −0.20 V vs. RHE.Moreover, it should be noted that the apparent catalytic activity decreased after tuning particle size of Au species dispersed on TiO2 ranging from nanometer down to sub-nanometer sizes.This work also indicated that the isolated precious metal onto oxide supports provided a well-defined system.The proposed pathway for the NH3 synthesis using Au/TiO2 catalyst was shown in Figure 4d, displaying a distal hydriding pathway.This group continued to explore the effectiveness of Au and proposed CeOx-induced amorphous Au nanoparticles on reduced graphite oxide (a-Au/CeOx-rGO) as e-NRR electrocatalysts [52].As shown in Figure 4e,f, it was found that the CeOx played an important role in transferring the crystalline Au NPs into the amorphous ones.Compared with its crystalline counterpart, a-Au/CeOx-rGO achieved a higher 10.10% FE with an NH3 yield of 8.3 µg h −1 mgcat.−1 at −0.20 V vs. RHE, because of their higher concentration of active sites and more structural distortion.In another study, Wang and co-workers [53] reported the Au/N-doped nano-porous graphitic carbon membrane (NCM) electrocatalyst.The synergistic effect between NCM and Au promoted the N2 adsorption and thereby improved the conversion of N2 to NH3.
The precious metal Ru is also a hot research subject in the electroreduction of N2.Similarly, the crystal structure and the particle size of Ru also have a great influence on e-NRR activity.Wang et al. [54] studied Ru nanoparticles as e-NRR electrocatalysts in 0.01 This group continued to explore the effectiveness of Au and proposed CeO x -induced amorphous Au nanoparticles on reduced graphite oxide (a-Au/CeO x -rGO) as e-NRR electrocatalysts [52].As shown in Figure 4e,f, it was found that the CeO x played an important role in transferring the crystalline Au NPs into the amorphous ones.Compared with its crystalline counterpart, a-Au/CeO x -rGO achieved a higher 10.10% FE with an NH 3 yield of 8.3 µg h −1 mg cat.
−1 at −0.20 V vs. RHE, because of their higher concentration of active sites and more structural distortion.In another study, Wang and co-workers [53] reported the Au/N-doped nano-porous graphitic carbon membrane (NCM) electrocatalyst.The synergistic effect between NCM and Au promoted the N 2 adsorption and thereby improved the conversion of N 2 to NH 3 .
The precious metal Ru is also a hot research subject in the electroreduction of N 2 .Similarly, the crystal structure and the particle size of Ru also have a great influence on e-NRR activity.Wang et al. [54] studied Ru nanoparticles as e-NRR electrocatalysts in 0.01 M HCl aqueous solution.The maximum yield rate of 5.50 mg h −1 m −2 was achieved at −0.10 V vs. RHE, whereas the highest FE was 5.40% at 0.10 V vs. RHE.The DFT calculations indicated that the efficient e-NRR activity at the low overpotential was attributed to instantaneous N 2 adsorption on Ru (001) surfaces and the spontaneous hydrogenation process by a dissociative mechanism.In another study, isolating Ru single atoms in N-doped porous carbon as electrocatalyst greatly promoted N 2 -to-NH 3 conversion (Figure 5a,b), affording an NH 3 formation rate of 3.665 mg h −1 mg Ru −1 at −0.21 V vs. RHE [55].It was found that the addition of ZrO 2 can effectively suppress the competitive Nanomaterials 2023, 13, 2580 10 of 24 HER, reaching a high FE of 21% at a low overpotential.From calculation results, the e-NRR mainly occurred at Ru sites with O vacancies, which was permitted through the stabilization of *NNH (low overpotential), destabilization of *H (high e-NRR/HER selectivity), and enchantment of N 2 adsorption (to initiate the e-NRR process).
In addition to the above catalysts, Rh, Ag, and Pd have also been studied for e-NRR, due to their strong adsorption energy and low overpotentials [33].Surfactant-free atomically ultrathin Rh nanosheets (Rh NSs) were synthesized and used as an effective e-NRR in a 0.10 M KOH solution [54].Benefiting from their unique ultrathin two-dimensional (2D) structure with abundant surface and modified electronic structure, Rh NSs exhibited an excellent e-NRR performance with a high NH 3 yield rate (23.88 µg h −1 mg cat.
−1 ) and selectivity (no N 2 H 4 generation) at −0.20 V vs. RHE.But, the FE at the same potential was only 0.217%, due to the dominant HER process.Yin et al. [56] reported Ag triangular nanoplates (Ag TPs) as e-NRR catalysts with efficient activity of NH 3 generation.The e-NRR activity of Ag TPs was much more efficient than circular Ag nanoparticles, owing to the more anchored atoms at sharp edges and corners on Ag TPs.Single Ag sites with the Ag-N 4 coordination (SA-Ag/NC) were synthesized massively by targeting the admolecules (Figure 5c), confirming that abundant Ag SAs exist in the carbon matrix by TEM and HAADF-STEM (Figure 5d-f).SA-Ag/NC achieved a record-high NH 3 yield rate (270.9 µg h −1 mg cat. −1 or 69.4 mg h −1 mg Ag −1 ) and a desirable FE (21.9%) in HCl aqueous solution [57].Through 20 consecutive cycle tests, the stability of SA-Ag/NC was maintained.Furthermore, to eliminate or quantify the sources of contamination, a rigorous reduction experiment was recommended by the isotopic labeling experiment using 15 N 2 , reliably confirming the ammonia production only from the N 2 source [4,58].As expected, through the isotopic labeling experiment, the NH 3 generation was verified from the gaseous N 2 over SA-Ag/NC during the e-NRR process.Based on first principles calculations (Figure 5g-j), the emergence of vertical end-on *N 2 and oblique end-on *NNH admolecules on single metal sites in succession were energetically favorable for e-NRR.
