Yi-Han
Wang
b,
Ji-Hong
Dong
a,
Zhenquan
Tan
a,
Xiao-Feng
Wang
c and
Xue-Zhi
Song
*a
aState Key Laboratory of Fine Chemicals, School of Chemical Engineering, Dalian University of Technology, Dalian 116024, China. E-mail: songxz@dlut.edu.cn
bLeicester International Institute, Dalian University of Technology, Panjin 124221, China
cKey Laboratory of Materials Modification by Laser, Ion and Electron Beams, Ministry of Education, School of Physics, Dalian University of Technology, Dalian 116024, China
First published on 1st May 2023
Ammonia (NH3) is the most important chemical and carbon-free energy carrier, which is currently mainly produced by the energy-intensive Haber–Bosch process at high temperatures and pressures. In the spirit of global carbon neutrality and sustainable development, electrochemical nitrogen reduction reaction (NRR) is expected as a promising alternative to the traditional Haber–Bosch process for industrial ammonia production due to both low energy consumption and zero emission of carbon dioxide. However, further breakthroughs in electrochemical NRR strongly depend on the efficiency and selectivity of electrocatalysts. The element Fe has been found to be an important component in nitrogenase enzymes, generating extensive research interest. As one of the most cost-effective and abundant transition metals, iron-based materials have presented great potential for electrochemical NRR. In this review, we summarize recent advances in iron-based materials (including its oxides, chalcogenides, borides, single Fe-atom catalysts, and bimetallic catalysts especially for Fe–Mo-based compounds) towards highly efficient N2-to-NH3 conversion under ambient conditions. Based on recent theoretical and experimental studies, we focus on the structure–property relationship. Strategies to boost NRR performances and perspectives for future developments are comprehensively presented in synthetic protocols, structure modification, activity/selectivity enhancement, and reaction mechanisms to provide insightful guidance for the field of NRR studies.
Currently, it has been proven that biocatalysis, photocatalysis, and electrocatalysis are clean pathways for ammonia synthesis from nitrogen reduction.12–15 Electrochemical synthesis of ammonia is considered the most promising alternative to the Haber–Bosch process, due to its mild and stable reaction conditions, abundant hydrogen source (direct extraction from electrolyte instead of fossil fuels), zero CO2 emissions, and flexible and easily-controlled production process by adjusting external parameters.16 In the electrochemical ammonia synthesis, NH3 is produced from N2 and H2O under ambient conditions by inputting electrical energy, with the catalyst playing a decisive role. In fact, the breakthrough development of the electrochemical ammonia synthesis is hampered by some inherent limitations: (1) the ultra-low solubility of N2 in water (6.8 × 10−4 M, 1.01 × 105 Pa, 298 K) is not conducive to maintaining high reactant concentrations, resulting in the unsatisfying ammonia production rate. (2) Low Faraday efficiency (FE). The improvement of FE is hindered by competing hydrogen evolution reactions (HER).17 The electrochemical NRR and HER are both thermodynamically available under reaction conditions. Not only that but HER is preferred from a kinetic point of view because it is a feasible two-electron process. By comparison, electrocatalytic NRR is a multi-step kinetic retardation process, involving six protons and six electrons as well as the activation of inert nitrogen bonds.18 (3) The stability and ammonia yield rate of the catalyst is challenging to fulfill the requirements of industrial production.19
From the molecular mechanistic level, the primary challenge of NRR results from the inertness of the NN triple bond, which is the strongest among the diatomic molecules and lacks a permanent dipole.20 A high energy input of 945 kJ mol−1 is required to cleave the NN triple bond, which well explains the harsh conditions of the Haber–Bosch process.21 The kinetically difficult electron transfer reaction and the thermodynamically forbidden protonation of N2 to N2H+ also make the electrochemical NRR more difficult.22 Researchers have developed numerous catalysts with good performances to address the above challenges of electrochemical NRR, including noble metals, transition metal (TM) compounds, and metal-free materials.23–33 Among these materials, iron-based catalysts stand out as a non-noble metal protagonist with the advantages of low cost, large-scale industrial synthesis, and high catalytic efficiency. Iron (Fe) is the only element that coexists among the three biological nitrogen-fixing enzymes, suggesting that iron plays an important role in NRR.21 Moreover, Fe is an active metal easy to be compounded with other elements, such as modifying it into nitrides, hydroxides, oxides, sulfides, and phosphides, which may exhibit distinguished catalytic activation and stability. More importantly, the electronic configuration of Fe provides good solutions to activate the N2 molecule. As a transition metal, the Fe atom has an electronic configuration of 3d64s2, indicating four unpaired d-spin electrons. These abundant d-orbital electrons and unoccupied orbitals activate the inert NN triple bond by driving the well-known donation-acceptance process.34–40 To be more precise, the empty or half-occupied orbitals in Fe allow σ donation to accommodate electrons from N2 bonding orbitals (σ orbitals). Meanwhile, the affluent d-electrons in Fe atoms could be injected into the antibonding orbitals (π* orbitals) in N2 by back-donation. As a result, the loss of electrons from the bonding orbitals and the injection of electrons into antibonding orbitals greatly destabilize N2 and expedite the break of the NN bond. It is acknowledged that the NRR activity of the transition metals is usually closely related to the occupancy of eg and t2g orbitals in the catalyst. Fe atom has a manipulable local spin state of t2g4eg2. Both experimental and theoretical analysis indicate that the heightened eg electrons penetration to the antibonding π-orbital of N2 and regulation of the adsorption strength of nitrogen species can be realized by customized spin-polarization configuration.41 Moreover, the electron jump between high-spin iron and ferrous in the appropriate Fe lattice provides good electronic conductivity and poor adsorption of protons, which may lead to more ideal efficiency and selectivity towards NRR.42,43 From a macroscopic view in the design of catalysts, Fe is easily modified for various construction strategies, including vacancy engineering, heteroatom doping, facet control, single atom dispersion, and even metal–organic frameworks (MOFs).21,44–46 These strategies usually enhance N2 adsorption by redistribution of charge density in the catalyst and provide more active sites to improve catalytic efficiency of NRR.47,48
Based on these facts, this review introduces the recent progress of iron-based catalysts for electrocatalytic NRR under ambient conditions. Firstly, we systematically present the recent advances on iron catalysts for NRR according to the types of iron compounds, which are iron oxides, iron sulfides, Fe-containing bimetallic materials, especially iron-molybdenum compounds, iron borides, and single iron atoms catalysts (Fig. 1). Furthermore, we analyze the available approaches to design advanced electronic and interface structures for different types of iron-based catalysts and relationships between structures and properties from experimental and theoretical perspectives. Finally, we prospect the challenges and potential strategies for further enhancement of NRR performance in terms of FE, ammonia yield, and selectivity.
Fig. 1 Scheme organization of the main contents of this review. Iron oxides: reproduced with permission: copyright 2021, Elsevier.49 Copyright 2022, Wiley-VCH.50 Copyright 2022, Royal Society of Chemistry.51 Iron chalcogenides: reproduced with permission: copyright 2021, Elsevier.52 Copyright 2021, Wiley-VCH.53 Fe–Mo-based catalysts: reproduced with permission: copyright 2019, Elsevier.54 Copyright 2020, Royal Society of Chemistry.55 Copyright 2020, American Chemical Society.56 Iron borides: reproduced with permission: copyright 2020, Elsevier.57 Copyright 2019, Royal Society of Chemistry.58 Fe-based SACs:reproduced with permission: copyright 2022, Wiley-VCH.59,60 |
Fig. 2 (A) Diagram of N atomic orbitals and N2 molecular orbital diagram. Reproduced with permission: copyright 2018, Wiley-VCH.66 (B) Binding modes of N2 molecules with transition metals. Reproduced with permission: copyright 2018, American Association for the Advancement of Science.79 (C) Reaction pathways for the electrochemical nitrogen reduction to ammonia. Reproduced with permission: copyright 2021, Elsevier.12 |
This will provide guidance for the rational design of catalytic active sites starting from addressing the intrinsic inertness of N2.
In acidic media
N2 + 6H+ + 6e− → 2NH3 |
In basic media
N2 + 6H2O + 6e− → 2NH3 + 6OH− |
The detailed NRR process mainly involves the following three steps: (i) the chemisorption of N2 molecules on the active sites; (ii) the dissociation and hydrogenation of nitrogen; and (iii) the desorption of the ammonia molecules from the catalyst surface. According to the adsorption configurations and the hydrogenation reaction sequences of nitrogen, two basic pathways have been proposed (i) the dissociative mechanism and (ii) the associative mechanism (Fig. 2C).69–73
In addition to the classical NRR pathways mentioned above, novel pathways have been discovered in practice by in situ spectroscopies.77,78 Shao's group proposed a new two-step reaction pathway for NRR, based on surface-enhanced infrared-absorption spectroscopy and differential electrochemical mass spectrometry data.78 In such a pathway, the absorbed nitrogen molecule reduces to by a two-electron transfer process, followed by its desorption and decomposition to form NH3 (3N2H2 → 2NH3 + 2N2). These characterization techniques and findings provide important tools and new insights into the reaction mechanism of NRR on metal surfaces, which can inspire further understanding of the NRR mechanism and rational design of the more advanced electrocatalysts.
