Concurrent Mechanisms of Hot Electrons and Interfacial Water Molecule Ordering in Plasmon‐Enhanced Nitrogen Fixation

The participation of high‐energy hot electrons generated from the non‐radiative decay of localized surface plasmons is an important mechanism for promoting catalytic processes. Herein, another vital mechanism associated with the localized surface plasmon resonance (LSPR) effect, significantly contributing to the nitrogen reduction reaction (NRR), is found. That is to say, the LSPR‐induced strong localized electric fields can weaken the intermolecular hydrogen bonds and regulate the arrangement of water molecules at the solid–liquid interface. The AuCu pentacle nanoparticles with excellent light absorption ability and the capability to generate strong localized electric fields are chosen to demonstrate this effect. The in situ Raman spectra and theoretical calculations are employed to verify the mechanism at the molecular scale in a nitrogen fixation process. Meanwhile, due to the promoted electron transfer at the interface by the well‐ordered interfacial water, as well as the participation of high‐energy hot electrons, the optimal catalyst exhibits excellent performance with an NH3 yield of 52.09 µg h−1 cm−2 and Faradaic efficiency (FE) of 45.82% at ─0.20 V versus RHE. The results are significant for understanding the LSPR effect in catalysis and provide a new approach for regulating the reaction process.


Introduction
Ammonia (NH 3 ) is an inorganic chemical indispensable in industrial processes for the synthesis of fertilizers and pharmaceuticals. [1]Due to the large volume energy density, NH 3 is also considered an ideal hydrogen-rich clean fuel. [2]he Haber-Bosch process (N 2 + 3H 2 → 2NH 3 ), a common method for industrially producing NH 3 , requires high pressure and temperature due to the particularly large bond energy (941 kJ mol −1 ) of N 2 , [3] which consumes the substantial global annual energy supply and emits a considerable amount of the greenhouse gas CO 2 . [4]As a mild approach, aqueous electrochemical nitrogen reduction reaction (NRR) for NH 3 production (N 2 + 6H + + 6e − → 2NH 3 ) is a promising technology substituting the energy-intensive Haber-Bosch process, which utilizes hydrogen atoms in solution as an alternative source to reduce the consumption of fossil fuels and the emission of CO 2 . [5]Nevertheless, the cleavage of the non-polar N 2 triple bond needs a large amount of energy.The severe kinetic barrier in this activation process hinders the energy coupling between the dinitrogen species and the electrons. [6]As a result, the mass transfer efficiency at the solid-liquid reaction sites is retarded, leading to a challenge to achieve efficient nitrogen fixation in practice. [7]herefore, exploring a strategy for activating inert N 2 molecules is crucial to facilitate the reaction kinetics in catalytic processes under ambient conditions.
Plasmonic metal nanoparticles, as a potential novel photosensitizer in the nitrogen fixation process, exhibit two recognized energy transfer mechanisms.It can exert a localized surface plasmon resonance (LSPR) effect to collect photon energy and generate high-energy hot electrons that can achieve indirect energy transfer and promote the breaking of N≡N bonds. [8]On the other hand, plasmonic metal nanoparticles can also lead to direct energy transfer by constructing complexes with adsorbates, in which the hot electrons can be motivated at the metal-adsorbate hybrid interface, and then, dissociate N≡N bonds. [9]Therefore, the hot electrons generated by the LSPR effect play a crucial role in promoting the catalytic reaction process.
However, it is reported that the LSPR effect can also disorder the hydrogen bonding between water molecules in the solution, [10] which is applied in elevating the content of dissolved O 2 and restraining NO release from inflammatory cells in the biology field.Regarding the NRR, such a process involves the activation of N≡N bonds and the transfer of multi-proton supplied by water dissociation.It is most likely that the state of water molecules can play a significant role in the hydrogenation process during the NRR.Therefore, investigating the water molecule behavior at the solid-liquid interface in the NRR enhanced by the LSPR effect is highly valuable.It may remarkably contribute to the NRR performance, which is rarely studied.
