Dual‐Metal Sites Boosting Polarization of Nitrogen Molecules for Efficient Nitrogen Photofixation

Constructing nitrogen (N2) adsorption and activation sites on semiconductors is the key to achieving efficient N2 photofixation. Herein, Mn–W dual‐metal sites on WO3 are designed toward efficient N2 photoreduction via controlled Mn doping. Impressively, the optimal 2.3% Mn‐doped WO3 (Mn‐WO3) exhibits a remarkable ammonia (NH3) production rate of 425 µmol gcat.−1 h−1, representing the best catalytic performance among the ever‐reported tungsten oxide‐based photocatalysts for N2 fixation. Quasi in situ synchrotron radiation X‐ray spectroscopy directly identifies that the Mn–W dual‐metal sites can enhance the polarization of the adsorbed N2, which is beneficial to the N2 activation. Further theoretical calculations reveal that the increased polarization is originated from the electron back‐donation into the antibonding orbitals of the adsorbed N2, hence lowering the reaction energy barrier toward the N2 photofixation. The concept of dual sites construction for inert molecule activation offers a powerful platform toward rational design of highly efficient catalysts for nitrogen fixation and beyond.


Introduction
Ammonia (NH 3 ) is one of the most important chemical products in modern industry to serve as the precursor of fertilizers, guaranteeing the subsistence of billions of people. [1][2][3] The N≡N bond to boost NH 3 synthesis. [24] Although some progress has been achieved on N 2 photofixation, active photocatalysts with efficient N 2 adsorption and activation sites are still urgently desired.
Since the essence of the molecule activation is stemmed from the electron back-donation into the antibonding orbitals of the adsorbate, constructing superior adsorption sites is the key to achieve highly efficient N 2 photofixation. [25,26] Generally, N 2 molecules are adsorbed on the single-metal sites of semiconductors through side-on or end-on configuration ( Figure 1A). [27,28] The adsorption of N 2 molecules is through the electronic coupling between the d orbitals of the metal sites and the p orbitals of the N 2 molecules, where N 2 molecules could accept electrons from metal sites and thus induce polarization for the adsorbed N 2 . [29,30] Based on the adsorption configuration, the impact of the single metal on the molecular polarization might not be high enough for the activation of inert N 2 molecules. In this regard, dual-metal sites with different metallicities and d orbitals can induce polarization in N 2 molecules to a larger extent ( Figure 1B). [31][32][33] Therefore, constructing dual-metal sites could provide new opportunities to develop highly efficient catalysts for N 2 fixation.
Tungsten oxides have been widely used toward photocatalytic N 2 fixation for low cost and stability. [34][35][36] However, tungsten oxides generally lack efficient reaction sites for activating N 2 . Meanwhile, Mn element has been regarded as a relatively active transition metal for N 2 reduction due to the coexistence of empty and occupied d orbitals. [37,38] Herein, we designed and fabricated Mn-W dual catalytic sites on WO 3 semiconductor toward N 2 photoreduction by Mn doping. Impressively, the optimal 2.3% Mn-doped WO 3 (Mn-WO 3 ) delivers an NH 3 production rate of 425 µmol g cat.
−1 h −1 , which is the best catalytic activity among the reported tungsten oxide-based photocatalysts for N 2 fixation. More importantly, quasi in situ X-ray photoelectron spectroscopy (XPS) and X-ray absorption near-edge spectroscopy (XANES) directly demonstrate that N 2 molecules are highly polarized at Mn-W dual sites, which greatly contributes to the activation of N 2 . Further theoretical analysis reveals that the increased molecular polarization is stemmed from the electron back-donation from the Mn-W dual sites to the antibonding orbitals of the adsorbed N 2 molecules, thus lowering the energy barrier of the N 2 fixation.

