Electrochemical reduction conditioning modified Fe-based catalysts with structural disorders for efficient ammonium production from nitrite reduction

Abstract

Electrochemical reduction of NO₂⁻ to NH₄⁺ offers a direct, energy-efficient pathway for sustainable ammonia production by circumventing the rate-determining NO₃⁻-to-NO₂⁻ conversion that constrains the traditional NO₃⁻ reduction reaction (NO₃RR). Herein, we introduce an electrochemical reduction conditioning (ERC) strategy to control Fe₂O₃ at different reduction potentials, generating a series of catalysts with tunable Fe³⁺/Fe²⁺/Fe components and lattice strain. Comprehensive ex situ and in situ characterization studies reveal that more negative ERC potentials induce greater structural disorder (i.e., tuned Fe/FeO/Fe₂O₃ components and pronounced lattice strain), which collectively enhance NO₂⁻ adsorption, water dissociation and hydrogenation of intermediates, while suppressing competing H₂ evolution. Theoretical calculations support that these defective catalyst surfaces lower the energy barriers for NO₂⁻ adsorption. As a result, the optimized ERC-treated Fe₂O₃ catalyst achieves a high NH₄⁺ production rate of 153 nmol s⁻¹ cm⁻², a faradaic efficiency of 93% and a partial current density of ∼96.5 mA cm⁻² at −1.0 V vs. RHE. Integration with plasma-generated NO₂⁻-rich electrolytes further demonstrates stable, decentralized NH₄⁺ production, yielding 32 nmol s⁻¹ cm⁻². This work clarifies the mechanistic role of ERC-induced structural disorders in the NO₂RR and provides design principles for next-generation metal–oxide catalysts enabling sustainable nitrogen-cycle management.

---

1 Introduction

Ammonia (NH₃), one of the world's most produced chemicals, was produced at a scale of 200.7 million tonnes in 2024 and is projected to grow by ∼5% over the next decade.^1,2 It underpins global food security through its dominant use in fertilizers and also serves as a key feedstock for various manufacturing sectors, including pharmaceuticals, plastics, and textiles.^3–6 Recently, NH₃ has also attracted interest as a hydrogen carrier because liquid NH₃ has a higher energy density (33.5 kWh L⁻¹) than gaseous hydrogen (2.4 kWh L⁻¹ H₂),^3,7 offering a more efficient way to store and transport hydrogen.⁶ It also benefits from its existing, well-developed global production, storage, and transport infrastructure,^6,8 making it central to the emerging global hydrogen economy and power-to-X pathways.⁸ However, over 96% of NH₃ is still made via the century-old Haber–Bosch (HB) process,⁹ which reacts nitrogen and hydrogen gases under extreme conditions, at temperatures ranging from 400 to 650 °C and pressures ranging from 150 to 200 bar.¹⁰ The hydrogen used in this process is primarily derived from steam methane reforming, an energy-intensive process that consumes significant amounts of fossil fuels.^11,12 The HB process is estimated to consume ∼1% to 2% of the world's total energy and contribute to substantial greenhouse gas emissions of ∼1.5% of global carbon dioxide emissions.^11,12 While the HB process has been instrumental in increasing agricultural yields by enabling mass production of NH₃, it is economically feasible only on a large, centralized scale.^9,13 This dependence on large-scale, fossil-based production conflicts with the envisioned role of NH₃ in a low-carbon economy and its potential as a renewable energy carrier for hydrogen, highlighting the urgent need for scalable, decentralized, and renewable-powered alternatives to produce “green” NH₃.

The electrochemical conversion of NO_x species (i.e., NO₃⁻ and NO₂⁻) to NH₃/NH₄⁺ using renewable energy has emerged as a promising green alternative to conventional NH₃ synthesis. The NO_x reduction reaction (NO_xRR) can achieve higher NH₃ yield and faradaic efficiency than other proposed green ammonia production methods, such as the direct N₂ electroreduction reaction (eNRR), which suffers from sluggish kinetics and low selectivity due to the strong N≡N bond in N₂.^14–17 Another proposed method, Li-mediated N₂ reduction (Li-NRR), despite achieving high yield and nearly 100% faradaic efficiency, is limited by high overpotentials, system instability and costly feedstocks (i.e., pure N₂ and H₂, as well as scarce Li metal).^6,8,18 NO_xRR also offers the benefit of closing the NO_x loop cycle by valorizing pollutants from sources such as power plants, industries and agriculture, producing NH₃ under mild conditions and thereby contributing to pollution reduction and resource recycling.⁸ Moreover, NO_x can be synthesized from atmospheric nitrogen using plasma technology,^6,19–21 and when combined with the NO_xRR process, this presents the potential for a self-sustaining, decentralized, green NH₃ production system powered entirely by renewable energy sources.⁶

