The Importance of Molybdenum(IV) Active Sites in Promoting Electrochemical Reduction of N₂ to NH₃ with MoFe Bimetallic Catalysts

Ammonium iron(II) sulfate (Fe(NH₄)₂·(SO₄)₂), Sodium molybdate dihydrate (Na₂MoO₄·2H₂O), Hexadecyl trimethyl ammonium Bromide (C₁₉H₄₂BrN, CTAB), concentrated hydrochloric acid (HCl), ammonium chloride (NH₄Cl), salicylic acid (C₇H₆O₃), sodium citrate dehydrate (C₆H₅Na₃O₇·2H₂O), sodium nitroferricyanide dihydrate (C₅FeN₆Na₂O·2H₂O), sodium hypochlorite solution (NaClO), and Nafion (5 wt.%) sodium were all purchased from Aladdin Ltd in Shanghai. Para-(dimethylamino) benzaldehyde (p-C₉H₁₁NO), hydrazine hydrate (N₂H₄·H₂O), sodium hydroxide (NaOH), hydrochloric acid (HCl), and ethanol (CH₃CH₂OH) were bought from Beijing Chemical Corporation. All the reagents and chemicals were used as received (without further purification). The de-ionized (DI) water used throughout the experiments was self-made via a Millipore system in the lab.

To prepare solution A, 0.5 mmol of Fe(NH₄)₂·(SO₄)₂ (0.1961 g) was added to 30 mL of DI water and then stirred for 0.5 h. For solution B, 0.5 mmol of Na₂MoO₄ (0.1210 g) and 0.1 g of PVP 10000 were added into 30 mL of DI water and stirred for 0.5 h; afterwards, solution A was added dropwise into solution B while vigorously stirred for 1 h. The obtained sample was then collected and washed with DI water and anhydrous ethanol repeatedly by centrifugation several times. The obtained sample was named MoFe-RT and then sintered in a muffle furnace at 500 °C for 2 h, and the obtained product after sintering was named MoFe-O. The MoFe-RT was also annealed separately in an Ar or NH₃ atmosphere at 500 °C for 2 h. The generated products were denoted as MoFe-Ar and MoFe-N, respectively.

The surface morphology of the catalyst was investigated with field emission-scanning electron microscopy (FE-SEM) (Zeiss Supra 55). High-resolution transmission electron microscopy (HR-TEM) characterization was performed on a TECNAI F30 TWIN equipped with an energy dispersive X-ray (EDX) accessory. Powder X-ray diffraction (XRD) patterns were collected on a Rigaku Corporation Ultima IV using Cu Kα radiation (λ = 1.54 Å) and operated at 30 mA and 40 kV. The scanning 2θ values ranged from 10° to 90° at a scan rate of 5° min⁻¹. X-ray photoelectron spectroscopy (XPS) signals were recorded on a K-Alpha + photoelectron spectrometer with monochromatic Al Kα X-ray. The binding energy of C1 s (284.8 eV) was used as a calibrated standard. The UV–vis absorption spectra were generated on a spectrophotometer (UV-1200, MAPADA). N₂ temperature-programmed desorption (N₂-TPD) was carried out with a Micrometeritics AutoChem Ⅱ 2920, where 100 mg catalysts were heated at 400 °C for 1 h in He (40 mL min⁻¹); after being cooled to 50 °C, the sample was swept with N₂ atmosphere for 1 h. Then, the pretreated sample was heated from 50 °C to 400 °C at a ramp of 10 °C min⁻¹ in He.

The electrochemical performance measurements were performed with a CHI 660E electrochemical analyzer (CH Instruments, Inc., Shanghai) using a standard three-electrode system with MoFe compounds/carbon paper as the working electrode, Pt foil as the counter electrode, and saturated Ag/AgCl as the reference electrode. All the potentials measured were calibrated to the reversible hydrogen electrode (RHE) using the following equation: E (RHE) = E (Ag/AgCl) + 0.278 (V). All experiments were performed at room temperature. For N₂ reduction experiments, the 0.05 M H₂SO₄ solution was purged with N₂ for 30 min to form a N₂-saturated solution. The chronoamperometry test was conducted in a N₂-saturated 0.05 M H₂SO₄ solution (40 ml) in a two-compartment cell, which was separated by a Nafion 117 membrane. Linear sweep voltammetry (LSV) was performed at the scan rate of 5 mV s⁻¹. Chronoamperometric tests were carried out at different potentials for 2 h.

