Reaction-driven formation of anisotropic strains in FeTeSe nanosheets boosts low-concentration nitrate reduction to ammonia

Structural characterizations of catalysts

The formation of reaction-driven anisotropic strains in FeTeSe nanosheets involves three steps (Fig. 1a). Fe(Te_xSe_1−x) is a typical layered material that crystallizes in a tetragonal structure (P4/nmm), with adjacent layers connected through the van der Waals (vdW) interactions (Supplementary Fig. S1), enabling the facile synthesis of 2D Fe(Te_xSe_1−x) nanosheets from bulk crystals. First, bulk Fe_0.90(Te_0.60Se_0.40) crystals were prepared via a high-temperature solid-state reaction, showing high crystallinity and a closely stacked lamellar layered architecture (Fig. 1b, e). Subsequently, a chemical lithium intercalation method (Supplementary Fig. S2) was employed to synthesize 2D wrinkled FeTeSe nanosheets with an average thickness of 1c, f and Supplementary Figs. S3, 4). The tetragonal phase of the exfoliated nanosheets is well preserved, as evidenced by the XRD pattern which displays the characteristic (001), (101), (002), (003), and (004) peaks without noticeable impurity peaks (Fig. 1f). The relative intensity of the (101) peak to the (001) peak increases, suggesting that some exfoliated nanosheets are no longer flat on the substrate, thus exposing more edge sites beyond the (001) basal plane (Fig. 1f). This finding is further corroborated by the transmission electron microscopy (TEM) and scanning TEM (STEM) images (Fig. 1c and Supplementary Figs. S3, 4), which show both flat and wrinkled nanosheets. Inductively coupled plasma optical emission spectroscopy (ICP-OES) confirms the composition as Fe_0.92(Te_0.70Se_0.30) (Supplementary Table S1). The increased surface area of the exfoliated 2D nanosheets is demonstrated by the enhanced adsorption of oxygen-containing species during the water-involved exfoliation and subsequent material characterization processes²⁷. X-ray photoelectron spectroscopy (XPS) analysis (Supplementary Fig. S5) reveals increased oxygen content in nanosheets compared to their bulk counterpart, as evidenced by the increased peak areas for Fe³⁺, Te^x+, and Se^x+ relative to the Fe²⁺, Te²⁻, and Se²⁻ peaks that are characteristic of the FeTeSe lattice, along with the overall shift of all peaks to higher binding energies^28,29. This increased oxidation is further supported by the Fe K-edge X-ray absorption near edge structure (XANES) spectra, which show a positive shift in the Fe K absorption edge with a general shift of its white line toward higher energy in the 2D FeTeSe nanosheets (Supplementary Fig. S6a). Wavelet transform (WT, Fig. 1h), Fourier-transformed (FT) k²-weighted extended X-ray absorption fine structure (EXAFS) spectra (Supplementary Fig. S6b), and the quantitative fitting results (Supplementary Fig. S6c–e and Table S2) further demonstrate an obvious increase in the Fe−O coordination, along with a decrease in the Fe−Se/Te coordination, which is attributed to the generation of Se/Te vacancies and the increased specific surface area during the exfoliation process^30,31.