V vs. RHE, whereas the highest FE was 5.40% at 0.10 V vs. RHE.The DFT calculations indicated that the efficient e-NRR activity at the low overpotential was attributed to instantaneous N2 adsorption on Ru (001) surfaces and the spontaneous hydrogenation process by a dissociative mechanism.In another study, isolating Ru single atoms in N-doped porous carbon as electrocatalyst greatly promoted N2-to-NH3 conversion (Figure 5a,b), affording an NH3 formation rate of 3.665 mg h −1 mg Ru −1 at −0.21 V vs. RHE [55].It was found that the addition of ZrO2 can effectively suppress the competitive HER, reaching a high FE of 21% at a low overpotential.From calculation results, the e-NRR mainly occurred at Ru sites with O vacancies, which was permitted through the stabilization of *NNH (low overpotential), destabilization of *H (high e-NRR/HER selectivity), and enchantment of N2 adsorption (to initiate the e-NRR process).
In addition to the above catalysts, Rh, Ag, and Pd have also been studied for e-NRR, due to their strong adsorption energy and low overpotentials [33].Surfactant-free atomically ultrathin Rh nanosheets (Rh NSs) were synthesized and used as an effective e-NRR in a 0.10 M KOH solution [54].Benefiting from their unique ultrathin two-dimensional (2D) structure with abundant surface and modified electronic structure, Rh NSs exhibited an excellent e-NRR performance with a high NH3 yield rate (23.88 µg h −1 mgcat.−1 ) and selectivity (no N2H4 generation) at −0.20 V vs. RHE.But, the FE at the same potential was only 0.217%, due to the dominant HER process.Yin et al. [56] reported Ag triangular nanoplates (Ag TPs) as e-NRR catalysts with efficient activity of NH3 generation.The e-NRR activity of Ag TPs was much more efficient than circular Ag nanoparticles, owing to the more anchored atoms at sharp edges and corners on Ag TPs.Single Ag sites with the Ag-N4 coordination (SA-Ag/NC) were synthesized massively by targeting the admolecules (Figure 5c), confirming that abundant Ag SAs exist in the carbon matrix by TEM and HAADF-STEM (Figure 5d-f).SA-Ag/NC achieved a record-high NH3 yield rate (270.9 µg h −1 mgcat.−1 or 69.4 mg h −1 mg Ag −1 ) and a desirable FE (21.9%) in HCl aqueous solution [57].Through 20 consecutive cycle tests, the stability of SA-Ag/NC was maintained.Furthermore, to eliminate or quantify the sources of contamination, a rigorous reduction experiment was recommended by the isotopic labeling experiment using 15 N2, reliably confirming the ammonia production only from the N2 source [4,58].As expected, through the isotopic labeling experiment, the NH3 generation was verified from the gaseous N2 over SA-Ag/NC during the e-NRR process.Based on first principles calculations (Figure 5g-j), the emergence of vertical end-on *N2 and oblique end-on *NNH admolecules on single metal sites in succession were energetically favorable for e-NRR.From the point of view of morphology control, crystallographic tailoring, structural manipulation, and defect engineering, the noble metal catalysts were summarized, aiming to enhance the e-NRR activity.Combined with discussing the relationship between the struc-ture and e-NRR activity based on experimental and theoretical results, they are expected to provide a reference for the rational design of e-NRR electrocatalysts in a targeted manner.

Non-Noble-Metal-Based Catalysts
There is a growing desire to explore resource-rich metal in the earth, in order to reduce cost and improve the applicability of e-NRR technology.Due to the important role of the FeMo cofactor in biological nitrogen fixation and Fe-based catalysts in Haber-Bosch technology, Mo-and Fe-based electrocatalysts have been explored toward e-NRR.
To explore the effect of crystal phase orientations for Mo catalysts, Yang et al. [59] prepared four kinds of Mo-based nanofilms with different facet orientations and surface morphology.Mo (110) plane can adsorb N adatoms more strongly than H adatoms, while Mo (211) dominantly follows competitive HER.As a result, Mo (110) was more efficient with FE of 0.72% at a low overpotential of −0.49V vs. RHE.This study showed that morphology control was also a feasible way to improve the catalytic activity of pure nonnoble metal toward e-NRR.A study of DFT simulation predicted that single Mo atom fixed on a defective boron nitride (BN) monolayer could be potentially used as a N 2 fixation electrocatalyst, where dispersed Mo atoms bonded to N atoms contributed to activate N 2 molecules, selectively stabilize N 2 H*, or destabilize NH 2 * during e-NRR [60].Based on this study, Han et al. [61] reported single Mo atoms anchored onto N-doped porous carbon (SA-Mo/NPC) as e-NRR electrocatalysts.Benefiting from the optimized abundance of active sites and 3D hierarchically porous carbon frameworks, SA-Mo/NPC achieved a high NH 3 yield rate (34.0 ± 3.6 µg h −1 mg cat.−1 ) and a high FE (14.6 ± 1.6%) in 0.10 M KOH electrolyte at −0.30 V vs. RHE.Similarly, efficient e-NRR activity and durability were also obtained by SA-Mo/NPC in 0.10 M HCl acid electrolyte.The authors also concluded that Mo-N sites of atomically dispersed Mo atoms bonding to N were the catalytic active sites.As shown in Figure 6a,b, the stabilized single Mo atoms anchored on holey N-doped graphene (Mo/HNG), with a continuous porous skeleton and plenty of edges containing N-coordination sites, were synthesized through a potassium saltassisted activation process [62].As plotted in Figure 6c, at −0.05 V vs. RHE, Mo/HNG exhibited an exceptional FE of 50.2% for NH 3 production (partial reduction current density: 17.0 µA cm −2 ) and a NH 3 production yield rate of 3.6 µg h −1 mg cat −1 .During continuous electrolysis (20,000 s), Mo/HNG still maintained over 50% FE (−0.05 V vs. RHE), with only 0.0125% of Mo on the electrode dissolved (ICP-MS test), exhibiting good stability.The isotopic labeling experiments were measured, respectively, using abundant natural 14 N 2 and 15 N 2 as feed gas [63].As shown in Figure 6d, the NH 4 + splitting patterns in 1H nuclear magnetic resonance were consistent with the corresponding resultant electrolyte using isotopic 14 N 2 or 15 N 2 source, presenting a specific double peak for 15 NH 4 + and three peaks for 14 NH 4 + [64].Through theoretical calculations, it is unveiled that the edge coordinated Mo atoms and the existence of vacancies on holey graphene jointly contribute to the intriguing e-NRR activity.
