Ammonia (NH₃) as very important chemical has been widely applied in fuel of vehicles, agricultural, plastic and textile industries [1-4]. Although abundant nitrogen (N₂) exists in the atmosphere (~78%), efficient utilization of N₂ is still very difficult due to high energy barrier for N≡N bond cleavage (~941 kJ/mol), large energy gap (~10.82 eV), and the lowest unoccupied molecular orbital [5]. The synthesis of NH₃ by N₂ hydrogenation is a complex reduction process with multiple reactions [6-8]. Industrially, the NH₃ production from N₂ and H₂ was performed at high temperature (400−600 ℃) and pressure (20−40 MPa) by the well-known Haber-Bosch process [9-12], consuming enormous energy and concurrently releasing a large amount of CO₂. Therefore, development of environmentally friendly NH₃ synthesis techniques at ambient conditions is highly desirable.

Electrocatalytic N₂ reduction reaction (NRR) at ambient conditions has been regarded as a promising NH₃ synthesis method for replacing the traditional energy- and capital-intensive Haber-Bosch process [9-12]. To date, varieties of electrocatalysts have been synthesized and applied in electrocatalytic NRR to synthesize NH₃, such as noble metal catalysts [13, 14], non-precious metal catalysts [15, 16] and metal-free carbon catalysts [17-19], demonstrating high NRR activities. Among non-precious metal catalysts, transition metal oxides have been proven to be efficient catalysts toward electrocatalytic NRR to NH₃, confirming the oxygen vacancies (V_o) as the active centers [20-24]. As we know, niobium oxides have become important materials for catalysis applications [25] because of their superior catalytic activity and high physical/chemical stability. Recently, Kong et al. [26] reported Nb₂O₅ nanowires array as high-performance NRR electrocatalyst, demonstrating an NH₃ yield rate of 1.58×10⁻¹⁰ mol s^-1 cm^-2 with a Faradaic efficiency (FE) of 2.26%. In their work, this good NRR activity can be ascribed to the important role of (001) facet of Nb₂O₅. The Nb₂O₅ nanofiber with high conductivity was also synthesized by Han and co-workers for electrocatalytic NRR to NH₃ [27], showing an NH₃ yield rate of 43.6 μg h^-1 mg_cat.^-1 with an FE of 9.26%. In their study, this high NRR activity is mainly resulted from Nb-edge atoms of Nb₂O₅ (181) surface to polarize and activate N₂ molecules. It is well known that pseudo hexagonal phase Nb₂O₅ has been confirmed to contain rich V_o [28], which could be a promising candidate as the electrocatalyst for NRR to NH₃. In particular, pseudo hexagonal phase Nb₂O₅ nanostructured arrays (e.g., nanochannel array) could provide more excellent electron transfer pathways with exposed V_o active sites for high-efficiency NRR [29, 30].

Herein, highly ordered Nb₂O₅ nanochannel film (Nb₂O₅-NCF) was fabricated on niobium foil substrate by a facile anodization method. After thermal treatment in air, the as-fabricated Nb₂O₅-NCF with pseudo hexagonal phase contains rich oxygen vacancy defects, as the free-standing electrode, displaying high electrocatalytic NRR activity with an NH₃ yield rate of 2.52×10⁻¹⁰ mol cm^-2 s^-1 and an FE of 9.81% at −0.4 V (vs. RHE) in 0.1 mol/L Na₂SO₄ solution (pH 3.2). During electrocatalytic NRR, it was found that the crystalline phase transformation of Nb₂O₅-NCF from pseudo hexagonal phase to hexagonal phase resulted in the electrochromism (EC) phenomenon and the decrease of NRR activity. This is mainly due to the decrease of V_o in hexagonal phase Nb₂O₅. The used Nb₂O₅-NCF electrode can be readily regenerated by low-temperature thermal treatment or applying an anodic potential, regaining high NH₃ yield rate and FE with superior recycling reproducibility.

After anodization, the surface of niobium (Nb) foil showed dark brown and the surface colour of Nb foil became white after further treating in air at 450 ℃ (Fig. S1 in Supporting information). The XRD patterns (Fig. 1a) of the thermally treated sample exhibit a pseudo hexagonal phase (TT-phase) Nb₂O₅ (JCPDS No. 28-0317) [28]. The SEM images (Figs. 1b and c) show that the surface of the sample possesses porous structure, and a highly ordered nanochannel film with a thickness of ~2.0 μm forms on the Nb foil substrate (denoted as Nb₂O₅-NCF). The TEM image and selected area electron diffraction (SAED) patterns of the sample (Fig. 1d) confirm that the nanochannel walls are polycrystalline structure and the lattice fringes with interplanar distance of 3.12 Å can be ascribed to the (100) plane of TT-phase Nb₂O₅. The surface survey XPS spectrum of Nb₂O₅-NCF sample was provided in Fig. S2a (Supporting information), exhibiting Nb and O elements. The high-resolution Nb 3d XPS spectrum (Fig. S2b in Supporting information) indicates that the Nb 3d peaks at 209.6 and 206.9 eV can be assigned to Nb₂O₅ [31, 32].

