High-performance and stable NH₃ production using a TiO₂-protected Si photocathode and patterned Au loading

Abstract

Crystalline silicon (c-Si) is a promising material for photoelectrochemical (PEC) ammonia (NH₃) production from nitrate (NO₃⁻) reduction owing to its appropriate band gap and optimal charge-transport properties. However, c-Si is not stable in aqueous solutions, causing the detachment of catalysts from the c-Si photoelectrode and resulting in a dramatic decrease in the performance. Furthermore, electrocatalysts on c-Si block light, therby reducing the PEC NH₃-production efficiency. Herein, we stabilized and increased the efficiency of the c-Si photocathode by TiO₂ deposition and loaded an optimized amount of Au using an e-beam patterning, respectively. We found that TiO₂ not only protects the c-Si photoelectrode from the electrolyte but also promotes strong bonding between Au and the c-Si photoelectrode. Notably, TiO₂ showed a synergistic effect with the Au electrocatalyst in increasing the faradaic efficiency (FE) of NO₃⁻ reduction for NH₃ production, which was further confirmed by density functional theory calculations. Overall, the Au-loaded TiO₂-protected c-Si photoelectrode showed a stable and record-high NH₃-production rate of 1590 ± 40 μg_NH3 cm⁻² h⁻¹ with an FE of 83.4% ± 5.6% at −0.35 V vs. the reversible hydrogen electrode.

---

Broader context

Developing a photoelectrochemical (PEC) device for the conversion of nitrate to ammonia is a promising approach for decentralized ammonia production and sustainable wastewater treatment process. In this work, we developed a highly stable and high-performance photocathode by depositing a gold (Au) pattern on the surface of a silicon (c-Si) wafer using a photolithography technique for NH₃ production. Our photoelectrode exhibited a record-high NH₃-production rate of 1590 ± 40 μg_NH3 cm⁻² h⁻¹ with an 83.4% ± 5.6% faradaic efficiency (FE) and stable long-term operation under 44 h continuous illumination. The high production rate of the Si–TiO₂–Au photocathode can be attributed to the role of the Au pattern in circumventing the rate-determining step (RDS) in NO₃⁻ reduction (NO₃⁻ → NO₂⁻) and the excellent catalytic role of the ALD-TiO₂ layer in NH₃ production. The synergistic effect of Au and TiO₂ could be scaled up through photolithography. This offers a promising photocathode for the PEC conversion of nitrate ions into ammonia.

1. Introduction

The exponential growth in the global population over the past two decades has posed significant challenges to the expansion of food production.^1,2 Nitrogen-based fertilizers, which primarily rely on ammonia (NH₃) that comprises 80% of these fertilizers by volume serve as the primary source of nutrition for the agricultural produce within the food supply chain.^3,4 Additionally, NH₃ offers promise as an energy carrier owing to its notably high volumetric energy density (15.6 MJ L⁻¹), surpassing that of Li-ion batteries (1.73 MJ L⁻¹) and compressed H₂ (5.5 MJ L⁻¹ at 70 MPa).^3–6 NH₃ is predominantly produced using the energy-intensive Haber–Bosch (HB) process that operates under high temperature and pressure over 250 °C and 300 atm, respectively.^7,8 This process accounts for more than 1–2% of the world's energy consumption and 1.6% of worldwide CO₂ emissions.^7,9 Hence, recent efforts have been dedicated to developing more energy-efficient and less CO₂-emitting methods for NH₃-production processes.^10–14 Photoelectrochemical (PEC) NH₃ production is a promising technique, because it can be operated at ambient temperature and pressure without the need for expensive H₂ and without emitting CO₂.^15,16

Crystalline silicon (c-Si) is a promising material for PEC NH₃ production because it can absorb a wide range of the solar spectrum due to its small band gap of 1.1 eV and high carrier mobility.^17–19 Recently, several works have reported on the application of c-Si photoelectrodes for the PEC nitrogen (N₂)-reduction reaction (NRR).^20–23 For example, MacFarlane et al. utilized Au-electrocatalyst-loaded c-Si and achieved an NH₃-production rate of 1.33 μg_NH3 cm⁻² h⁻¹ under 2 suns illumination.²⁰ Zheng et al. applied the aerophilic polytetrafluoroethylene (PTFE) to a Au-loaded c-Si photocathode to increase the Faraday efficiency (FE) for the NRR, and achieved a NH₃-production rate of ∼18.9 μg cm⁻² h⁻¹ with 37.8% NH₃ selectivity at −0.2 V vs. the reversible hydrogen electrode (RHE).²⁴

To commercialize PEC NH₃ production, the NH₃-production rate needs to be more than ∼1000 μg NH₃ cm⁻² h⁻¹, which is equivalent to a current density of approximately 10 mA cm⁻².^25,26 However, the NRR partial current density in aqueous electrolytes is usually less than 1 mA cm⁻², which correspond to less than a 100 μg NH₃ cm⁻² h⁻¹ NH₃ production rate (about 10-times lower compared to the NO₃-reduction reaction), because less N₂ (0.0001 M) is dissolved in the water and N≡N bonds are hard to selectively dissociate.^27,28 Replacing N₂ with nitrate (NO₃⁻) as a nitrogen-containing source can address some of the drawbacks of the NRR, because NO₃⁻ is very soluble in water (0.5–2 M) and NO₃⁻ can be easily converted into NH₃ through nitrite (NO₂⁻) with a high FE.² Thus NH₃ production from NO₃⁻ may be a promising way to not only maximize the NH₃ productivity but also for the remediation of NH₃ from waste streams and seawater.²⁹

Lee et al. used Au-decorated c-Si nanowires for the PEC reduction of NO₃⁻ to NH₃, achieving a high NH₃ FE of 95.6% at a positive applied potential (0.2 V vs. RHE).¹⁶ However, the NH₃-production rate remained very low (>10 μg_NH3 h⁻¹ cm⁻²), and both the FE and NH₃-production rate gradually decreased due to the detachment of Au nanoparticles (NPs) from the electrode surface.¹⁶ Au-loaded c-Si photocathodes showed low stability and performance for PEC NH₃ production, due to the severe photocorrosion and chemical etching of the c-Si photoelectrode, especially in alkaline media. The continuous etching of c-Si in the corrosive electrolyte not only diminished its performance but also led to the detachment of the electrocatalyst from the photoelectrode.¹⁶

In the present study, we demonstrated high-performance and stable NH₃ production by protecting the c-Si photocathode with a TiO₂ layer deposited by atomic layer deposition (ALD), followed by patterned Au deposition. The TiO₂ layer could not only protect the c-Si surface from electrolyte degradation, but could also improve the adhesion between the Au catalyst and the Si. By optimizing the amount of Au loading through the photolithography technique, we were able to minimize the light absorption by the Au catalyst while maintaining a high selectivity for NH₃ production. Our Au-loaded TiO₂ protected c-Si photocathode (Si/TiO₂/Au) exhibited a record-high production rate of 1590 ± 40 μg_NH3 cm⁻² h⁻¹, which is higher than the NH₃-production rate criteria for the practical application of PEC NH₃ production. Moreover, the FE of NH₃ was as high as 83.4 ± 5.6%, and Si/TiO₂/Au demonstrated stability during long-term operation under 44 h continuous illumination. The high production rate of the Si/TiO₂/Au photocathode could be attributed to the role of Au in circumventing the rate-determining step (RDS) in NO₃⁻ reduction (NO₃⁻ → NO₂⁻) and the catalytic role of the ALD-TiO₂ layer in NH₃ production, which was further verified by density functional theory (DFT) calculations. These results indicate that TiO₂ not only protects the c-Si surface from Au electrocatalyst detachment, but also plays an important role in the selective NO₃⁻ reduction reaction to produce NH₃.