Stoichiometric Fe3O4 and Fe2O3 and their derived structures have been extensively studied as iron oxides for NRR electrocatalysis. Hu and coworkers reported a Fe/Fe3O4 hybrid catalyst via thermal oxidization of Fe foil and subsequent in situ electrochemical reduction (Fig. 3B).100 X-ray photoelectron spectroscopy (XPS) (Fig. 3C) revealed the composition and chemical state changes of the surface layer in the three samples. The Fe0 on the Fe foil surface was completely oxidized and then reduced to both Fe0 and oxidic Fe components by the subsequent electrochemical reduction. Furthermore, the presence of α-Fe and Fe3O4 crystallites in the as-prepared catalyst was confirmed by X-ray diffraction (XRD) (Fig. 3D). Such Fe/Fe3O4 catalyst demonstrated significantly better NRR selectivity than several commercial iron materials (Fig. 3A), reaching an NH3 yield of 0.19 μg cm−2 h−1 and a FE of 8.29% in the phosphate buffer solution. Similarly, Xie et al. developed a one-step green synthesis of Fe–Fe3O4 catalyst via the reduction of FeSO4 in an aqueous solution using KBH4.50 High-resolution transmission electron microscope (HRTEM) image (Fig. 3E) depicted the presence of atom-vacancy defects in Fe–Fe3O4 nanocrystals, which resulted in the excellent NRR performance by enhancing the nitrogen adhesive ability. Electrochemical impedance spectroscopy (EIS) suggested that Fe–Fe3O4 had a significantly enhanced charge transfer capacity. During five cycle tests, no obvious changes in the current density curves (Fig. 3F), NH3 yield rate, and FE (Fig. 3G) were observed, indicating the excellent stability of Fe–Fe3O4 for NRR. Porous structures with high specific surface area have been built to enhance the catalytic performance by increasing active surface area. In a recent study, Ying et al. reported porous Fe3O4 nanosheets via regeneration from deep eutectic solvent (DES) and subsequent annealing (Fig. 3H).49 A series of annealing temperatures was applied, sample Fe3O4-300, which was sintered at 300 °C had the largest specific surface area (88.30 m2 g−1), exhibiting an NH3 yield rate of 12.09 μg h−1 mgcat−1 and a remarkable FE of 34.38% (Fig. 3I).
Fig. 3 (A) FE of Fe/Fe3O4 and several commercial iron materials. (B) SEM images of the Fe foil, the 300 °C oxidized sample, and the sample after the reduction. XPS spectra of the Fe 2p peaks (C) and grazing-incidence XRD patterns (D) of the three samples. Reproduced with permission: copyright 2018, American Chemical Society.100 (E) HRTEM images of Fe–Fe3O4 catalyst. Chronoamperometry curves (F), NH3 production rate and FE (G) of Fe–Fe3O4 catalyst during the 5 cycles tests. Reproduced with permission: copyright 2022, Wiley-VCH.50 (H) Synthetic route schematic of porous Fe3O4 nanosheets. (I) Average NH3 production rate and FEs (at −0.1 V) of samples were annealed at different temperatures. Reproduced with permission: copyright 2021, Elsevier.49 |
Except for the widely studied aqueous electrolyte systems, advances of iron-based catalysts have also been made in non-aqueous electrolyte systems to address some inherent difficulties.18,101,102 Ammonia electro-synthesis in aqueous systems suffer from low N2 solubility and competitive HER.103 Typically, the reduction of water to hydrogen occurs in the same region of potential with NRR, resulting in low FEs.104 Calculations and experiments showed that nitrogen is much more soluble in some ionic liquids (ILs) than in water.105,106 The improved N2 solubility in IL electrolytes is an effective way to increase N2 adsorption at the active sites.11 Besides, ILs are hydrophobic and can serve as excellent non-aqueous substances because they can lower the water content at an optimal level to effectively restrain the evolution of H2.104 MacFarlane and coworkers practically explored the NRR performance of Fe deposited working electrode substrates (including fluorine-doped tin oxide glass, nickel foam, and stainless steel cloth) in a hydrophobic, high N2-solubility IL electrolyte ([C4mpyr][eFAP] and [P6,6,6,14][eFAP], respectively).104 The strong interaction between N2 and [eFAP]− was proposed by DFT calculations, which explained the high solubility of N2. This work also suggests the influence of the viscosity of ILs on electrochemical performance, that lower viscosity can support higher mass transport to produce higher current densities and yields. Although this NRR system achieved an incredibly high FE of 60%, the reduction currents obtained on low surface area substrates are too small and the yield needs to be improved for practical applications. In another work, MacFarlane's group further focused on the rational electrode–electrolyte design and demonstrated the effectiveness of using highly fluorinated solvents to realize efficient electrochemical ammonia synthesis.107 The mixture of 1H,1H,5H-octafluoropentyl-1,1,2,2-tetrafluoroethyl ether and [C4mpyr][eFAP] were chosen and characterized to determine the optimum solvent–IL ratio to achieve a maximum conductivity of 1.95 mS cm−1 and electrochemical window of 3.40 V, which was ideal for NRR. The core–shell α-Fe@Fe3O4 catalyst was applied in such an IL mixture and accomplished an ammonia yield rate of ∼2.35 × 10−11 mol s−1 cmECSA−2 and NRR selectivity of ∼32%.
Thus far, most reported studies of iron oxide catalysts also focused on Fe2O3.100 High-index facets are usually represented by a set of Miller indices {hkl} with at least one index greater than 1. These facets have a high density of exposed low-coordination atoms and are therefore more open and active than the tightly packed {111} and {100} low-index facets.108 The energy barrier of HER can be elevated at some specific HIFs, while a more feasible NRR pathway is opened to improve NRR selectivity.109 Xie et al. combined the facet modulation and nanoarray structure design to improve the NRR performance of α-Fe2O3.110 The α-Fe2O3 nanorod arrays (NAs) with exposed (104) HIF offered abundant unsaturated Fe atomic dangling bonds as the NRR active sites. Moreover, the nanoarray structure achieved an enhanced electrochemical surface area (ECSA) to fully display the active sites and maintain sufficient mass diffusion. Thus, the selectively exposed (104) surface coupled with the high ECSA of the nanoarray structure achieves the superior electrocatalytic performance of α-Fe2O3 NAs. Unfortunately, current studies on HIFs to optimize the electrocatalytic selectivity and activity are popular in noble metal-based materials and their alloys, such as Pt and Au.87,111,112 The lattice facets control engineering on iron materials still needs to be explored.
Given the fact that adjusting the OV concentration is an effective method for improving the oxide catalyst properties for NRR, various strategies have been applied for this target, such as annealing and heteroatom doping. High-temperature annealing is a feasible way to increase the concentration of OVs. Cui et al. reported an o-Fe2O3–Ar/CNT catalyst that was annealed under argon (Ar) utilizing commercial hematite nanopowder as raw material.44 It was found that the NRR performance of o-Fe2O3–Ar with the highest OV concentration (12.26%) was better than that of o-Fe2O3-air and Fe2O3-untreated. After 16 hours of the chronoamperometry test, the NH3 production rate and FE decreased significantly, suggesting that the stability of the structure needs to be improved. However, high-temperature annealing is an energy-intensive process, which is typically carried out in a reducing atmosphere. Thus, the introduction of OVs by doping with low valence metals is highly desirable to be developed as a relatively environmentally friendly synthesis method. Yu et al. developed zinc-doped Fe2O3 nanoparticles via a wet chemical route at only 160 °C.113 Zinc has a similar ion radius to that of Fe3+, which is able to induce OVs in Fe2O3 nanoparticles. In neutral media, it offers a maximum FE of 10.4% and an NH3 formation rate of 15.1 μg h−1 mgcat−1 at −0.5 V vs. RHE. Furthermore, Zhang et al. were the first to propose a task-specific ionic liquid strategy for OV-rich α-Fe2O3 nanocubes (denoted as Fe2O3-IL), achieving an NH3 production rate of 32.13 μg h−1 mgcat−1 with a FE of 6.63% at 0.3 V vs. RHE.114 The ionic liquid n-octylammonium formate (OAF) simultaneously worked as the reaction medium and structure-directing template for growing α-Fe2O3 nanocubes while the reductant formic anion of ionic liquid drove rich OVs in situ in α-Fe2O3. This innovative approach allows the OV concentration to be adjusted under mild conditions. Additionally, the low surface tensions of OAF compared with water resulted in a higher nucleation rate of Fe2O3, than the growth rate. Consequently, Fe2O3-IL had a much smaller particle size of 23.2 ± 1.8 nm than that of Fe2O3 synthesized in a water medium. As a result, a larger specific surface area and more active sites were delivered in the catalytic process.