The ammonia concentration in the product was determined with an ultraviolet spectrophotometer via the indophenol blue method. In detail, 2 mL of 1.0 M NaOH solution containing 5 wt.% salicylic acid, 5 wt.% sodium citrate, 1 mL of 0.05 M NaClO, and 0.2 mL of 1 wt.% C₅FeN₆Na₂O were added into a 2 mL electrolyte and allowed to react for a while. Then, the indophenol blue concentration was detected at a wavelength of 655 nm using a UV–vis absorption spectrum after the solution stood at room temperature for 2 h. The working curves of different concentrations-absorbance were generated by measuring the absorbance of ammonium chloride solutions at different concentrations. The obtained working curves (y = 0.1174x + 0.0159, R² = 0.9998) had a strong linear relationship between the concentration and the absorbance.

On the other hand, the concentration of hydrazine, a by-product, was spectrophotometrically determined via the Watt and Chrisp method. The details are below: 5 mL of the prepared color reagent (5.99 g of p-C₉H₁₁NO, 30 mL of HCl and 300 mL of C₂H₅OH) were mixed first. Then, with the solution standing at room temperature for ∼10 min, the absorption spectrum was detected at a wavelength of 455 nm using the UV–vis absorption spectrum. Curves of different concentrations-absorbance were generated by measuring the absorbance of hydrazine solutions at different concentrations. The obtained working curves (y = 0.08x−0.002, R² = 0.999) had a good linearity for the concentration and the absorbance.

The yield rate for NH₃ production was calculated with the following equation:

$$\begin{eqnarray}&&{R}_{N{H}_{3}}=\left({c}_{N{H}_{3}}\times V\right)/\left(17\times t\times A\right)\end{eqnarray} \tag{ 1 }$$

where ${c}_{N{H}_{3}}$ is the measured NH₃ concentration, V is the volume of electrolyte, t is the time of the whole reduction reaction, and A is the effective area of the working electrode.

The Faradaic efficiency (FE) of NH₃ production was calculated using the following equation:

$$\begin{eqnarray}&&{\rm{FE}}=3\times {c}_{N{H}_{3}}\times V\times F/Q\end{eqnarray} \tag{ 2 }$$

where ${c}_{N{H}_{3}}$ is the measured NH₃ concentration, V the volume of electrolyte, F the Faradaic constant, and Q the total electricity of the whole NRR process.

A series of MoFe catalysts were successfully prepared via the facile one-step thermal method (please see Supporting Information for the preparation details). The obtained precipitation of sodium molybdate and ferrous ammonium sulfate hexahydrate in aqueous solution under ambient conditions was named MoFe-RT and then calcined at 500 °C under different atmospheres. The primary preparation process is illustrated in Fig. S1 (available online at stacks.iop.org/JES/168/126518/mmedia). Based on the thermal treatment atmosphere (O₂, Ar and NH₃), the final products were labeled MoFe-O, MoFe-Ar and MoFe-N. The scanning electron microscopy (SEM), transmission electron microscopies (TEM) and dispersive X-ray spectroscopy (EDS) analysis were used to investigate the morphology of the MoFe compound catalysts and spatial distribution of Mo, Fe, N and O atoms in MoFe compound catalysts. SEM and TEM images of MoFe-RT, MoFe-O, MoFe-Ar, and MoFe-N compounds (Fig. 1) indicated that they were irregular particles with diameters of about 100 nm. The HRTEM image of MoFe-N is shown in Fig. S2. As seen, there are no obvious lattice fringes or material characteristic peaks. The energy-dispersive X-ray (EDX) elemental mapping images (Figs. 1i–1l)) revealed the homogeneously distributed Mo, Fe, N, and O elements in the nanoparticles. The X-ray diffraction (XRD) patterns of four MoFe compound catalysts are shown in Fig. S3. As observed, only the MoFe-O sample shows diffraction peaks at 21.6°, 22.9°, 24.9°, 26.7°, 27.4°, 31.0° and 34.0°, corresponding to the (102), (031), (131), (221), (202), (311) and (240) planes of Fe₂(MoO₄)₃ (JCPDS No. 33–0661). The other MoFe compound catalysts did not show any characteristic peaks of the substances. Additionally, the results showed that a bimetallic catalyst (MoFe-N) with the homogeneous mixture of Mo and Fe was prepared.