2. Results and discussion

2.1. Fabrication process and PEC performance of Si/TiO₂/Au

A thin layer of TiO₂ (∼20 nm) was deposited on the surface of a cleaned HF-treated p-type silicon (p-Si) wafer using the ALD technique (ESI,† Fig. S1). Photolithography was conducted using a Cr photomask to print square patterns with different sizes and densities on the surface of p-Si.^30,31 After spin-coating and soft baking of the photoresist, an MIF 300 developer was used to imprint the desired pattern on the Si/TiO₂ substrate. Finally, a 100 nm metallic film (Au) was deposited by an e-beam evaporator (ESI,† Fig. S2), and the desired pattern was lifted by washing the resist in acetone solution for 60 s (Fig. 1(a)). To remove any extra photoresist, an oxygen plasma treatment was utilized to clean the photocathode surface. Fig. 1(b) and ESI,† Fig. S3 show top-view microscopy and angled scanning electron microscopy (SEM) images of the Au pattern with a 4% density on the surface of Si/TiO₂ (Si/TiO₂/Au-4). Energy-dispersive spectroscopy (EDS) elemental mapping indicated that the Au patterns, without any imperfection or missing squares, occupied about 4% of the photoelectrode surface, leaving 96% of the surface free for light penetration into the p-Si/TiO₂ substrate and for electron–hole generation (Fig. 1(b) and ESI,† Fig. S4). Using this photolithography method, we also loaded other patterned metals, such as Ag, Cu, and Pd, on Si/TiO₂ photoelectrode. Since Au was the most effective in the NO₃⁻ reduction reaction to produce NH₃, we selected Au as the main electrocatalyst (Fig. 1(c), (d), and ESI,† Fig. S5). The determined band alignments of p-Si with and without metal catalyst doping are depicted in ESI,† Fig. S6. A large downward band bending of 0.4 eV was observed for the p-Si photoelectrode without Cu or Au catalyst doping. The photogenerated electrons are transferred to the conduction band of silicon upon light irradiation. However, due to the small band gap, the photoexcited electrons recombine, likely leading to an inactive surface for NO₃⁻ reduction (ESI,† Fig. S6a). After Cu metal doping, the band bending was slightly decreased, resulting in an improvement in the photocurrent density and onset potential (ESI,† Fig. S6b). When Au catalyst was deposited on the surface, the Schottky barrier created due to its high work function facilitated electron transfer to the surface. This barrier reduced the backward flow of electrons from the Au to the p-Si, improving the NITRR kinetics and suppressing the recombination of photogenerated electron–hole pairs (ESI,† Fig. S6c). However, due to the weak Cu work function, some of the photogenerated electrons likely returned to p-Si, reducing the PEC performance.

Fig. 1 (a) Schematic of the Si/TiO₂/Au-4 photocathode fabrication process. An ultrathin layer of TiO₂ was deposited on the surface of p-Si by atomic layer deposition (ALD), and then a low density 20 μm × 20 μm pattern was printed onto the Si/TiO₂ substrate using a photolithography process, and then, 100 nm Au was deposited and lifted off in acetone. (b) SEM image and elemental mapping of Au patterns on the Si/TiO₂ surface, red (Si) and green (gold). (c) J–V curves for the photocathodes for patterns with different metals in 0.5 M NO₃⁻ and 1.0 M KOH (pH = 14). (d) NH₃-production rates and FEs for NH₃ and NO₂⁻ with the Si/TiO₂/M-4 photocathode (M: Au, Pd, Cu, Ag) with different metal pattern deposition at −0.35 V vs. RHE after 2 h in 1.0 M KOH and 0.5 M NO₃⁻, (e) J–V curves for SiTiO₂ and Si/TiO₂/Au-4 for the electrolyte containing 0.5 M KNO₃ (solid line) and KNO₂ (grid line) at pH = 14. (f) FE% and ¹⁵NH₄⁺ concentration for isotopically labelled 0.5 M ¹⁵NO₃⁻ as a nitrogen-containing source in 1.0 M KOH. (g) ¹H-NMR graphs for a control electrolyte containing the same molar ratio of ¹⁴NO₃⁻ and ¹⁵NO₃⁻.

We first investigated the role of the TiO₂ layer in the PEC NO₃⁻ reduction reaction to produce NH₃. As shown in Fig. 1(c) and ESI,† Fig. S7a, the onset potential was shifted by 0.21 V toward a more positive potential, and the PEC performance was improved by nearly 27% for Si/TiO₂ compared to the Si photoelectrodes. Furthermore, Si/TiO₂ demonstrated a higher NH₃ FE of 60% compared to 40% for the bare Si (ESI,† Fig. S7b). Therefore, the ALD-deposited TiO₂ layer could not only protect the Si surface from chemical etching, but also improved the PEC performance and selectivity for NO₃⁻ reduction to NH₃. The role of Au on bare p-Si and Si/TiO₂ was also examined. In both cases, Au loading improved the onset potential and photocurrent density due to the catalytic activity of Au active sites for the reduction of NO₃⁻ ions (ESI,† Fig. S7a). Subsequently, with an optimized Au pattern on the TiO₂ layer, the NH₃ FE was improved to 83.4 ± 5.6%. This improvement could be attributed to the critical role of the Au pattern in converting NO₃⁻ to NO₂⁻ ions, which is well-known to be the RDS (eqn (1)).³² To elaborate, the PEC performance was recorded for Si/TiO₂ and Si/TiO₂/Au-4 in the presence of NO₃⁻ and NO₂⁻ aqueous solutions. As shown in Fig. 1(e), the onset potential clearly shifted for Si/TiO₂ in the presence of NO₂⁻ compared to NO₃⁻. In contrast, the onset potential and PEC performance for Si/TiO₂/Au-4 changed only slightly, indicating the crucial role of Au in suppressing the activation energy for the sluggish conversion of NO₃⁻ to NO₂⁻ in the RDS steps.³² Element mapping of the Au pattern after 24 h continuous NO₃⁻ conversion to NH₃ was performed for the elements N, K, Si, and Au. As shown in ESI,† Fig. S8, only on the surface of the Au pattern were both N and K detected, corresponding to the attachment of KNO₃ on the surface, while on the Si surface these elements were missing. This clearly demonstrated the critical role of the Au pattern in the adsorption and conversion of NO₃⁻ ions.

Subsequently, it could be concluded that the Au pattern facilitated the sluggish 2e reduction of NO₃⁻ to NO₂⁻ (eqn (1)). In the next step (eqn (2)), intermediate NO₂⁻ ions are converted to NH₃ mainly on the TiO₂ surface. Therefore, the kinetics and thermodynamics of NO₃⁻ reduction to NH₃ (eqn (3)) are accelerated due to the synergistic effect of both the Au and TiO₂ layers, which enhances the RDS 2e⁻ reduction of NO₃⁻ to NO₂⁻ (eqn (1)) and the subsequent 6e⁻ reduction reaction to NH₃ (eqn (2)). The results and explanations above were further validated through DFT calculations, as reported in the final section.