As one of the most earth-abundant metals, Fe-based catalysts have also shown great potential as excellent e-NRR electrocatalysts.For instance, Wang et al. [65] theoretically proposed the catalytic mechanisms of single Fe atom embedded N-doped graphene for e-NRR.The results indicated that the magnetic moment of the Fe atom increased with the increase in coordination of the neighboring N atom, resulting in a lower overpotential of N 2 reduction.In experiment, Wang et al. [66] recently used a single-atom catalyst (iron on N-doped carbon, Fe SA -N-C) as an e-NRR electrocatalyst, enabling a dramatically enhanced FE.Here, the DFT calculations suggested that the Fe SA -N-C structure could effectively attract the access of N 2 molecules with a small energy barrier, which benefits preferential N 2 adsorption instead of H adsorption.The isotope-labeling experiments and control experiments indicated that the generated NH 3 entirely comes from the e-NRR process catalyzed by Fe SA -N-C.Careful characterization and consecutive recycling electrolysis were preformed, suggesting its excellent stability.In another study, an Fe-N/C-carbon nanotube catalyst (Fe-N/C-CNTs) was designed, through carbonizing a metal-organic framework and carbon-nanotube-based composite [67], with built-in Fe−N 3 sites.The corresponding synthesis process was shown in Figure 6e.In 0.10 M KOH electrolyte, the optimal NH 3 formation rate was 34.83 µg h −1 mg cat.
−1 with an FE of 9.28% at −0.20 V vs. RHE.The favorable e-NRR activity was attributed to Fe−N 3 species as active sites.The theoretical results further revealed that the e-NRR reaction proceeded preferentially via the distal pathway.
FE. Here, the DFT calculations suggested that the FeSA-N-C structure could effectively attract the access of N2 molecules with a small energy barrier, which benefits preferential N2 adsorption instead of H adsorption.The isotope-labeling experiments and control experiments indicated that the generated NH3 entirely comes from the e-NRR process catalyzed by FeSA-N-C.Careful characterization and consecutive recycling electrolysis were preformed, suggesting its excellent stability.In another study, an Fe-N/C-carbon nanotube catalyst (Fe-N/C-CNTs) was designed, through carbonizing a metal-organic framework and carbon-nanotube-based composite [67], with built-in Fe−N3 sites.The corresponding synthesis process was shown in Figure 6e.In 0.10 M KOH electrolyte, the optimal NH3 formation rate was 34.83 µg h −1 mgcat.−1 with an FE of 9.28% at −0.20 V vs. RHE.The favorable e-NRR activity was attributed to Fe−N3 species as active sites.The theoretical results further revealed that the e-NRR reaction proceeded preferentially via the distal pathway.Recently, Leung's group was dedicated to non-noble bimetals on nitrogen-doped carbons, selecting from either side of the theoretical volcano plot for the e-NRR.Dispersed Mo-Co bimetallic nanoparticles immobilized on N-doped porous carbon (Mo-Co/NC) were developed and exhibited the enhanced activity and selectivity of e-NRR electrocatalysis with an ammonia yield, in comparison to single-metallic Co/NC [68].Additionally, to overcome the sluggish kinetics of the proton-coupled electron transfer on the single-atom site, Leung et al. [69] synthesized atomically dispersed Co-Mo pairs anchored on N-doped carbon frameworks (Co-Mo-SA/NC) through calcinating Co-Mo-doped zinc-based zeolite imidazole framework precursors.Revealed by Bader charge analysis and charge density difference analysis, 0.35 e − and 0.30 e − were, respectively, transferred to N1 on the Mo-end and N 2 on the Co-end from Co-Mo active sites; simultaneously, Co and Mo atoms with two occupied d orbitals possess the capability to donate their electrons to the empty p* orbital of N, ultimately forming triple bonds.Nevertheless, the N 2 on a single active site follows the electron acceptance-donation concept, resulting in a significant increase in energy required for the initial activation of N 2 .Consequently, the Co-Mo-SA/NC catalyst achieves outstanding e-NRR performance in 0.1 M Na 2 SO 4 solution with 37.73 µg h −1 mg cat.
−1 and a desirable FE of 23.18% at −0.1 V vs. RHE, which are twofold higher than those of the isolated single-atom Co (Co-SA/NC) or Mo (Mo-SA/NC) catalyst.