Most of the catalysts we mentioned above were prepared with the additive of surfactants through solution methods. However, the surfactants had a passivation effect on the catalyst surface, which decreased the activity of the catalyst. Therefore, developing a strategy to construct a clean surface could be critical for achieving high activity.115 Herein, Wang et al. reported porous Fe2O3 nanorods grown on carbon cloth (p-Fe2O3/CC).116 The EDX element mapping image suggested that Fe and O elements with a ratio of about 2/3 are uniformly distributed throughout the nanorods. With the assistance of Na2SO4 as a structure-directing agent, the self-supported structure of p-Fe2O3/CC with a large surface area provided abundant active sites and accessible channels for promoting mass/charge transfer during the NRR process. Furthermore, Wang et al. successfully synthesized clean-surface Fe2O3 nanoparticles without any surfactants or structure-directing agents.115 Fe(NO3)3 was dispersed in glycerine for prior hydrothermal treatment, followed by high-temperature calcination. This clean-surface catalyst exhibited better performance with a high yield of NH3 22 μg h−1 mgcat−1 and a FE of 3.5% at −0.5 V vs. RHE. The as-synthesized catalyst showed good catalytic and structural stability even without a binder.
The conventional preparation of iron oxide materials mostly requires high-temperature annealing, which is not convenient and highly energy intensive. Preparation strategies at room temperature need more exploration. In addition, iron oxide suffers from poor electrical conductivity, which is against the high catalytic efficiency. Good electron donors can be introduced to modulate the electronic structure of the catalyst. Many of the current iron oxide electrocatalysts for NRR struggle with low ammonia yields or FEs. Further theoretical studies are recommended to explore their catalytic intrinsic principles and thus guide the catalyst improvement.
Fig. 4 (A) Schematic of the formation of Fe3S4 in the PEG based DES. Reproduced with permission: copyright 2018, Royal Society of Chemistry123 (B) chronoamperometry curve of Vs-FePS3 in N2-saturated electrolyte with the addition of SCN− ion. Inset: FEs with and without the SCN− ion. (C) XPS spectra of S 2p of Vs-FePS3 before and after NRR. Reproduced with permission: copyright 2021, Elsevier.52 (D) Schematic diagram of coordination configuration in FeS2. (E) Free energy profile for NRR in Fe atoms as the active sites. Reproduced with permission: copyright 2020, American Chemical Society.126 (F) Nyquist plots and (G) ammonia yield and FE of FeS2@GO, GO and FeS2 at −0.3 V in N2-saturated 0.1 M HCl solution. (H) Six-time recycling test of FeS2@GO. Reproduced with permission: copyright 2021, American Chemical Society.128 |
Inducing surface defects is an available strategy to increase the active sites of NRR via heteroatom doping. In a recent work, a surface-defective FeS2 catalyst was obtained by sulfurization of stainless-steel foil at 500 °C.125 The fitting of the XPS spectrum sulfur S 2p peaks and the HRTEM image proved the existence of abundant SVs. Meanwhile, such defects were considered to be the active sites for the stable and efficient adsorption of dinitrogen molecules. It was strongly advised that the generation of an active hydrogen atom was primarily caused by the exposed (210) facet of iron pyrite. The stainless-steel-based catalyst possessed a considerably higher FE of 14.6% in contrast to the pure iron-based catalyst in 0.1 M Li2SO4 at −0.2 V. Because of the existence of Cr species, the SS-based electrode with anti-corrosion properties presented a superior catalytic stability under acidic conditions. Hou et al. reported a porous exfoliated FePS3 nanosheet with rich SVs (Vs-FePS3 NSs), which was formed through electrochemical exfoliation and hydrogenation.52 Vs-FePS3 NSs showed significant activity in the electrocatalytic NRR with a high NH3 yield rate of 3.88 μg h−1 cm−2 at −0.25 V and a FE of 12.36% at −0.20 V versus RHE due to their rich SV and unique 2D morphology. To identify the actual active sites of Vs-FePS3 NSs for NRR, SCN− ion was used to act as a poisoning reagent for metal-centered active sites. A significant decrease appeared in the current density and FE (from 12.36% to 5.31%), indicating that the active Fe site was effectively shielded (Fig. 4B). The shift of S 2p main peaks to higher binding energies in high-resolution XPS spectra suggested that the SVs boosted the hydrogenation of N2 (Fig. 4C).
The suitable ligand configuration of the active sites plays a significant role in enhancing the catalytic performance. It is found that a metal sulfur cluster with four permanent ligands in a twisted tetrahedral ligation shell is an effective active center of NRR. This coordination structure is well demonstrated in FeS2, where the NRR process proceeds preferably on Fe atoms (Fig. 4D). In a recent study, a FeS2 NRR catalyst was synthesized via the one-step hydrothermal method.126 The catalyst exhibited an outstanding NRR performance, achieving a high NH3 yield rate of 37.2 μg h−1 mgcat−1 and a FE of 11.2% at −0.5 V versus RHE. The current density and original structure of FeS2 had no significant change after 12 h of electrolysis, indicating excellent durability and stability. Fe atoms were proven to be the active sites instead of S atoms by poisoning experiments and free energy profile for NRR in Fe sites and S sites. The reaction mechanism was investigated by DFT calculations that showed that the associative distal pathway was preferred and the conversion from *NH2 to *NH3 was regarded as the potential-determining step (PDS) (Fig. 4E).
The heterostructure of iron sulfide-based catalysts has also been developed to enhance their catalytic performance. The interface within heterostructures can usually regulate the electronic structure and expose extra active sites simultaneously. Chen and coworkers designed a hybrid catalyst of crystalline Fe2O3 with amorphous FeS (Fe2O3/FeS) via a DES-regulated fabrication, followed by an annealing process.127 The hybrid catalyst exhibited abundant channels and a high specific surface area of up to 103.94 m2 g−1, providing more exposed electrochemical active sites for N2 fixation and accelerating mass convection. Interface engineering between crystalline Fe2O3 and amorphous FeS optimized electron distribution and surface structure, being conducive to the absorption and activation of the N2 molecules. In another work, Gao et al. created a graphene oxide-wrapped pyrite nanoparticle (FeS2@GO) via Fe(NO)3, NH3·H2O, and graphene oxide as a precursor.128 The coupling effect between the catalyst and the substrate accelerated the electron transport property, FeS2@GO exhibited the smallest charge transfer resistance, highest NH3 production, and FE, much better than those of pure FeS2 and GO (Fig. 4F and G). Meanwhile, the NH3 yield rate and FE remained almost constant during the six-time recycling tests (Fig. 4H). Importantly, the low-coordinated Fe atoms are activated as highly active sites, where the free energy change ΔG (*N2) of N2 binding was reduced to −0.57 V, leading to stronger binding of N2 and longer NN bond of absorbed *N2 with respect to pristine FeS2 (100).
Further progress in iron sulfide can be made by modulating the electronic structures of electrocatalysts by doping, tailoring surface vacancies, defect engineering, and/or heterointerface engineering. Meanwhile, refining and creating more active sites are also proposed to optimize the adsorption and activation of the nitrogen species.
Fig. 5 (A) TEM image of FeTe2/RGO. (B) Optimized structures of *N2 over Fe3c sites. (C) Free energy profiles of three NRR pathways on Fe3c site at U = 0. Reproduced with permission: copyright 2021, Elsevier.129 (D) Side views of the defective Fe3GeTe2 monolayer. (color code: red, Fe; bule, Ge; gray, Te) The blue arrows stand for local magnetic moments of Fe atoms up aligned. (E) Possible reaction pathways for NRR. Reproduced with permission: copyright 2021, Royal Society of Chemistry.130 (F) Electron location function of FeMoSe (left), FeMoS (right), in which the Fe, Mo, Se and S atoms are denoted by red, green, yellow and cyan balls, respectively. (G) Schematic illustration for N2 binding to FeMoSe and FeMoS. NH3 yields and FEs of FeMoSe, FeMoS (H) and other comparative samples (I).132 |
2D layered double hydroxide (LDH) with evenly distributed ferric ions exhibited N2 fixation activity, but the intrinsic resistivity and poor charge transfer properties astricted its further application. A NiFe-LDH-derived NiFe-selenide (Ni0.75Fe0.25Se2) was designed to solve this weakness, the introduction of selenium conspicuously accelerated the p–d coupled electron transfer during nitrogen reduction.131 Through theoretical analysis, the selenization of iron caused lattice distortion, which may arise from the uneven occupation of t42g and e2g orbitals in Fe2+. More N2 molecule adsorption sites were introduced by lattice distortion regulation. A bio-inspired joint electron pool was realized in FeMo(Se, Te) by replacing sulfur with selenium or tellurium.132 Notably, Se and Te had a lower electron-negativity as compared with S, allowing a more expansive electron cloud to form the joint electron pool, which originated from the delocalized electronic structure (Fig. 5F). The density of state (DOS) profiles of the top valence band near the Fermi level in FeMoSe is mainly composed of Fe d, Mo d, and Se p valence orbitals, illustrating a higher electron density and charge transport ability than those of FeMoS and MoSe2. Theoretical models for the joint electron pool displayed stronger nitrogen adsorption, which is critical for the subsequent hydrogenation steps. As these, FeMoSe had a greater ability to inject electrons into N2 (Fig. 5G), exhibiting an NH3 yield rate of 20.14 μg h−1 mgcat−1 and a FE of 13.65%, which were far better than those of FeMoS and other comparative samples (Fig. 5H and I).