NO₃⁻ + H₂O + 2e⁻ → NO₂⁻ + 2OH⁻ (1)

NO₂⁻ + 6H₂O + 6e⁻ → NH₄OH + 7OH⁻ (2)

NO₃⁻ + 7H₂O + 8e⁻ → NH₄OH + 9OH⁻ (3)

To elaborate the competitive reactions of the NITRR and HER, which can occur simultaneously on the photocathode surface, the performance of the Si/TiO₂/Au-4 photocathode was measured in the presence and absence of NO₃⁻ ions under 1 sun illumination (ESI,† Fig. S9a). The onset potential was shifted about 0.25 V toward a more positive potential in the presence of NO₃⁻ ions, attributed to the less thermodynamically demanding NO₃⁻ reduction reaction (NITRR) compared to the HER.¹⁶ Moreover, an increased photocurrent density of about 20% was observed for the NITRR compared to that of the HER. To further assess the performance of Si/TiO₂/Au-4, we quantitatively measured the products in the cathode compartment in the presence and absence of NO₃⁻ ions (ESI,† Fig. S9b). For this control experiment, an H-cell configuration was used, with the catholyte connected to a gas chromatography (GC) instrument to measure the H₂ evolution. Additionally, a syringe sampling system was employed to take samples from the catholyte solution to measure the NH₃ and NO₂⁻ concentrations using indophenol blue and Griess procedures, respectively (ESI,† Fig. S10–S12). As shown in ESI,† Fig. S9b, in the absence of NO₃⁻ ions, only H₂ was evolved. In contrast, in the presence of NO₃⁻, the HER reaction was effectively suppressed and NH₃ was produced continuously without H₂ gas bubbling. The influence of the product evolution during the PEC performance of Si/TiO₂/Au-4 is shown in ESI,† Fig. S9c. It can be seen an stable NH₃ production rate with no hydrogen gas evolution.

Another critical point is that we precisely identified the source of NH₃ production, attributing it to the reduction of NO₃⁻ ions rather than contamination (e.g., NH₃ in air and human breath) or formation from other nitrogen-containing sources (e.g., N₂, nitrite, amines, and nitrogen oxides, as shown in ESI,† Fig. S13). We next conducted ¹⁵NO₃⁻ isotope labelling, which allowed us to carefully identify and quantify the source and amount of ¹⁵NH₃ production via a ¹H-NMR calibration curve (ESI,† Fig. S14a and b). Specifically, we observed that the ¹H-NMR doublet peak intensity increased over time, as shown in ESI,† Fig. S14c, leading to a consistent increase in ¹⁵NH₄⁺ concentration, proving that ¹⁵NO₃⁻ ions were the only source of NH₃ production (Fig. 1(f)). Furthermore, a control experiment was carried out with a mixture of ¹⁵NO₃⁻ and ¹⁴NO₃⁻ ions as nitrogen-containing sources. The ¹H-NMR spectra presented a mixed spectrum of doublet and triplet peaks attributed to the presence of ¹⁴NH₄⁺ and ¹⁵NH₄⁺ (Fig. 1(g)).

2.2. Optimization of the Au loading amount on Si/TiO₂

As depicted in ESI,† Fig. S15, an optimal surface coverage by the Au pattern is critical for NH₃ production. When only a small portion of the photoelectrode's surface was covered by the Au pattern, the reduction of NO₃⁻ to NH₃ was not efficient; while when the coverage percentage of Au was too high, Au blocked the light, resulting in a dramatic loss in PEC performance (ESI,† Fig. S15). Thus, we optimized the amount of the Au in Si/TiO₂/Au-X (X = 0.0, 0.5, 4.0, 25, 50, 100) photocathodes (X correlates to the surface percentage of the photoelectrode that was covered by Au, e.g., X = 100 means the entire surface of the photoelectrode was covered by Au, while X = 4 corresponds to only 4% surface coverage, leaving 96% free for light absorption). The morphologies of the different Au-patterned configurations and their surface coverages are shown in Fig. 2(a). It can be seen that with the increase in Au-pattern density, a larger portion of the Si/TiO₂ surface was covered, varying between 0.5% to 50%. Fig. 2(b) shows the influence of the Au-pattern coverage on the photocurrent density and onset potential for the Si/TiO₂/Au-X photoelectrodes. It can be that the photocurrent density was improved for Si/TiO₂/Au-0.5 compared to that of Si/TiO₂, and was a maximum for Si/TiO₂/Au-4. As shown in Fig. 2(b), the photoelectrodes with a higher density of Au patterns (e.g., Si/TiO₂/Au-25 and Si/TiO₂/Au-50) showed reduced photocurrent densities due to the decreased carrier generation by Si because of the light blockage by the Au pattern (ESI,† Fig. S16). The roles of the Au-pattern configurations on the FE of NH₃ and NO₂⁻ and on the selectivity of NH₃ are shown in Fig. 2(c) and (d). As shown in Fig. 2(c), the NH₃ FE increased with increasing the Au-pattern density, reaching a maximum value of 89% NH₃ FE for Si/TiO₂/Au-50, while the FE of NO₂⁻ continuously decreased to 11%. In contrast, the NH₃-production rate was maximized for the 4% Au-pattern photoelectrode (Si/TiO₂/Au-4) with an average of 1590 ± 40 μg cm⁻² h⁻¹ (Fig. 2(d)). This obtained NH₃-production rate is competitive among other reported PEC NO₃⁻-conversion rates for silicon photocathodes used in PEC NH₃ production (ESI,† Table S1).

Fig. 2 (a) SEM images of Au pattern with different pattern densities on the surface of Si/TiO₂ substrate, (b) J–V curves for photocathodes with different patterns in 0.5 M NO₃⁻ and 1.0 M KOH (pH 14), (c) FE%s for NH₃, and NO₂⁻ for photocathodes with different Au-pattern densities, (d) selectivity and NH₃ production rate for different Au-pattern densities at −0.35 V vs. RHE after 2 h in 1.0 M KOH and 0.5 M NO₃⁻, (e) NH₃-production rate vs. FE (%), comparison of the current research with previous studies for PEC NH₃ production, for more information see ESI,† Table S1 and the references in the supporting file.

Although the NH₃ FE and FE_NH3/FE_NO2 were increased to 88% and 7 for Si/TiO₂/Au-50, respectively, the NH₃-production rate decreased abruptly to 535 μg cm⁻² h⁻¹. This could be attributed to the 50% reduction in light absorption by Si/TiO₂/Au-50, consequently leading to deficient electron–hole generation (ESI,† Fig. S16). Next, we quantified the NH₃ concentration using both H-NMR and indophenol blue for different Au patterns in order to compare the reliability of the detection procedure. ESI,† Fig. S17 shows there was a good correlation between both detection methods for quantifying NH₃ in solution.