Zoom In Zoom Out Reset image size

Additionally, the elemental valence states of surface elements of MoFe catalysts were investigated by X-ray photoelectron spectroscopy (XPS). The survey spectra of MoFe compound catalysts revealed the presence of Fe, Mo, O and N elements in the catalysts (Fig. S4). All spectra were calibrated according to the C 1 s binding energy at 284.8 eV (Fig. S5). The elemental compositions in MoFe-O, MoFe-Ar and MoFe-N are shown in Table S1. The molar ratio of Mo/Fe in the MoFe samples was essentially 1:1. However, the content of Mo in MoFe-O was obviously decreased, which could be due to the partial covering of the Mo element by the oxide layer on the surface of the materials. In the Mo 3d region (Fig. 2a), two independent peaks are located at 235.8 and 232.6 eV, which can be assigned to Mo 3d_3/2 and Mo 3d_5/2, respectively. ³⁰ Generally, Mo⁶⁺ originates from surface-oxidized Mo species. ³² Unlike the MoFe-O and MoFe-Ar catalysts, the Mo 3d region of MoFe-N clearly indicated the presence of two low valence chemical states, i.e. Mo⁴⁺ (234.7 eV) and Mo^δ+ (230.3 eV). ³³ The atomic ratio of Mo in different valence states is Mo⁶⁺: Mo⁴⁺: Mo^δ+ = 0.4: 0.4: 0.2, and the reductive NH₃ atmosphere can result in an additionally lower oxidation state of the Mo element. The reason why Mo (IV) can act as an active site for nitrogen fixation is because elements with unoccupied d orbital can accept electrons from N₂, thus facilitating the adsorption of N₂. ^34,35 From the high-resolution XPS spectra of Fe 2p for MoFe-O, MoFe-Ar, and MoFe-N (Fig. 2b), the coexistence of Fe²⁺ and Fe³⁺ mixed valence states is evidenced by the observed peaks at 711.1, 724.3, 712.6, and 725.5 eV, and the atomic ratio of Fe²⁺ and Fe³⁺ does not change with the thermal treatment conditions of MoFe catalysts. ³⁶ The peaks of MoFe-Ar and MoFe-N in Fe 2p_3/2 (711.1 and 712.6 eV) are positively shifted ∼ 0.7 eV in comparison with those of MoFe-O (710.4 and 711.9 eV). Typically, the signals at 530.4 eV and 531.2 eV for O 1 s correspond to the typical lattice oxygen and chemisorbed oxygen, respectively, indicating the presence of metal oxides on the catalysts (Fig. S6). ^31,37 Also, MoFe-O has the highest oxygen content and lowest nitrogen content among the MoFe compounds as a result of the thermal treatment in an air atmosphere. The N 1 s region of MoFe-N (Fig. 3c) showed three peaks at 397.08, 398.73 and 400.4 eV, corresponding to the Mo–N bond, pyridinic N and pyrrolic-N, respectively. ^38–40 The peak for the metal-N bond was detected in MoFe-N, which indicated a strong electron interaction between the surface nitrogen species and the metal induced by the reductive NH₃ atmosphere and Mo-bearing species in the surroundings. However, the peak of the metal-N bond did not appear in the MoFe-O and MoFe-Ar compounds, which signals that the thermal treatment under an NH₃ atmosphere is indeed an effective measure for the construction of a metal-N bond, while N of PVP mainly forms a pyrrolic-N bond during the thermal process (NH₃ and Ar atmosphere). In addition, the N contents of MoFe-O, MoFe-Ar and MoFe-N are 17.77, 23.78 and 26.22 at.%, respectively. It is worth noting that the increase in nitrogen content in MoFe-N is due to not only the nitrogen in metal-N bond, but also the nitrogen in the surface nitrogen species. In addition, there is a strong electronic interaction between the Mo-N and the Fe species. Therefore, the Mo-N-Fe was formed in the catalyst of MoFe-N, which further optimized the electronic structure of Mo elements, and thus benefitted the NRR catalysis. The surface bonding configuration of MoFe compound catalysts were investigated with Fourier transform infrared (FTIR) spectroscopy. As shown in Fig. 2d, the IR signals at 3447 and 1654 cm⁻¹ correspond to the stretching vibration and the bending vibration of water that was physically adsorbed on the surface of the catalysts. ⁴¹ The spectra of PVP showed three peaks at 2958, 1423 and 1291 cm⁻¹, which can be attributed to the C-H, N–H and C–N bonds of PVP, respectively. ^41,42 The FTIR spectra of MoFe-O and MoFe-N showed no characteristic peak of PVP, which suggests that no PVP remained in the catalyst. ¹⁹ The IR signal of the N–H bonds of MoFe-N may originate from NH₃ atmosphere. The characteristic peak of the N–H bond was found in the IR spectra of MoFe-Ar, which was believed to be the N contained in PVP. The FTIR spectra of MoFe-O showed a characteristic peak of Mo–O bond at 819 cm⁻¹. The four different MoFe compound catalysts actually have similar FTIR characteristic peaks; that is, the introduction of N did not significantly affect the chemical structure of the samples.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The adsorption capacity of N₂ is one of the prerequisites for an electrocatalyst to have excellent NRR performance. Fig. S7 shows the N₂ temperature-programmed desorption (N₂-TPD) of MoFe compound catalysts. As seen, MoFe-N (green line) had two peaks at temperatures of ∼100 °C (which is attributed to the physisorption of N₂) and 340 °C (which corresponds to the chemisorption of N₂). As such, the strong N₂ adsorption on MoFe-N denotes the strong binding of N₂, which is favorable for electrochemical NRR.