Briefly, the novel metal chalcogenides may be promising electrocatalysts with high efficiency for ammonia synthesis from gaseous dinitrogen. There is a large space for scientists to apply multi-element doping and heterostructure engineering similar to that used in iron sulfides to further promote NRR electrocatalysis. More in-depth research about iron selenide/iron telluride is expected to promote this electrocatalytic NRR field.
Fig. 6 Structure and projected density of states (pDOS) for the clean Fe2Mo6S8 surface (A), the Fe2Mo6S8 surface with *N2 (B), *NNH (C), *NH3 (D), and *NNH (E) with the Fe surface atom fixed at its position on the clean surface. Reproduced with permission: copyright 2022, American Chemical Society.136 (F) Illustration of the proposed binding patterns for N2 absorption and subsequential conversion to NH3. (G) Comparison of FE and NH3 yield for Mo6S8 and M2Mo6S8 (M = Fe, Mn, and Cu) electrocatalysts. Reproduced with permission: copyright 2021, American Chemical Society.137 (H–J) Charge density difference of MoS2–Fe, MoS2, MoS2–V (from left to right). Electron excess and deficiency are denoted by red and green isosurfaces, respectively. Isosurface was set as 0.003 Å−3. (K) Free-energy profiles of the NRR of MoS2–Fe, MoS2, and MoS2–V. Reproduced with permission: copyright 2022, reproduced with permission: copyright 2021, Springer.140 |
Progress has also been made in bionic design from FeMoS-nitrogenases.141 In recent work, Du and coworkers designed a Mo(IV)-doped FeS2 (FeS2–Mo) via the hydrothermal reaction of MoCl5, Fe(NO3)2, and thioacetamide.56 Energy Dispersive Spectrometer (EDS) mapping indicated that Mo, Fe, and S elements were distributed uniformly throughout the nanosheets. Electrochemical measurements suggested that pure FeS2 presented HER negative and a relatively low yield rate for NH3. However, the NRR activity was significantly promoted after Mo doping and increased gradually with the Mo content, proving that Mo could act as active sites. The Mo-doped-modified catalyst exhibited a high FE of 14.41% and an NH3 yield rate of 26.15 μg h−1 mgcat−1. Afterwards, the optimal pathways for NRR in FeS2 and FeS2–Mo were proposed by DFT calculations. Moreover, the doping of Mo opened up an associative distal pathway in FeS2–Mo, instead of the associative alternating pathway in FeS2. The reaction energy barrier of the critical dinitrogen hydrogenation step (NN* + H → NNH*) was reduced from 1.24 eV to 0.62 eV. Therefore, the FeS2 substrate was regarded to suppress the competitive HER, while the doping of Mo favored the adsorption and activation of nitrogen. Similarly, Su et al. achieved the suppression of HER by designing atomically dispersed Fe-decorated MoS2 (Fe–MoS2).142 The edge S atoms in MoS2 were generally considered active sites for HER, while the Mo atoms acted as active centers for NRR.135 Therefore, better selectivity for NRR could be accomplished by Fe modification of the S sites without shielding the Mo sites to achieve high FE (18.8% at −0.3 V versus RHE). DFT calculations proved that the energy barriers of HER in Fe–MoS2 (0.15 eV) were obviously magnified with respect to pure MoS2 (0.03 eV). It is worth mentioning that our group predicted and achieved Fe–O–Mo subunits for highly active NRR in Mo-doped Fe2O3 (Fig. 7A).143 The introduction of Mo destroyed the ordered crystal structure and enriched lattice defects, creating more active sites. It also tuned the electronic structures of Fe2O3 to enhance the electron injection into N2, resulting in an elongated and weakened nitrogen bond (Fig. 7C and D). The as-synthesized material showed much higher ammonia yields (21.3 μg h−1 mgcat−1) and FE (11.2%) than pure Fe2O3, proving the significant role of Mo doping (Fig. 7B). An optimized reaction pathway with a lower energy barrier was confirmed on Mo-doped Fe2O3(110) in comparison with pristine α-Fe2O3(110) (Fig. 7E). Unfortunately, this work could not achieve the doping amount control. It needs further effort to reveal the phase transition mechanism of the catalytic material during the long-term electrocatalytic process.
Fig. 7 (A) Diagrammatic representation of the formation of Mo-doped Fe2O3 porous nanospheres. (B) Comparison of NH3 production rates and FEs at −0.6 V for Mo-doped Fe2O3 and Fe2O3. The charge density difference of N2-adsorbed α-Fe2O3(110) (C) and Mo-doped α-Fe2O3(110) (D) with an isosurface value of 0.006 e Å−3. (E) Free energy diagram for the NRR process on α-Fe2O3(110) and Mo-doped α-Fe2O3(110), respectively. Reproduced with permission: copyright 2022, American Chemical Society.143 (F) Schematic illustration for the formation routes of Fe–MoS2/CC. (G) NH3 yield and FE with different mass ratios of Fe–MoS2. Loading mass: 2.15 mg cm−2. Reproduced with permission: copyright 2019, Royal Society of Chemistry.152 |
Doping heteroatoms is effective to promote the electrocatalytic activity of basic materials. Further endeavors should be devoted to disclosing the position of the doping exogenous atom, the coordination and valent state, and the corresponding function during the whole catalytic process.
A FeS2 and MoO2 nanocomposite anchored on a three-dimensional (3D) graphene aerogel (MoO2/FeS2/GA) was developed by Liu and coworkers.146 FeS2 and MoO2 nanoparticles were embedded as active sites to absorb and activate N2 in an interconnected 3D graphene-based mesoporous aerogel, which provided good conductivity and large accessible specific surface areas. Synergistic effects of such a ternary microstructure exhibited superior electrochemical performance over either single structure, endowed the hybrid catalyst with an impressively high FE of 37.44% and NH3 yield of 40.18 μg h−1 mgcat−1 at −0.25 V in 0.1 M HCl. Zhao et al. reported Fe nanodot-decorated MoS2 nanosheets on carbon cloth (Fe–MoS2/CC) via a chemical reduction strategy (Fig. 7F).152 The strong electronic interaction of Fe and MoS2 was confirmed by the shift of XPS binding energies of Mo and S and the red shifts in Raman spectra. Furthermore, the effect of Fe loading on the NRR performance was also explored via electrochemical measurements. The best NH3 yield reached 12.5 μg h−1 cm−2 with a FE of 10.8% at a Fe/MoS2 mass ratio of 0.077 (Fig. 7G). When the Fe content got too large, the recession in NRR performance was speculated to be attributed to the degraded structures. In another work, FeS-dotted MoS2 nanosheets covered carbon fiber cloth (CFC) (FeS@MoS2/CFC) was prepared through a facile single-step hydrothermal preparation.54 FeS and MoS2 uniformly supported the highly conductive CFC and successfully simulated the composition of the active site in FeMoS-cofactor. Notably, it exhibited considerable NRR catalytic activity over a wide pH range.
Chen et al. reported a one-step pyrolysis–phosphating process from MoFe-MOF precursors to fabricate phosphorus-doped carbon microspheres (MoFe-PC), as a cost-effective and highly active catalyst for N2 fixation under ambient conditions.170 Electrochemical measurements confirmed that P-doped MoFe-PC (34.23 μg h−1 mgcat−1, FE: 16.83%) showed more excellent activity than MoFe–C (24.73 μg h−1 mgcat−1, FE: 12.47%). On the one hand, the introduction of P to form a P–C connection increased the conductivity of the hybrids by optimizing the electronic structure. On the other hand, the synergistic interaction between bimetallic oxides and P-doped carbon ensured abundant active sites for the adsorption and activation of nitrogen. Chen and coworkers applied a nanostructure tailoring strategy to successfully synthesize unique hollow NiαFe1−α–MoS2 nanocubes through vulcanization treatment of Ni–Fe PBA by ammonium thiomolybdate.171 The as-prepared material showed high stability during the two-hour NRR process at various applied potentials (Fig. 8A). The charge redistribution of the catalyst in an optimal atomic ratio of the NiFe (α = 0.3) as compared to the pure MoS2 was presented by the calculated electron density difference (EDD) mappings (Fig. 8B–C). Based on the electron transfer pathways of NiαFe1−α–MoS2 catalysts, the spin-polarized density of states, and the corresponding partial density of states, it was suggested that the coupling of S 2p and Mo 3d orbitals generated the Ni0.3Fe0.7 active sites, which could easily donate electrons to the absorbed N2 molecules. Furthermore, a possible pathway of the NRR on the NiFe–MoS2 NCs catalyst was proposed by electrochemical-FTIR spectroscopy analysis and DFT calculations (Fig. 8D and E). Another Fe-containing MOF precursor, famous as MIL, has also been demonstrated to synthesize Fe2O3@MoS2 composites for NRR electrocatalysis and achieved a remarkable NH3 yield rate of 112.15 μg h−1 mgcat−1 and a FE of 8.62%.150
Fig. 8 (A) Time-dependent current density (j) of Ni–Fe–MoS2 at various potentials. Electron density difference of the optimized structures of MoS2 (B) and Ni0.3–Fe0.7@MoS2 (C) (electron excess and deficiency are denoted by red and blue, respectively). (D) In situ electrochemical-FTIR spectra of the NRR process on the Ni–Fe–MoS2 NCs catalyst. (E) Proposed electrocatalytic mechanism on Ni–Fe@MoS2. Reproduced with permission: copyright 2020, Royal Society of Chemistry.171 (F) Schematic diagram of the preparation routes of Fe1.89Mo4.11O7/FeS2@C and FeMoO4/FeS2@C composites. (G) Comparison of NH3 production rate and FE of FeS2@C, FeMoO4@FeS2@C, and Fe1.89Mo4.11O7@FeS2@C (1.6 wt%) at the optimum voltage, respectively. Reproduced with permission: copyright 2020, Royal Society of Chemistry.55 |
Polyoxometalate-based metal–organic frameworks (POMOFs) combine the advantages of both MOFs and POMs to provide specific target structures derived from organic ligands and serve as a source of metals such as Co, Fe, Ni, Mo.169,172,173 Lately, Ma et al. made an exploration using PMo12@MOF-100(Fe)@PVP as a precursor to synthesize Fe1.89Mo4.11O7/FeS2@C for efficient electrocatalysis of nitrogen reduction (Fig. 8F).55 Experiments and characterization analysis concluded that the ratio of Fe and Mo could be controlled by accurately regulating the sulfuration time and temperature. Besides, a certain amount of PVP was convinced to be converted into a defected graphitization configuration, according to the intensity ratio of the D band and G band in Raman spectroscopy. Most of the FeS2 was successfully encapsulated in the graphitic carbon layers leading to obtaining good conductivity. Fe1.89Mo4.11O7/FeS2@C exhibited an extremely high NH3 yield of 105.3 μg h−1 mgcat−1 and a FE of 54.7% at −0.4 V (Fig. 8G), as well as long-period electrolytic stability. The above works open new schemes for the rational design and fabrication of bimetallic MoFe-composites derived from MOFs for NRR. It can be further envisioned that the MOFs and even COFs (COF = covalent organic framework) containing Fe and Mo simultaneously may be used as electrocatalysts and through coordination modulation to regulate the electrocatalytic NRR performance. Moreover, MOFs and COFs may also be utilized as templates or precursors to structurally direct functional nanomaterials with special nano-morphologies, multi-component, and even heterostructures for high-performance NRR electrocatalysis. Thus, Fe,Mo-MOF/COF should be paid much attention to for the materials synthesis and structure design.