Then, in-situ anisotropic strains were induced in the 2D FeTeSe nanosheets by subjecting them to electrolysis in an electrolyte containing 1 M KOH and 10 mM KNO₃ under negative bias (Fig. 1d, g). The emergence of lattice strain is demonstrated by conducting operando synchrotron-based XRD using a homemade electrolysis cell at an X-ray wavelength of 0.41 Å (Fig. 2a–c and Supplementary Figs. S7–9). Two characteristic peaks of the FeTeSe lattice are focused, i.e., the (001) peak at ~3.86° and the (101) peak at ~7.33°, which represent the planar distances of the out-of-plane (c-axis) and in-plane (ac-axis), respectively (Fig. 2b, c). The XRD patterns in Fig. 2a-c depict the average patterns obtained from multiple scans conducted before and during NO₃⁻RR under the open-circuit potential (OCP) (Supplementary Figs. S8, 9), with chronoamperometry being performed in between under negative bias. Notably, an intriguing anisotropic strain phenomenon is observed. During electrolysis, the (001) peak shifts rightward to ~3.88°, while the (101) peak shifts leftward to ~7.31°, corresponding to a 0.52% compressive strain in the (001) plane and a 0.27% tensile strain in the (101) plane (Fig. 2b, c). Beyond these two peaks, other peaks corresponding to the (110) and (111) planes also exhibit slight leftward shifts (Fig. 2a), supporting the emergence of in-plane tensile strain within the FeTeSe nanosheets during electrolysis. Furthermore, ex-situ XRD using an X-ray wavelength of 1.54 Å was conducted on the catalyst-loaded carbon cloths taken out from two electrolyte conditions (i.e., 1 M KOH and 1 M KOH + 10 mM KNO₃) after chronoamperometry. The results are consistent with those obtained using operando synchrotron-based XRD, showing a slight left shift of the peak corresponding to the (101) plane and a notable right shift of the peak indexed to (002) (Supplementary Fig. S10). Interestingly, the peaks continue to shift after electrolysis in 1 M KOH followed by 1 M KOH + 10 mM KNO₃, suggesting that both OH⁻ and NO₃⁻ ions contribute to the strain formation (Supplementary Fig. S10b). Additionally, selected area electron diffraction (SAED) patterns in TEM characterization obtained from a bunch of nanosheets demonstrate increased lattice spacings after NO₃⁻RR for the (101), (110), and (111) planes (Fig. 2d, e), further confirming the emergence of in-plane tensile strain. Besides, the ex-situ XRD pattern of FeTeSe nanosheets after electrolysis reveals that the tetragonal phase is well preserved, showing no apparent crystalline impurities (Supplementary Fig. S11). The morphology and crystal structure of the FeTeSe nanosheets before and after NO₃⁻RR were also examined using electron microscopy. TEM images of the pristine (Fig. 2f) and post-electrolysis (Fig. 2n) catalysts indicate that the wrinkled nanosheet morphology is well maintained. Both the high-resolution TEM (HRTEM) images (Fig. 2g, o) reveal a clear atomic lattice, and the corresponding fast Fourier transform (FFT) patterns (Fig. 2h, p) display a characteristic set of diffraction spots along the tetragonal [001] zone axis, confirming the preservation of the tetragonal phase after electrolysis. Energy-dispersive X-ray spectroscopy (EDS) elemental mappings confirm the homogeneous distribution of Fe, Te, Se, and O elements throughout the entire FeTeSe nanosheets before (Fig. 2i–m) and after (Fig. 2q–u) electrolysis. These findings collectively indicate the formation of in-plane tensile strain and out-of-plane compressive strain in the 2D FeTeSe nanosheets during NO₃⁻RR, highlighting that the catalyst undergoes spontaneous strain engineering during electrolysis (Fig. 2v).

a Operando synchrotron-based XRD patterns of FeTeSe nanosheets before and during catalysis. OCP denotes open-circuit potential. b, c Enlarged regions of (a), showing the shift of (001) and (101) peaks. d, e SAED patterns of a bunch of FeTeSe nanosheets before (d) and after (e) catalysis. f, g TEM (f) and HRTEM (g) images of FeTeSe nanosheets before catalysis. h FFT pattern of (g). i–m STEM image (i) and the corresponding EDS elemental mappings (j–m) of FeTeSe nanosheets before catalysis. n, o TEM (n) and HRTEM (o) images of FeTeSe nanosheets after catalysis. p FFT pattern of (o). q–u STEM image (q) and the corresponding EDS elemental mappings (r–u) of FeTeSe nanosheets after catalysis. v Schematic illustration of the FeTeSe crystal structure along the b and c directions, where the strain directions are indicated by the orange arrows. Scale bars, d, e 2 1/nm; f 50 nm; g 1 nm; i 1 µm; n 100 nm; o 2 nm; q 500 nm. Source data are provided as a Source Data file.

