Typically, a stoichiometric ratio of the precursors, molybdenum chloride (MoCl₅) and urea (CO(NH₂)₂) were mixed together with 10 ml ethanol and aged on a hot plate at 80°C for 2–3 h. The as-prepared samples were further pyrolyzed at 800°C for 3 h under N₂ flow. Further, the samples were prepared by varying the urea-metal ratio to 2:1 and 3:1, termed Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C, respectively.

The crystal structure and phase formation of the samples were examined by powder X-ray diffraction (XRD; Rint 1000, Rigaku, Japan) with Cu Kα radiation (λ = 1.5418 Å). X-ray photoelectron spectroscopy (XPS; K-ALPHA+, Thermo Scientific) was used to determine the oxidation states of the elements in the electrocatalyst. The morphology and microstructure of the electrocatalyst was investigated by scanning electron microscopy (SEM; JSM-7500F, JEOL, Japan) and transmission electron microscopy (TEM; JEM-2100F, JEOL, Japan), respectively.

The slurry prepared using PVDF, carbon black and prepared electrocatalysts in the ratio 1:1:8 was loaded on carbon cloth (1 × 1 cm²) and dried in vacuum oven for further electrochemical studies. The electrochemical NRR studies were carried out in a three-electrode cell configuration using the VMP3-Biologic multi-channel potentiostat under 0.1 M KOH electrolyte with and without N₂ purging. Graphite rod and Hg/HgO were used as the counter and reference electrodes, respectively. The Hg/HgO potential (E_(Hg/HgO)) was converted into reversible hydrogen electrode (RHE) potential (E_RHE) using Equation 1. E R H E = E H g / H g O + 0.118   V + 0.0591   x   p H (1)

The ammonia content in the electrolyte after the chronoamperometry (CA) tests for 2 h at various applied potentials were measured using the colorimetric method. 5 ml of the collected electrolyte after CA tests was sampled with colouring agents and catalysts using the standard ammonia test kits. Then the intensity of the coloured solution was measured using the UV-vis spectrophotometer. The intensity of the respective UV curves at 680 nm were compared to the standard calibration graphs of ammonia to estimate the concentration of ammonia present in the solution. The calibration graphs were derived from UV curves prepared by various known concentration of standard NH₄Cl salt. The linear curves from the UV plots yield the calibration graphs for ammonia. Hence, the concentration of ammonia can be estimated from the UV curves for samples after CA tests and calibration graphs (Liu et al., 2020).

The ammonia yield rate and Faradaic efficiency (FE) of the prepared electrocatalysts can be estimated using Equation 2 and Equation 3, respectively (Liu et al., 2020). Y i e l d   r a t e   μ g   h − 1 c m − 2 = N H 3   μ g m l × V m l t   h × A   c m − 2 (2) F E   % = 3 × Y i e l d   μ g   h − 1 c m − 2 × F   C   m o l − 1 17.03   g m o l × j   m A   c m − 2 (3) where, [NH₃] represent the concentration (μg/ml) of the ammonia present in the corresponding electrolyte, V is the volume of the electrolyte (ml), t is the time of electrolysis (hours), A is the area of the electrode (cm^-2), F is the Faradaic constant (96485 C mol^-1), and j is the current density at CA tests (mA cm^-2), respectively.

Mo₂C-Mo₂N_0.92 is synthesized by the simple urea glass route (Figure 1A). Initially, the MoCl₅ precursors are mixed with alcohols which primarily forms Mo-ethoxide (Mo(OC₂H₅)₅) and the Cl¯ ions are released as HCl (Giordano et al., 2008). Further, mixing urea (CO(NH₂)₂) with the above solution replaces the ethoxide groups in Mo(OC₂H₅)₅ due to the high solubility of urea under the presence of metal ions (Equation 4). Urea is used as both C and N precursors for the formation of Mo₂C and Mo₂N, respectively. The C=O bond cleavage in urea replaces the ethoxide molecule in Mo(OC₂H₅)_5, and the urea moiety coordinates with the Mo⁵⁺ atom forming metal-urea complexes, Mo-(OCN₂H₄)₅ (Equation 5). (Sardar et al., 2005) After heat treatment, the metal-urea complex is reduced into the desired Mo₂N-Mo₂C nanocomposites (Equation 6). By varying the urea-to-metal ratio (R), dominant phases of Mo₂N and Mo₂C can be achieved, respectively. Thereby, a lower urea content in the urea-metal ratio results in the dominant phases of Mo₂C, and high urea content favors Mo₂N formation respectively.