Apart from the aforementioned Fe-and Mo-based catalyst, other transition metals have also emerged as electrocatalysts for N 2 fixation [70][71][72][73].For example, Co single-atomembedded N-doped porous carbon (CSA/NPC) was synthesized as an electrocatalyst for e-NRR [74].At a low overpotential of −0.20 V, CSA/NPC presented a high NH 3 yield rate of 0.86 mmol cm −2 h −1 and a FE of 10.50%, attributed to the positive effects of Co single atoms, N-doping, and porous structure.In another work, Wang et al. [75] reported atomically dispersed Ni sites on a carbon framework with nitrogen-vacancy (Ni x -N-C) as an effective non-noble-metal electrocatalyst for the e-NRR, synthesized from a Ni-doped ZIF-8 precursor.Compared with Ni clusters supported on the N-doped carbon framework, significant e-NRR activity was observed on Ni x -N-C with an NH 3 production rate of 115 µg cm −2 h −1 at −0.80 V (vs.RHE) and FE of 20% at −0.60 V (vs.RHE) in LiClO 4 solution.From calculation results, Ni-N x sites were responsible for the experimentally observed activity and the potential determining step was the hydrogenation during the e-NRR.Cu as a common and low-cost metal has also been studied for e-NRR by Zang and co-workers [76].In this experiment, a Cu single atom on a porous N-doped carbon network (NC-Cu SA) was studied for catalytic performance toward e-NRR in both alkaline and acidic solutions.The NC-Cu SA exhibited a high NH 3 yield rate and FE, specifically ~53.30µg h −1 mg cat.
−1 and 13.80% under 0.10 M KOH, ~49.30 µg h −1 mg cat.−1 and 11.70% under 0.10 M HCl.Similarly, the experimental analysis and DFT calculations indicated that the local Cu−N 2 coordination was identified as the efficient sites and responsible for the outstanding e-NRR performance.
Up to now, non-noble-metal-based materials have been reported as efficient e-NRR electrocatalysts, most investigations mainly focused on their composites including nitrides, carbides and oxides.More discussion about metal compounds will be summarized in the following section.

Metal Sulfide and Metal Nitride Catalysts
In recent years, a range of metal sulfides and metal nitrides have been taken into consideration for e-NRR.Although the intrinsic catalytic activity of MoS 2 for water reduction suppressed the e-NRR process, MoS 2 is still considered and utilized as an electrocatalyst to catalyze the N 2 reduction reaction.Sun's group theoretically predicted and experimentally confirmed that MoS 2 as an active e-NRR electrocatalyst achieved a high NH 3 yield rate (8.08 × 10 −11 mol s −1 cm −1 ) and FE (1.17%) at −0.50 V vs. RHE [77].Impressively, this study further indicated that MoS 2 was still active for e-NRR, where a strong HER occurs.Soon thereafter, they found that defect-rich MoS 2 (DR MoS 2 ) nanoflower could greatly boost electrocatalytic N 2 reduction to NH 3 39 .Compared with the defect-free counterpart, a high FE (8.34%) and high NH 3 yield (29.28 µg h −1 mg cat. −1 ) were obtained at −0.40 V vs. RHE.DFT calculations revealed that the potential-determining step was *NH 2 → *NH 3 , and the barrier of DR MoS 2 (0.60 eV) was lower than the barrier of the defect-free catalyst (0.68 eV).In another theoretical study, Fe-doped MoS 2 through an associative distal pathway revealed that the presence of a vicinal Fe atom enabled highly selective chemisorption of N 2 , which was conducive to the efficient activation of the N≡N bonds [78].This investigation provides some new ideas for designing active metal sulfides for the electrochemical synthesis of NH 3 .
Under the comprehensive theoretical investigation on a range of transition metal nitrides (TMN) for e-NRR, metal nitrides are believed to offer the potential advantages for N 2 fixation [79].Skúlason et al. [80] studied the possibility of nitrogen activation for electrochemical NH 3 formation on a range of (111) TMN surfaces (ScN, TiN, VN, CrN, MnN, YN, ZrN, NbN, MoN, HfN, TaN, WN, and ReN).It was found that VN, CrN and MnN were the most promising candidates, which were expected to catalyze e-NRR at the relatively low onset potential (from −0.80 V to −0.50 V vs. SHE).However, the possibility of poisoning toward MnN and WN was found in an electrochemical environment.Only NbN with the (111) plane can be regenerated itself and can activate N 2 to NH 3 , with active and stable activity.
To date, only Mo-and V-based nitrides have been experimentally studied and proved to enable efficient catalytic activity toward e-NRR.Based on the theoretical investigations, vanadium nitride nanosheet [81], vanadium nitride nanowire array [82], and vanadium nitride nanoparticles [83] have been fabricated and tested for e-NRR activity.Comparatively, vanadium nitride (VN) nanoparticles exhibited better catalytic performance for e-NRR, with an NH 3 production rate and FE of 3.30 × 10 −10 mol s −1 cm −2 and 6.0%, respectively [83].According to a combination of ex situ and operando characterizations, multiple vanadium oxide, oxynitride and nitride species were present on the surface.Among them, VN 0.7 O 0.45 was identified as the active phase in the e-NRR, and the conversion of VN 0.7 O 0.45 to VN phase was proposed as the deactivation pathway.
Except for vanadium nitride, concerns regarding molybdenum nitride have been recently discussed.Li et al. [84] theoretically studied the 2D layered molybdenum nitride nanosheets (MoN 2 ) as NH 3 synthesis catalysts at room temperature.According to calculations, MoN 2 exhibited excellent performance for adsorption and activation of N 2 molecules, but large energy input was requested to regenerate the MoN 2 surface.However, the e-NRR performance can be remarkably promoted after Fe-doping, with ∆G max = 0.47 eV for the rate-determining step.The conclusion about Fe-doping agreed with the recent report regarding Fe-doped MoS 2 [78].Experimentally, the MoN nanosheet array on a carbon cloth (MoN NA/CC) was explored as a high-performance catalyst towards e-NRR in 0.10 M HCl under ambient conditions [81].This catalyst achieved an NH 3 yield of 3.01 × 10 −10 mo1 s −1 cm −2 and an FE of 1.15% at −0.30 V vs. RHE.Moreover, N 2 H 4 was not detected, and therefore, MoN NA/CC showed excellent selectivity to NH 3 .The potential-determining step of this catalyst was the second protonation of the surface N, confirmed by DFT calculations.In another study, this group reported a Mo 2 N nanorod as an efficient electrocatalyst to electrochemically convert N 2 to NH 3 [85].Mo 2 N nanorods were prepared by nitriding of the precursor MoO 2 in an NH 3 atmosphere.Compared with MoO 2 , the NH 3 yield of Mo 2 N was much higher.When tested in 0.10 M HCl, Mo 2 N could enhance the FE to 4.5% at an applied potential of −0.30V vs. RHE, which was higher than MoN NA/CC in the previous report.DFT calculations also confirmed that the free energy barrier of the potential-determining step for the Mo 2 N catalyst was dramatically lower than MoO 2 .Based on the studies above, the FE in e-NRR still needs to be further improved in the future.