To shed more light on the interconnected role of light absorption and pattern density, a COMSOL Multiphysics simulation was carried out.^33–35 Here, we calculated the transmitted irradiance and assumed that all transmitted light could participate in charge-carrier generation by p-Si. The detailed computational method presented herein can be found in the “Simulation Method”. ESI,† Fig. S18 and S19 demonstrate that the transmitted light was negligible for the Si/TiO₂/Au-100 nm photocathode, leading to the termination of charge-carrier generation by p-Si. However, in the case of Si/TiO₂/Au-4 after light illumination, the onset potential was shifted to 0.15 V vs. RHE and the photocurrent density increased to −30 mA cm⁻² in −0.9 V vs. RHE, representing a dramatic change with the dark condition (ESI,† Fig. S20). Furthermore, we checked NH₃ production under light and dark conditions for Si/TiO₂/Au-4 (ESI,† Fig. S21). Under light irradiation, NH₃ production followed the theoretical values, under dark conditions the NH₃ production dropped to nearly zero. The current densities of the Si/TiO₂/Au-4 photoelectrode and the Au electrode alone are shown in ESI,† Fig. S22. The Si/TiO₂/Au-4 photoelectrode exhibited an onset potential of 0.15 V vs. RHE for NO₃⁻ conversion to NH₃, while the Au electrode only showed activity at potentials as low as −0.4 V vs. RHE. Furthermore, the absolute reflectances of the Si/TiO₂/Au-X (X = 0, 0.5, 4, 25, 50, 100) photoelectrodes were measured and are shown in ESI,† Fig. S23. It was observed that Si/TiO₂ reflected about 31.7% of the incident light due to the shiny surface of both the Si and TiO₂ layers. The Au pattern with only 0.5% surface coverage slightly increased the light reflectance, while the Au patterns with 4%, 25%, and 50% coverages increased the incident light proportionally increased the light reflectance in the visible region. Furthermore, 100% surface coverage with Au deposition could reflect 100% of the incident light, effectively acting like a mirror and preventing light transmission into the Si photoelectrode. We next investigated the effect of the applied potential and pH on the activity and NH₃ FE of Si/TiO₂/Au-4 as the most effective photoelectrode for the conversion of NO₃⁻ to NH₃. All the possible reduction reactions products, including H₂, NH₃, and NO₂ were measured and are shown in ESI,† Fig. S24a at different applied potentials. ESI,† Fig. S25a indicates that at low cathodic applied potentials, NH₃ and NO₂ were the main products on the surface of the Si/Ti/Au-4 photocathode. Although the NH₃ FE at 0.044 V vs. RHE was comparable to that at −0.35 V vs. RHE, the photocurrent density for the NITRR was much lower, resulting in a lower NH₃-production rate (ESI,† Fig. S24b). At an applied potential of −0.35 V vs. RHE, Si/TiO₂/Au-4 showed a maximum NH₃ FE of 83.4 ± 5.6%. With increasing the applied potential to more negative regions (e.g., −1.05 or −1.35 V vs. RHE), the HER reaction became more competitive with the NITRR, leading to an FE of about 40% corresponding to H₂ and 30% related to NH₃. One of the most influential operational parameters in the PEC conversion of NO₃⁻ to NH₃ is the pH of the electrolyte. Consequently, we measured the LSV for Si/TiO₂/Au-4 in three different pH media: acidic (0.5 M H₂SO₄), neutral (0.5 M Na₂SO₄), and alkaline (1.0 M KOH) with the same concentration of NO₃⁻ (0.5 M KNO₃). As can be seen in ESI,† Fig. S24c and d, the Si/TiO₂/Au-4 photoelectrode showed the highest performance in the alkaline medium. The FEs for the possible products were measured at −0.35 V vs. RHE in both acidic and alkaline media, and are shown in ESI,† Fig. S24d. The results indicated that in the acidic medium, HER became a more competitive reaction with the NITRR, while in the alkaline medium, NH₃ was the main product. In the acidic medium, only 25% of the photocurrent was utilized for transforming NO₃⁻ to NH₃ (FE < 25%), while about 70% of the photocurrent was utilized for the conversion to H₂ gas. In the basic solution (KOH 1.0 M), NO₃⁻ reduction to NH₃ predominantly occurred, with an NH₃ FE of >88% ± 5%, while the HER reaction was effectively suppressed and no gas bubble were observed on the surface of the photocathode.

The concentration of NO₃⁻ ions can also influence the performance and NH₃ FE of Si/TiO₂/Au-4, as shown in ESI,† Fig. S25. It could be observed that when the NO₃⁻ concentration was decreased to 0.1 M (i.e., 5 times less), the current density decreased by 50% compared to that of the 0.5 M concentration (ESI,† Fig. S25a and b). Surprisingly, the Si/TiO₂/Au-4 photoelectrode maintained a high NH₃ FE even at the lower concentration of 0.025 M (ESI,† Fig. S25c). This demonstrates the reliability of the photoelectrode for converting low concentrations of NO₃⁻ ions into NH₃. Furthermore, we measured the performance of Si/TiO₂/Au-4 in the presence of SO₄²⁻ and NO₃⁻ with different NO₃⁻ concentrations at neutral pH (ESI,† Fig. S26). The results indicated that even in the presence of interfering ions, such as SO₄²⁻, the performance of the photoelectrode was only slightly changed with decreasing the NO₃⁻ concentration.

2.3. Stability of the Si/TiO₂/Au-4 photoelectrode

In order to investigate the stability of the Si/TiO₂/Au-4 photoelectrode, its top-view surface morphology was analyzed by SEM before and after PEC NO₃⁻ reduction (Fig. 3(a) and (b)). The SEM images of Si before PEC measurement showed it had a smooth surface, while after immersing the photocathode in the electrolyte and starting the experiments in KOH solution, a 3D pyramidal structure was observed on the surface of p-Si. This could be attributed to the anisotropic chemical etching of the silicon surface by the KOH aqueous solution, which is a phenomenon that has been known about for a long time.^36–38 After the deposition of a thin TiO₂ protective layer, the Si/TiO₂ photocathode surface remained unchanged and could sustain its integrity after the PEC NO₃⁻-reduction experiments (Fig. 3(a) and (b)). Moreover, ESI,† Table S2 summarizes the concentrations of Si, Au, and Ti ions in the catholyte solution, in which the Si concentration reached 11.2 ppm for the Si photoelectrode compared to 0.1 ppm for Si/TiO₂ (ESI,† Fig. S27). Fig. 3(a) shows the detachment of part of the Au pattern in the Si/Au photoelectrode after the PEC experiments, resulting in an increase in Au ion concentration approaching 0.8 ppm. In contrast, the Au patterns remained unchanged on the surface of Si/TiO₂/Au-4 after PEC NO₃⁻ reduction, and there were no empty or damaged arrays, leading to a very low Au ion concentration of 0.001 ppm. Furthermore, the ICP results in ESI,† Table S2 indicate there was a very low concentration of Ti ions in the catholyte, highlighting the role of the TiO₂ layer as a protective surface. The impact of Au-pattern detachment on the PEC performance and NH₃ FE could also be observed in Fig. 3(c) and (d). Chronopotentiometry measurements were next performed, with a constant −0.2 V vs. RHE obtained for the Si/TiO₂/Au-4 photoelectrode, while the voltage continuously decreased for Si/Au to sustain the −10 mA cm⁻² current density. The NH₃ FE measurement also showed a stable trend for Si/TiO₂/Au-4, while it decreased for the Si/Au-4 photoelectrode. Finally, the long-term stability for PEC NO₃⁻ reduction to NH₃ was investigated for Si/TiO₂/Au-4 at −0.35 V vs. RHE for 44 h (Fig. 3(e)). A stable NO₃⁻-reduction photocurrent density was recorded, as well as an almost linear increase in NH₃ concentration over time. Fig. 3(f) shows that the NH₃ FE remained constant over 24 h of operation in the range of 83.4 ± 5.6%. Moreover, we investigated the surface morphology of the Si/TiO₂/Au-4 photoelectrode after 44 h continuous operation (ESI,† Fig. S28). Both the AFM and SEM images showed the clean and smooth surface the Au pattern with a height of about 100 ± 5 nm before starting the PEC performance tests. After 44 h PEC conversion of nitrate in the catholyte, the surface of the Au pattern became slightly rough (±10 nm), while the height of the Au pattern stayed in almost the same range.