In this work, we report the LSPR effect for enhanced NRR.We find a new mechanism of the LSPR effect beyond the known, generating hot electrons in the NRR process; that is, under the liquid environment and light conditions, surface plasmons can induce water molecules to an ordered distribution at the solid-liquid interface, facilitating the NRR.This phenomenon is evidenced at the molecular level by in situ Raman spectra and theoretical calculations.Twinned AuCu pentacle nanoparticles, which can effectively trap light, are utilized.The structure of the twin crystals can facilitate the rate-determining step (RDS, *N 2 -to-*NNH), which is proved by density functional theory (DFT) calculations.More importantly, owing to the modulation of the interface water molecular states and the injection of the generated high-energy hot electron by the LSPR effect, the optimized twinned AuCu pentacle nanoparticles achieve the supreme NH 3 yield of 52.09 μg h −1 cm −2 and Faradaic efficiency (FE) of 45.82% at −0.20 V versus RHE.

Characterization and Hot-Electron Generation of AuCu Pentacle Nanoparticles
The AuCu pentacle nanoparticles were manufactured by an oil bath.As shown in Figure 1a, the transmission electron micro-scope (TEM) image (inset) and the scanning electron microscope (SEM) image manifest a pentacle nanostructure with four branches of 35 nm in average length and a shorter one around 17 nm.The branch angles are ≈72°.The pentacle nanoparticles with long branch lengths of 50 and 100 nm were additionally prepared by changing the experimental condition (Figure S1, Supporting Information).The X-ray diffraction (XRD) was carried out to examine the composition of the nanoparticles (Figure S2, Supporting Information), in which the characteristic signals located between the planes of pure Au and Cu could be indexed to a face-centered cubic nanostructure. [11]Meanwhile, high-resolution TEM (HRTEM) was performed to investigate the crystalline structure of the central (selected area b in Figure 1a) and the branch (selected area d in Figure 1a) of the pentacle nanoparticles (Figure 1b-e).With the observation along the [110] axis, the well-ordered lattice fringes with a lattice distance of 0.23 nm corresponded to the {111} planes.Figure 1c,e demonstrate the corresponding selected area electron diffraction (SAED) patterns, in which the diffraction spots in circles and triangles belong to the {111} and {200} planes, respectively, further indicating that the {111} twinning planes extended outward from the center of the nanoparticle to the ends of the branches.The scanning TEM (STEM) and energy-dispersive X-ray spectroscopy (EDXS) were then performed to investigate the spatial distribution of the elements.As shown in Figure 1f, the elemental mapping reveals the homogeneous loading of the Au and Cu on the alloy structure.Moreover, by changing the ratio of Au and Cu sources during the preparation process, pentacle nanoparticles with different atomic ratios of Au and Cu could be obtained.The values were confirmed by the inductively coupled plasma atomic emission spectroscopy (ICP-AES).The atomic ratios of Au:Cu were collected to be 0.96:0.04,0.91:0.09,0.87:0.13,and 0.83:0.17,denoted as Au 0.96 Cu 0.04 , Au 0.91 Cu 0.09 , Au 0.87 Cu 0.13 , and Au 0.83 Cu 0.17 , respectively.The high-resolution X-ray photoelectron spectroscopy (XPS) of Au 4f and Cu 2p are displayed in Figure 1g; Figure S3, Supporting Information.Due to the spinorbit coupling, Au 4f spectra demonstrated 4f 7/2 and 4f 5/2 at ≈84.2 and 87.9 eV doublet peaks; while, the Cu 2p spectra were composed of 2p 3/2 and 2p 1/2 at ≈931.7 and 951.2 eV, testifying that both Au and Cu existed in the metallic states in the alloy. [12]Significantly, the binding energy of Au 4f exhibited a slight negative shift with an increase in Cu content.At the same time, Cu 2p displayed a slight positive shift, which was attributed to the charge transfer from Cu to Au, [13] and such charge migration could lead to electron-deficient Cu sites; therefore, enhancing the activation of the adsorbed N 2 molecules (Figure S4, Supporting Information).