Metal Carbide Catalysts
The metal carbides are an interesting class of catalysts.According to the d orbital theory, transition metal carbides with unoccupied d orbitals should have good adsorption ability for electron-enriched reactants [86,87].In order to investigate the viability of using molybdenum carbide as an e-NRR electrocatalyst, a computational study was conducted by Matanovic and co-workers [88].The comparison between two competing reactions (HER and NRR) revealed that MoC (111) was the only surface that suppressed the adsorption of H-atoms at low overpotentials, among various crystallographic surfaces.Additionally, the e-NRR in MoC (111) surface could take place at small negative potentials of −0.30V vs. SHE, and followed an associative reaction pathway.The authors also illustrated that introducing carbon vacancies could mitigate hydrogen evolution and H-adatom accumulation.Recently, molybdenum carbide nanodots embedded in ultrathin carbon nanosheets (Mo 2 C/C) were designed by molten salt synthesis, and used as a catalyst candidate for e-NRR [89].The obtained Mo 2 C/C nanosheets exhibited efficient e-NRR catalytic activity with an NH 3 production rate of 11.3 µg h −1 mg −1 and FE of 7.8%.Based on the experiments and DFT calculations, the catalytic active center of Mo 2 C nanodots was favorable for adsorbing N 2 , and their unique electronic structure was feasible for N 2 activation and hydrogenation.MoS 2 , MoO 3 , MoN and Mo 2 N have been reported as e-NRR electrocatalysts with relatively lower FE of 1.17%, 1.9%, 1.15%, and 4.5% [77,81,85,90], respectively.To continuously enhance the performance, Sun's group reported Mo 2 C nanorod as a catalyst for electrocatalytic N 2 reduction to NH 3 production [90].At the applied potential of −0.30V vs. RHE, such a catalyst achieved a high FE of 8.13% and NH 3 yield rate of 95.10 µg h −1 mg cat.
−1 in 0.10 M HCl electrolyte.To date, metal carbide nanocomposites such as the e-NRR catalyst are rarely reported.
MXenes, a group of 2D layers of transition metal carbides, are promising catalysts for e-NRR [91,92].A large number of studies have focused on theoretical calculations.In 2019, Luo et al. [93] firstly reported that the MXene (Ti 3 C 2 T x ) nanosheets attached to a vertically aligned metal host could achieve a high NH 3 FE (5.78%) at an ultralow overpotential of −0.10 V vs. RHE.From the combined experimental and theoretic results, a greater number of exposed edge sites and a metal host with poor HER activity were responsible for higher e-NRR activity.In another work, a Ti 3 C 2 T x MXene nanosheet was used as both a conductive and Ti source toward the in situ hydrothermal growth of TiO 2 nanoparticles [94].The combination of TiO 2 and Ti 3 C 2 T x led to a synergistically active Ti-based nanohybrid catalyst with enhanced activity.As a result, such a TiO 2 /Ti 3 C 2 T x hybrid catalyst exhibited an NH 3 yield of 26.32 µg h −1 mg cat.
−1 with an FE of 8.42% in 0.10 M HCl electrolyte (−0.60 V vs. RHE).It is universally known that 3D porous MXenebased aerogel architectures are beneficial for rapid mass diffusion, higher exposure of electrochemically active sites, and faster mass diffusion and charge/electron transport.Herein, Li et al. [95] designed a functional 3D MXene-based composite heterojunction aerogel (MS@S-MAs) for e-NRR, fabricating metal sulfide nanoparticles confined in 3D S-doped MXene sheets (Figure 7a), via divalent metal-ion-induced assembly following the thermal sulfidation method.Remarkably, CoS@S-MAs gave the best reactivity among metal sulfide nanoparticles (M = Co, Fe, Cu, Ni), showing an NH 3 yield rate and a FE of 12.4 µg h −1 mg cat.
−1 and 27.05% at the lower potential of −0.15 V vs. RHE in Na 2 SO 4 solution.Additionally, CoS@S-MAs, after 50 h of e-NRR, displayed a slight loss in FE and NH 3 yield rate (Figure 7b), indicating the excellent long-term stability of the catalyst.This study offers a new prospect for 3D porous aerogel materials for application in e-NRR metal oxide catalysts.