Results and Discussion
With regard to the synthesis of Mn-WO 3 , 1.0 g of commercial W 50 Mn 50 powders were etched by sulfuric acid solution (H 2 SO 4 , 0.31 m) at 353 K for 2 h, followed by thermal treatment in muffle furnace at 873 K for 1 h, and 2.3% Mn-WO 3 was obtained with the mass percentage of Mn element being 2.3%. In addition, WO 3 , 0.7% Mn-WO 3 , 2.7% Mn-WO 3 , and 5.0% Mn-WO 3 were also prepared by the similar synthetic procedure by varying the concentration of H 2 SO 4 solution. The scanning electron microscopy images of as-obtained samples show similar morphology with aggregated nanocrystalline structure ( Figure S1, Supporting Information). Figure 2A displays the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of 2.3% Mn-WO 3 . The fringes with interplanar spacing of 3.6 Å are assigned to the (200) plane of WO 3 nanocrystals. Besides, STEM energy-dispersive X-ray elemental mapping images of 2.3% Mn-WO 3 indicate the homogeneous distribution of W, O, and Mn elements ( Figure 2B). Moreover, the 3D image of 2.3% Mn-WO 3 was revealed by soft X-ray computed tomography, exhibiting the morphology of irregular nanoparticles (Figure 2C). The X-ray diffraction (XRD) patterns of WO 3 , 0.7% Mn-WO 3 , and 2.3% Mn-WO 3 can be well indexed to the monoclinic WO 3 (JCPDS#71-2141) ( Figure 2D). As for 2.7% Mn-WO 3 and 5.0% Mn-WO 3 , new peaks attributed to monoclinic MnWO 4 (JCPDS#13-0434) appear. As a result, WO 3 , 0.7% Mn-WO 3 , and 2.3% Mn-WO 3 are regarded as WO 3 nanocrystals doped by Mn atoms, while MnWO 4 phase is formed for 2.7% Mn-WO 3 and 5.0% Mn-WO 3 . Figure S2, Supporting Information, shows the XPS spectra of W 4f for different catalysts, where the peaks at 35.3 and 37.4 eV correspond to W 4f 7/2 and 4f 5/2 of W 6+ species, respectively. [39] As for the XPS spectra of O 1s, the peaks located at 530.1 and 532.8 eV are assigned to lattice oxygen and surface hydroxyl group, respectively ( Figure S3, Supporting Information). [40] The W species are in W 6+ state for all samples, while these catalysts mainly contain the lattice oxygen species. Meanwhile, the XPS analysis also reveals that the nitrogen impurity is negligible in these catalysts ( Figure S4, Supporting Information). When it comes to the XPS spectra of Mn 2p, Mn 2+ species with two peaks located at 641.4 and 653.1 eV can be clearly observed for 2.3% Mn-WO 3 , 2.7% Mn-WO 3 , and 5.0% Mn-WO 3 ( Figure S5, Supporting Information). [41] XANES was further measured to explore the electronic properties of these samples. Mn species in 0.7% Mn-WO 3 , 2.3% Mn-WO 3 , 2.7% Mn-WO 3 , and 5.0% Mn-WO 3 exhibit the characteristic of oxidation state in Mn K-edge XANES spectra ( Figure 2E). [42] In addition, the W L 3 -edge XANES spectra of the samples are similar to that of commercial WO 3 ( Figure 2F), which is consistent with the XPS results.