Origins and formation mechanisms of strains during alkaline NO₃
⁻RR

The possible origins and formation mechanisms of the anisotropic strains developed during NO₃⁻RR were investigated through a series of operando and ex-situ measurements, alongside theoretical calculations. First, operando synchrotron-based XAFS measurements were performed using a homemade electrolysis cell to reveal changes in the valence state and coordination environment of Fe during NO₃⁻RR (Supplementary Fig. S12). The results are presented in the Fe K-edge XANES spectra (Fig. 3a–c and Supplementary Fig. S13) and the FT k²-weighted EXAFS spectra (Fig. 3d–f and Supplementary Fig. S14) for 2D FeTeSe nanosheets under varying reaction conditions: (i) 1 M KOH at OCP, (ii) 1 M KOH + 10 mM KNO₃ at OCP, and (iii-v) during NO₃⁻RR at −0.4 V vs. RHE for 20, 40, and 60 min.

a Operando Fe K-edge XANES spectra of FeTeSe nanosheets in 1 M KOH under OCP, in 1 M KOH + 10 mM KNO₃ under OCP, during NO₃⁻RR at −0.4 V for 20, 40, and 60 min. b, c Enlarged regions of (a). d, e FT k²-weighted Fe K-edge EXAFS spectra (d), and the corresponding contour map (e). The color bar represents intensity with unit of Å⁻³. f Changes of the normalized X-ray absorption coefficient under different conditions for Fe−O, Fe−Se, Fe−Te, and second-shell Fe−Te scattering paths. g Total X-ray scattering of the FeTeSe nanosheets before (red) and after (blue) electrolysis, as well as their difference pattern (green). h Scheme illustrating the in-situ formation of strained catalyst due to chalcogen vacancies and oxygen dopants in 2D FeTeSe nanosheets. i O 1 s spectra of the pristine FeTeSe nanosheets and those after catalysis in 1 M KOH under OCP, in 1 M KOH + 10 mM KNO₃ under OCP, in 1 M KOH + 10 mM KNO₃ at −0.55 V for 1, 2, and 4 h. OCP denotes open-circuit potential. Source data are provided as a Source Data file.

As shown in Fig. 3a–c, even without applying any potential bias (i.e., under OCP), the addition of 10 mM KNO₃ into 1 M KOH results in a shift of the Fe K absorption edge toward higher energy, accompanied by an increase in the intensity of the white line peak. Subsequently, when a cathodic potential of −0.4 V vs. RHE is applied for 20, 40, and 60 min, the absorption edge continues to shift toward higher energies (Fig. 3a–c), indicating a distinct increase in the Fe valence state in FeTeSe nanosheets as the NO₃⁻RR progresses over time. Notably, the positions of the absorption edges for samples measured under operando conditions still lie between those of the ex-situ-measured standard FeSe and Fe₂O₃ samples (Supplementary Fig. S13), suggesting that the average valence state of the Fe cations increases, while remaining within the range of +2 and +3 (i.e., Fe^δ+, where 2 3⁻RR at a cathodic potential of −0.4 V indicates a continuous incorporation of oxygen species from the electrolyte (e.g., OH⁻ and NO₃⁻), which possess a higher electronegativity (3.44 for O) than Se (2.55) and Te (2.10) chalcogens³². The interplay between the applied negative bias, which tends to lower the valence, and the favorable incorporation of the electronegative oxygen species, which tends to raise the valence, results in a net increase in the nominal Fe valence state. This finding challenges the stereotype that metal cations in transition metal chalcogenides tend to be reduced under cathodic overpotentials³³. Only a few reports document unusual increases in metal valence states during cathodic reactions, but all at relatively positive potentials (e.g., −0.1 to 0 V)^21,34,35.

The changes in the coordination environment of Fe are revealed by the Fe FT k²-weighted EXAFS spectra. Upon the addition of NO₃⁻, the amplitude of the shell peak at ~1.44 Å, corresponding to the Fe−O scattering, shows a significant increase, while the peak intensities associated with Fe−Se at ~1.87 Å and Fe−Te at ~2.30 Å are reduced (Fig. 3d-f and Supplementary Fig. S14a). This trend continues during the NO₃⁻RR at −0.4 V over a duration of 60 min, demonstrating an increase in the Fe−O coordination at the expense of Fe−Se/Te coordination (Fig. 3d-f and Supplementary Fig. S14b). The amplitude of the second-shell Fe−Fe scattering peak at ~2.76 Å also increases (Fig. 3e, f), further corroborating the loss of Se/Te chalcogens that likely reduces the electronic shielding effect. Additionally, an obvious loss of Se than Te during electrolysis is also noticed (Fig. 3e, f). As further confirmed by the ICP-OES analysis of the post-reaction electrolyte (Supplementary Table S3), Se and Te anions leach by 15.3% and 8.6% during electrolysis, respectively, resulting in the formation of Se/Te vacancies, which in turn provide abundant sites for the adsorption and diffusion of oxygen species in thin FeTeSe nanosheets³⁶, thus promoting their incorporation. Furthermore, the Fe−O peaks exhibit a high-R shift from ~1.44 Å to ~1.47 Å, while the Fe−Se peaks display a low-R shift from ~1.87 Å to ~1.84 Å, indicating a distorted coordination environment for the Fe atoms (Fig. 3d)^35,37.

