Nitrogen Photoelectrochemical Reduction on TiB2 Surface Plasmon Coupling Allows Us to Reach Enhanced Efficiency of Ammonia Production

Ammonia is one of the most widely produced chemicals worldwide, which is consumed in the fertilizer industry and is also considered an interesting alternative in energy storage. However, common ammonia production is energy-demanding and leads to high CO2 emissions. Thus, the development of alternative ammonia production methods based on available raw materials (air, for example) and renewable energy sources is highly demanding. In this work, we demonstrated the utilization of TiB2 nanostructures sandwiched between coupled plasmonic nanostructures (gold nanoparticles and gold grating) for photoelectrochemical (PEC) nitrogen reduction and selective ammonia production. The utilization of the coupled plasmon structure allows us to reach efficient sunlight capture with a subdiffraction concentration of light energy in the space, where the catalytically active TiB2 flakes were placed. As a result, PEC experiments performed at −0.2 V (vs. RHE) and simulated sunlight illumination give the 535.2 and 491.3 μg h–1 mgcat–1 ammonia yields, respectively, with the utilization of pure nitrogen and air as a nitrogen source. In addition, a number of control experiments confirm the key role of plasmon coupling in increasing the ammonia yield, the selectivity of ammonia production, and the durability of the proposed system. Finally, we have performed a series of numerical and quantum mechanical calculations to evaluate the plasmonic contribution to the activation of nitrogen on the TiB2 surface, indicating an increase in the catalytic activity under the plasmon-generated electric field.

cavitation field in the ultrasonic reactor for 60 min. According to the mechanical design of the ultrasonic horn and the input power of the ultrasonic generator (1 kW) at atmospheric pressure, the output intensity was calculated to be 100 W cm -2 at an amplitude of 80 μm. Centrifugation (3000 rpm for 10 min) was then used to remove the poorly dispersed material.
Preparation of TiB 2 @AuNPs composites. To a 4 mL aqueous suspension of TiB 2 flakes (1 g in 100 mL) was added 12 mL of HAuCl 4 (0.1 mM). The mixture was sonicated for 30 min. The resulting composite mixture was then purified three times by centrifugation, washing with methanol, re-dispersion, and subsequent dispersion in methanol.
Au grating preparation. The periodic surface of the DVD with a pattern area of 0.7x2 cm 2 was coated with gold film about 30 nm thick by vacuum sputtering (DC Ar plasma, gas purity of 99.995 %, gas pressure of 4 Pa, discharge power of 7.5 W, sputtering time 300 s).
Deposition of TiB 2 @AuNPs on Au grating. TiB 2 @AuNPs were deposited on the surface of an Au grating using an improved technique, resulting in a thin and homogenous coating of Au grating on TiB 2 @AuNPs (the prepared samples are referred to as Au grating/TiB 2 @AuNPs). The optimization was carried out in accordance with the approach described in 1 , and as a final step, we utilized spin-coating deposition from a TiB 2 @AuNPs solution in methanol (1.1 mg.mL -1 ) at 500 rpm.

Measurement Techniques
The peak force AFM measurements were conducted using the Icon (Brucker) microscope.
The HRTEM-EDX mapping was performed using Jeol 2200 FS microscope (Jeol, Japan). TEM images of TiB 2 and TiB 2 @AuNPs flakes were obtained with the help of a JEOL JEM-1010 transmission electron microscope with a SIS MegaView III digital camera. SEM-EDX photos and maps were obtained on Lyra3 GMU (Tescan, CR) microscope with an accelerating voltage of 2 kV. The X-ray photoelectron spectroscopy (XPS) was performed using a Thermo Fisher Scientific XPS NEXSA spectrometer with a monochromated Al K Alpha X-ray source working at 1486.6 eV.
X-Ray diffraction measurements were carried out using the Empyrean, Malvern Panalytical diffractometer with Cu K <α> radiation source in 2θ-θ diffraction mode, the Bragg-Brentano geometry.
Raman spectra were collected using a Thermo Scientific DXR Raman microscope equipped with a 532 nm excitation wavelength. Raman mapping was performed across a surface area of 1.5×0.7 mm 2 , having 70×40 points spaced by a gap of 0.025 mm.
UV-Vis absorbance spectra of the samples were measured using a Lambda 25 spectrometer (PerkinElmer, USA). Reflection spectra of the samples were obtained by using a HR2000 (Ocean Optics) spectrometer using the AvaLight-DHS light source (Avantes).
3 1 H NMR were recorded on Bruker Avance III™ (500 MHz) spectrometer. The rate of H 2 evolution was estimated by GC-7920 (Agilent) gas chromatography system.