The as-obtained samples were applied as the photocatalysts toward N 2 photofixation. Each reaction was performed in 20 mL of water under 1 atm of N 2 at 25°C with 10 mg of catalysts. The generated NH 3 was analyzed spectrophotometrically by using Nessler's reagent, ion chromatography, and 1 H nuclear magnetic resonance ( 1 H-NMR) methods (Figures S6-S8 and Table S1, Supporting Information). [43] In the absence of catalysts, no product was observed (Table S2,  −1 h −1 , respectively. Besides, the NH 3 production rate for 2.3% Mn-WO 3 under visible light still reaches 5.7, 2.1, 1.7, and 4.1 times as high as that for WO 3 , 0.7% Mn-WO 3 , 2.7% Mn-WO 3 , and 5.0% Mn-WO 3 , respectively. Thus, 2.3% Mn-WO 3 exhibits superior catalytic activity for N 2 photoreduction, which also represents the best catalytic performance among the ever-reported tungsten oxide-based photocatalysts ( Figure S9 and Tables S3 and S4, Supporting Information). Furthermore, controlled experiments were carried out to confirm the authenticity of the reaction. As shown in Table S2, Supporting Information, no NH 3 was detected without light irradiation over 2.3% Mn-WO 3 , confirming that the N 2 reduction process is driven by light. Besides, replacing H 2 O with CH 3 CN also inhibits the formation of NH 3 , suggesting that H 2 O is the proton source of NH 3 (Table S2, entry 3, Supporting Information). In addition, the disappearance of NH 3 after replacing N 2 with Ar proves that NH 3 originates from N 2 fixation (Table S2, entry 4, Supporting Information). Moreover, neither N 2 H 4 nor NO 3 − was detected during the photocatalytic process. Furthermore, we also simultaneously detected the production of H 2 and O 2 during the N 2 fixation process ( Figure S10, Supporting Information).
To further corroborate the origin of obtained NH 3 , we designed an isotopic labeling study by using 15 N 2 as the purging gas. [44] The obtained 15 NH 4 + was measured by NMR spectroscopy. As shown in 1 H NMR spectra ( Figure 3B), the isotopically labeled sample exhibits the doublets of 15 NH 4 + , reaffirming that the generated NH 3 indeed originates from N 2 fixation. To evaluate the light utilization efficiency, the apparent quantum efficiency (AQE) of 2.3% Mn-WO 3 was measured under the irradiation of monochromatic light ( Figure 3C). Specifically, the AQE is calculated to be 0.18% at 450 nm for 2.3% Mn-WO 3 . We further assess the solar-to-ammonia (STA) efficiency of 2.3% Mn-WO 3 in pure water under simulated AM1.5G irradiation. The STA efficiency was calculated to be ≈0.019%, which was comparable to the results reported recently. [45,46] Furthermore, the stability of 2.3% Mn-WO 3 was studied by recycling the catalyst ( Figure 3D). After five successive reaction rounds, nearly ≈100% of initial activity was retained, indicating remarkable catalytic stability.
To explore the catalytic mechanism, we investigated the energy band structures of the obtained catalysts. As indicated by diffuse reflectance ultraviolet-visible (UV-vis) spectra, the intrinsic UVvis absorption edges of prepared samples are similar ( Figure S11, Supporting Information). In addition, the values of band gaps are determined to be 2.60 and 2.58 eV for WO 3 and 2.3% Mn-WO 3 , respectively ( Figure S12, Supporting Information). The positions of valence band (VB) for these samples were measured by synchrotron radiation photoemission spectroscopy with a photon energy of 169.96 eV (Figures S13 and S14, Supporting Information). As a result, the values of VB for WO 3 Figure 4A. Apparently, Mn doping elevates the CB of WO 3 nanocrystals, ensuring sufficient driving force to trigger N 2 reduction. By contrast, commercial WO 3 is thermodynamically unfavorable for N 2 reduction (Figures S16-S19, Supporting Information).