In this work, the fabricated Nb₂O₅-NCF was directly used as the cathodic electrode for NRR evaluation in 0.1 mol/L Na₂SO₄ solution (pH 3.2) using an H-type electrochemical cell. The linear sweep voltammograms (LSV) results (Fig. 2a) show that the cathodic current density of Nb₂O₅-NCF electrode in N₂-saturated electrolyte is obviously larger than that obtained in Ar-saturated electrolyte when increasing applied potential from −0.3 V to −0.9 V, confirming good NRR activity of Nb₂O₅-NCF. Next, we investigated the influence of applied potential on the NRR performance of Nb₂O₅-NCF. During electrocatalytic NRR, the produced NH₃ and/or N₂H₄ were determined by the indophenol blue [33] and Watt/Chrisp method [34], respectively. In this work, only NH₃ product can be detected (Fig. S3 in Supporting information) while N₂H₄ is undetectable (Fig. S4 in Supporting information). Fig. 2b shows the NH₃ yield rate and Faradaic efficiency (FE) of Nb₂O₅-NCF at different potentials, and corresponding chronoamperometric curves at each potential are shown in Fig. S5 (Supporting information). According to the UV–vis absorption spectra analysis (Fig. S3a) of NRR samples, the largest NH₃ yield rate can reach 2.52×10⁻¹⁰ mol cm^-2 s^-1 with an FE of 9.81% at −0.4 V (vs. RHE) with the NRR time of 60 min (Fig. 2b). When the applied potential is more negative than −0.4 V (vs. RHE), the NRR performance of Nb₂O₅-NCF obviously decreases. This is mainly due to the competitive hydrogen evolution reaction (HER) concurrently happened on Nb₂O₅-NCF at more negative potentials [27]. To eliminate the environmental interference on the NH₃ yielded during NRR, several control experiments were performed. The results (Fig. 2c) show that no NH₃ product can be detected in 0.1 mol/L Na₂SO₄ solution without catalyst, with Nb₂O₅-NCF but without applied potential, or Ar-saturated electrolyte with Nb₂O₅-NCF at −0.4 V (vs. RHE), indicating no environmental interference on the produced NH₃ by NRR process. In contrast, the UV–vis absorption spectra show stronger absorption peak at ~695 nm ascribed to the yielded NH₃ determined by the indophenol blue method. To further confirm this, the isotopic labelling experiments using ¹⁵N₂ as the feeding gas was also conducted. The ¹H nuclear magnetic resonance (NMR) spectra (Fig. 2d) reveal that the chemical shift of doublet coupling can be observed for the NRR sample, due to¹⁵N in ¹⁵NH₄⁺, suggesting the formed NH₃ indeed originated from the Nb₂O₅-NCF catalyzed NRR process.

Subsequently, we evaluated the durability of Nb₂O₅-NCF electrode for NRR. It was found that the yielded NH₃ amount during NRR initially rapidly increased. After 20 min, the increase trend of NH₃ yield became slow (Fig. S6 in Supporting information). This results in an initial increase then decrease of NH₃ yield rate, and the largest NH₃ yield rate was obtained to be 3.38×10⁻¹⁰ mol cm^-2 s^-1 with an FE of 11.2% at -0.4 V (vs. RHE) (Fig. 3a) when NRR time of 20 min. The above experimental results suggested that the NRR activity of Nb₂O₅-NCF decreased after 20 min of reaction, which deserves a further investigation. During electrocatalytic NRR, the Nb₂O₅ electrode was found to transform to NbO₂ species from XRD results (Fig. 3b), and electrochromism (EC) phenomenon was also observed, namely, the electrode surface color changed from white to dark with NRR time (Fig. 3c). This is mainly attributed to H⁺ ions insertion into the interlayer spacing of TT-phase Nb₂O₅, resulting in the crystalline phase transformation of Nb₂O₅ from TT-phase to T-phase, concurrently accompanied by part of Nb⁵⁺ reduction to Nb⁴⁺, thus the generation of EC phenomenon [35]. Seen from the XRD results of Nb₂O₅-NCF electrode with NRR time (Fig. 3b), compared to the pristine Nb₂O₅-NCF, the (100) diffraction peaks of Nb₂O₅-NCF with different NRR times apparently shift toward low angel direction in the XRD patterns, suggesting H⁺ ions insertion into interlayer spacing of Nb₂O₅ [36]. As we know, TT-phase Nb₂O₅ possesses more oxygen vacancies (V_o) than T-phase Nb₂O₅ [28]. These oxygen vacancies have been proven to be the catalytic active sites for N₂ adsorption, activation and hydrogenation [26, 27]. The N₂ firstly adsorbed on V_o sites of N₂O₅, and V_o-induced bands in Nb₂O₅ concurrently provided defect states for trapping excited electrons. Then these electrons were transferred and exchanged to the empty antibonding orbitals (p*) of the adsorbed N₂. These adsorbed N₂ (NN*) would be severely weakened and exhibited high activation due to the electron injection. It is conducive to hydrogenation for the N≡N bond to NH₃. During NRR, the crystalline phase transformation from TT-phase to T-phase leads to the decrease of V_o concentration in Nb₂O₅, thus the decreased NRR performance.