In the NO_xRR pathway, the NO₃⁻ reduction reaction (NO₃RR) to NH₃/NH₄⁺ is an eight-electron, multi-step process in which the initial adsorption and deoxygenation of NO₃⁻ to NO₂⁻ has been reported to involve a high-energy barrier, and depending on the catalysts and reaction conditions, may constitute the rate-determining step (RDS).^22–24 These limitations, combined with the relatively weak binding energy of nitrate and nucleophilicity compared with nitrite when interacting with transition metals,^25–28 restrict NO₃RR efficiency. A promising route to improve the viability of NH₃/NH₄⁺ production is to start directly from NO₂⁻, which bypasses one of the most energetically demanding steps and requires fewer electrons, thus offering a more direct and energy-efficient pathway to NH₃. Achieving this goal requires careful control of NO_x feedstocks. Plasma technology facilitates this by enabling precise control over the relative production of NO₂⁻ and/or NO₃⁻ species in solution, thereby providing a pathway to leverage the NO₂⁻ reduction reaction (NO₂RR) directly. Recent studies have explored the optimization of plasma parameters, such as reactor design, discharge power, operational modes, and plasma activation duration, to enhance NO_x yields and fine-tune the NO₂⁻/NO₃⁻ ratios in plasma-activated solutions.^6,21 However, the impact of variations in reaction medium conditions (e.g., pH or ion types) on the final NO_x yields, NO₂⁻/NO₃⁻ ratios, and their subsequent influence on NO_xRR activity remains an area for further investigation.

Despite the advantages of the NO₂RR, the factors governing its electrocatalytic activity remain unclear. In particular, catalysts that efficiently reduce NO₃⁻ are not always effective for NO₂⁻ reduction, indicating that the active sites and mechanistic requirements for the NO₂RR differ from those of the NO₃RR. This discrepancy arises from several factors: (1) the adsorption behaviors of NO₃⁻ and NO₂⁻ driven by differences in their distinct molecular structures and electronic configurations, where catalysts optimized for NO₃⁻ adsorption may not interact as effectively with NO₂⁻; (2) differences in electron–proton transfer steps, as the NO₃RR involves an additional electron–proton transfer step to convert NO₃⁻ to NO₂⁻, necessitating different active sites or catalytic properties than those required for the direct reduction of NO₂⁻; and (3) the specificity of surface active sites, as those effective for NO₃⁻ binding and reduction may not exhibit similar efficacy for NO₂⁻ reduction. These factors collectively underscore the complexity in designing catalysts capable of efficiently facilitating both the NO₃RR and the NO₂RR, highlighting the importance of tailored strategies to optimize NO₂RR activity.

Among various transition-metal catalysts, Fe-based materials have shown considerable promise due to their cost-effectiveness and catalytic performance.^29–31 Fe₂O₃, in particular, has been extensively studied for the NO₃RR, with well-characterized mechanisms and RDS. Research indicates that the key RDS for Fe₂O₃ catalysts involves the reduction of NO₃⁻ to NO₂⁻, and that modifying Fe₂O₃ by alloying with other transition metals and inducing defects can shift the RDS toward hydrogenation processes.^29,30 These findings underscore the dynamic role of Fe species, where Fe³⁺ and Fe²⁺ facilitate electron transfer to overcome the high energy barriers in the NO₃RR. However, the established catalytic efficiency of Fe₂O₃ in the NO₃RR does not guarantee similar performance in the NO₂RR. The NO₂RR may involve different reaction intermediates and electron transfer dynamics, potentially requiring alternative active site properties or crystal structures.^26,28,32 Consequently, the behavior of Fe₂O₃ in the NO₂RR remains underexplored, representing a significant knowledge gap in understanding the role of surface chemistry and the electronic interactions that govern NO₂RR activity. Addressing this gap requires detailed investigations into the surface chemistry, electron transport mechanisms, and active site engineering of Fe₂O₃ specific to the NO₂RR.