The proposed reaction mechanisms are elucidated as follows. M o C l 5 + C 2 H 5 O H   →   M o O C 2 H 5 5 + H C l (4) M o O C 2 H 5 5 + C O N H 2 2   →   M o − O C N 2 H 4 5 (5) M o 5 + + R   O C N 2 H 4   →   M o 2 C − M o 2 N + N H 3 + H C N + C O 2 (6)

The structural analysis of the prepared samples was evaluated by X-ray diffraction (XRD) analysis. XRD patterns obtained for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C display crystalline peaks for the prepared samples (Figure 1B). The obtained peaks are well-matched to the respective phases of Mo₂C (JCPDS: 98–007–7158) and Mo₂N_0.92 (JCPDS: 98–015–8843). Similarly, XRD patterns for the bare Mo₂C samples reveal the respective well-matched phases. The simultaneous existence of the XRD patterns of the two phases without any peak shift or peak broadening can ascribe to the heterostructure formation. XRD patterns for Mo₂C-Mo₂N_0.92 prepared with 2:1 ratio reveals the dominant phases of Mo₂C with relatively smaller presence of Mo₂N_0.92 phases. Similarly, XRD patterns for Mo₂N_0.92-Mo₂C prepared with 3:1 urea to metal ratio displays the presence of dominant Mo₂N_0.92 phases with feeble Mo₂C phases, respectively. Thereby, a low urea content favors Mo₂C formation and a high urea content forms Mo₂N_0.92, respectively. Hence, Mo₂N_0.92 and Mo₂C phases exist together in a single system exhibiting differences in relative intensity of the XRD patterns without any peak shift or peak broadening. The relative intensity of the XRD patterns for the samples prepared at different urea to metal ratio explain the dominant phases of the Mo₂C and Mo₂N_0.92 moieties in the prepared electrocatalysts. Thus, the urea-metal ratio is the phase-determining factor for Mo₂N and Mo₂C.

The chemical states of the prepared Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C samples were evaluated by the X-ray photoelectron spectroscopy (XPS) analysis. The Mo 3d spectra in Figure 1C can be deconvoluted into two peaks at 232.8 eV and 236 eV for Mo 3d_5/2 and Mo 3d_3/2, respectively. The small peak arising at the lower binding energy around 228.8 eV corresponds to the metallic Mo species for both Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C samples. The C 1s spectra in Figure 1D display three deconvoluted peaks at 284.6 eV, 286.8 eV, and 289 eV corresponding to the sp-carbon (C=C) bonds (Yue et al., 2021), N doped carbon (C-N), and feeble C=O signals due to incomplete decomposition of urea (CO(NH₂)₂), respectively. The deconvoluted N 1s spectra in Figure 1E display major peaks at 397.4 eV, 398.7 eV, and 400 eV, corresponding to pyridinic-N, pyrrolic-N, and graphitic-N, respectively. Besides, the feeble peak around 394.8 eV suggests the presence of Mo-N bonds in the prepared samples (Zhang et al., 2022). Evidently, the Mo₂N_0.95-Mo₂C samples reveal relatively higher intensity for Mo-N bonds compared to Mo₂C-Mo₂N_0.92, which validates the dominant presence of Mo₂N_0.92 in the prepared sample.