As seen in Figure 1h; Figure S5, Supporting Information, the extinction spectra are used to identify the unique LSPR properties of the pentacle nanoparticles.The characteristic absorption peaks reveal a significant blue shift as the particle size decreases.For the 35 nm-branch nanoparticles, three well-defined peaks are located at ≈ 1100, 750, and 550 nm.Femtosecond transient absorption spectroscopy (fs-TAS) is then performed to verify the generation of plasmonic-induced hot electrons in the Au 0.87 Cu 0.13 pentacle nanoparticles under light illumination.As shown in Figure 1i, the fs-TAS spectra collected under different pump-probe time delays show significant absorption bleaching at ≈550 and 750 nm, corresponding well to the steady-state absorption peaks in Figure 1h.These results illustrate the existence of abundant hot electrons formed through the LSPR effect: under 325 nm pump beam illumination, the d-band electrons near the Fermi level of gold and copper can be excited to the higher-lying energy states in the sp band through single-photon absorption processes, [14] and these photoexcited electrons will then decay non-radiatively into the two LSPR bands with an exponentially decreased population redistribution; and hence, promote hot-electron-mediated absorption of the white-light probe beam at the two LSPR bands. [15]In addition, the shorter the pump-probe time delay (>0.2 ps), the stronger the absorption bleaching (i.e., the deeper the TAS spectral dips).Fitting the fs-TAS decay kinetics profiles at 550 and 750 nm reveals two characteristic decay time constants for each wavelength: the fast components are ≈1.7 and 2.1 ps; corresponding hot electrons rapidly thermalize through electron-phonon scattering, and then, lose energy to the lattice on the few picosecond time scale, such as recombination with the holes created in the d band. [16]The slow components are ≈62.4 and 75.3 ps, which is energy transfer to the environment process, [16b] probably corresponding to the ultrafast transfer of hot electrons to some electron acceptors in the solution.Such robust generation of hot electrons in these binaryalloyed nanoparticles indicates the feasibility of light-triggered catalytic reactions such as NRR.Considering their stronger absorption than the other samples in the wavelength range of the solar simulator, the 35 nm-branch Au 0.87 Cu 0.13 nanoparticles will be selected to investigate the catalytic performance later.

Plasmon Field Modulated Interfacial Water Molecules
In addition to the generation of hot electrons, LSPR induces strong near-field localization and enhancement.To disclose this effect, which is favorable for NRR, we perform the finitedifference time-domain (FDTD) calculations to simulate the electric field distribution profiles in Au 0.87 Cu 0.13 pentacle nanoparticles.As shown in Figure 2a,b, the results at the two dominant absorption peaks demonstrate significant electric field intensity enhancements, which tend to increase from the center of the pentacle nanoparticle to the edge of each branch and reach the maximum at the branch tip.This manifests that the as-prepared Au 0.87 Cu 0.13 pentacle nanoparticles can produce a favorable surface-enhanced Raman scattering (SERS) effect, which is well suitable for probing the ordering behavior and vibrational state of water molecules at the solid-liquid interface.As displayed in Figure 2c, the in situ Raman measurements are then performed in an N 2 -saturated 0.1 m Na 2 SO 4 electrolyte to explore the bonding and ordering of interfacial water molecules and related intermediates during NRR at the molecular level.The Raman peaks at 1355 and 1585 cm −1 belong to the characteristic D and G vibrational modes of the glassy carbon electrode; the 983 cm −1 peak is assigned to the stretching mode of SO 4 2− , and the two broad bands located within ≈250-1100 cm −1 and ≈3000-3800 cm −1 correspond to libration and stretching modes of H 2 O. [17] The signal intensity of the band at ≈250-1100 cm −1 is enhanced with the increase of laser intensity, which is related to the ordered arrangement of water molecules at the solidliquid interface. [18]Gaussian fitting is carried out further to confirm the behavior of the interfacial water molecules (Figure 2d; Figure S6, Supporting Information).The O─H stretching band can be decomposed into three peaks, which are assigned to three states of the interfacial water molecules, including two types of strong hydrogen bonding (denoted as SHB 4 ─H 2 O with four hydrogen bonds and SHB 2 ─H 2 O with two hydrogen bonds) and a kind of weak hydrogen bonding (denoted as WHB─H 2 O with Na + union).Meanwhile, the two peaks centered at 1153 and 1533 cm −1 in the Raman spectra of the N 2 -saturated solution (Figure 2c-e) can be assigned to ─NH 2 and ─HNNH, [19] respectively, manifesting the occurrence of N 2 -to-NH 3 conversion, whereas no nitrogen related peaks can be observed in the Raman spectra measured in the Ar-saturated electrolyte (Figure S7, Supporting Information).