Fig. 3 (a) and (b) Top-view SEM images of photoelectrodes before and after 24 h PEC nitrate (NO₃⁻)-reduction measurements, (c) chronopotentiometry (−10 mA cm⁻²) curves for Si/Au-4 and Si/TiO₂/Au-4 photocathodes in 0.5 M KNO₃ at pH 14, (d) FE (%) for Si/Au-4 and Si/TiO₂/Au-4 photocathodes in 0.5 M KNO₃ at pH = 14 J–V curves, (e) long-term chronoamperometry stability PEC NO₃⁻ reduction of Si/TiO₂/Au-4 at −0.35 V vs. RHE in 1 M KOH with 0.5 M NO₃⁻, and (f) evolution of the FEs and concentration of NH₃ and NO₂⁻ for 24 h continuous PEC NITRR.

2.4. Theoretical insight into the NO₃⁻-reduction mechanism

Next, we sought to gain an atomistic-scale understanding of the NO₃⁻-reduction reaction using density functional theory (DFT). The stable structures of rutile TiO₂(110), Au(111), and Si(111) were taken into account in our simulation (more information is provided in the computational method and ESI,† Fig. S29–S32). Electrochemical NO₃⁻ reaction involves an 8 proton–electron coupled mechanism (NO₃⁻ + 6H₂O + 8e⁻ → NH₃ + 9OH⁻), leading to many intermediate steps and by-products, as well as the competing hydrogen evolution reaction (HER). NO₃⁻ adsorption on the catalyst surface is the primary step for NH₃ synthesis ahead of the deoxygenation/hydrogenation steps. There are two adsorption modes for NO₃⁻ on the surface in which, namely, one or two metal-to-oxygen (M–O) bonds can be formed. As shown in Fig. 4(a), the preferred configuration for NO₃⁻ adsorption is on TiO₂, while Au is formed with two and one M–O bond, respectively, and Si strongly adsorbs NO₃⁻ (ΔG = −1.67 eV) with one M–O bond. The Au surface is a better adsorption site for NO₃⁻ than TiO₂, according to the free-energy diagram. The charges transferred from the substrate to adsorbed *NO₃ on Au and TiO₂ are 0.61 e and 0.51 e (Fig. 4(b)), respectively, which is further evidence of the potential of Au to adsorb NO₃⁻ ions. The *NO₃ protonation step (*NO₃ to *HNO₃) is downhill for all the catalysts, while on the Si surface the intermediate structure of *NO₂ + *OH is generated. The strong *NO₃ adsorption on Si causes an easy first protonation step, but the second step (*NO₂ + *OH to *NO₂) is the potential determining step (PDS) with a barrier energy of 1.46 eV, representing the highest barrier among the three catalysts, as well. First principle calculations disclosed that the desorption of *NO₂ species on the Au surface requires overcoming a large barrier of 1.8 eV, ensuring subsequent reactions towards the final product (NH₃). The free-energy change of the NO molecule (NO_(g) to *NO) on TiO₂ (−0.33 eV) is more negative than on Au (−0.12 eV), as shown in Fig. 4(c), demonstrating a better potential of TiO₂ for the subsequent reactions (*NO, *NOH, *N, *NH, *NH₂, *NH₃), most of which are exothermic processes. The free-energy change of *NO to *NOH acted as the PDS on TiO₂ with an ultralow limiting potential of −0.14 V, which is much lower than that of Au at −1.12 V. It is worth noting that even employing the DFT+U scheme did not significantly alter the results, in which the free-energy change for the protonation step of *NO to *NOH was downhill as well, with a small change of 0.09 eV for the adsorption energy of NO₃⁻ on TiO₂.

Fig. 4 (a) Free-energy diagram for electrochemical NO₃⁻ reduction to ammonia (NH₃) on TiO₂(110), Au(111), and Si(111) (b) Charge-density difference for Au(111) and TiO₂(110), where the yellow and cyan colours depict charge accumulation and charge depletion, respectively. (c) Adsorption free-energy changes of NO(g) to *NO on TiO₂(110) and Au(111). Colour code, Ti: light blue, Au: gold, O: red, N: blue, H: white.

The suppression of the competing HER reaction is imperative for an efficient NO₃⁻ reaction. We calculated the adsorption free energy of *H on the catalysts. The ΔG_*H (* + H⁺ + e⁻ → *H) on Au(111) was +0.26 eV, which was about two times larger than the PDS (0.14 eV) of the NITRR (ESI,† Fig. S33), demonstrating that NH₃ was by far the most generated product compared to H₂. Furthermore, the results showed that the ΔG_*H on TiO₂ as the main catalyst was 2.39 eV (metal site), indicating that the HER would be strongly suppressed. The more negative limiting potentials of the HER indicate that the NITRR would be the dominant reaction, leading to outstanding selectivity. In addition, protons (H⁺) are better nucleophiles than NO₃⁻ in the electrolyte, resulting in a better adsorption of NO₃⁻ at the electron-deficient metal centres.

In the first step, both Au and TiO₂ surfaces can adsorb NO₃⁻, with the adsorption free energy on Au (1.16 eV) being much lower than on TiO₂ (1.95 eV). Once the reaction is triggered, subsequent reactions can easily proceed on TiO₂ and Au, while NO gas can be chemisorbed on TiO₂ with a low limiting potential of −0.14 V for NH₃ production. In fact, the Au surface is effective at capturing NO₃⁻ ions and transforming them into *NO species, while TiO₂ mainly catalyzes NO₃⁻ to NH₃ with a largely reduced onset potential (0.14 eV). The PDS of Si is very high (1.46 eV), which prevents the efficient reduction of NO₃⁻ to NH₃, aligning with our experimental results. These DFT results demonstrate the synergistic effect of Au and TiO₂ layers for selective NH₃ production from NO₃⁻ reduction.

3. Conclusion

The synergistic role of Au pattern and TiO₂ layer was elaborated for the PEC conversion of NO₃⁻ ions into NH₃ using Si photocathodes. The ALD-TiO₂ layer demonstrated multifunctional duties, including chemical etching protection, serving as an adhesion layer for the catalyst/Si interface, and catalyzing NO₃⁻ reduction to NH₃. The Au pattern facilitated the 2e⁻ RDS of NO₃⁻ to NO₂⁻ reduction, resulting in enhanced PEC performance for Si/Au and Si/TiO₂/Au after printing the Au pattern on the electrodes. The ideal pattern density of Au was also investigated by determining the NH₃ production rate and FE for different pattern configurations. The Si/TiO₂/Au-4 photoelectrode, which had a 4% Au-pattern coverage of the Si/TiO₂ surface, demonstrated a high NH₃ production rate of 1590 ± 40 μg_NH3 cm⁻² h⁻¹ at −0.35 V vs. RHE, which is the highest value reported yet for PEC NH₃ production with silicon photocathodes.