Metal oxides have been widely applied in chemical research and also exhibited great potential in e-NRR.To find the viability of electrocatalysts for catalyzing NH 3 formation electrochemically at ambient conditions, 11 types of transition metal dioxides (NbO 2 , TaO 2 , RuO 2 , ReO 2 , TiO 2 , OsO 2 , RhO 2 , MnO 2 , CrO 2 , IrO 2 , and PtO 2 ) in the rutile structure were investigated by DFT calculations on their (110) lattice planes [96].The predicted onset potentials as a function of the binding energy of NNH were given in Figure 7c, with only two potential-determining steps.Among 11 types of transition metal dioxides, ReO 2 , TaO 2 , and OsO 2 , required an overpotential similar to, or lower than, the overpotential required for reducing nitrogen through nitrogensase, but that is believed to be approximately 0.63 V. Metal oxides have been widely applied in chemical research and also exhibited great potential in e-NRR.To find the viability of electrocatalysts for catalyzing NH3 formation electrochemically at ambient conditions, 11 types of transition metal dioxides (NbO2, TaO2, RuO2, ReO2, TiO2, OsO2, RhO2, MnO2, CrO2, IrO2, and PtO2) in the rutile structure were investigated by DFT calculations on their (110) lattice planes [96].The predicted onset potentials as a function of the binding energy of NNH were given in Figure 7c, with only two potential-determining steps.Among 11 types of transition metal dioxides, ReO2, TaO2, and OsO2, required an overpotential similar to, or lower than, the overpotential Moreover, the (110) facets of ReO 2 and TaO 2 were found to favor NNH adsorption over H adsorption, whereas IrO 2 and NbO 2 surfaces might be poisoned by adsorbed hydrogen atoms.Huang et al. [98] experimentally verified the potential of NbO 2 nanoparticles as an efficient e-NRR electrocatalyst.Compared to Nb 2 O 5 with a similar crystal structure but a different linkage style, the Nb 4+ cation of NbO 2 enabled effective N 2 adsorption by proving empty d-orbitals and subsequent activation by back donation.Consequently, the NbO 2 nanoparticles presented both an efficient NH 3 production rate (11.60 µg h −1 mg cat. −1 ) at −0.65 V and FE (32%) at −0.60 V, significantly higher than those of Nb 2 O 5 nanoparticles under similar conditions.
Due to the industrial application of Fe-based catalysts in the Haber-Bosh process, Fe-based oxide materials were also considered as an efficient candidate in the field of e-NRR.Ever since nano-Fe 2 O 3 was reported as an e-NRR electrocatalyst by Licht and coworkers, Fe-based oxides have attracted wide attention.Later, a complementary theoretical study demonstrated the chemical formation process of NH 3 on two kinds of hematite (γ-Fe 2 O 3 ) surfaces.Compared with single-iron (Fe-O 3 -Fe-), double-iron (Fe-Fe-O 3 -) needed a smaller applied bias for proton transfer, owing to the two available reactive Fe sites on this surface [99].Kong et al. [7] firstly investigated the e-NRR activity of nanosized γ-Fe 2 O 3 electrocatalysts at low temperature (<65 • C), in basic aqueous solution and in the membrane electrode assembly (MEA)-based reactors, respectively.Compared with the half-cell, the e-NRR activity in MEA-based reactors was observed with a dramatical increase to 55.90 nmol h −1 mg −1 .The enhanced catalytic performance may be attributed to the efficient utilization of γ-Fe 2 O 3 after it is coated on the porous carbon paper.Furthermore, the Fe 2 O 3 -CNT and oxygen-vacancy-enriched-Fe 2 O 3 /CNT catalysts were also reported [100,101].In addition to Fe 2 O 3 , Fe 3 O 4 was also reported to be catalytically active for e-NRR.A spinel Fe 3 O 4 nanorod on a Ti mesh (Fe 3 O 4 /Ti) was fabricated as a catalyst for electrochemical N 2 conversion to NH 3 , with long-term electrochemical durability [102].Hu et al. [103] investigated the Fe-based materials for electrocatalytic NH 3 production and revealed the effect of different chemical states of Fe on the e-NRR activity.The Fe/Fe 3 O 4 catalyst was fabricated via in situ growth on the Fe foil.In particular, the activity and selectivity of Fe/Fe 3 O 4 were superior to those of Fe, Fe 3 O 4 and Fe 2 O 3 nanoparticles.It has been concluded that the e-NRR catalytic performance was related to Fe/Fe oxide ratio.
Much attention has also been focused on developing Mo-based oxides as high-performance e-NRR electrocatalysts.Sun et al. [90] discovered that MoO 3 nanosheets exhibited remarkable e-NRR activity with excellent selectivity in 0.10 M HCl electrolyte (NH 3 yield: 29.43 µg h −1 mg cat. −1 and FE: 1.9%).It was found that the outermost Mo atoms served as the active sites for effective N 2 adsorption, by DFT calculations.To further tailor the performance of Mo oxides, a hybrid catalyst of MoO 2 on reduced graphene oxide (MoO 2 /RGO) was fabricated to catalyze the e-NRR.In 0.10 M Na 2 SO 4 electrolyte, an enhanced e-NRR performance was obtained, with an NH 3 yield of 37.4 µg h −1 mg −1 and FE of 6.6% at the potential of −0.35 V (vs.RHE) [104].Relative to MoO 2 alone, MoO 2 /RGO hybrid promoted the electronic interactions with *N 2 H, and enabled the donation of more electrons from the active Mo sites to *N 2 H, leading to the enhanced e-NRR activity.Based on the vacancy and heterostructure engineering, O-vacancy-rich MoO 3-x anchored on Ti 3 C 2 T x -MXene (MoO 3-x /MXene) was explored, as a highly efficient and selective e-NRR electrocatalyst, obtaining an exceptional e-NRR performance with an NH 3 yield of 95.8 µg h −1 mg −1 at −0.4 V and a FE of 22.3% at −0.3 V [97].MoO 3−x /MXene produce steady NH 3 yields and FEs during consecutive seven cycles of electrolysis, while just a very small change compared to the initial one.In Figure 7d, OV-rich MoO 3 and MoO 3 /MXene achieved higher e-NRR activities with respect to their corresponding OV-rich MoO 3 and MoO 3 /MXene, indicating the critical role of OVs for substantially improving e-NRR performance.Through in situ Raman spectroscopy adopted in a tailor-made electrolytic cell (Figure 7e), the 3D plots for the time-dependent Raman spectra traces of various catalysts at −0.4 V were shown in Figure 7f-i, to track the changes in surface chemical bonds of considerable catalysts.Together with molecular dynamics simulations and DFT computations, the synergistic effects of OVs and MXene on the e-NRR of MoO 3−x /MXene were confirmed.SnO 2 , known for its low cost and high chemical stability, was initially developed by Zhang et al. [105] as an e-NRR electrocatalyst in the form of cubic sub-micron SnO 2 particles loaded on carbon cloth (SnO 2 /CC).However, to enhance conductivity and active sites of such catalyst, Chu et al. [106] developed a novel fluorine-doped SnO 2 mesoporous nanosheets on carbon cloth (F-SnO 2 /CC) as an e-NRR electrocatalyst.From the calculations, F-doping contributed to readily enhance the conductivity and increase the positive charge density on active Sn sites, resulting in reduced reaction energy barriers and enhanced e-NRR activities.This group also investigated the e-NRR performance of supporting the ultrasmall SnO 2 QDs on RGO [106].Similarly, the experimental and theoretical results confirmed that coupling SnO 2 QDs and RGO could readily tailor the electronic structure of SnO 2 , leading to fascinating e-NRR activity.