The correlation between the Raman peak area proportion of water molecules in different states and the variation tendency of product signal intensities are then combined to study the effect of water molecule arrangement on the NRR process.As shown in Figure 2f, when the laser intensity is increased, the component of SHB 4 ─H 2 O reduces from 18.1% to 1.3%, SHB 2 ─H 2 O decreases from 81.3% to 74.9%, and WHB─H 2 O rises from 0.6% to 23.8%.This is because the increased laser intensity induces a stronger LSPR effect and larger electric near-field enhancement at the interface.Under co-interaction with the Na + ions and the enhanced interfacial electric fields, the water molecules exhibit a more distinct ordered distribution with hydrogen atoms anchoring to the metal surface.Further, based on the normalized extinction spectra in Figure 1h, we select an excitation laser source with a wavelength of 488 nm to minimize the excitation of the LSPR effect of the catalyst and investigate the water molecules in different states (Figures S8-S10, Supporting Information).The peaks related to the ordered arrangement of water molecules located between ≈250 and 1100 cm −1 display no noticeable change (Figure S8, Supporting Information); meanwhile, the component of SHB 4 ─H 2 O, SHB 2 ─H 2 O, and WHB─H 2 O maintains stability as the laser intensity increases (Figure S10, Supporting Information).This phenomenon proves that the LSPR effect makes the water molecules orderly arranged on the catalyst surface.The interaction between the water molecules and the catalyst can also be demonstrated by in situ Fourier transform infrared (FTIR) spectra (Figure S11, Supporting Information).Meanwhile, the signal intensity of ─NH 2 rises from 466.3 to 479.8, and that of ─HNNH varies from 474.5 to 481.0 as the ratio of WHB─H 2 O increases.This result is further confirmed by theoretical calculations (Figure S12, Supporting Information).In our calculations, a WHB─H 2 O cluster is placed on the Au 0.87 Cu 0.13 twin boundary (TB) under an applied electric field.The metal─H bond length decreases with increasing the electric field strength, which can narrow the energy difference between the metal Fermi level and the empty antibonding orbital of the water molecule.This can enhance the interaction of the metal─H bond, [18] thereby promoting the generation of protons that directly participate in the hydrogenation reaction and ultimately facilitating the NRR process.In addition, the adsorbing hydrogen on the catalytic surface is energetically unfavorable, so the adsorbed hydrogens are easily desorbed, avoiding being used to enhance HER competitive reactions (Figure S13, Supporting Information).Therefore, the LSPR effect is confirmed to promote water split, and free energy for hydrogen adsorption proves the catalyst surface has poor hydrogen adsorption capacity, so, the increased adsorbed hydrogen can easily desorb to form protons and then participate in the hydrogenation in NRR.Moreover, this strategy of gently providing protons differs from the method of adjusting pH to directly change the proton concentration in the solution.It can control the proton concentration within a suitable range that is beneficial to the NRR reaction.
Combining the above experiment and calculation results, we propose the possible schematic diagram depicted in Figure 2g.