The photoinduced charge-separation efficiency was further explored for these photocatalysts. Figure S20, Supporting Information, shows the transient photocurrent responses spectra of WO 3 , 2.3% Mn-WO 3 , and 5.0% Mn-WO 3 under Ar atmosphere, where 2.3% Mn-WO 3 exhibits the strongest photocurrent density. The high photoinduced charge-separation efficiency of 2.3% Mn-WO 3 is further proved by room-temperature photoluminescence spectroscopy ( Figure S21, Supporting Information). Meanwhile, 5.0% Mn-WO 3 displays the highest intensity among all the samples, which may suggest the rapid recombination of the charge carrier for excessive Mn doping. In addition, as can be seen from the O K-edge X-ray absorption spectroscopy ( Figure S22, Supporting Information), the peak located at 533 eV is related to W/Mn-O hybridization, originating from the electron transfer from O to W/Mn atoms. 2.3% Mn-WO 3 displays enhanced covalency between metal and O atoms due to the strongest intensity of the peak at 533 eV. As the metal-oxygen covalency promotes the electron transfer from metal active sites to adsorbed molecules, [34,36] the adsorbed N 2 molecules are expected to facilely accept electrons from 2.3% Mn-WO 3 . This point is confirmed by transient photocurrent responses of the catalysts ( Figure 4B). When the electrolyte was saturated N 2 , the transient photocurrent responses declined, which can be attributed to the competitive interfacial electron transfer to adsorbed N 2 molecules. [47] Apparently, the photocurrent responses of 2.3% Mn-WO 3 under N 2 decreases by ≈39.1% compared with those under Ar atmosphere, which further confirms the active electron transfer from 2.3% Mn-WO 3 to adsorbed N 2 molecules over the photocatalysis process. [36] To further clarify the promotion effect of Mn-W dual sites on N 2 activation, quasi in situ XPS and quasi in situ XANES were performed. After treated with N 2 gas, the binding energies of Mn 2p spectrum for 2.3% Mn-WO 3 displays a positive shift of ≈0.5 eV ( Figure 4C). Meanwhile, a similar shift of 0.4 eV to higher energy region is also observed in Mn L-edge spectra. As displayed in Figure 4D, the Mn L-edge spectra of 2.3% Mn-WO 3 own two photon energy peaks of L 3 and L 2 , located at 640.8 and 652.3 eV, corresponding to the electron transition from Mn 2p 3/2 and 2p 1/2 to 3d, respectively. [48] The variation of the electronic state of Mn indicates that the Mn sites spontaneously inject electrons into the adsorbed N 2 molecules. [49] We further applied N K-edge spectra to monitor the evolution of adsorbed N 2 molecules. As shown in Figure 4E, the peaks in the range of 395 to 405 eV are assigned to multi-electron excitations associated with the prominent N 1s→ * transition, while the peak at ≈418 eV is called * shape resonance. [50] A fact which has been discussed extensively is that the energy position and strength of the * resonance are very sensitive to the internuclear distance, namely bond length. Specifically, the longer bond length results in lower position and higher intensity of * resonance for first-row diatomics. [51] Apparently, the * resonance of adsorbed N 2 for 2.3% Mn-WO 3 shows a sharp position movement and an increased intensity compared with that for WO 3 . It clearly indicates that the N≡N triple bond is asymmetrically elongated and efficiently polarized over 2.3% Mn-WO 3 , as illustrated in Figure 4F.
Furthermore, density functional theory (DFT) calculations were conducted to gain in-depth insights into the reaction process over Mn-WO 3 . Based on the experimental results shown in Figure 2A,D, the WO 3 (200) and Mn-doped WO 3 (200) models were built as shown in Figure S23, Supporting Information. The stability of the Mn-doped model was first studied. As displayed in Figure S24, Supporting Information, the model of Mn-WO 3 (200) still maintained its geometric structure even after 500 steps in 5ps simulation. Moreover, the adsorption configurations and energies of N 2 molecules on these models were calculated. As for WO 3 (200), N 2 molecule is weakly adsorbed on WO 3 (200) through end-on configuration with an adsorption energy of −0.56 eV (Figure 5A). As for side-on configuration of N 2 on WO 3 (200), the corresponding adsorption energy is determined to be endothermic www.advancedsciencenews.com www.advancedscience.com ( Figure 5B). When it comes to Mn-WO 3 (200), N 2 molecule is adsorbed on Mn-WO 3 (200) with the adsorption energies of −1.19 and −1.22 eV for end-on configuration at W and Mn sites, respectively ( Figures 5C and 5D). In addition, the presence of Mn-W dual sites enables the adsorption of N 2 molecules with sideon configuration ( Figure 5E). For better understanding the polarization effect of Mn-W dual sites on adsorbed N 2 molecules, we tried to use the bond length as the descriptor for polarization of N 2 molecules. Specifically, we define the degree of polarization of N-N bond in N 2 (1.09 Å) and PhN=NPh (1.25 Å) as 0 and 1, respectively, and calculate the polarization degree of each adsorption configuration according to the above functional relationship. As displayed in Figure S25, Supporting Information, the adsorbed N 2 through side-on configuration in Mn-WO 3 has the highest degree of polarization, which is 6, 3, and 2 times as high as that for end-on in WO 3 , end-on at W site in Mn-WO 3 , and end-on at Mn site in Mn-WO 3 , respectively.