As one of the classical semiconductors, TiO 2 -based materials have been firstly investigated as efficient photocatalysts in the photo-reduction of N 2 to NH 3 .Lately, Sun et al. [107] explored the TiO 2 nanosheets array on the Ti plate (TiO 2 /Ti) for electrochemical N 2 conversion to NH 3 .When measured in 0.10 M Na 2 SO 4 , TiO 2 /Ti achieved a high NH 3 yield of 9.16 × 10 −11 mol s −1 cm −2 with an FE of 2.50% at −0.70 V vs. RHE, due to the enhancement of adsorption and activation of N 2 by in situ-generated oxygen vacancies.To further enhance electronic conductivity, a TiO 2 nanoparticle-reduced graphene oxide hybrid (TiO 2 -rGO) was fabricated as an e-NRR electrocatalyst, by Sun's group [108].The FE of TiO 2 -rGO was enhanced to 3.30% at −0.90 V vs. RHE.
Inspired by the enhanced activity of nitrogenases with Mn 2+ , Wang et al. [109] reported MnO particles on Ti mesh (MnO/TM) as a robust e-NRR catalyst.In 0.10 M Na 2 SO 4 electrolyte, such catalyst achieved a high FE up to 8.02% and a large NH 3 production of 1.11 × 10 −10 mol s −1 cm −2 at −0.39 V (vs.RHE).Theoretical calculations further revealed that the MnO(200) surface preferentially adsorbed N atoms instead of H atoms, and the potential-determining step was *N 2 → *N 2 H transformation.Additionally, a spinel LiMn 2 O 4 nanofiber could act as a noble-metal-free electrocatalyst for NH 3 synthesis with an excellent FE of 7.44% [110], much higher than that of the previous Mn 3 O 4 nanocube (3.00%) [111] and Mn 3 O 4 nano-particles-reduced graphene oxide (3.52%) [112].Besides, Cr 2 O 3 nanofiber was fabricated as a non-noble-metal e-NRR electrocatalyst [113].This catalyst achieved an efficient performance in both FE and NH 3 formation, with favorable electrochemical durability.

Conclusions and Outlook
In conclusion, we discuss the recent advances of metal-based e-NRR electrocatalysts using the structure-function relationship, concluding that noble-metal-based catalysts, non-noble-metal-based catalysts and metal compound catalysts provide a fundamental basis for rational electrocatalyst design.Additionally, the challenges and prospects for e-NRR were proposed.Although the encouraging progress on e-NRR electrocatalysts has been achieved with favorable performance, the reported studies of e-NRR still have a long distance to go in contrast with the industrial Haber-Bosch process, from the point of view of industrialization and commercialization.Research in e-NRR still faces several key challenges in the near future.
(1) Selectivity of catalysts is a much larger issue for improving FE, due to the competitive reactions.The hydrogen and hydrazine simultaneously generated during the ammonia production resulted in a relatively low selectivity towards e-NRR [114][115][116].
The designed catalysts are required to have a much stronger binding energy of *N compared to the *H.In addition, the strategy of enhancing the solubility of N 2 in the electrolyte also needs to be developed [117][118][119][120]. (2) In-depth studies of the e-NRR mechanism are still limited and plain.Most of the research only simulated the possible reaction pathways and the energy barriers using theoretical calculations.Most reported theoretical studies for identifying the research direction toward electrocatalyst design were performed on appropriate and simplified models.However, the real-time operation of e-NRR is always in combination with different reaction conditions and parameters (pH, environmental electrolyte, voltage, environmental temperature, and pressure, etc.) that ought to be considered in further calculations [3,28,121].
(3) The stability of the catalyst is as important as catalyst activity and selectivity.After longtime electrolysis operation, the electrocatalyst may undergo decomposition and deactivation [4,122].Therefore, the electrocatalysts should be designed with a stable structure.Additionally, the prolonged periods for the stability tests are recommended to screen active electrocatalysts for e-NRR [28,123].(4) The relationship between structure and activity for e-NRR is of significance to provide a guideline on the rational design of novel catalysts.Despite the great efforts on developing advanced materials for e-NRR, it remains challenging to reveal the relationship between the structure and activity under the reaction conditions [124,125].
In situ analytical techniques and theoretical experiments are highly desirable and beneficial in providing evidence of catalyst surface reconstruction and generation of key intermediates under real-time reaction conditions, as well as in achieving a comprehensive understanding of the kinetic mechanism [25,124,126,127].