The electrochemical NRR activities of MoFe compounds were measured in an N₂-saurated 0.05 M H₂SO₄ electrolytic using an H-type electrolytic cell separated by a Nafion 117 membrane. The prepared Fe-Mo compound that was coated on carbon paper was used as the working electrode, the Ag/AgCl saturated KCl electrolyte was used as the reference electrode, and Pt was used as the counter electrode. Figure S8 shows the LSV curves of MoFe-N in an N₂-saturated (black line) and an Ar-saturated (red line) 0.05 M H₂SO₄ electrolyte. The current density obtained in an N₂-saturated electrolyte is obviously higher than that of an Ar-saturated electrolyte at given potentials, which indicates that MoFe-N is a more effective catalyst for the electrochemical NRR. Subsequently, the electrocatalytic activities of the prepared Fe-Mo compounds toward NRR were investigated at different potentials from −0.2 to −0.6 V in an N₂-saturated 0.05 M H₂SO₄ electrolyte. It was observed that after 2 h of the NRR process at the given potentials, the produced NH₃ could be detected in the electrolyte via the indophenol blue method, and a possible by-product N₂H₄ was also detected via the Watt and Chrisp method. The standard curves of NH₃ and N₂H₄ are shown in Fig. S9. The chronoamperometric (i-t) curves of MoFe-N at different potentials are plotted in Fig. S10. Further, the NH₃ yield rates and FEs of the prepared MoFe compounds were obtained and plotted in Fig. 3a. As seen, the NH₃ yield rates and the FEs of the MoFe catalysts shared the same tendency, which generally increased first and then decreased with the potential becoming more negative. Also, the MoFe-N exhibited the largest NH₃ yield rate of 33.31 μg h⁻¹ mg⁻¹ _cat. At −0.3 V, which was slightly higher than that of other Mo-Fe compound catalysts.