The morphological and microstructure studies were observed using scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM) analyses. SEM images for the prepared Mo₂C-Mo₂N_0.92 samples reveal the uniform distributions of the nanocomposites anchored to a substrate. SEM images of Mo₂C-Mo₂N_0.92 sample reveal the average sizes of the nanocomposites around ∼50 nm (Figure 2A). HRTEM analysis reveals the lattice spacing of about 0.26 nm for Mo₂C-Mo₂N_0.92 nanocomposites corresponding to (010) lattice planes of Mo₂C (Figure 2B). The inset of Figure 2B shows the FFT patterns for the Mo₂C-Mo₂N_0.92 sample, which reveal the crystalline phases of Mo₂C. Similarly, the SEM images in Figure 2C reveal the average sizes of the nanocomposites around ∼110 nm for Mo₂N_0.92-Mo₂C. The HRTEM analysis for Mo₂N_0.92-Mo₂C sample reveals the lattice spacing of about 0.23 nm corresponding to (101) lattice planes of Mo₂C (Figure 2D). Furthermore, the intensity-line profiles for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C samples, as shown in the inset of Figures 2B, D, display an average interplanar distance of nearly 0.26 nm for Mo₂C-Mo₂N_0.92 and 0.23 nm for Mo₂N_0.92-Mo₂C, respectively.

Electrochemical NRR performances of the prepared electrocatalysts were performed under constant N₂ purging. Linear sweep voltammetry (LSV) performed under Ar- and N₂-saturated electrolyte reveals a lower onset with higher current density under N₂ saturation for Mo₂C-Mo₂N_0.92 electrocatalyst due to the occurrence of NRR (Figure 3A). NRR performances were quantified by the chronoamperometric (CA) measurements performed for 2 h under N₂-saturation at various applied potentials. The electrolytes were sampled for ammonia quantification by UV-colorimetric analysis using a commercial ammonia test kit. The ammonia calibration graphs were derived using known ammonia concentrations with a UV-vis spectrometer (Supplementary Figure S1). UV-vis spectra for the sampled electrolytes under various potentials were observed to display varied intensities indicating the concentration of ammonia. The ammonia concentration was calculated from the calibration graphs, and the ammonia yield rate and FE from Equation 2 and Equation 3, respectively. Figures 3B, C represent the CA curves and UV plots for Mo₂C-Mo₂N_0.92 electrocatalyst at various applied potentials, respectively. Similarly, the LSV curves for Mo₂N_0.92-Mo₂C electrocatalyst reveal a higher current density under N₂-saturation, indicating NRR (Figure 3D). Figures 3E, F represents the CA plots and corresponding UV curves for the prepared Mo₂N_0.92-Mo₂C electrocatalyst for ammonia quantification.

Figure 4A represents ammonia yield rates and FE of Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C electrocatalysts for various applied potentials, respectively. A maximum ammonia yield of about 9.6 μg h^-1 cm^-2 and 6.61 μg h^-1 cm^-2 at −0.3 V vs. RHE was observed for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C electrocatalyst, respectively. Consequently, a maximum FE of about 10.15% and 4.21% at −0.1 V vs. RHE was observed for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C electrocatalyst, respectively. A comparatively high current density was observed under N₂ flow but a lower FE for ammonia of about 4.21% was observed at −0.1 V vs. RHE for Mo₂N_0.92-Mo₂C electrocatalyst. HER is kinetically favorable due to the comparatively low potential required for H₂ evolution than to dissociate the stable triple bonds of N₂. Thereby, the low FE for ammonia can be attributed to the undesired HER with increasing negative potentials. Moreover, the particle sizes of Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C from SEM images were observed to be 50 nm and 110 nm, respectively. It is suggested that a smaller size of the particles enables higher surface area and high exposure of active sites to drive the required electrochemical reactions. Consequently, Mo₂C-Mo₂N_0.92 electrocatalysts with a particle size of about 50 nm exhibits a higher NRR performance compared to Mo₂N_0.92-Mo₂C electrocatalysts with a higher particle size of about 110 nm. Therefore, particle sizes play an important role in determining the effective NRR performances for Mo₂C-Mo₂N_0.92 electrocatalysts.