Fig. 9 (A) Optimized structures for N2 adsorption on various types of Fe and Mo atoms. Reproduced with permission: copyright 2020, American Chemical Society.175 (B) Schematic illustration of the synthesis for FMO-NPs. (C) Variation of free energy for the distal and alternating pathways at the Fe2(MoO4)3 (110) surface. Reproduced with permission: copyright 2020, American Chemical Society.176 (D) The p-DOS and schematic illustrations of 4d orbitals of Mo atoms on Au/Fe2(MoO4)3, 2p-orbitals of the N2 gas molecule, and their interaction within -Au/Fe2(MoO4)3. (E) Synthetic route of Au/Fe2(MoO4)3. (F) TEM image of Au/Fe2(MoO4)3. Reproduced with permission: copyright 2021, Wiley-VCH.178 |
The deposition of noble metal nanoparticles is universal to improve the electrocatalytic activity. The above-mentioned iron molybdate materials demonstrate satisfying NRR activities, meanwhile, Au has generally low activity of HER.177 Therefore, it is a reasonable scheme to introduce Au to optimize the atomic structure of Fe2(MoO4)3 and enhance the performances of NRR. Yao et al. supported Au on the Fe2(MoO4)3 substrate to construct an Au/Fe2(MoO4)3 hybrid via a facile room-temperature reduction approach, achieving an excellent FE of 18.79% at −0.4 V (Fig. 9E and F).178 Compared with Fe2(MoO4)3 and Au/Fe2(MoO4)3via the analysis of temperature-programmed desorption of N2, the later one showed stronger binding strength of N2, which originated from the Au regulating promotion. To get a profound understanding of the activation of NN, the projected density of states plots of Mo atom and N2 were calculated. The d-σ and d-π* orbitals hybridization facilitated the electron transfer from Mo 4d orbitals to N2 2p orbitals (Fig. 9D), thus effectively weakening the inert NN triple bond. As a result, the reaction energy barrier of RDS of the distal pathway at Mo active centers was reduced from the former 0.98 to 0.6 eV after Au loading. Unfortunately, these catalytic ferric molybdate materials with large particle sizes suffer from insufficient active sites. In the future, material scientists may develop novel synthetic routes to decrease the size to expose more sites. The electrocatalytic performance can be further enhanced by simultaneous heteroatom doping and heterostructure engineering.
Due to the coupling effect, the multicomponent nanostructures generally demonstrate more advanced NRR performance than their individual counterparts.32,184 Introducing the second metal compound to a Fe-based material will be a compelling strategy to attain further improved performance. For example, doping Co elements into Fe-based materials can be an effective way to rationally regulate the electronic structure of catalysts due to the small atomic size and electronegativity of Co atoms, and the d–d coupling reduces the free energy barrier for NN triple bond dissociation. This idea was confirmed experimentally by Chen and coworkers by designing the Co-doped Fe3S4 nanoflowers (Fig. 10A). The introduced Co behaved as a new active center to promote N2 absorption and reduce the energy barrier, achieving a good catalytic performance (Fig. 10B).185
Fig. 10 (A) Schematic of Co-doped Fe3S4 nanoflowers. (B) NH3 yields and FEs of Co-doped Fe3S4 nanoflowers at a series of applied potentials. Reproduced with permission: copyright 2020, American Chemical Society.185 (C) TEM image of Au–Fe3O4 NPs. (D) N2 temperature-programmed desorption curves of different catalysts. (E) The geometric structure of various intermediate states during NRR, in which the hydrogen, oxygen, nitrogen, iron, and gold atoms are denoted by white, red, blue, slate blue, and yellow balls, respectively. (F) Free energy diagram of the NRR pathways on Au–Fe3O4 (101), Fe3O4 (101), and Au (111). Reproduced with permission: copyright 2020, Wiley-VCH184 (G) Illustration of synthesis and electronic interactions of dendritic Fe2O3/Cu electrocatalyst. (H) XPS valence band spectra of Fe2O3/Cu, Cu, and Fe2O3. Reproduced with permission: copyright 2019, Elsevier.186 |
Zhang et al. prepared heterogeneous Au–Fe3O4 nanoparticles (NPs) (Fig. 10C) via a one-pot wet-chemical reaction of HAuCl4 and diironnonacarbonyl (Fe2(CO)9).184 Such Au–Fe3O4 NPs exhibited good performance with a high ammonia yield of 21.42 μg h−1 mgcat−1 and a favorable FE of 10.54%. The N2 temperature-programmed desorption (TPD) (Fig. 10D), and surface valence band spectra together confirmed that Au–Fe3O4 NPs had a strong synergistic effect for adsorbing reaction species and suitable interface for charge and proton transfer. The theoretical calculations showed that Fe performed as an active site to fix N2 into *N2H while the introduction of Au boosted the adsorption of NRR intermediates and provided an energetic-favorable pathway for NRR (Fig. 10E and F). Huang et al. developed a Fe2O3/Cu composite catalyst by electro-deposition in the presence of FeCl2 and CuCl2 as electrolytes (Fig. 10G), which reached its highest yield of 15.66 μg h−1 mgcat−1 with a FE of 24.4% at −0.1 V vs. RHE.186 The composite improved the electron transfer on Fe2O3, as copper is a good conductor of electrons. As d-band theory suggests, the closer the d-band center approaches the Fermi level, the stronger the adsorption is. The electronic interactions between Fe2O3 and Cu elevated the d-state center (Fig. 10H), enabling strong adsorption and back-bonding of N2 molecules, leading to high catalytic activity.
Fig. 11 (A) Schematic of π-backdonation between boron and N2. Reproduced with permission: copyright 2019, American Chemical Society.192 Top and side views of FeB (B) and FeB2 (C). Fe and B are represented as purple and pink balls, respectively. Models and bonding network for (D) FeB6 (α) and (E) FeB6 (β). The length of Fe–B5c (short) and Fe–B6c (long) bonds are labeled in units of Å. Reproduced with permission: copyright 2019, Royal Society of Chemistry.58 (F) Illustration of the phase transformation from a-FeB2 to c-FeB2. (G) Charge density difference: the electron accumulation and depletion are represented as red and blue regions, respectively. (H) Free energy profiles of *N2 and *N2H adsorption on c-FeB2 and a-FeB2. Reproduced with permission: copyright 2021, Elsevier.57 |
In a theoretical study, four iron-borides were constructed and explored as potential NRR catalysts, including FeB, FeB2, FeB6(α), and FeB6(β) under the framework of DFT (Fig. 11B–E).58 The authors proposed that Fe–B bonding could remarkably affect Fe–N interaction by examining the free energy differences in reaction pathways on the Fe-sites of FeB2 and FeB, respectively. Additionally, the NH3 release issue had been solved by establishing the hypercoordinated FeB6(α) and FeB6(β). Among these, FeB6(β) offered the lowest ΔGmax of 0.68 eV, owing to the low oxidized Fe in the β-phase. Thus, N2 could acquire electrons as a newly oxidizing reagent and be activated on the lightly oxidized Fe sites. In the evaluation of selectivity, the calculated ΔG(H*) for the four catalysts is less than the ΔGmax of NRR, but H* adsorption is weak in contrast to N2 adsorption. The heterostructure construction was also applied to design high-performance iron boride-based electrocatalysts. In the current work, the process of NRR was studied on iron atoms supported by Bn-doped graphyne (GY).196 A distal pathway was successfully identified with a minimum limiting potential of −0.53 V using DFT calculations. Additionally, the FeB@GY surface could effectively promote electron transfer efficiency and significantly improve the stability of intermediate adsorption during NRR. The substitution of B and N atoms for the C atoms in the doped GY structure is suitable for improving surface activity. From the results of the Bader charge analysis and charge density difference, it was detected that the reaction intermediates could not only accept electrons but also extract electrons from FeB@GY. The analysis of the partial density of states indicated a strong interaction between Fe-3d and N-2p in addition to B-2p, which was caused by the existence of the overlapped orbitals.