Additionally, it was found that the FEs of MoFe compound catalysts also decreased as more negative potentials were applied. Noticeably, MoFe-N showed the highest FE of 33.26 % at −0.3 V, which is significantly higher than that of the other MoFe compound catalysts. In addition, the obtained NH₃ yield rate and FE for MoFe-N are much better than those reported for single Mo-, Fe- catalysts and even better than some of the noble metal-based catalysts (see Table S2). As observed, the current densities of the four MoFe compounds at −0.3 V had no obvious decrease for at least 10 h, illustrating that the electrocatalysts had excellent stability (Fig. S11). In addition, the accelerated degradation test (ADT) in Fig. S12 also demonstrated a good electrochemical stability. It was also observed that after 5 cycles, the NH₃ yield rate and the FE of MoFe-N had no obvious decrease, indicating its excellent stability for electrochemical NRR (Fig. 3b). We believe that the reason the MoFe catalysts exhibited FE loss at negative potentials lower than −0.3 V is due to the increased competition from the HER. Also, from Fig. 3c, one can see that the current density of MoFe-N is the smallest of all the MoFe compound catalysts prepared. However, as the HER is the dominant reaction in the acidic electrolyte, ⁴³ the improved FE of MoFe-N must be due to the effective inhibition of HER. Therefore, we speculate that Mo(IV) might be able to suppress the activation of H₂ significantly. To verify the above speculation, we designed an experiment to collect the hydrogen gas generated by the four different MoFe catalysts during a 10 h chronoamperometric electrolysis using the Drainage Method. The schematic diagrams of the drainage and gas collection process and of the MoFe catalysts are shown in Fig. S13 of the Support Information. The results showed that there is not much of a difference in H₂ production rates among MoFe-RT, MoFe-O and MoFe-Ar, and all of them exhibit significantly higher H₂ production rates than MoFe-N. Based on the above experiments, it was verified that the Mo(IV) active sites not only benefit the NH₃ formation, but also play a critical role in suppressing the HER. Furthermore, no N₂H₄ was detected in the electrolyte after the NRR electrolysis, revealing that the MoFe catalysts have exclusive selectivity for NH₃.

Tafel slope was further used to evaluate the HER kinetics. Figure S14 shows the Tafel slopes for the catalysts of MoFe-RT, MoFe-O, MoFe-Ar and MoFe-N, which are calculated from the corresponding polarization curves obtained in a 0.05 M H₂SO₄ electrolyte. As seen, the slope of MoFe-N was 65 mV dec⁻¹ for HER, higher than that of the other MoFe catalysts, indicating the poor intrinsic HER activity of MoFe-N. Also, it was demonstrated that the low valence of Mo(IV) produced under the NH₃ atmosphere can effectively inhibit the HER kinetics. ⁴⁴ The cyclic voltammogram curves at varying scan rates and electrochemical double layer capacitances (C_dl) of MoFe-RT, MoFe-O, MoFe-Ar and MoFe-N are shown in Fig. S15. As observed, the C_dl of MoFe-N was slightly higher than that of MoFe-RT, MoFe-O and MoFe-Ar, respectively. The electrochemical impedance spectroscopy (EIS) measurements were performed at −0.3 V in an N₂-saturated electrolyte to get more insight into the NRR kinetics. As seen, MoFe-N possessed a lower charge transfer resistance and a lower mass transfer resistance than MoFe catalysts, which implied improved NRR kinetics (Fig. S16). ⁴⁵ Further, to confirm that the detected NH₃ in the electrolyte was actually coming from the electrochemical reduction of N₂ rather than from the MoFe catalysts themselves, a series of control experiments were performed, including: (a) Replacing N₂ gas with Ar gas at −0.3 V for MoFe-N NRR electrolysis; (b) Applying open-circuit potential to the MoFe-N in an N₂-saturated electrolyte; (c) Using a bare carbon paper as the working electrode in an N₂-saturated electrolyte at −0.3 V. The testing results are presented in figure 4(a). As seen, no NH₃ was detected in the electrolyte in any of the above control tests, indicating that the produced NH₃ does indeed come from electrochemical N₂ fixation.