Besides, a comparatively lower ammonia yield and FE was observed for Mo₂N_0.92-Mo₂C electrocatalyst, which can be due to the less activity of Mo₂N_0.92 phases for N₂ reduction. The NRR performances for bare Mo₂C and bare Ni foam were analyzed, which reveal a maximum ammonia yield of about 4.53 μg h^-1 cm^-2 and 0.07 μg h^-1 cm^-2 and FE of about 0.14% and 0.03%, respectively (Supplementary Figure S2). This poor NRR performance for bare Mo₂C and bare Ni foam substrate indicates the synergistic activity of the Mo₂C-Mo₂N_0.92 electrocatalyst. Figure 4B illustrates the NRR mechanism in the prepared Mo₂C-Mo₂N_0.92 heterostructures. The Mo₂C phases in the Mo₂C-Mo₂N_0.92 electrocatalyst generate ammonia by the electrochemical associative pathway by protonating the external N₂ gas and the Mo₂N phases following the MvK pathway by protonating the lattice-N to ammonia.

Mo₂C-Mo₂N_0.92 electrocatalyst has revealed considerably higher NRR performances due to the combined activity of the Mo₂C and Mo₂N_0.92 phases. N₂ adsorption and activation over an electrocatalyst surface occur at the exposed active sites. Previous studies have observed a strong N₂ adsorption behavior for Mo₂C, which reveals a promising NRR electrocatalyst (Wan et al., 2022). Especially, the electron-rich Mo atoms with moderate H adsorption energy and high N₂ affinity are the main electrocatalytic active sites for NRR (Ma et al., 2020; Liu et al., 2021; Zhang et al., 2018a).

NRR takes place by the ‘acceptance-donation’ mechanism at the Mo active sites is explained as follows. The lone pair of electrons from the σ-orbitals of N₂ are donated into the empty d-orbitals of Mo, resembling acceptance, followed by the back-donation of the d-orbital electrons into the π* orbitals of N₂, resembling donation (Fan et al., 2022; Wan et al., 2022). This acceptance-donation mechanism over Mo active centers results in the activation of the highly stable triple bonds of N₂ and realizes an increase in bond length ((Mo)-N=N•), as shown in Equation 7. The chemisorbed N₂ on Mo active sites is further protonated by six proton-coupled electron transfer steps (PCET) to form ammonia (Equation 8) by associative mechanism (Cheng et al., 2018; Ren et al., 2019). A c t i v a t i o n   s t e p : N 2 + M o   →   M o − N = N • (7) O v e r a l l   r e a c t i o n : M o − N = N • + 6   H 2 O + 6   e – →   2   N H 3 + 6   O H – (8)

The associative mechanism for ammonia production over the Mo electrocatalytic active sites can be suggested as follows. Further, the activated N₂ can follow either an alternating protonating pathway or a distal protonating pathway. M o − N = N • + H 2 O + e – ¯ →   M o − N = N H + O H – ¯ (9)

Distal associative pathway M o − N = N H + H 2 O + e – →   M o − N − N H 2 + O H – (10) M o – N – N H 2 + H 2 O + e – →   M o – N – N H 3 + O H – →   M o – N + N H 3   ↑ (11) M o − N + H 2 O + e   – →   M o − N H + O H – (12) M o − N H + H 2 O + e   – →   M o − N H 2 + O H – (13) M o − N H 2 + H 2 O + e   – →   M o − N H 3 + O H – →   M o + N H 3   ↑ (14)

Alternating associative pathway M o − N = N H + H 2 O + e – →   M o − N H − N H + O H – (15) M o − N H = N H + H 2 O + e – →   M o − N H − N H 2 + O H – (16) M o − N H = N H 2 + H 2 O + e – →   M o − N H 2 − N H 2 + O H – (17) M o − N H 2 − N H 2 + H 2 O + e – →   M o − N H 2 − N H 3 + O H – →   M o − N H 2 + N H 3   ↑ (18) M o − N H 2 + H 2 O + H 2 O + e – →   M o − N H 3   →   M o + N H 3   ↑ (19)