Compared with most water molecules with a chaotic arrangement, under the interaction of the strong electric field generated from the LSPR effect and the Na + ions, the water molecules demonstrate weakened hydrogen bonds, along with the ordered distribution at the solid-liquid interface.The distance of water molecules to the catalyst surface is also shortened in such an entropy decrease process.These enhance the electron transfer efficiency and the generation of protons, which is beneficial for + samples and electrolytes with 15 N 2 and 14 N 2 as the feed gases.h) The quantitative comparison of NH 3 yield rate and FE recorded by UV-vis and NMR methods.i) The performance comparison of Au 0.87 Cu 0.13 -PEC (marked with a star) with previously reported works for NRR, in which S means literature in Supporting Information.
reducing the hydrogenation barriers, and eventually, for facilitating kinetics in the NRR process.Overall, the redistribution of interfacial water molecules is another mechanism other than the hot electrons induced by the LSPR effect, remarkably contributing to the effective NRR reaction.

NRR Performances
We further characterize the performance of AuCu pentacle nanoparticles for both photocatalytic (PC)-and photoelectrochemical (PEC)-NRR without any sacrificial agent.The PC-NRR performance of Au 0.87 Cu 0.13 (Au 0.87 Cu 0.13 -PC) is first investigated by calculating the apparent quantum efficiency (AQE) at different wavelengths.As shown in Figure 3a; Figure S14, Supporting Information, the AQE values of Au 0.87 Cu 0.13 -PC for NH 3 production are closely related to the LSPR extinction spectra, demonstrating a maximum efficiency of 0.84% at 750 nm.Under 750 nm illu-mination, the AQE rises from 0.83% to 0.85%, 0.96% and 1.17%; while, the NH 3 yield rate increases from 7.35 to 15.05, 25.50, and  41.44 μg h −1 mg −1 at the light intensity varying from 10 to 20, 30, and 40 mW cm −2 (Figure 3b).These results indicate that the LSPR effect significantly promotes the NRR process, which is attributed to the facilitated charge transfer at the interface due to the ordered arrangement of interface water molecules.As the light intensity further increases, the growth of AQE slows down significantly (1.23% at 50 mW cm −2 , 1.24% at 60 mW cm −2 ), which may be restricted to the sluggish kinetics at the solidliquid-gas interface. [20]he promotion effect of the LSPR is further demonstrated in PEC tests.As shown in linear sweep voltammetry (LSV) curves in Figure 3c, apparent enhancement in current density under light illumination suggests an improved catalytic activity toward NRR.In Figure 3d, the corresponding Tafel slope of 135 mV dec −1 under light illumination (Au 0.87 Cu 0.13 -PEC) is lower than that only in electrocatalytic (EC)-NRR (Au 0.87 Cu 0.13 -EC, 255 mV dec −1 ), which suggests the accelerated kinetics in NRR because of the participation of hot electrons and more unobstructed charge transfer caused by well-ordered water molecules at the solidliquid interface.In addition, as the light intensity increases, the current density of Au 0.87 Cu 0.13 -PEC rises (Figure S15a, Supporting Information), and the most sensitive light response is displayed at 100 mW cm −2 illumination (Figure S15b, Supporting Information), manifesting a more substantial LSPR effect is excited and more hot electrons are generated boost the NRR reaction.