4. Experimental section

4.1. Preparation of p-type silicon

As-supplied p-type boron-doped 〈100〉 silicon wafers (p-Si) with a resistivity of 10–20 Ohm m and 525 μm thickness (Atecom Ltd, Taiwan) were utilized as substrates for the fabrication of the photocathode. To remove the oxide layer (SiO₂), the p-Si wafers were immersed in 10% HF aqueous solution just before a thin layer of TiO₂ was deposited using the atomic layer deposition (ALD) technique.

4.2. Atomic layer deposition

A 20 nm TiO₂ layer was coated by ALD (Lucida D100) onto an HF-treated p-Si wafer using a TTiP precursor and H₂O at 150 °C at a flow rate of 6.5 ml min⁻¹. To achieve adequate vapour pressure, the C₈H₂₄N₄Ti was heated to 75 °C. A sequential process was utilized and adjusted for 20 nm deposition of the TiO₂ layer onto the p-Si substrate.

4.3. Photolithography process for the fabrication of Au patterns

The metal frame formation process included the standard photolithography process, metal deposition, and subsequent lift-off. In the photolithography process, the Cr photomask was designed to make patterns of different sizes and densities (5 × 5 μm, 20 × 20 μm low density, 20 × 20 μm high density, and 75 × 75 μm). The photoresist (PR, AZ5214E, positive) was spin-coated at 4000 rpm for 60 s, followed by soft baking at 95 °C for 60 s. The sample was illuminated by a UV lamp in a mask aligner through a Cr mask with an exposure dose of 90 mJ cm². After the exposure, the sample was developed in MIF 300 (metal-ion tree) developer for more than 50 s.^30,31,39

A 100 nm metallic film was deposited by an e-beam evaporator (KVE-E4000) without rotation at a rate of 0.3 Å s⁻¹ to fill the holes of the photoresist patterns. A thin 3 nm layer of Cr was evaporated before the metallic film to increase the adhesion of the metallic layer and TiO₂ frame. Finally, the desired pattern was obtained by lifting off the resist in acetone and by oxygen plasma treatment using a plasma asher system to remove extra PR residue.

In the Si/TiO₂/Au-X photoelectrode, where X is correlated to the percentage of the photoelectrode surface covered by Au: for example, X = 100 means the entire surface is covered by gold, which leads to minimal light absorption because gold typically absorbs light in the visible and near-infrared range. In the case of 0 < X < 100, as the gold coverage (X) decreases, a larger portion of the surface is exposed, allowing for increased light absorption by the underlying Si material. Finally, for X = 0, with no gold coverage, the photoelectrode has maximum light absorption, but the silicon surface is inactive for nitrate conversion to NH₃.

4.4. Characterization

The Si surface and Au pattern morphologies were analyzed by field emission scanning electron microscopy (FE-SEM, S4800, HITACHI) and energy-dispersive X-ray spectroscopy. UV-visible-diffuse reflectance spectroscopy (UV-DRS) was used to measure the reflectance of p-Si and the different patterns on the p-Si substrate using a UV-3600 instrument (Shimadzu). The Si and Au contents in the catholyte solutions were evaluated by inductively coupled plasma optical emission spectrometry (Varian 700-ES).

4.5. Photoelectrochemical nitrogen reduction

The PEC measurements were conducted using a customized H-type cell configuration with a Nafion membrane (Fuel Cell Store) to separate the catholyte and anolyte. The catholyte compartment was equipped with a gas-sealed Ar flow inlet and outlet, and was connected online to a gas chromatography system (GC, X) to detect the evolution of H₂ during the PEC measurements. Additionally, a gas-sealed syringe connection was installed in the catholyte for taking liquid samples for measurement of the NH₃ and NO₂⁻ concentrations using indophenol blue and Griess protocols, respectively. A potentiostat (IviumStat, Ivium Technologies) was used to collect and analyze the PEC data in a three-electrode system with Hg/HgO (or Ag/AgCl) as the reference electrode, Pt wire as the counter electrode, and Si/Ti/Au photocathodes as the working electrode. To simulate 1 sunlight illumination (100 mW cm⁻²), a 1.5 G filter was placed in front of a Xe lamp (300 W) and the light intensity was calibrated with a reference certified by NREL, USA. The pH of the electrolyte was adjusted using 1.0 M KOH, 0.5 M H₂SO₄, and K₂SO₄ for obtaining alkaline, acidic, and neutral electrolytes, respectively. Also, 30 ml of electrolyte containing NO₃⁻ at various concentrations was used in the catholyte for the PEC measurements. Before each measurement, the surface of the photocathodes was scanned by J–V to ensure a stable surface of the Si–Ti–Au photocathodes. The Faraday efficiencies (FEs) for NH₃ and NO₂ and the selectivity for NH₃ were calculated as follows:

4.6. Ammonia detection process

The ammonia concentration was measured using the well-known indophenol blue spectroscopy method.^7,8 Samples were appropriately diluted and mixed with 1 ml NaOH (1 M) containing 5 wt% salicylic acid and 5 wt% sodium citrate solution. Next, 0.5 ml of 0.05 M NaClO was added to the solution, followed by 0.1 ml sodium nitroferricyanide solution (1.0 wt%). The mixture was kept in the dark for 1 h before being analyzed with a UV-Vis spectrophotometer. The peak absorbance at 655 nm wavelength was recorded, and the concentrations were calculated using a standard concentration–absorbance calibration curve with NH₄Cl solution (ESI,† Fig. S10).

Furthermore, a 500 MHz nuclear magnetic resonance (NMR) spectrometer was used to measure the NH₃ concentration according to a previously reported procedure. After diluting the samples to fall within the detection range, they were mixed with 0.1 ml DMSO. The sample was measured with a Liquid Bruker Avance NMR spectrometer (ESI,† Fig. S11).

4.7. Nitrite-detection process

The NO₂⁻ concentration was also calculated using a spectroscopy procedure as previously reported.^7,8 Here, a 2.0 g solution of p-aminobenzenesulfonamide was dissolved in 50 ml and kept at low temperature (T < 5 °C). Another solution was then prepared by mixing 0.1 g ethylenediamine dihydrochloride with 3.94 ml phosphoric acid in 50 ml deionized water. Before the UV-vis measurements, the volumes of the prepared solutions were mixed and kept in the fridge for detection. Then 1.0 ml of the samples was diluted to fall within the detection range, mixed with 2.0 ml 0.5 M H₂SO₄ solution, and then combined with 1 ml of the colour reagent. The peak intensity at 540 nm was measured by UV-vis spectrophotometry (UV-2600). The concentration was calculated using the concentration–absorbance calibration curve fitting equation (ESI,† Fig. S12).