Furthermore, we also calculated the adsorption energies of H 2 O*, H*, and OH* on the surface of WO 3 (200) and Mn-WO 3 (200), respectively. As displayed in Table S5 Figure 5F shows the partial density of states (PDOS) of WO 3 (200) and Mn-WO 3 (200) after the adsorption of N 2 molecules. For Mn-WO 3 (200), d-orbitals of metals and p-orbitals of N 2 molecules overlap to a large extent, contributing to the electron transfer from metal sites to adsorbed N 2 molecules. [52][53][54] The p-orbitals of N 2 over Mn-WO 3 become more broadened compared with that of WO 3 , which is caused by the hybridization between nitrogen p-orbitals and metal dorbitals. Meanwhile, the broadened nitrogen p-orbitals over Mn-WO 3 have both empty and occupied orbitals around Fermi level, which could synergistically accept electron and back-donate to metal sites. It could accelerate the electron transfer between metal d-orbitals and nitrogen p-orbitals, as well as favor subsequent hydrogenation reactions. [55] Besides, the formation of Mn-W dual sites is also beneficial for the polarization of adsorbed N 2 molecules. The difference of electron density between two nitrogen atoms in adsorbed N 2 molecule over Mn-WO 3 (200) is much higher than that over WO 3 (200). Thus, N 2 molecules are obviously polarized by Mn-W dual sites, leading to elongated bond length of N≡N triple bond. Moreover, the reaction pathways over WO 3 (200) and Mn-WO 3 (200) were also screened. The simultaneous activation of two nitrogen atoms through side-on configuration greatly optimizes subsequent hydrogenation steps over Mn-WO 3 (200) ( Figure S26, Supporting Information). Meanwhile, the other possibilities of reaction paths were also considered and studied in Figures S27 and S28, Supporting Information. [56] As a result, the N 2 fixation over WO 3 (200) undergoes distal pathway, while Mn-WO 3 (200) favors associative alternating pathway (Figure 5G). More importantly, due to the inertness and stability of the N 2 molecules, the hydrogenation from N 2 * to NNH* with an energy of 0.810 eV is the rate-limiting step over WO 3 (200). While in Mn-WO 3 (200), the polarization effect from Mn-W dual sites greatly weaken the adsorbed N 2 molecules, hence the energy barrier of subsequent hydrogenation is reduced and the rate-limiting step is changed to the hydrogenation of HNNH 2 to H 2 NNH 2 with an energy of 0.591 eV.

Conclusion
In summary, we developed a highly active and stable photocatalyst toward N 2 photofixation by constructing Mn-W dual sites on WO 3 semiconductor. An impressive NH 3 production rate of 425 µmol g cat.
−1 h −1 is achieved over the optimal 2.3% Mn-WO 3 catalyst, which represents the best values among the tungsten oxide-based photocatalysts for N 2 fixation. Meanwhile, nearly ≈100% of initial activity can be retained after five successive reaction rounds. Quasi in situ XPS and XANES reveal that Mn-W dual sites can highly polarize the adsorbed N 2 molecules through interfacial electronic coupling. Further theoretical calculations demonstrate that the formation of Mn-W dual sites plays a pivotal role in promoting the adsorption and activation of N 2 molecules. This work not only opens an avenue to develop efficient photocatalysts for N 2 fixation under mild conditions via the construction of dual-metal sites, but also provides atomic-level insights into understanding the catalytic process.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.