In the distal associative pathway, the protonation occurs from the farther side of the activated N atom (Equations 9-11). The protonation of the N atom bonded to the active site initiates only after complete protonation and ammonia evolution of the farthest N atom (Equations 12–14). In the alternating pathway, the protonation occurs alternatively on both N atoms and ammonia is generated at the final successive steps (Equations 15–19). Typically, the charge transfer reaction takes place via the charge accumulation and depletion stages in both Mo and N atoms attributing to better NRR performances (Wan et al., 2022).

Besides, Mo₂N_0.92 is suggested to generate ammonia by decomposition of the Mo₂N lattice rather than by catalytic mechanism (Zhang et al., 2018b; Hu et al., 2019). Thereby, the Mo₂N_0.92 entity present in the Mo₂C-Mo₂N_0.92 electrocatalyst generates ammonia by the Mars-van-Krevelen (MvK) mechanism. By the MvK mechanism, the lattice N atoms in metal nitride electrocatalysts are protonated rather than the conversion of gaseous N₂ to form ammonia (Equations 20–22) (Yang et al., 2019). The detailed MvK mechanism over Mo₂N phases is elucidated as follows. M o 2 N + H 2 O + e – →   M o 2 N H + O H – (20) M o 2 N H + H 2 O + e – →   M o 2 N H 2 + O H – (21) M o 2 N H 2 + H 2 O + e – →   M o 2 N H 3 + O H – →   M o 2 * + N H 3   ↑ (22) M o 2 * + N 2   →   M o 2 N ≡ N   S u r f a c e   r e p l e n i s h m e n t (23) M o 2 N ≡ N + H 2 O + e – →   M o 2 N ≡ N H 2 + O H – (24) M o 2 N = N H + H 2 O + e – →   M o 2 N − N H 2 + O H – (25) M o 2 N – N H 2 + H 2 O + e – →   M o 2 N – N H 3 + O H – ¯ → M o 2 N + N H 3   ↑ (26)

Where Mo₂* represents the nitrogen vacancy formed due to the MvK mechanism. The surface replenishment of the nitrogen vacancies from N₂ occurs, which generates ammonia (Equation 23) (Abghoui et al., 2016). In cases due to low N₂ solubility in aqueous electrolytes, the surface nitrogen vacancies are replaced by the N atoms from the bulk resulting in the degradation of the electrocatalyst over time. Thereby, the surface replenished N is then protonated to release ammonia (Equations 24–26). Besides, the MvK mechanism demands high energy for NRR, which facilitates HER at higher potentials (Ren et al., 2018). Nitrogen vacancies affect the neighboring N atoms in the lattice resulting in poor electrocatalytic performances (Ren et al., 2018). As a result, an increased ammonia yield due to the heterostructure formation can be ascribed to the combined contribution from electrochemical N₂ reduction and the MvK mechanism. The NRR performances in the prepared Mo₂C-Mo₂N_0.92 electrocatalysts reveal the combined activity of the Mo₂C and Mo₂N_0.92 phases. Hence, the combined electrochemical associative mechanism for N₂ reduction via Mo₂C and the MvK pathway by lattice-N protonation via Mo₂N_0.92 resulted in a high ammonia yield compared to the pristine counterparts.

Electrochemical impedance spectroscopy (EIS) analysis was undergone to elucidate the impedance characteristics of the prepared electrocatalysts. The Nyquist plots for real part (Z′) and imaginary part (Z”) of impedance elucidate the charge transfer and diffusion characteristics of the prepared electrocatalysts. The initial intercept on Z’ axis at the high frequency region denotes the solution resistance (R_s) in the electrode and electrolyte interface. Besides, the diameter of the semi-circle in the mid-frequency region denotes the charge transfer resistance (R_ct) for the redox reactions between the active sites of the electrode and the electrolytic ions. Additionally, the linear curve in the low frequency region ascribes to the diffusion characteristics of the prepared electrocatalysts.