The findings from XAFS measurements were further corroborated by total X-ray scattering analysis of the 2D FeTeSe nanosheets before (red line) and after (blue line) NO₃⁻RR, along with their difference pattern (green line), offering comprehensive structural information at the atomic scale (Fig. 3g). For the 2D FeTeSe nanosheets before catalysis, the first peak top is located at ~1.94 Å, corresponding to the Fe−O pair. The second through fifth peak tops can be assigned as the collectivity of Fe−Fe/Se/Te or Se/Te−Se/Te bonding. The difference pattern reveals that the relative intensity of the first peak increases, while the intensities of the second through fifth peaks decrease. Moreover, the scattering pattern broadens after catalysis, and new peak tops emerge between the original peak tops in the difference pattern. This suggests a wider range of interatomic distances, which can be attributed to several factors, including the positional disorder of atoms³⁸, the loss of Se/Te atoms that alters the relative bonding ratios of Fe/Te/Se−Fe/Te/Se, and the formation of new Fe−O bonds due to the replacement of Se/Te with O atoms. These observations correlate with the formation of Fe−O(H) species indicated by the operando XAFS results (Fig. 3a–f) and the dissolution of Se (by 15.3%) and Te (by 8.6%) as determined by the ICP-OES (Supplementary Table S3).

Based on the aforementioned synchrotron-based X-ray techniques, the electrochemistry-driven formation of chalcogen vacancies and incorporation of oxygen species could be linked to strains observed in the operando XRD patterns (Fig. 2a–c). DFT calculations were resorted to investigating the formation mechanism of strains (Supplementary Figs. S15–17). We first calculated the formation energies of chalcogen vacancies in FeTeSe, which show similar values for the formation of Se and Te vacancies (Supplementary Fig. S15). The chalcogen vacancies tend to be repaired by oxygen species from the electrolyte due to the stronger Fe−O bond compared to Fe−chalcogen bonds, rendering the coordination with oxygen species thermodynamically favorable^39,40. Then, the formation energies of various oxygen species (i.e., O, OH, and NO₃) at both substitutional and interstitial positions of the FeTeSe lattice were calculated (Supplementary Fig. S16). The doping of OH at both substitutional ( −0.73 eV) and interstitial ( −0.16 eV) positions is thermodynamically favorable, suggesting that OH⁻ ions from the alkaline electrolyte contribute to the strain formation via doping. On the other hand, when the NO₃⁻ molecule enters the interstitial position, it spontaneously dissociates to form O-doped FeTeSe, inducing the formation of a strained lattice³⁹. Moreover, theoretical investigations demonstrate that the type, concentration, and position of chalcogen vacancies and oxygen dopants can distort the original FeTeSe lattice in different directions and to different degrees (Supplementary Fig. S17). Therefore, combining the findings from operando measurements and DFT calculations, we conclude that the lattice strain arises from the formation of chalcogen vacancies in the thin 2D FeTeSe nanosheets, which facilitate the filling and diffusion of oxygen species from the electrolyte into the lattice vacancy sites³⁶, creating the strained 2D FeTeSe-O(H) (Fig. 3h). The observed anisotropic strains likely result from a combined effect of Se/Te vacancies, interstitial/substitutional O(H) dopants, and distortion caused by *NO₃ adsorption followed by dissociation (Supplementary Fig. S16, 17). In contrast, the electrochemistry-driven formation of strain is unobvious in FeTe nanosheets without Se doping (Supplementary Fig. S18) and FeTeSe bulk crystals with huge material volume (Supplementary Fig. S19). This indicates that both the presence of Se and 2D morphology contribute to the reaction-driven strain development. On one hand, the favorable leaching of Se promotes the creation of chalcogen vacancies during electrolysis, which in turn facilitates the backfill of oxygen, thereby promoting strain development. On the other hand, the high specific surface area and the capacity of 2D materials to sustain significantly larger mechanical strains compared to their bulk counterparts⁴¹ also promote the in-situ strain formation.