Chu's group pioneered the experimental synthesis of amorphous FeB2 porous nanosheets (α-FeB2 PNSs) through a facile reflux method.57 Amorphization engineering produced vast unsaturated coordination sites and surface-exposed defects, which served as active sites to promote the electrocatalytic reaction energetics and kinetics. In 0.5 M LiClO4 electrolyte, α-FeB2 PNSs showed a maximum NH3 yield rate of 39.8 μg h−1 mgcat−1 at −0.3 V and a FE of 16.7% at −0.2 V. Furthermore, the chrono-potentiometric response showed no obvious decays for 20 h of consecutive electrolysis, with a 92.8% retention of the initial NH3 yield, indicating excellent long-term stability. In this work, crystalline FeB2 PNSs (c-FeB2 PNSs) were also prepared by annealing at 600 °C for 3 h (Fig. 11F). When comparing the NRR performance, c-FeB2 showed smaller ECSA and weaker N2 absorption than a-FeB2. Meanwhile, a-FeB2 had strong charge transfer between *N2 and active substrates and more injected electrons to activate the N2 molecule to a greater extent, as observed in Fig. 11G. A lower G*N2H (0.45 eV) on a-FeB2 was provided to stabilize *N2H and reduce the initial reduction barrier (Fig. 11H). Currently, most research on TMBs (TM = transition metal) especially iron borides as NRR electrocatalysts is focused on the theoretical calculations.58,197–200 There is a deficiency of reports about experimental fruits on iron borides for NRR electrocatalysis, probably due to the harsh preparation conditions.
Nitrogen-doped carbon is one of the most encouraging candidates due to its enhanced electronic conduction, optimized charge distribution, and electronic structure of active sites, which is caused by the interfacial interactions between iron atoms and the surrounding nitrogen/carbon atoms.210 As revealed by DFT calculations, atomically dispersed Fe–N3 site on nitrogen-doped graphene (Fe–N3/graphene) can substantially activate the inert NN bond of N2 with high spin-polarization and a localized magnetic moment.215 Yang et al. anchored Fe single-atoms on a nitrogen-doped carbon substrate (Fe SAC/N–C) to realize the coordination structure of Fe–N3.210 The HAADF-STEM images indicated that the majority of Fe objects were downsizing to single atoms with a range of 0.1 to 0.2 nm, and there were no obvious Fe nanoparticles or clusters. The catalyst achieved an enhanced NH3 yield rate of 53.12 μg h−1 mgcat−1 and a high FE of 39.6% in 0.1 M KOH with ignored decay in seven consecutive electrolysis cycles. Computational studies provided direct evidence of the robust binding capability of Fe–N3 catalytic active sites to N2 molecules to achieve the high selectivity of Fe SAC/N–C for NH3 formation. The calculated Gibbs free energy diagrams illustrated that the alternating pathway was a more convincing mechanism on highly dispersed Fe sites compared with the distal pathway, as its lower energy barrier of ΔG = 0.915 eV (Fig. 12A).
Fig. 12 (A) Gibbs free energy profiles of NRR on Fe–N3 and Fe–N4 through alternating and distal pathways. The inset shows the optimized structure of Fe–N3 coordination and the corresponding adsorption of the NRR intermediates. Reproduced with permission: copyright 2020, American Chemical Society.210 (B) SEM images of carbon substrate. Reproduced with permission: copyright 2019, American Chemical Society.216 (C) Schematic illustration of the synthesis routes of Fe–N/C-CPs. (D) HAADF-STEM image and magnified image of Fe–N/C. The Fe-SA centers are marked by red circles. Reproduced with permission: copyright 2021, American Chemical Society.213 (E) NH3 yield rates and FEs at each given potential of FeSA–N–C. (F) Free energy profiles for NRR on the FeSA–N–C catalyst at U = 0 V. The inset represents the corresponding adsorption models (H, C, N, O, and Fe atoms are represented with gray, orange, blue, red and green spheres, respectively). Reproduced with permission: copyright 2019, Springer.218 (G) NH3 yields and FEs during six-recycling tests at −0.45 V vs. RHE. (H) Charge density difference and the N–N bond length after N2 adsorption (I) The density of states for C 2p, B 2p, N 2p and Fe 3d for Fe–B/NC. (J) Gibbs free energy profiles of the H2 evolution on the as-prepared sample. Reproduced with permission: copyright 2022, Elsevier.222 |
Fe–N4 configurations are also developed for electrocatalytic ammonia synthesis. He et al. reported an atomically dispersed iron phthalocyanine (FePc) molecules on porous carbon with Fe–N4 coordination structure, which achieved a dramatically high NH3 yield rate of 137.95 μg h−1 mgcat−1 at a low potential of −0.3 V (vs. RHE).216 The SEM image showed the comb-like structure of 3D carbon substrate with massive nanopores, which contributed to the high loading ability of 7.43 wt% (the loading of FePc) (Fig. 12B). The current density decreased significantly in poisoning treatment by SCN− suggests that the Fe center of FePc is the dominant active site for NH3 formation. Liu et al. explored the realization of high Fe–N4 loading capacity.213 Fe-TPP, C5H8N2, and N,N-diethylformylamide (DEF) etc. were milled in a zirconia-ball grinding mill to form the Fe-TPP⊂rho-ZIF precursor, and Fe–N/C was obtained by pyrolyzing in a tube heating furnace at 950 °C for 3 h under an N2 atmosphere (Fig. 12C). The HAADF-STEM image indicated that the Fe atoms were uniformly distributed on the carbon substrate, in which the Fe single-atom loading reached 3.5 wt% (Fig. 12D). The sublimation of Zn in the pyrolysis process of the precursor was favorable to expose the catalytic sites and increase the specific surface area,217 which reached up to 1088.96 m2 g−1. Wang et al. reported a FeSA–N–C catalyst with an extremely high FE of over 56.55% (Fig. 12E).218 Single Fe atoms were uniformly loaded throughout the N-doped carbon support, and no Fe nanoparticles or aggregations were obviously detected. The adsorption behavior of N2 toward Fe active sites was illustrated by molecular dynamics simulations, which suggested a lower energy barrier of 2.38 kJ mol−1 at approximately 0.54 nm for efficient N2 access. As depicted in Fig. 12F, the alternating pathway with lower energy uphill of 0.28 eV was more favorable than the distal pathway. Additionally, the high energy barriers (2.91 eV) for *H adsorption resulted, in particular, sluggish HER kinetics. Based on the above characteristics, the desired selectivity and FE were achieved by enhancing nitrogen reduction and suppressing hydrogen evolution.
Fe SACs that anchored on strain-engineered graphene substrates were modeled and systematically examined by first-principles calculations, suggesting that curvature strongly influences the catalytic activity and selectivity of SACs for NRR.219 Bu et al. studied the NRR pathways for a dual-atom Fe2 species on graphitic carbon nitride (g-C3N4) by changing external conditions.220 It was found that applying an appropriate range of tensile to the graphitic carbon nitride (g-C3N4) substrate could reduce the overpotential values, which arose from a break in the intrinsic scaling relationship of the adsorption energy between the *N2 and *HNN adsorbates. More influences of structural distortion on the NRR catalytic properties of Fe, N codoped carbon (Fe–N–C) were explored.221 The NRR intermediate species *NNH could be strongly bound at the FeN4 and FeN3 active sites embedded in strained graphene layers, contributing to the enhanced NRR activity. In contrast to *NNH, *H showed a greater adsorption energy barrier, which indicated the less favored competitive HER and higher selectivity of NRR. Their computational work provides evidence for improving the catalytic activity and selectivity of Fe–N–C materials towards NRR by tuning the degree of compressive strain in their graphene layers. Although the theoretical studies of strain engineering on Fe-SACs have achieved promising results, the predictions need to be further confirmed by experimental works.