From the Nyquist plots in Figure 5A, Mo₂C-Mo₂N_0.92 electrocatalyst exhibits a low R_s value of about 11 Ω compared to Mo₂N_0.92-Mo₂C (15 Ω) electrocatalyst. Besides, the Mo₂C-Mo₂N_0.92 electrocatalyst reveals R_ct value of about 4.5 Ω which is very less compared to 6.5 Ω for Mo₂N_0.92-Mo₂C electrocatalyst. Thereby, the Mo₂C-Mo₂N_0.92 electrocatalyst is observed to exhibit faster electrocatalytic redox reactions superior to Mo₂N_0.92-Mo₂C electrocatalyst. The superior electrocatalytic activity of the dominant Mo₂C phases in Mo₂C-Mo₂N_0.92 electrocatalyst explains the reduced resistance for redox reactions. Additionally, a higher angle for the linear curve with respect to x-axis was observed for Mo₂C-Mo₂N_0.92 in the low frequency region compared to Mo₂N_0.92-Mo₂C which indicate the superior diffusion characteristics of the Mo₂C-Mo₂N_0.92 electrocatalyst.

The Nyquist admittance plots explain the conductivity of the prepared electrocatalysts as a function of frequency. The admittance plot is characterized by the knee frequency which elucidate the corresponding charge transfer resistances of the prepared electrocatalysts. From Figure 5B, the Mo₂C-Mo₂N_0.92 electrocatalyst displays a higher knee frequency at 58.52 Hz compared to Mo₂N_0.92-Mo₂C (26.68 Hz). The higher knee frequency substantiates the lower charge transfer resistance for Mo₂C-Mo₂N_0.92 electrocatalyst which agrees with the Nyquist plots. The Bode phase plot in Figure 5C for the prepared electrocatalysts were obtained from the phase angle vs. frequency plots to elucidate the frequency response of the electrocatalyst. The phase angle denotes the phase difference between the current and voltage of the electrochemical cell. A phase angle of −90° denotes the ideal capacitor behavior elucidating a double-layer capacitance (C_dl). The curves in the low frequency region of the Bode phase plot denote the double layer capacitance(C_dl) of the electrocatalyst and the curves in high frequency region denote the capacitance in the redox reactions. Thereby, the Bode phase plots suggest a slight phase angle shift towards high frequencies for Mo₂C-Mo₂N_0.92 electrocatalyst than Mo₂N_0.92-Mo₂C indicating faster reaction kinetics. Hence, the dominant phases of Mo₂C in Mo₂C-Mo₂N_0.92 electrocatalyst drives a faster reaction kinetics than the Mo₂N_0.92 phases which validate the previous electrochemical studies.

Electrochemical active surface area (ECSA) of the prepared electrocatalysts were measured to estimate the available surface area for electrochemical reactions. ECSA was estimated by calculating the double layer capacitance (C_dl) of the prepared electrocatalysts by performing CV tests for various scan rates in the non-faradaic region, as shown in Figure 5D (Supplementary Figure S3). The current response in the non-faradaic region is purely due to the charge/discharge of the double layer. Thereby, the C_dl value (mF) was estimated from the slope of the linear plot of scan rate vs. current density derived from the CV curves. Consequently, the ratio of the C_dl value and the specific capacitance (40 μF cm^-2) yields ECSA value of the prepared electrocatalyst. As per the discussion, the electrocatalysts exhibit C_dl values of about 32.8 mF and 14.9 mF for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C, respectively. Consequently, the electrocatalysts exhibit a high ECSA value of about 820 cm² and 372.5 cm² for Mo₂C-Mo₂N_0.92 and Mo₂N_0.92-Mo₂C electrocatalyst, respectively. Thereby, a high ECSA value attributes to higher exposure of active sites to the electrolytic ions for redox reactions.