The surface oxygen environment of 2D FeTeSe nanosheets was investigated using XPS after reactions under various electrochemical conditions: (i) pristine catalyst, (ii) 1 M KOH at OCP, (iii) 1 M KOH + 10 mM KNO₃ at OCP, and (iv-vi) after NO₃⁻RR at −0.55 V for 1, 2, and 4 h (Fig. 3i and Supplementary Figs. S20–24). All samples display a broader O 1 s peak after immersion in the aqueous electrolyte (Fig. 3i). Deconvolution was performed to identify the different sources of oxygen: peaks at ~530.0 eV, ~531.5 eV, and ~533.0 eV are indexed to M − O, M − OH, and adsorbed water, respectively⁴². The relative percentages of these oxygen species were calculated and shown in Supplementary Fig. S22. The concentration of M − O slightly increases during NO₃⁻RR, while the percentages of M − OH (adsorbed hydroxyl) and adsorbed water increase and reach their maxima when the catalyst is immersed in 1 M KOH electrolyte. This phenomenon occurs because in the absence of KNO₃, the adsorption of H₂O and OH⁻ dominates. When KNO₃ is introduced, its preferential adsorption on the catalyst surface reduces the amounts of surface-adsorbed H₂O and OH⁻, thereby facilitating the NO₃⁻RR. Furthermore, the dominant presence of M − OH based on the XPS analysis also indicates a significant abundance of surface-adsorbed hydroxyl groups (Supplementary Fig. S22), which form as a result of rich chalcogen vacancies, leading to the creation of a di-anionic surface (i.e., chalcogen vacancies and surface hydroxyl groups).

Electrochemical performance of catalysts

The catalytic performance of 2D FeTeSe nanosheets with in-situ developed strains toward NO₃⁻RR-to-NH₃ was evaluated and compared to two counterparts: FeTeSe bulk crystals and FeTe nanosheets without Se doping. Targeting a NO₃⁻ concentration relevant to industrial wastewater (10 mM)², the electrochemical NO₃⁻RR was conducted under ambient conditions in a standard three-electrode H-type cell (Fig. 4). The desired product, NH₃, was quantified using the spectroscopic indophenol blue method (Supplementary Fig. S25). The main gaseous by-product, H₂, was quantified via the online gas chromatography, while the main liquid by-product, nitrite (NO₂⁻) (Supplementary Fig. S26a–c), and the reactant, NO₃⁻ (Supplementary Fig. S26d–f), were quantified via the ion chromatography. All potentials are referenced to the RHE. The NO₃⁻RR activity of all catalysts was first assessed via linear sweep voltammetry (LSV) tests (Fig. 4a). All catalysts show a notable increase in cathodic current density upon the addition of 10 mM KNO₃ into 1 M KOH electrolyte, with FeTeSe nanosheets demonstrating the highest current density and the most positive onset potential, suggesting the more effective reduction of NO₃⁻ ions. Product quantifications were performed using chronoamperometry over the potential range of −0.35 to −0.75 V vs. RHE (Fig. 4b-e and Supplementary Figs. S27, 28). The NH₃ FE for FeTeSe nanosheets presents a volcano-type curve and peaks at 98.7 ± 2.2% at −0.55 V, with a corresponding NH₃ yield rate of 21.16 ± 1.49 mg h⁻¹ mg_cat⁻¹ (Fig. 4b). With more negative overpotentials, the NH₃ FE starts to decline due to the stronger HER competition (Fig. 4c). The NO₂⁻ FE reaches a maximum of 18.6 ± 9.8% at −0.35 V, and continuously decreases to 4c). The NH₃ yield rate for FeTeSe nanosheets increases monotonically with more negative potentials, peaking at 42.14 ± 2.06 mg h⁻¹ mg_cat⁻¹ at −0.75 V (Fig. 4b). In comparison, FeTeSe bulk crystals (maximum NH₃ FE of 83.5 ± 2.9% with the corresponding yield rate of 9.97 ± 0.91 mg h⁻¹ mg_cat⁻¹), FeTe nanosheets (maximum NH₃ FE of 83.5 ± 4.2% with the corresponding yield rate of 8.18 ± 1.28 mg h⁻¹ mg_cat⁻¹), and commercial Fe₂O₃ (maximum NH₃ FE of 81.2% with the corresponding yield rate of 15.90 mg h⁻¹ mg_cat⁻¹) exhibit inferior catalytic activities (Fig. 4d, e and Supplementary Fig. S29). Additionally, the performance of FeTeSe nanosheets exceeds that of most previously reported electrocatalysts for low-concentration ( ~ 10 mM) NO₃⁻RR, excelling in both NH₃ FE and yield rate (Supplementary Fig. S30 and Table S4).