The performance of the samples with different atomic ratios of Au:Cu was also investigated.The NH 3 yield rate and FE of samples were examined by the indophenol blue method in N 2saturated 0.1 m Na 2 SO 4 electrolyte without any sacrificial agent.The concentration of the produced NH 3 could be quantitatively detected according to the calibration curves in Figure S16, Supporting Information.As the ratio decreased (i.e., the Cu content increased), the current density, NH 3 yield rate, and FE increased until the ratio dropped to 0.87:0.13(Figure S17, Supporting Information), which was consistent with the electrochemical impedance spectroscopy (EIS) shown in Figure S18, Supporting Information.The smaller semicircle diameter at the medium frequency and the greater slope at the low frequency of the Au 0.87 Cu 0.13 -PEC manifested the faster charge and mass transfer kinetics compared with the control samples. [21]Nevertheless, excessive Cu atoms could create severe disorder in the crystal lattice, resulting in stronger incoherent scattering of electron waves, which retarded the electron transport kinetics and led to performance degradation in NRR. [22]he NH 3 yield rate and FE of Au 0.87 Cu 0.13 -PEC and Au 0.87 Cu 0.13 -EC were then studied in N 2 -saturated 0.1 m Na 2 SO 4 electrolyte without any sacrificial agent further to evaluate the promotion of the LSPR effect in NRR. [23]The chronoamperometry test for 2 h under different potentials from −0.10 to −0.40 V versus RHE is shown in Figure S19, Supporting Information.For the Au 0.87 Cu 0.13 -EC in Figure 3e, the NH 3 yield rate and FE achieved the maximum of 5.74 μg h −1 cm −2 and 13.09% at −0.30 V versus RHE.Nevertheless, the NH 3 yield rate and FE decreased with more negative potentials because the protons competitively adsorbed on the nanoparticle surfaces at a large potential to produce hydrogen. [24]In contrast, the performance of Au 0.87 Cu 0.13 -PEC also suggests similar trends but a much more significant improvement at each of these potentials.As mentioned, the LSPR effect was proven to induce water molecules with ordered distribution at the solid-liquid interface.Such tuned water with a shortened metal─H bond length could facilitate the hydrogenation reaction and decrease the reaction barrier in the NRR process.Consequently, the highest NH 3 yield rate and corresponding FE of Au 0.87 Cu 0.13 -PEC reached 52.09 μg h −1 cm −2 and 45.82% at a more positive potential of −0.20 V versus RHE.Further, to exclude photothermal effects from evaluating the contribution of the LSPR effect in NRR, the infrared thermal imaging of the Au 0.87 Cu 0.13 pentacle nanoparticles was carried out (Figure S20a, Supporting Information).As shown in Figure S20b, Supporting Information, the temperature increased rapidly, and then, maintained at ≈58 °C.Performance evaluation was then conducted at this temperature (Figure S21, Supporting Information).Compared with the NH 3 yield rate obtained at room temperature in EC-NRR, the enhanced NH 3 yield rate in thermocatalysis may be because of the improved kinetics of diffusion processes at an increased temperature. [25]he NH 3 yield rate in thermocatalysis process at all potentials indicated obvious gaps compared to PEC (Figure S21a, Supporting Information).Meanwhile, only the temperature increase did not significantly impact the NH 3 FE (Figure S21b, Supporting Information).These phenomena further manifested that the predominant improvement in performance stemmed from the non-thermal effects in the LSPR effect excited by light illumination.In addition, the by-product of hydrazine was determined based on the Watt-Chrisp method (Figure S22, Supporting Information), but almost no hydrazine could be detected within the potential range. [26]As displayed in Figure 3f, the Au 0.87 Cu 0.13 -PEC revealed a stable NH 3 yield rate and FE of ≈49.67 μg h −1 cm −2 and 44.56% during ten-time cycling (2 h for each cycle).
Further, to confirm the source of the N in N 2 -to-NH 3 , isotopelabeled measurements were performed by employing 14 N 2 and 15 N 2 as feed gases. [27]After a 2 h chronoamperometry test at −0.20 V versus RHE, the production was analyzed by 1 H nuclear magnetic resonance (NMR).As shown in Figure 3g, three characteristic peaks with a spacing of 52 Hz belonging to 14 NH 4 + could be observed, [28] which corresponds to the standard 14 NH 4 + sample.After replacing 14 N 2 with 15 N 2 as the feed gas, only two typical split signals with a coupling constant of 72 Hz indexed to the 15 NH 4 + could be obtained, [29] in accordance with the standard curves.This result excluded interference from other nitrogen sources in the environment and certified that the detected NH 3 was generated from the NRR rather than by contamination.The quantitative NMR was also carried out by using maleic acid as an internal standard (Figure S23, Supporting Information), in which the 15 NH 3 yield rate and FE were collected at 51.13 μg h −1 cm −2 and 44.49% (Figure 3h), in agreement with the UV-vis measurements, further testifying the source of the produced NH 3 during NRR experiments.To the best of our knowledge, the performance of the Au 0.87 Cu 0.13 -PEC exceeds most previously reported works for EC-, PC-, and PEC-NRR (Figure 3i; Tables S1-S3, Supporting Information).