Electrochemical and photoelectrochemical nitrogen reduction
Electrochemical measurements were carried out using a Palm Sens 4 potentiostat (Palm Instruments, Netherlands) controlled by the PSTrace 5.9 program using a three-electrode twocompartment electrolytic cell (H-type), which was separated by a Nafion 117 membrane. Before experiments, the Nafion membrane was immersed in 3 wt. % H 2 O 2 solution at 90 o C for one h. Then it was cleaned in deionized water, boiled in 5 wt. % H 2 SO 4 water solution, and cleaned again in deionized water for 2 h. The Au grating and Au grating/TiB 2 @AuNPs samples were used as the working electrode (active surface area -0.7×0.6 cm 2 ). An Ag/AgCl (sat. with 3M KCl) electrode (BVT Technologies, CZ), placed in the same part of the cell as the working electrode, was used as a Moreover, electrochemical measurements were performed without and with the illumination of the sample's surface using the solar simulator (Solar Simulator SciSun-300, Class AAA, and the intensity on the sample surface were adjusted to 100 mW cm −2 ). After measurements, all potentials were converted into the reversible hydrogen electrode (RHE) potential. All experiments, except the stability tests, were replicated five times and calculated values of standard deviation were subsequently used as error bars.

Photocatalytic nitrogen reduction
The photocatalytic (PC) nitrogen reduction was conducted at ambient temperature and pressure. A solar simulator (Solar Simulator SciSun-300, Class AAA, and the intensity on a sample surface was adjusted to 100 mW cm −2 ) was used as an artificial source of sunlight. The 0.1 M Na 2 SO 4 mixture solution containing 20 wt. % ([C4mpyr][eFAP]) was purged with "pure" nitrogen for 30 min before the experiment and then continuously purged with nitrogen during the PC experiment. The Au grating/TiB 2 @AuNPs was immersed in a clear glass beaker filled with electrolyte, and the unit was illuminated for one hour using a simulator. At the end of the experiment, the sample was removed, and the reaction solution was analyzed by a photometric kit test.

Quantification of NH 3
The NH 3 produced was quantitatively determined by the ammonia photometric kit. According to the procedure described in the ammonia test documents, the following steps were performed: 5 mL of the reaction solution (after electrochemical measures) was mixed with 0.6 mL of reagent #1 (containing sodium hydroxide). Then some amount of reagent #2 (containing thymol) was added to the resulting solution and shaken vigorously until the reagent was completely dissolved. After 5 min, reagent #3 (containing 2-propanol) was added to the reaction solution and stirred. The resultant solution was left to stand for 5 min at room temperature and then analyzed by UV-Vis absorption spectroscopy. For the creation of the calibration curve, the known concentration of NH 4 Cl was added to 0.1 M Na 2 SO 4 and 0.1 NaOH (to introduce the similar values of pH for "real" and calibration solutions) mixture solution containing 20 wt. % ([C4mpyr][eFAP]) and analyzed by the method described above. The absorbance intensity at ∼ 692 nm was utilized to estimate the yield of ammonia based on the standard curve.
The NH 3 yields (as a function of catalyst loafing or electrode surface area) were calculated by the equations 1: where is the total amount of NH 3 (measured by photometric test), is the volume of the

Impact of N 2 gas(es) impurity
Nitrate (NO 3 − ) and nitrite (NO 2 − ) concentrations were quantitatively determined by the nitrate and nitrite photometric kit tests, respectively.
The following steps were taken according to the instructions described in the nitrate photometric kit test document: after purging with 15 N 2 (refer to section 3.3 Control experiments), 0.5 mL of 0.1 M Na 2 SO 4 solution was added to 4.0 mL of reagent #1, which contains sulfuric and phosphoric acids. The resulting solution was subsequently mixed with 0.5 mL of reagent #2, which contains 2,6-dimethylphenol, and left to stand for 10 min at room temperature before being analyzed by UV-Vis spectroscopy analysis. The absorbance intensity at ∼ 507 nm was utilized to estimate the yield of nitrate based on the standard curve.
According to the procedure described in the nitrite photometric kit test document, the following steps were performed: some amount of reagent #1 (containing sulfanilic acid) was added to 5 mL 0.1 M Na 2 SO 4 solution after purging with 15 N 2 (refer to section 3.3 Control experiments) and shaken vigorously until the reagent was completely dissolved. The resulting mixture was left to stand for 10 min and then analyzed by UV-Vis spectroscopy. The absorbance intensity at ∼ 540 nm was utilized to estimate the yield of nitrite based on the standard curve.
For the creation of the calibration curves, the known concentrations of NaNO 3 or NaNO 2 were added to 0.1 M Na 2 SO 4 solution and analyzed according to the procedure described above.