4.8. COMSOL multiphysics simulation

We calculated the transmitted irradiance at the Au/Si and water/Si interfaces. Assuming that all the transmitted light is eventually absorbed in the Si substrate, the transmitted irradiance can be regarded as the absorbed irradiance. The transmitted irradiance I_t can then be written as follows:where E_t and E₀ are the transmitted and incident electric fields, respectively, n_Si is the real part of the refractive index of Si, n_water is the refractive index of water, and I₀ is the incident irradiance. We set n_water = 1.33 and I₀ is given by the reference spectral data in ‘ASTM G173-03’, while |E_t|²/|E₀|² can be determined by considering the Fresnel coefficients and multiple reflections.³³ For the Au/Si interface,where ñ_Au and ñ_Si are complex refractive indexes of Au and Si, respectively, k₀ is the free-space wavenumber, and h is the Au film thickness. For ñ_Au and ñ_Si, we used the previously tabulated data.³⁵ For the water/Si interface,

The simulation for the energy flux |S_z/S₀| distribution was conducted using COMSOL Multiphysics 6.0 at a wavelength of 480 nm where the transmitted irradiance at the water/Si interface is the maximum. The Au film width was scaled down to 1 μm to reduce the computational burden, but to represent the physical implication yet.

4.9. Computational details

The spin-polarized density functional theory (DFT) implemented in the Vienna ab initio simulation package (VASP) was used for all the calculations.⁴⁰ The Perdew–Burke–Ernzerhof (PBE)⁴¹ functional within the generalized gradient approximation (GGA) was used to model the exchange–correlation interactions, while the van der Waals (vdW) corrections were described by the Tkatchenko–Scheffler (TS) scheme.⁴² The projector augmented wave method (PAW) was employed to describe the electron–nucleus interactions. The energy cutoff was 500 eV and at least a 15 Å vacuum space was applied to avoid interactions between adjacent layers in the perpendicular direction to the surface. Nine, four and six layers of rutile TiO₂(110), Au(111), and Si(111) were utilized to model our slab geometries (Fig. S1, ESI†). Among the catalysts, except for the two bottom layers of TiO₂ which were fixed, all other layers were fully relaxed using 2 × 2 × 1, 4 × 4 × 1, and 1 × 1 × 1 Monkhorst–Pack k-point grids for TiO₂(110), Au(111), and Si(111) (a supercell of 96 atoms), respectively. The energy convergence and residual force criteria in the self-consistent field (SCF) were set to 10⁻⁵ eV and 0.02 eV Å⁻¹, respectively. The Gibbs free energy was calculated based on the computational hydrogen electrode (CHE).⁴³ The free energy change was calculated by ΔG = ΔE + ΔZPE − TΔS + ΔG_U, where ΔE is the electronic energy difference, ΔZPE is the difference in the zero-point energy, T is the temperature (298.15 K), ΔS is the entropy change, and ΔG_U (−eU) is the free energy related to the applied electrode potential, while U and e are the applied potential and the charge of the electron, respectively. The limiting potential (U_L) was employed to evaluate the catalytic activity. Here, U_L was obtained by considering the potential determining step (U_L = −ΔG_PDS/e). Since dealing with charged NO₃⁻ species in periodic DFT calculation is challenging, we chose natural HNO_3(g) as the reference value as follows:^44,45

* + HNO₃(g) → *NO₃ + H⁺ + e⁻ (10)

ΔG_*NO3 = G_*NO3 − G_HNO3 − G_* + 0.5G_*H2 + ΔG_correction

where G_*, G_HNO3, G_*NO3, and G_H2 are the Gibbs free energy of the substrate, HNO₃, NO₃⁻ adsorbed on substrates, and H₂ molecules in the gas phase, respectively. In addition, ΔG_correct indicates the correction of the adsorption energy, which is added to eqn (S1) (ESI†) (0.392 eV).⁴⁶

The elementary steps for the NITRR can be described as follows (in adsorbed *NO, N is attached to the metal):

* + NO₃⁻ → *NO₃ (11)

*NO₃ + H⁺ + e⁻ → *HNO₃ (12)

*HNO₃⁻ + H⁺ + e⁻ → *NO₂ + H₂O (13)

*NO₂ + H⁺ + e⁻ → *HNO₂ (14)

*HNO₂ + H⁺ + e⁻ → *NO + H₂O (15)

*NO + H⁺ + e⁻ → *NOH (16)

*NOH + H⁺ + e⁻ → *N + H₂O (17)

*N + H⁺ + e⁻ → *NH (18)

*NH + H⁺ + e⁻ → *NH₂ (19)

*NH₂ + H⁺ + e⁻ → *NH₃ (20)

*NH₃ → NH_3(g) (21)

Author contributions

A. T. and J.-W. J. conceived and proposed this work. A. T. and J. J. performed and designed the experiments and analysis. M. H. fabricated the Au pattern on Si wafer directed by D. K. J. J. and A. T. measured the PEC performance for NITRR with quantification of products. A. T., J. J., M. T., and Ch. S. characterised and interpreted the photoelectrodes physiochemical and morphological properties. M. Z. and G. L. performed the DFT calculations and wrote the corresponding section. D. L. conducted the COMSOL Multiphysics simulations directed by D. K. A. T. and J.-W. J. co-wrote the manuscript. M. G. L. revised the manuscript and H.-J. G. performed experiments regarding the effect of light. All authors read and commented on the manuscript.

Data availability

All the experimental data are included in the main text or in the ESI.†

Conflicts of interest

The authors declare no competing interests.

Acknowledgements

This work was supported by the National Research Foundation (NRF; RS-2023-00222006, 2019H1D3A1A01103006, 2022H1D3A3A01081140, and RS-2023-00257666), as well as the Research Fund (1.240005.01) of the UNIST, sponsored this study (Ulsan National Institute of Science and Technology). A. T. and M. G. L. acknowledges financial support from the European Union (Marie Skłodowska-Curie Actions, 101107294).