a LSV curves of FeTeSe nanosheets, FeTeSe bulk crystals, and FeTe nanosheets with and without the addition of 10 mM KNO₃ in 1 M KOH under Ar at a scan rate of 10 mV s⁻¹. b NH₃ yield rates and FEs at different potentials over FeTeSe nanosheets. c FEs of NH₃, NO₂⁻, and H₂ at different potentials over FeTeSe nanosheets. d, e Comparisons of FeTeSe nanosheets with FeTeSe bulk and FeTe nanosheets toward electro-reduction of 10 mM NO₃⁻ in terms of the NH₃ FE (d) and yield rate (e) at different potentials. f NH₃ yield rates and FEs at −0.55 V under different concentrations of KNO₃ over FeTeSe nanosheets. g NH₃ FEs and yield rates for 10 consecutive cycles. h Concentrations of NO₃⁻, NO₂⁻, and NH₃ with increasing time, as well as the NH₃ FEs at different time points. Experimental conditions: A catalyst mass loading of 0.1 mg cm⁻² on the Toray H-090 carbon paper (geometry area of 1×1 cm²) in a H-cell with Nafion 117 membrane separating two chambers under ambient conditions for nitrate electrolysis. Each chamber has 30 mL of Ar-saturated 1 M KOH and 10 mM KNO₃ (pH = 13.98 ± 0.07). All potentials are not iR corrected. The solution resistance in 1 M KOH + 10 mM KNO₃, determined by potentiostatic electrochemical impedance spectroscopy, was 2.53 ± 0.03, 2.39 ± 0.08, and 2.48 ± 0.27 for FeTeSe nanosheets, FeTeSe bulk, and FeTe nanosheets, respectively. Error bars denote the standard deviations calculated from three independent measurements. Source data are provided as a Source Data file.

The catalytic performance of the best-performing 2D FeTeSe nanosheets for NO₃⁻RR was further evaluated to assess its adaptability to various wastewater sources. Given the wide range of NO₃⁻ concentrations in real-world scenarios², the performance of FeTeSe nanosheets was examined across a NO₃⁻ concentration spectrum of 1 to 100 mM (Fig. 4f and Supplementary Fig. S31). As KNO₃ concentration increases, the current density also rises (Supplementary Fig. S31), indicating improved NO₃⁻RR activity with greater reactant availability. Specifically, the FeTeSe nanosheets deliver impressive NH₃ FEs of >90% within the KNO₃ concentration range of 10-100 mM at −0.55 V, and the corresponding NH₃ yield rate monotonically increases with rising NO₃⁻ concentration (Fig. 4f). The stability of the FeTeSe nanosheets was evaluated through long-term chronoamperometric test (Supplementary Fig. S32) and 10 consecutive cycle tests at −0.55 V (Fig. 4g and Supplementary Fig. S33). While the current density, NH₃ yield rate, and NH₃ FE fluctuate slightly in each cycle, the average results remain stable, confirming its remarkable stability. Additionally, the effectiveness of FeTeSe nanosheets for low-concentration NO₃⁻ remediation was evaluated by performing a batch conversion test using an initial concentration of 10 mM. Figure 4h illustrates that as the electrolysis time increases, the NO₃⁻-N concentration gradually decreases, along with the gradual increase in NH₃-N concentration, while the NO₂⁻-N concentration remains negligible. The catalyst reduces NO₃⁻ levels to below the World Health Organization’s recommended maximum of 50 ppm for drinking water⁴³ within 5 h, showing a high conversion rate of 93.4%.