Mechanism Investigations and DFT Calculation
To dynamically monitor N 2 -to-NH 3 conversion at a molecular level, the in situ attenuated total reflectance FTIR (ATR-FTIR) spectroscopy was investigated on the Au 0.87 Cu 0.13 electrode at −0.2 V versus RHE in N 2 -saturated 0.1 m Na 2 SO 4 .As shown in Figure 4a,b, the stretching vibration peak belonging to the adsorbed reactant (*N 2 ) was located at 1670 cm −1 .Meanwhile, the intensities of some positive-going bands associated with the reaction intermediates were obviously improved by increasing the electroreduction time from 0 to 300 s.The peaks at 1110, 1342, and 1516/1551 cm −1 were assigned to ─N─N stretching, ─NH 2 wagging, and ─H─N─H bending vibrations, indicating the generation of *N 2 H y intermediates and manifesting the NRR process follows the associative mechanism. [30]To further investigate the reaction mechanism and clarify the origin of the excellent performance of Au 0.87 Cu 0.13 pentacle nanoparticles in the NRR process, DFT calculations were subsequently carried out.The electric field of 1.0 × 10 8 V m −1 was employed to simulate the high-intensity localized electric fields originating from the LSPR effect of Au 0.87 Cu 0.13 alloy excited by light illumination.The free energy diagrams and the corresponding optimized structures of the reaction intermediates on the Au 0.87 Cu 0.13 TB and (111) plane are demonstrated in Figure 4c,d.Two association reaction mechanisms, including distal and alternating pathways, were considered, in which the protonation sequence of the former was the remote N priority; while, the remote and proximal N atoms were alternately hydrogenated in the latter pathway.Nevertheless, the alternate pathway was more likely to occur due to the high energy barrier of *NNH 2 to *N in the distal pathway, and Au 0.87 Cu 0.13 TB with an electric field greatly reduced this barrier, accelerating the conversion of intermediates and reducing the accumulation of *NNH 2 ; thus, promoting this pathway.In both pathways, the sample applied with an electric field revealed smaller adsorption energy of N 2 on the surface of catalysts compared to the counterparts, indicating a stronger interaction on the active sites.Generally, the uphill free energy variation of hydrogenation steps in both pathways revealed that the Au 0.87 Cu 0.13 TB displayed a smaller slope than that of the Au 0.87 Cu 0.13 (111) plane, suggesting that the TB could enhance the catalytic activity in the NRR.Further, for the rate-determining step (RDS) from *N 2 to *NNH in associative distal and alternating pathways, the Au 0.87 Cu 0.13 TB with electric field manifested the smallest variation of the free energy, indicating that the TB and LSPR effect could greatly reduce the energy barrier of the reaction and significantly promote the NRR process.

Conclusion
In summary, we find that, except for the hot electron effect for LSPR in promoting the catalytic reaction, another mechanism exists.This is proved by both experimental and theoretical studies with the twinned AuCu pentacle nanoparticles in an NRR process as model material and reactions.Under light illumination, strong localized electric fields are generated near the surface of the catalyst.Due to interfacial electrostatic interaction, the water molecules bonding with cations in the solution achieve weakened intermolecular hydrogen bonding, approaching the catalyst surface and forming an ordered structure.This entropy decrease process can enhance energy conversion and promote charge transfer efficiency and reaction rate.Moreover, with the participation of hot electrons generated by the LSPR effect and the existence of material twins that can adjust the protonation process of the reaction intermediates, the optimal catalyst exhibits supreme NH 3 yield of 52.09 μg h −1 cm −2 and FE of 45.82% at −0.20 V versus RHE.This work elucidates another essential mechanism of the LSPR effect in promoting the catalytic reaction process and achieves an efficient nitrogen fixation under ambient conditions.The results can be generalized to regulate the interfacial water state to promote catalytic reaction rates.