References

1. Y. Kojima and M. Yamaguchi, Ammonia as a hydrogen energy carrier, Int. J. Hydrogen Energy, 2022, 47, 22832–22839.
2. B. H. R. Suryanto, et al., Challenges and prospects in the catalysis of electroreduction of nitrogen to ammonia, Nat. Catal., 2019, 2, 290–296.
3. M. Wang, et al., Can sustainable ammonia synthesis pathways compete with fossil-fuel based Haber–Bosch processes?, Energy Environ. Sci., 2021, 14, 2535–2548.
4. D. R. MacFarlane, et al., A Roadmap to the Ammonia Economy, Joule, 2020, 4, 1186–1205.
5. J. Zheng, L. Jiang, Y. Lyu, S. P. Jian and S. Wang, Green Synthesis of Nitrogen-to-Ammonia Fixation: Past, Present, and Future, Energy Environ. Mater., 2021, 452–457,  DOI:10.1002/eem2.12192.
6. V. Rosca, M. Duca, M. T. DeGroot and M. T. M. Koper, Nitrogen Cycle Electrocatalysis, Chem. Rev., 2009, 109, 2209–2244.
7. S. Ghavam, M. Vahdati, I. A. G. Wilson and P. Styring, Sustainable Ammonia Production Processes, Front. Energy Res., 2021, 9, 1–19.
8. V. Kyriakou, I. Garagounis, A. Vourros, E. Vasileiou and M. Stoukides, An Electrochemical Haber–Bosch Process, Joule, 2020, 4, 142–158.
9. C. Arnaiz del Pozo and S. Cloete, Techno-economic assessment of blue and green ammonia as energy carriers in a low-carbon future, Energy Convers. Manage., 2022, 255, 115312.
10. P. M. Krzywda, et al., Electroreduction of NO₃− on tubular porous Ti electrodes, Catal. Sci. Technol., 2022, 3281–3288,  10.1039/d2cy00289b.
11. D. Liu, et al., Photoelectrochemical Synthesis of Ammonia with Black Phosphorus, Adv. Funct. Mater., 2020, 30, 1–7.
12. N. C. Kani, A. Prajapati and M. R. Singh, Sustainable Routes for Photo-Electrochemical Synthesis of Ammonia Using Various Nitrogen Precursors, ACS ES&T Eng., 2022, 2, 1080–1087.
13. R. Daiyan, et al., Nitrate reduction to ammonium: From CuO defect engineering to waste NO_x-to-NH₃ economic feasibility, Energy Environ. Sci., 2021, 14, 3588–3598.
14. H. Okamoto, et al., Effects of annealing conditions on the oxygen evolution activity of a BaTaO₂N photocatalyst loaded with cobalt species, Catal. Today, 2019, 354, 204–210.
15. L. Barrera and R. Bala Chandran, Harnessing Photoelectrochemistry for Wastewater Nitrate Treatment Coupled with Resource Recovery, ACS Sustainable Chem. Eng., 2021, 9, 3688–3701.
16. A. Tayyebi, et al., Bias-free solar NH₃ production by perovskite-based photocathode coupled to valorization of glycerol, Nat. Catal., 2024, 7, 510–521.
17. J. Saari, et al., Interface Engineering of TiO₂ Photoelectrode Coatings Grown by Atomic Layer Deposition on Silicon, ACS Omega, 2021, 6, 27501–27509.
18. S. Hu, et al., Amorphous TiO₂ coatings stabilize Si, GaAs, and GaP photoanodes for efficient water oxidation, Science, 2014, 344, 1005–1009.
19. M. Hannula, et al., Improved Stability of Atomic Layer Deposited Amorphous TiO₂ Photoelectrode Coatings by Thermally Induced Oxygen Defects, Chem. Mater., 2018, 30, 1199–1208.
20. M. Ali, et al., Nanostructured photoelectrochemical solar cell for nitrogen reduction using plasmon-enhanced black silicon, Nat. Commun., 2016, 7, 1–5.
21. M. Xu, et al., Atomic layer deposition technique refining oxygen vacancies in TiO₂ passivation layer for photoelectrochemical ammonia synthesis, Compos. Commun., 2022, 29, 101037.
22. H. E. Kim, et al., Photoelectrochemical Nitrate Reduction to Ammonia on Ordered Silicon Nanowire Array Photocathodes, Angew. Chem., Int. Ed., 2022, 61, e202204117.
23. H. Huang, et al., Lithium-Mediated Photoelectrochemical Ammonia Synthesis with 95% Selectivity on Silicon Photocathode, ACS Energy Lett., 2023, 8, 4235–4241.
24. J. Zheng, et al., Photoelectrochemical Synthesis of Ammonia on the Aerophilic-Hydrophilic Heterostructure with 37.8% Efficiency, Chem, 2019, 5, 617–633.
25. N. S. Lewis and D. G. Nocera, Powering the planet: Chemical challenges in solar energy utilization, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 15729–15735.
26. P. Thangavel, et al., Graphene-nanoplatelets-supported NiFe-MOF: High-efficiency and ultra-stable oxygen electrodes for sustained alkaline anion exchange membrane water electrolysis, Energy Environ. Sci., 2020, 13, 3447–3458.
27. J. Li, et al., Efficient Ammonia Electrosynthesis from Nitrate on Strained Ruthenium Nanoclusters, J. Am. Chem. Soc., 2020, 142, 7036–7046.
28. X. Zhang, et al., Regulating intermediate adsorption and H2O dissociation on a diatomic catalyst to promote electrocatalytic nitrate reduction to ammonia, Energy Environ. Sci., 2024, 6717–6727,  10.1039/d4ee02747g.
29. J. M. McEnaney, et al., Electrolyte engineering for efficient electrochemical nitrate reduction to ammonia on a titanium electrode, ACS Sustainable Chem. Eng., 2020, 8, 2672–2681.
30. G. Choi, et al., Enhanced Terahertz Shielding of MXenes with Nano-Metamaterials, Adv. Opt. Mater., 2018, 6, 1–6.
31. W. Park, J. Rhie, N. Y. Kim, S. Hong and D. S. Kim, Sub-10 nm feature chromium photomasks for contact lithography patterning of square metal ring arrays, Sci. Rep., 2016, 6, 1–6.
32. W. Jung and Y. J. Hwang, Material strategies in the electrochemical nitrate reduction reaction to ammonia production, Mater. Chem. Front., 2021, 5, 6803–6823.
33. C. Schinke, et al., Uncertainty analysis for the coefficient of band-to-band absorption of crystalline silicon, AIP Adv., 2015, 5, 067168.
34. H. Li, Y. Hu, Y. Yang and Y. Zhu, Theoretical investigation of broadband absorption enhancement in a-Si thin-film solar cell with nanoparticles, Sol. Energy Mater. Sol. Cells, 2020, 211, 110529.
35. S. Babar and J. H. Weaver, Optical constants of Cu, Ag, and Au revisited, Appl. Opt., 2015, 54, 477.
36. H. Tanaka, S. Yamashita, Y. Abe, M. Shikida and K. Sato, Fast etching of silicon with a smooth surface in high temperature ranges near the boiling point of KOH solution, Sens. Actuators, A, 2004, 114, 516–520.
37. E. Van Veenendaal, K. Sato, M. Shikida and J. Van Suchtelen, Micro-morphology of single crystalline silicon surfaces during anisotropic wet chemical etching in KOH: Velocity source forests, Sens. Actuators, A, 2001, 93, 232–242.
38. R. Rao, J. E. Bradby and J. S. Williams, Patterning of silicon by indentation and chemical etching, Appl. Phys. Lett., 2007, 91, 1–4.
39. H. S. Kim, et al., Phonon-Polaritons in Lead Halide Perovskite Film Hybridized with THz Metamaterials, Nano Lett., 2020, 20, 6690–6696.
40. D. Joubert, From ultrasoft pseudopotentials to the projector augmented-wave method, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
41. J. P. Perdew, K. Burke and M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett., 1996, 77, 3865–3868.
42. A. Tkatchenko and M. Scheffler, Accurate molecular van der Waals interactions from ground-state electron density and free-atom reference data, Phys. Rev. Lett., 2009, 102, 6–9.
43. J. K. Nørskov, et al., Origin of the overpotential for oxygen reduction at a fuel-cell cathode, J. Phys. Chem. B, 2004, 108, 17886–17892.
44. J. X. Liu, D. Richards, N. Singh and B. R. Goldsmith, Activity and Selectivity Trends in Electrocatalytic Nitrate Reduction on Transition Metals, ACS Catal., 2019, 9, 7052–7064.
45. S. Guo, et al., Insights into Nitrate Reduction over Indium-Decorated Palladium Nanoparticle Catalysts, ACS Catal., 2018, 8, 503–515.
46. H. Niu, et al., Theoretical Insights into the Mechanism of Selective Nitrate-to-Ammonia Electroreduction on Single-Atom Catalysts, Adv. Funct. Mater., 2021, 31, 2008533.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ey00282b

‡ These authors contributed equally to this work.

---