Zoom In Zoom Out Reset image size

In order to study the reaction mechanism, the active sites of the MoFe catalysts need to be identified. Based on the above characterization, we believe that the low valence Mo species is very likely the exclusive active site for NRR. It is also known that the SCN⁻ ion has a high affinity to metal species, as it can poison the metal active sites toward electrochemical reaction. The EIS was measured at −0.3 V in a N₂-saturated electrolyte for both cases of with and without 5 mM KSCN. As seen in Fig. S17, the impedance of the catalyst in the electrolyte with 5 mM KSCN was significantly smaller than that of without 5 mM KSCN, which means that the addition of KSCN is beneficial to the conductivity of the catalyst. As shown in Fig. 4b, KCSN was added purposely to the electrolyte during the ongoing process. The inset of the figure shows the NRR activity of MoFe-N in the 0.05 M H₂SO₄ with 5 mM KSCN electrolyte. It is clearly seen that the current was significantly reduced after the addition of SCN⁻. Meanwhile, no NH₃ was detected in the 0.05 M H₂SO₄ solution after a 5 mM KSCN electrolyte was added. These results suggest that the decrease in current density for NRR can be ascribed to the blocking of low valence Mo species by SCN⁻, which further confirms that the low valence Mo species is the main active site of MoFe-N for NRR. Additionally, the current density of MoFe-N in an Ar-saturated electrolyte with and without KSCN was measured. As shown in Fig. S18, the current density of MoFe-N increased significantly after the addition of KSCN, and HER process occurred in the Ar-saturated electrolyte (no NRR). Therefore, we believe that the NRR activity of MoFe-N decreased after the addition of KSCN.

Moreover, in order to further verify the source of N in the produced NH₃, a 10 h chronoamperometry test in an Ar-saturated 0.05 M H₂SO₄ electrolyte was performed to check the variation of N in the MoFe-N catalysts before and after the test. The N 1 s XPS spectra and elemental compositions of MoFe-N after the 10 h chronoamperometry test in an Ar-saturated electrolyte are shown in Fig. S19 and Table S1. The results indicate that the content of N atoms in MoFe-N did not change before and after the 10 h test, but the content of Fe³⁺ and Mo⁶⁺ increased. This, we think, is probably due to the HER process mainly occurring in the Ar-saturated 0.05 M H₂SO₄ electrolyte during the 10 h chronoamperometry. The evolution of H₂ was observed with a large number of electrons transferred. The N₂-TPD results revealed that MoFe-N exhibited a better N₂ adsorption capacity than the other MoFe catalysts that were prepared. However, there is no obvious difference in NH₃ yield rates across the four different MoFe catalysts, indicating that N₂ adsorption on the active sites of MoFe catalysts is not the rate-determining step (RDS). It may not even be the first step for electrochemical NRR, which is distinctly different from the widely accepted associative and dissociative mechanisms. Recently, more attention has been paid to a new NRR mechanism of surface hydrogenation (SHM) proposed by Ling et al. ⁴⁶ As shown in Fig. 4c, the N₂ adsorption is replaced by the reduction of H⁺ as the foremost step in NRR (or the first elementary step of HER: H⁺ + e⁻ + * → *H). Subsequently, the gas phase N₂ can be directly activated and reduced into *N₂H₂. ¹⁰ Furthermore, the activation of N₂ can be reinforced through the cooperative effect of surface *H and active sites. Ling et. al also took an N-doped carbon-supported single Mo catalyst and a pure Mo catalyst as prototypes for calculations. Both catalysts demonstrated strong N₂-binding strength and relatively high activity toward the NRR. This shows that the surface hydrogenation mechanism is suitable not only for catalysts with weak N₂ binding strength, but also for catalysts with strong N₂ binding strength. We know while multiple mechanisms may coexist in a system, only one mechanism dominates. The NH₃ production rate of the MoFe-N is not different from that of the other three MoFe catalysts studied, but its FE is quite different from those of the other three, which is why we proposed that the catalyst conforms to the surface hydrogenation mechanism. According to the first-principle theory, it is only possible that the first step is the RDS, which means that in the case of insufficient amount *H, the N₂ adsorption will be stronger while the H reduction will be weaker. In other words, the MoFe-N catalyst with a strong N₂ adsorption capacity could inhibit HER, thus enhancing the NRR.