Figure 1 .
Figure 1.The morphology, composition, and plasmonic properties of AuCu binary alloyed nanoparticles.a) SEM image.The inset is a TEM image of an individual AuCu nanoparticle.b) HRTEM image of the selected area in the inset of (a) and c) the corresponding SAED pattern.d) HRTEM image of the selected area in the inset of (a) and e) the corresponding SAED pattern.In (c,e), the diffraction spots in circles and triangles belong to the {111} and {200} planes, respectively.f) STEM image and EDXS elemental mappings.g) High-resolution XPS of Au 4f of the samples with different Cu/Au ratios.h) Normalized extinction spectra of nanoparticles with various sizes measured in aqueous suspensions.i) Femtosecond transient absorption spectra (fs-TAS, pump beam at  = 325 nm) of 35 nm-branch Au 0.87 Cu 0.13 in solution and j) the corresponding transient kinetics decay spectra at two TAS dips in (550 and 750 nm) (i).Scale bar, 100 nm in (a) and 20 nm in the inset of (a).Scale bar, 2 nm in (b,d) and 2 1/nm in (c,e).Scale bar, 20 nm in (f).

Figure 2 .
Figure 2. Electric field simulation, in situ Raman characterization, and schematic diagram.Simulated electric field intensity enhancement |E/E 0 | 2 of the LSPR modes at a) 850 nm and b) 1160 nm for the 35 nm-branched Au 0.87 Cu 0.13 pentacle nanoparticle.c) In situ Raman spectra normalized by the signal intensity of carbon, d) Gaussian fitting of O─H stretching modes, and e) contour maps measured under different laser intensities in the N 2 -saturated electrolyte.f) Laser intensity-dependent population of interfacial water with various vibrational modes and Raman signal intensities of ─NH 2 and ─HNNH obtained from (c,d).g) Schematic of water molecular behavior in the NRR process without (left) and with (right) light illumination, indicating plasmon-triggered ordering of interfacial water molecules.In (d,g), O, H, Na, Au, Cu, and N atoms are indicated by black, white, blue, yellow, red, and purple, respectively.

Figure 3 .
Figure 3.The catalytic performance of Au 0.87 Cu 0.13 in NRR.a) The AQE for PC NRR and the UV-vis extinction spectrum of Au 0.87 Cu 0.13 .b) NH 3 yield rate and AQE of Au 0.87 Cu 0.13 -PC with different light intensities at 750 nm illumination.c) LSV curves collected in Ar-and N 2 -saturated electrolytes.d) Tafel plots, e) NH 3 yield rate and FE of Au 0.87 Cu 0.13 -PEC and Au 0.87 Cu 0.13 -EC at different potentials.The inset in (e) is the yield rate of Au 0.87 Cu 0.13 -EC at −0.10 and −0.15 V. f) NH 3 yield rate and FE of Au 0.87 Cu 0.13 -PEC at −0.20 V collected by ten-time cycling.g) 1 H NMR analysis of standard 15 NH 4 + and 14 NH 4+ samples and electrolytes with 15 N 2 and 14 N 2 as the feed gases.h) The quantitative comparison of NH 3 yield rate and FE recorded by UV-vis and NMR methods.i) The performance comparison of Au 0.87 Cu 0.13 -PEC (marked with a star) with previously reported works for NRR, in which S means literature in Supporting Information.

Figure 4 .
Figure 4. Mechanism investigations and DFT computational results of NRR process.a) Time-resolved in situ electrochemical ATR-FTIR spectra measured at −0.2 V versus RHE in N 2 -saturated 0.1 m Na 2 SO 4 and b) the corresponding contour maps.The free energy profile of associative c) distal and d) alternating pathways and the corresponding optimized structure simulated with the electric field (F) of 1.0 × 10 8 V m −1 , in which * is a surface-active site, TB is twin boundary, and H, Au, Cu, and N atoms are indicated by white, yellow, red and purple, respectively.