Figure 3 .
Figure 3. (a) Volcano plot of e-NRR for the flat (black lines) and stepped (red lines) transition metal, via dissociative (solid lines) and associative (dotted lines) mechanisms (the redox-potential-limiting step for each metal is highlighted with circles) (an asterisk, *, denotes the adsorption site; the vertical lines (a, b, c, and d) separate different parts and display which species are most strongly bound to the surface) [33], copyright 2012, Royal Society of Chemistry; (b) diagrams of N atomic orbitals and

Figure 3 .
Figure 3. (a) Volcano plot of e-NRR for the flat (black lines) and stepped (red lines) transition metal, via dissociative (solid lines) and associative (dotted lines) mechanisms (the redox-potential-limiting step for each metal is highlighted with circles) (an asterisk, *, denotes the adsorption site; the vertical lines (a, b, c, and d) separate different parts and display which species are most strongly bound to the surface) [33], copyright 2012, Royal Society of Chemistry; (b) diagrams of N atomic orbitals and their hybridization as N 2 molecular orbitals [36], copyright 2018, Wiley-VCH; schematic diagram of nitrogen reduction pathway on heterogeneous catalysts [37] ((c) dissociative pathway, associative pathways including (d) distal and alternating pathway, and (e) enzymatic pathway), copyright 2016, American Chemical Society.

Figure 6 .
Figure 6.(a) Schematic illustration of the synthetic process for the Mo/HNG catalyst [62], (b) atomic resolution HAADF-STEM images of Mo-HNG and magnified area with circled individual Mo atoms anchored on the carbon matrix at N-rich edges [62], (c) NH3 yields, FEs and partial current densities for e-NRR on Mo/HNG, Mo/NG, 2Mo/HNG catalysts determined from chronoamperometric measurements [62], (d) 1H NMR spectra of the resultant electrolyte obtained from the e-NRR measurement of Mo/HNG, respectively, using 14 N2 or 15 N2 as the isotopic nitrogen source at −0.05 V [62], copyright 2022, Wiley-VCH GmbH.(e) Schematic illustration of the synthesis of Fe−N/C−CNTs [67], copyright 2018, American Chemical Society.

Figure 6 .
Figure 6.(a) Schematic illustration of the synthetic process for the Mo/HNG catalyst [62], (b) atomic resolution HAADF-STEM images of Mo-HNG and magnified area with circled individual Mo atoms anchored on the carbon matrix at N-rich edges [62], (c) NH 3 yields, FEs and partial current densities for e-NRR on Mo/HNG, Mo/NG, 2Mo/HNG catalysts determined from chronoamperometric measurements [62], (d) 1H NMR spectra of the resultant electrolyte obtained from the e-NRR measurement of Mo/HNG, respectively, using 14 N 2 or 15 N 2 as the isotopic nitrogen source at −0.05 V [62], copyright 2022, Wiley-VCH GmbH.(e) Schematic illustration of the synthesis of Fe−N/C−CNTs [67], copyright 2018, American Chemical Society.

Figure 7 .
Figure 7. (a) Schematic illustration of the synthesis of CoS@S-MAs.(b) Chronoamperometry curve of CoS@S-MAs for 50 h electrolysis (−0.15 V vs. RHE), and corresponding NH3 yield and FEs before and after 50 h (inset), Copyright 2021, Wiley-VCH GmbH.(c) Volcano plot of plotting the predicted onset potentials for e-NRR on the (110) facet of transition-metal dioxides against the binding energy of NNH, ΔENNH, as the descriptor of catalytic activity [96].Copyright 2017, American Chemical Society; (d) NH3 yields/FEs at −0.4 V of MoO3, MoO3-x, MoO3/MXene and MoO3-x/MXene (pink star represents FE, dotted line separates OV-rich materials and OV-poor materials), (e) schematic of tailor-made electrolytic cell, (f-i) 3D plots of the time-dependent in situ Raman spectroscopy of different catalysts for e-NRR process at −0.4 V [97], Copyright 2021, Wiley-VCH GmbH.

Figure 7 .
Figure 7. (a) Schematic illustration of the synthesis of CoS@S-MAs.(b) Chronoamperometry curve of CoS@S-MAs for 50 h electrolysis (−0.15 V vs. RHE), and corresponding NH 3 yield and FEs before and after 50 h (inset), Copyright 2021, Wiley-VCH GmbH.(c) Volcano plot of plotting the predicted onset potentials for e-NRR on the (110) facet of transition-metal dioxides against the binding energy of NNH, ∆E NNH , as the descriptor of catalytic activity [96].Copyright 2017, American Chemical Society; (d) NH 3 yields/FEs at −0.4 V of MoO 3 , MoO 3-x , MoO 3 /MXene and MoO 3-x /MXene (pink star represents FE, dotted line separates OV-rich materials and OV-poor materials), (e) schematic of tailor-made electrolytic cell, (f-i) 3D plots of the time-dependent in situ Raman spectroscopy of different catalysts for e-NRR process at −0.4 V [97], Copyright 2021, Wiley-VCH GmbH.

Funding:
The authors acknowledge the support from the Natural Science Foundation of Shandong Province Youth Project (ZR2021QE067), Research Grants Council of the Hong Kong Special Administrative Region (CityU 11206520), Shenzhen Knowledge Innovation Program (Basic Research, JCYJ20190808181205752), Development Plan of Youth Innovation Team in Shandong Province (2022KJ213), the Outstanding Young Talents Project of Shandong University of Science and Technology (SKR21-3-A-011) and Innovation and Entrepreneurship Training Program for College Students of Shandong University of Science and Technology (X202210424006).