The doping of multi-element atoms into the substrate is an effective and well-explored strategy for adjusting the Fe active site configuration to enhance NRR. Wang et al. developed boron-induced electron-rich SACs (Fe–B/NC) by dispersing single Fe sites on boron, nitrogen-co-doped carbon.222 The desired selectivity and stability were proved by no detectable N2H4 and six recycling tests (Fig. 12G). The introduction of B atoms with low electronegativity (χ = 2.0) broke the symmetric electron distribution of the Fe–N4 coordination, engendering electron donor (B)/acceptor (N) pairs along the B–Fe–N bridge to enhance electron transfer. After being adsorbed on B–Fe sites, the triple bond in N2 was elongated by 0.115 Å with a comparison of free N2, due to back-donation from B and Fe to N2, which was confirmed by the Bader charge after N2 adsorption analysis (Fig. 12H). Moreover, the coupling with B 2p could delocalize the electrons in the Fe 3d orbital (Fig. 12I), increasing the electron conductivity for N2 adsorption/activation, thereby enhancing the NRR performance. In addition, the catalyst exhibited a hindered effect on HER relative to the boron-free sample, owing to a large ΔGH* (Fig. 12J). Another coordination mode for anchoring Fe SA has also been explored, typically as Fe-(O–C2)4 single-atom sites on lignocellulose-derived carbon with impressive catalytic activity (32.1 μg h−1 mgcat−1 equal to 5350 μg h−1 mgFe−1), by Zhao and coworkers.45
In addition to nitrogen-doped carbon, other substrates also can be applied to support Fe single atoms as SACs for NRR. Porous materials with large surface areas or vacancies are ideal supports that can accommodate ample active sites, realizing the high metal loading. Chen et al. prepared S-coordinated Fe SACs on mesoporous TiO2 (Fe1Sx@TiO2), which were constructed by a lattice-confined strategy.60 FeS2O2 sites were confirmed as active centers by theoretical calculations. No peak was observed for a Fe–Fe bond in the Fe-K edge EXAFS spectra of Fe1Sx@TiO2, indicating that Fe atoms are distributed individually (Fig. 13A). Abundant OVs and Ti–S coordination bonds were detected by Ti K-edge EXAFS spectra, resulting in high dispersion stability. The finite element analysis showed that confinement of opened and ordered mesopores could promote mass transfer and provide a large active surface area. As a result, the catalyst showed a good NH3 production rate of 18.3 μg h−1 mgcat−1 with 17.3% FE, and the origin of NH3 was verified via an isotopic-labeling experiment (Fig. 13B). Moreover, the relationship between Fe loading capability and NRR performance was also investigated, when the ratio of Fe:Ti is 6:100, NRR activity reached maximum without Fe aggregation. First-principles calculations revealed the existence of an enzymatic mechanism with a low overpotential of 0.21 V and side-on adsorption of N2 in MoS2-supported Fe2 clusters (Fe2/MoS2), which is conducive to effective NN bond activation.223 Li et al. creatively designed a single Fe atom supported on atomically thin molybdenum disulfide (SACs-MoS2–Fe) to form high curvatures with interfacial protrusion-like shape (Fig. 13C).224 Triggering interfacial polarization by curvature-rich surface and interface demonstrates potential universality in wide-scope electrocatalysis, including the promising and energy-effective breaking of NN bond. Fe single atoms worked as external adatoms conjugating with MoS2 to create 3D protrusions, instead of interior dopants substituting in MoS2 skeletons, which was distinct from extensively reported metal–Nx and defect-trapped-SACs. The unique geometry stimulated electron injection into NN efficiently enabled N2 polarization and reduction. When the atomic ratio of Fe and MoS2 reached 2.0, SACs-MoS2–Fe2 demonstrated the best catalytic performance of ammonia yield (97.5–6 μg h−1 cm−2) and FE (31.6–2%) at −0.2 V.
Fig. 13 (A) k3-weighted Fourier-transform EXAFS spectra of the Fe R-space for FeS2 and Fe foil, Fe1Sx@TiO2 and Fe–TiO2, the inset is the first peak of Fe1Sx@TiO2 and Fe–TiO2. (B) 1H-NMR spectra of 14NH4+ and 15NH4+ produced in the NRR process. Reproduced with permission: copyright 2022, Wiley-VCH.60 (C) 3D topographic potential distribution images of MoS2 and pure MoS2. Reproduced with permission: copyright 2020, Elsevier.224 (D) In situ FTIR spectra obtained on the surfaces of FeMo/NC, Fe/NC, and Mo/NC SACs, respectively (from left to right). (E) Proposed NRR pathways on FeMo/NC, Fe/NC, and Mo/NC surfaces. Reproduced with permission: copyright 2022, Springer.231 (F) NH3 yields and FEs on various applied potentials. (G) Fe and Co K edge EXAFS fitting curves of Fe/Co–O–C-1.0 at R space. Inset is the predicted bimetallic Fe–Co coordination configuration in Fe/Co–O–C-1.0. (C, O, Fe, and Co are denoted by brown, red, orange, and blue, respectively). Reproduced with permission: copyright 2022, Springer.233 (H) Differential charge densities of absorbed N2 on Pd and PdFe1, where electron accumulation and depletion are denoted by yellow and cyan, respectively. (I) 3D topographic atom images of PdFe1. Reproduced with permission: copyright 2022, Wiley-VCH.59 |
Diatomic catalysts (DACs) derived from single-atom catalysts have been widely explored for highly efficient electrocatalysis.225–228 According to related investigations, the catalytic activity of graphene-based Mn, Ni, Co, and Fe DACs exceeds that of the corresponding SACs.229,230 Zhang and coworkers reported FeMo/NC DSACs as a “dual-site” material by anchoring isolated Fe and Mo atoms on hierarchical N-doped carbon nanotubes.231 This catalyst demonstrated an NH3 yield of 26.8 μg h−1 mgcat−1 and a FE of 11.8, which is 1.6 and 2.5 times larger than those of Mo/NC and Fe/NC. Time-dependent in situ Fourier transform infrared spectroscopy (in situ FTIR) measurements were conducted for dual-site FeMo/NC DSACs and single-site Fe/NC and Mo/NC to investigate the synergy mechanism of Mo and Fe single atoms (Fig. 13D). Three positive-going bands of N–N stretching, –NH2 wagging, and H–N–H bending were appeared at the in situ FTIR spectra of FeMo/NC, suggesting the formation of *N2Hy intermediates during the NRR process. No positive-going NN stretching was observed on the FeMo/NC surface, which implied that the *N2Hx intermediates (0 ≤ x ≤ 2) had no obstacle to separate into a single bond N–N of *N2Hy intermediates (1 ≤ y ≤ 4). The NRR over FeMo/NC may follow an associative mechanism (Fig. 13E), due to the existence of the three main intermediates, as well as no N2H4 was detected. Ma et al. revealed the catalytic principle of Fe/Mo–N–C from the point of view of the spin moment of the active centers through theoretical calculations.232 The neighboring Fe could modulate the spin state of the Mo center in MoN4 to switch from a high-spin state to a medium-spin state, effectively relieving the strong overlap between the Mo 4d orbital and the NxHy intermediates, facilitating the desorption of NH3, and ultimately leading to a lower overpotential. Recent work reported that an extraordinary NRR performance of 579.2 ± 27.8 μg h−1 mgcat−1 NH3 yield rate and 79.0 ± 3.8% FE could be achieved on an atomically dispersed, bimetallic Fe–Co active site catalyst (Fe/Co–O–C-r) (Fig. 13F).233 The observation and fitting of EXAFS spectra of Fe/Co–O–C-1.0 suggested that the atomically dispersed Fe, Co atoms are anchored on carbonized BC (CBC) by a bimetallic Fe–Co arrangement anchoring to graphitic carbon through six O-bridging bonds, for which, [(O–C2)3Fe–Co(O–C2)3] is the likely bimetallic unit (Fig. 13G). Calculations pointed out that an energy-favorable transformation from [(O–C2)3Fe–Co(O–C2)3] to [(O–C2)3Fe–Co(O–C)C2] can occur at the end of an NRR cycle on desorption of *NH3. [(O–C2)3Fe–Co(O–C)C2] was further confirmed to be the actual active site of Fe/Co–O–C-r and resulted in remarkable NRR performance.
Traditional SACs have the inherent disadvantage that single metal atoms tend to agglomerate due to their large surface free energy. Avoiding the aggregation of individual atoms at the expense of low metal loading limits the improvement of NRR performance and practical applications. A new variety of SACs, single-atom alloys (SAAs), in which, the isolated metal atoms are anchored on the substrate metal has been proposed with integrated merits.234–236 Specifically, the extremely strong metal–support interactions endow the high thermodynamic stability to break limitations in metal loading. Moreover, the alloying effect of SAAs tunes the electronic structure of atomic metal sites to enable an advantageous balance between the appropriate binding of intermediates and the effective dissociation of reactants. Li et al. first developed a PdFe1 single-atom alloy metallene, which demonstrated an NH3 yield of 111.9 μg h−1 mgcat−1 and a FE of 37.8% at −0.2 V vs. RHE.59 Preliminary theoretical computations indicated the enhancement of N2 affinity on PdFe1 and the Lewis acid–base interaction facilitated the following N2 polarization and cleavage (Fig. 13H). In the free energy diagrams, the energy barrier of the rate-limiting step was reduced to 0.72 eV, due to the efficient N2 activation on Fe sites. Encouraged by the above theoretical results, the preconceived PdFe1 (Fe: 3.91 at%) was prepared via a one-step wet chemistry strategy. The low-peak-intensity spots in 3D topographic atom images suggested that Fe atoms successfully substituted the partial Pd atoms on the PdFe1 substrate (Fig. 13I). Thus, this stably distributed the SAA catalyst and presented a steady current density during the 100 h of continuous electrolysis in chronoamperometry, and no apparent recession in NRR yields and FEs within ten cycling tests were observed.