Control experiments
Detection of hydrazine.

Calculation of TiB 2 @AuNPs electronic structure
Calculation of the valence and conductive bands gap position of TiB 2 @AuNPs was performed using the combination of Tauc (created from UV-Vis measurements) and Mott-Schottky (created from EIS measurements) plots (see related discussion below).

Finite-Difference Time-Domain simulation
The simulation was carried out using MEEP software, which employs the finite-difference time-domain (FDTD) technique. The simulated system consists of a gold grating on a polymer substrate (AFM-measured profile, grating shape) with metal nanoparticles at the "top" of the grating separated by TiB 2 flakes. To ensure convergence, the simulation was carried out at a resolution of 2000 px/µm. A broadband Gaussian source was used to stimulate the simulated cell. Following the completion of the time-domain simulation, the data were converted to the frequency domain using the FFT technique 3 .

Calculation of free energy profile
Estimation of the electric field on the nanoparticle surface after irradiation with a 100 mW cm -2 laser was performed according following prerequisites: Let us assume that the laser radiation flux density: where is the volumetric energy density, is the speed of light .

Density functional calculation of plasmon-assisted NRR
The DFT calculations were carried out with the CP2K package. 5 The DFT calculations used the combination of the Gaussian and plane-wave (GPW) 6

Figs. S2, S3 -related discussion (appearance of AuNPs)
To check formation of all AuNPs on the TiB 2 surface, and possible overspending of relatively expensive gold, we carried out two control measurements. First, the TEM results ( Figure S2) showed that all features characteristic for AuNPs (visible as dark regions) lie in the TiB 2 spatial region (TiB 2, which is more transparent to electrons and visible as grey regions). Next, we performed precipitation of AuNPs and TiB 2 @AuNPs suspension under mild conditions (7800 rpm and 20 min) and subjected the remaining supernatants to UV-Vis absorption measurements ( Figure S3). In the case of the control sample (AuNPs suspension), precipitation under mild conditions did not lead to complete sedimentation of AuNPs and their characteristic bands remained visible. In the case of TiB 2 @AuNPs, UV-Vis spectroscopy showed the absence of previously clearly visible absorption bands (including the characteristic band of plasmon absorption), which indicates the complete sedimentation of the "heavier" structures (i.e., TiB 2 @AuNPs flakes) and the absence of "free" AuNPs, which, according to the control experiment, should stay in solution.

Figure S17
Control NMR measurement of ( 15 NH 4 ) 2 SO 4 -the chemical shift area, attributed to peaks from 14 N "impurities" is magnified in an insert.

Figure S18
Comparison of non-isotopic and isotopic ammonia yield rate and Faradaic efficiency quantified by photometric kit test.

Figure S27
Relationship between calculated bands structure of TiB 2 @AuNPs flakes and potentials of NRR proceeding.

Figs. S22 and S27 -description note
The band gap of TiB 2 @AuNPs nanostructures was calculated from the Tauc plot ( Figure   S26A), created from the UV-Vis spectrum ( Figure S3 (A)). The band gap value was found to be .
2.81 eV. The type of TiB 2 semiconductor was determined in previous work 13 . The flat-band potentials of TiB 2 @AuNPs (E fb vs. Ag/AgCl) were measured individually using the Schottky-Mott plot and found to be -0.8 V ( Figure S26B). Using a potential difference of 0.21 V, the potential measured relative to the Ag/AgCl (3M KCl) reference was converted to the normal hydrogen electrode (NHE) potential. As a consequence, the predicted flat-band locations of TiB 2 @AuNPs were 24 -0.6 V vs. NHE. Because the E fb for undoped n-type semiconductors is about 0.3 V below the conduction band minimum (CBM), the CBM of TiB 2 @AuNPs was determined to be -0.9 V vs. NHE.
The valence band maximum (VBM) was calculated using the location of the CBM and the band gap to be 1.91 V vs. NHE. Based on these results, Figure S27 presents the energy band values of TiB 2 @AuNPs compared to N 2 /NH 3 , H + /H 0 , or N 2 /N 2 H 4 reaction potentials.

Figure S28
Impact of spacer thickness of SPP-LSP coupling (local value of plasmon related electric field).