Control experiments were performed to confirm the N source in the electrochemically produced NH₃. NH₃ is almost undetectable in electrolytes with the absence of KNO₃ or potential bias, while the amount is trivial when pure carbon paper without the loading of FeTeSe catalyst is used as the working electrode (Supplementary Fig. S34a). Moreover, isotope labelling coupled with ¹H nuclear magnetic resonance (NMR) measurements demonstrate characteristic triple and double peaks, corresponding to ¹⁴NH₃ and ¹⁵NH₃, when K¹⁴NO₃ and K¹⁵NO₃ are used, respectively (Supplementary Fig. S34b). These findings verify the generated NH₃ derives from the FeTeSe nanosheet-catalyzed NO₃⁻RR process under negative bias.

Mechanism studies of the strain effect on NO₃
⁻RR

The reaction pathway of the NO₃⁻RR-to-NH₃ process was investigated using in-situ ATR-IR spectroscopy to capture potential reaction intermediates (Supplementary Fig. S35). Figure 5a presents the in-situ ATR-IR spectra collected over the FeTeSe nanosheets in an electrolyte containing 1 M KOH and 10 mM KNO₃, with IR signal acquisition ranging from 1050 to 3600 cm⁻¹ during a scan from OCP to −0.75 V vs. RHE. The spectra display characteristic peaks associated with the adsorbed NO_x, water, and NH₃. The upward and downward bands reflect the formation and consumption of the corresponding intermediates, respectively. Notably, a pair of downward peaks at ~1520 and 1340 cm⁻¹ represent the asymmetric and symmetric vibration modes of the absorbed NO₂ intermediate, respectively, while the downward peak at ~1120 cm⁻¹ is attributed to the N − O bond or Si−O from the substrate⁴⁴. The intensities of these downward peaks increase with more negative potentials, indicating the consumption of NO_x species during NO₃⁻RR. The emergence of upward bands corresponding to N − H bending and −NH₂ wagging modes at ~3220 and 1380 cm⁻¹, respectively, suggests the formation of NH₃ (ref. ^44,45). Additionally, the upward bands at ~3380 and 1640 cm⁻¹ arise from the O − H stretching and bending modes of water, respectively, implying water adsorption on the catalyst surface⁴⁴. These upward bands concurrently become more prominent with more negative potentials, suggesting intensified water adsorption to supply protons for the hydrogenation of N-involving intermediates. These findings demonstrate that the *NO_x deoxygenation process is followed by a series of protonation steps during the NO₃⁻RR-to-NH₃ process, with water dissociation being essential to provide adequate protons for the hydrogenation of N-involving intermediates.

a In-situ ATR-IR spectra of FeTeSe nanosheets under various potentials. OCP denotes open-circuit potential. b, c Calculated PDOS of strained FeTeSe-O/unstrained FeTeSe with NO₃^– (b) and H₂O (c). d, e Gibbs free energy diagrams via the minimum energy pathway during NO₃^–RR-to-NH₃ (d) and HER (e) over the strained FeTeSe-O (red) and unstrained FeTeSe (blue) surfaces. Source data are provided as a Source Data file.

Then, the effects of strain on both the NO₃⁻RR-to-NH₃ and water dissociation processes were explored theoretically using DFT calculations. Structural models of both the unstrained surface and strained surface with the presence of chalcogen vacancies and oxygen dopants were constructed (Supplementary Fig. S36). The projected density of states (PDOS) for *NO₃ (Fig. 5b) and *H₂O (Fig. 5c) indicate that the Fe 2p and O_ads 2p orbitals show greater overlap below the Fermi level on the strained FeTeSe-O surface, suggesting stronger binding between *NO₃/*H₂O and the Fe sites. This is further supported by the calculated binding free energies of NO₃⁻ and H₂O on both surfaces, which show more negative values of −0.67 eV and −0.50 eV for the adsorption of NO₃⁻ and H₂O, respectively, on the strained FeTeSe-O surface, as compared to those on the unstrained FeTeSe surface ( −0.39 eV for NO₃⁻ and 0.18 eV for H₂O) (Supplementary Fig. S37). Subsequently, the free energy diagrams based on the Gibbs free energy changes (ΔG) and the corresponding adsorption configurations for all possible pathways from NO₃⁻ to NH₃ on both the strained and unstrained surfaces were examined (Supplementary Figs. S38, 39a, b). Our calculations show that the minimum energy pathway begins with the adsorption of NO₃⁻, followed by its reduction to *NOH, then to *N, and the subsequent hydrogenation of *N to produce NH₃ on both surfaces (Fig. 5d and Supplementary Fig. S39a, b). The conversion from *NH₂ to *NH₃ is calculated to be the potential-determining step (PDS) for the strained FeTeSe-O surface, with a smaller uphill ΔG of 0.18 eV. In contrast, for the unstrained FeTeSe surface, the PDS is the conversion from *NH to *NH₂, which has a larger ΔG of 0.40 eV (Fig. 5d). These results underscore the critical role of the strained surface in facilitating the adsorption of reactants (NO₃⁻ and H₂O) and lowering the energy barrier of the PDS, thereby enhancing the NO₃⁻RR activity.