Catalyst | Electrolyte | NH3 yield | Quantification methods | FE (%) | Isotope labeling | Ref. |
---|---|---|---|---|---|---|
o-Fe2O3–Ar/CNT | 0.1 M KOH | 0.46 μg h−1 cm−2 | Nessler colorimetry | 6.0 | No | 44 |
Fe/Fe3O4-300 | 0.1 M phosphate buffer solution | 0.19 μg h−1 cm−2 | Indophenol blue colorimetry | 8.29 | No | 100 |
Fe–Fe3O4 nanoparticles | 0.1 M Na2SO4 | 24.6 μg h−1 mgcat−1 | Indophenol blue colorimetry | 53.2 | No | 50 |
Fe3O4-300 nanosheet | 0.1 M Na2SO4 | 12.09 μg h−1 mgcat−1 | Indophenol blue colorimetry | 34.38 | Yes | 49 |
Fe–Fe3O4 nanoparticles | 0.1 M Na2SO4 | 22 μg h−1 mgcat−1 | Indophenol blue colorimetry | 3.5 | No | 115 |
p-Fe2O3/CC | 0.1 M Na2SO4 | 6.78 μg h−1 cm−2 | Indophenol blue colorimetry | 7.69 | No | 116 |
Zn doped Fe2O3 | 0.1 M Na2SO4 | 15.1 μg h−1 mgcat−1 | Indophenol blue colorimetry | 10.4 | Yes | 113 |
α-Fe2O3 nanocubes | 0.1 M KOH | 32.13 μg h−1 mgcat−1 | Indophenol blue colorimetry | 6.63 | Yes | 114 |
Au–Fe3O4 nanoparticles | 0.1 M KOH | 21.42 μg h−1 mgcat−1 | Nessler colorimetry | 10.54 | Yes | 184 |
Fe2O3/Cu | 0.1 M KOH | 15.66 μg h−1 mgcat−1 | Indophenol blue colorimetry | 24.4 | No | 186 |
Fe3S4 nanosheets | 0.1 M HCl | 75.4 μg h−1 mgcat−1 | Indophenol blue colorimetry | 6.45 | Yes | 123 |
FeSx/Fe | 0.1 M KOH | 4.13 × 10−10 mol s−1 cm−2 | Ion chromatography, indophenol blue colorimetry, Nessler colorimetry | 17.6 | Yes | 124 |
Vs-FePS3 NSs | 0.1 M HCl | 3.88 μg h−1 cm−2 | Ion chromatography, indophenol blue colorimetry | 12.36 | No | 52 |
FeS2 | 0.1 M Na2SO4 | 37.2 μg h−1 mgcat−1 | Indophenol blue colorimetry | 11.2 | No | 126 |
FeS2 | 0.1 M Li2SO4 | 11.5 μg h−1 mgFe−1 | Nessler colorimetry | 14.6 | No | 125 |
Fe2O3/FeS | 0.1 M KOH | 34.31 μg h−1 mgcat−1 | Indophenol blue colorimetry | 18.06 | Yes | 127 |
FeS2@GO | 0.1 M HCl & 0.1 M Na2SO4 | 78.6 μg h−1 mgcat−1 in 0.1 M HCl, 27.9 μg h−1 mgcat−1 in 0.1 M Na2SO4 | Indophenol blue colorimetry | 4.7 in 0.1 M HCl, 6.8 in 0.1 M Na2SO4 | No | 128 |
FeTe2/RGO | 0.5 M LiClO4 | 39.2 μg h−1 mgcat−1 | Indophenol blue colorimetry | 18.1 | Yes | 129 |
FeMoSe | 0.1 M Na2SO4 | 20.14 μg h−1 mgcat−1 | Indophenol blue colorimetry | 13.65 | Yes | 132 |
FeS2–Mo | 0.1 M KOH | 26.15 μg h−1 mgcat−1 | Indophenol blue colorimetry, Nessler colorimetry, ion chromatography | 14.41 | Yes | 56 |
Fe2Mo6S8 | 0.5 M Na2SO4 mixed with 0.1 M sodium citrate buffer | 70 μg h−1 mgcat−1 | Indophenol blue colorimetry | 12.5 | Yes | 137 |
MoS2–Fe | 0.1 M Na2SO4 | 20.11 μg h−1 mgcat−1 | Indophenol blue colorimetry | 15.72 | No | 140 |
Fe–MoS2 | 0.5 M K2SO4 | 8.63 μg h−1 mgcat−1 | Indophenol blue colorimetry | 18.8 | Yes | 142 |
Mo-doped Fe2O3 | 0.1 M Na2SO4 | 21.3 μg h−1 mgcat−1 | Indophenol blue colorimetry | 11.2 | No | 143 |
MoO2/FeS2/GA | 0.1 M HCl | 40.18 μg h−1 mgcat−1 | Indophenol blue colorimetry | 37.44 | Yes | 146 |
FeS@MoS2/CFC | 0.1 M Na2SO4 | 8.45 μg h−1 cm−2 | Indophenol blue colorimetry | 2.96 | No | 54 |
Fe–MoS2/CC | 0.1 M KOH | 12.5 μg h−1 cm−2 | Indophenol blue colorimetry | 10.8 | Yes | 152 |
SACs-MoS2-Fe | 0.1 M KCl | 97.5 μg h−1 cm−2 | Indophenol blue colorimetry | 31.6 | Yes | 224 |
MoFe-PC | 0.1 M HCl | 34.23 μg h−1 mgcat−1 | Indophenol blue colorimetry | 16.83 | No | 170 |
Ni0.3–Fe0.7@MoS2 | 0.1 M Na2SO4 | 128.17 μg h−1 mgcat−1 | Indophenol blue colorimetry | 11.34 | No | 171 |
Fe1.89Mo4.11O7/FeS2@C | 0.5 M K2SO4 | 105.3 μg h−1 mgcat−1 | Nessler colorimetry | 54.7 | Yes | 55 |
FeMoO4 nanorods | 0.1 M Na2SO4 | 17.51 μg h−1 mgcat−1 | Indophenol blue colorimetry | 10.53 | No | 174 |
FeMoO4 nanorods | 0.5 M LiClO4 | 45.8 μg h−1 mgcat−1 | Indophenol blue colorimetry | 13.2 | Yes | 175 |
Fe2(MoO4)3 nanoparticles | 0.1 M Na2SO4 | 18.16 μg h−1 mgcat−1 | Indophenol blue colorimetry | 9.1 | No | 176 |
Au/Fe2(MoO4)3 | 0.2 M Na2SO4 | 7.61 μg h−1 mgcat−1 | 1H-NMR method | Yes | 178 | |
α-FeB2 PNS | 0.5 M LiClO4 | 39.8 μg h−1 mgcat−1 | Indophenol blue colorimetry | 16.7 | Yes | 57 |
Fe SAC/N–C | 0.1 M KOH | 53.12 μg h−1 mgcat−1 | Indophenol blue colorimetry | 39.6 | Yes | 210 |
Fe–N/C | 0.5 M K2SO4 | 2.27 μg h−1 mgcat−1 | Indophenol blue colorimetry | 7.67 | Yes | 213 |
FePc/C | 0.1 M Na2SO4 | 137.95 μg h−1 mgFePc−1 | Indophenol blue colorimetry | 10.05 | No | 216 |
10.25 μg h−1 mgcat−1 | ||||||
FeSA–N–C | 0.1 M KOH | 7.48 μg h−1 mgcat−1 | Indophenol blue colorimetry | 56.55 | Yes | 218 |
Fe–B/NC | 0.1 M KOH | 88.7 μg h−1 mgcat−1 | Indophenol blue colorimetry | 9.2 | Yes | 222 |
Fe1Sx@TiO2 | 0.1 M HCl | 18.3 μg h−1 mgcat−1 | Indophenol blue colorimetry | 17.3 | Yes | 60 |
FeMo/NC DSACs | 0.1 M Na2SO4 | 26.8 μg h−1 mgcat−1 | Indophenol blue colorimetry | 11.8 | No | 231 |
Fe/Co–O–C-r | 0.1 M Na2SO4 | 579 μg h−1 mgcat−1 | Indophenol blue colorimetry, 1H-NMR method | 79.0 | Yes | 233 |
PdFe1 | 0.5 M LiClO4 | 111.9 μg h−1 mgcat−1 | Indophenol blue colorimetry | 37.8 | Yes | 59 |
Fig. 14 Strategies to boost NRR performances and perspectives for future developments of iron-based catalysts. Regulate in-depth mechanism: reproduced with permission: copyright 2022, Wiley-VCH.59 Coupling strategy: reproduced with permission: copyright 2020, Royal Society of Chemistry.237 Copyright 2022, Elsevier.150 Copyright 2020, Elsevier.224 Theoretical calculation: reproduced with permission: copyright 2022, American Chemical Society.136 |
This journal is © The Royal Society of Chemistry 2023 |