Fig. 3d shows the electron paramagnetic resonance (EPR) spectroscopy of Nb₂O₅-NCF with different NRR times. In comparison with the pristine Nb₂O₅-NCF, the V_o concentration in Nb₂O₅ obviously decreased with the NRR time, primarily owing to H⁺ ions occupying the V_o sites thus resulting in the reduced active sites for N₂ adsorption and activation.

After 20 min of NRR, we replaced fresh 0.1 mol/L Na₂SO₄ electrolyte and conducted the recycling experiments using the used Nb₂O₅-NCF electrode and obtained obviously decreased NRR performance (Fig. 3e, each recycling experiment of 20 min). The above results further confirmed that with NRR time, the formation of T-phase Nb₂O₅ results in the decrease of V_o concentration, thus the reduced NRR performance. How to regenerate the electrode with high NRR activity is therefore critically important. We carried out the cyclic voltammetry (CV) experiments in 0.1 mol/L Na₂SO₄ electrolyte in the potential range from −1.2 V to 1.2 V (vs. RHE). The results (Fig. S7 in Supporting information) demonstrated that the oxidation peak lies at ~ −0.5 V (vs. RHE), corresponding to the extraction of H⁺ ions from the interlayer spacing of Nb₂O₅. The cathodic current density has a slight increase with the cycling number for Nb₂O₅-NCF and almost unchanged until 25^th cycle, exhibiting an increased capacity for H⁺ ions insertion. The chronoamperometric measurements were performed by applying potentials of 0.6 and −0.6 V (vs. RHE) for 100 s. The comparison between the anodic and cathodic current intensities indicates a faster kinetic extraction than the insertion process, exhibiting an approximate bleaching time of 15 s while the colored time of 20 s (Fig. 4a). The reversible EC results mean that the used Nb₂O₅-NCF for NRR may be regenerated to recover its high NRR activity. For this, we tried to apply an anodic potential to realize this. As expected, when an anodic potential of 1.0 V (vs. RHE) was applied to the used Nb₂O₅-NCF electrode for 5 min, the T-phase Nb₂O₅ with part of Nb⁴⁺ can be readily transformed to TT-phase Nb₂O₅ (curve Ⅱ in Fig. 4b). This means that high NRR performance may be recovered by using the TT-phase Nb₂O₅ with more V_o sites. Utilizing such regeneration approach, the Nb₂O₅-NCF electrode displays superior recycling reproducibility with high NH₃ yield rate and current efficiency (Fig. 4c).

Furthermore, we found that low-temperature treatment approach (60 ℃ for 0.5 h) is also feasible for regeneration of the used Nb₂O₅-NCF electrode for high-performance NRR (curve Ⅲ in Fig. 4b). The recycling reproducibility with high NH₃ yield rate and current efficiency was shown in Fig. 4d. The EPR measurements of Nb₂O₅-NCF sample after 120 min NRR and subsequently low temperature annealing results also confirm this (Fig. S8 in Supporting information).

In summary, highly ordered Nb₂O₅ nanochannel film (Nb₂O₅-NCF) with rich V_o was fabricated on niobium foil substrate by a facile anodization method. As a free-standing electrode for NRR, the Nb₂O₅-NCF demonstrated high electrocatalytic activity toward NH₃ synthesis. However, the NRR activity of Nb₂O₅-NCF decreased when the reaction time was over 20 min, due to a crystalline phase transformation of Nb₂O₅ from TT-phase to T-phase, resulting in the decrease of V_o concentration in Nb₂O₅, thus the reduced active sites for N₂ adsorption and activation. However, the used Nb₂O₅-NCF with decreased NRR activity could be regenerated by applying an anodic potential or low-temperature thermal treatment approach, indicating superior recycling reproducibility with high NH₃ yield rate and current efficiency. This work provides an efficient solution to the inactivation issue of an electrocatalyst for high-performance NRR to NH₃.

The authors report no declarations of interest.

This work is financially supported by the National Key Research and Development Program of China (No. 2017YFA0207203) and the Natural Science Foundation of China (No. 51872292).

Supplementary material related to this article can befound, in the online version, at doi:https://doi.org/10.1016/j.cclet.2021.01.020.