Additionally, considering the critical role of water dissociation in supplying the necessary protons for the hydrogenation of N-involving intermediates during the alkaline NO₃⁻RR, we also calculated the free energy diagrams for the alkaline HER on both the strained and unstrained catalyst surfaces (Fig. 5e and Supplementary Fig. S39c, d). For both the strained FeTeSe-O and unstrained FeTeSe surfaces, the PDS of the alkaline HER is the conversion from *H to H₂, with an energy barrier of 0.42 eV and 0.51 eV, respectively, indicating the strong inhibition of H − H dimerization to H₂. Meanwhile, on the strained FeTeSe-O surface, the binding free energy of H₂O ( −0.50 eV) is less negative than that of NO₃^– ( −0.67 eV) (Supplementary Fig. S37), and the energy barrier for the PDS in alkaline HER (0.42 eV) is higher than that in NO₃^–RR (0.18 eV) (Fig. 5d, e). These findings suggest that, on the strained FeTeSe-O surface, the adsorption and activation of NO₃^– are energetically more favorable than HER, while the favorable dissociation of water into protons ensures a sufficient supply of protons for the hydrogenation process. In contrast, for the unstrained FeTeSe, the unfavorable H₂O adsorption (0.18 eV, Supplementary Fig. S37) indicates its poorer ability to provide protons, which limits the hydrogenation process.

Other factors also contribute to the superior activity of 2D FeTeSe nanosheets, in addition to (i) the strain effect (Fig. 5). (ii) The presence of abundant hydroxyl groups on the surface, resulting from the rich chalcogen vacancies (Fig. 3i), facilitates the adsorption of NO₃⁻ by forming hydrogen bonds, as evidenced by the more negative adsorption energy of NO₃⁻ on the catalyst surface with hydroxyl groups (Supplementary Fig. S40), thereby creating a reactant-friendly interface that helps drag NO₃⁻ ions more effectively to the catalyst surface. Similar phenomena have been observed in metal (oxyhydr)oxide catalysts, where surface hydroxyl groups enhance the adsorption of NO₃⁻ (ref. ⁴⁶) and CO₂ (ref. ⁴⁷) through hydrogen bonding. (iii) Upon intercalation and exfoliation, 2D FeTeSe nanosheets exhibit an increased surface area with more exposed active sites, as indicated by a larger electrochemical active surface area (ECSA) compared to the bulk counterpart (Supplementary Fig. S41a–c). After normalizing the LSV curves and NH₃ partial current densities over ECSA, the FeTeSe nanosheets still exhibit higher NO₃⁻RR-to-NH₃ activity (Supplementary Fig. S41d), indicating that its superior performance is attributable not only to their larger surface area but also to their enhanced intrinsic activity. (iv) The charge transfer kinetics of the 2D FeTeSe nanosheets improve after exfoliation, as evidenced by the smaller charge transfer resistance compared to those of FeTeSe bulk crystals and FeTe nanosheets in the fitted electrochemical impedance spectroscopy (EIS) plots (Supplementary Fig. S42). Overall, a series of electrochemical studies, in-situ measurement, and theoretical investigations suggest that the superior catalytic activity of 2D FeTeSe nanosheets arises from the synergistic effects of strained Fe sites exposed due to the formation of Se/Te vacancies, the presence of surface hydroxyl groups that create a reactant-friendly interface via hydrogen bonding, and the increased surface area due to their 2D structure. Collectively, these factors enhance the adsorption and activation of NO₃⁻, lower the energy barrier of the PDS, and promote the water dissociation process to supply protons while suppressing the HER side reaction, thereby boosting the low-concentration NO₃⁻RR-to-NH₃ activity of FeTeSe nanosheets.