Strain engineering has emerged as a promising strategy to enhance catalytic activity by altering crystal structures, inducing lattice distortions, and optimizing active sites.^33,34 In fact, variations in electrochemical reduction conditions can introduce defects that lead to non-stoichiometry, resulting in significant structural changes and creating lattice strains in defect regions.^33,35,36 These strains can modulate the local electronic structure at the catalyst surface, thereby tuning interactions between reaction species and the catalyst.³⁵ While strain has been shown to enhance catalytic activity in reactions like the NO₃RR,^35,37,38 its specific effects on NO₂RR performance remain poorly understood. In particular, the detailed relationship between strain-induced electronic structure changes and the resulting local physicochemical properties of NO₂RR intermediates on Fe₂O₃ catalysts is unclear. Additionally, the influence of variables such as applied potential on the creation and maintenance of strain has not been thoroughly examined. Furthermore, the operational stability of strained catalysts is not well-documented, as mechanical relaxation or structural changes over time could diminish the benefits of strain. Therefore, mechanistic studies that correlate the degree of strain in Fe₂O₃ with its catalytic activity and selectivity in the NO₂RR are essential to guide the rational design of strained Fe₂O₃ catalysts.

This study employs an electrochemical reduction conditioning (ERC) strategy to control Fe₂O₃ catalysts at different reduction potentials, producing a series of catalysts (pristine Fe₂O₃, LR-Fe₂O₃ and HR-Fe₂O₃) for investigating their activity for NO₂RR-to-NH₄⁺ under alkaline conditions. This approach enables a systematic evaluation of how Fe³⁺ → Fe²⁺/Fe reduction, with tunable Fe/FeO/Fe₂O₃ components and strain induction, influences NO₂RR performance. Through a suite of ex situ and in situ characterization techniques, it is revealed that more negative ERC potentials induced greater structural disorder (i.e., tuned Fe/FeO/Fe₂O₃ components and lattice strain), thereby enhancing NO₂⁻ adsorption, promoting hydrogenation of intermediates and suppressing competing H₂ evolution. Density functional theory (DFT) calculations on the pristine and defective hematite surfaces provide further corroboration of these findings. Specifically, the structural defects induced by electrochemical reduction act as bifunctional active sites that simultaneously enhance NO₂⁻ binding and activate water dissociation to provide the surface protons essential for the NO₂⁻ reduction process, offering valuable insights for designing next-generation metal–oxide catalysts. These ERC-treated catalysts are further tested in plasma-generated, optimized NO₂⁻-rich electrolytes obtained by tuning solution pH and plasma reactor voltage to produce NH₄⁺, showcasing substantial advantages for green ammonium production through bypassing the NO₃⁻-to-NO₂⁻ step and reducing overall energy costs.

2 Results and discussion

In this work, Fe₂O₃ catalyst powders were prepared through a precipitation–calcination route. Briefly, Fe₂O₃ powders were obtained by reacting iron(III) nitrate with sodium carbonate, followed by calcination at 400 °C in flowing air. The resulting Fe₂O₃ powder was drop-cast onto carbon fibre paper to prepare the pristine Fe₂O₃ electrode. The characteristics of the Fe₂O₃ electrodes were further modified using an electrochemical reduction conditioning (ERC) strategy at different applied potentials, with the resulting electrodes denoted as eR-Fe₂O₃. Specifically, prior to electrochemical testing, the Fe₂O₃ electrodes were pre-treated at reduction potentials of −1.0 V or −2.0 V vs. the reversible hydrogen electrode (RHE) for 0.5 h. The Fe₂O₃ electrodes conditioned at −1.0 V are referred to as Low-Reduction Fe₂O₃ (LR-Fe₂O₃), whereas those treated at −2.0 V are designated as High-Reduction Fe₂O₃ (HR-Fe₂O₃) (Experimental section).

The NO₂RR activity of pristine Fe₂O₃ and eR-Fe₂O₃ electrodes treated with different ERC potentials (i.e., LR-Fe₂O₃ and HR-Fe₂O₃) was evaluated within a two-compartment H-cell system using an electrolyte composed of 0.1 M KOH and 0.1 M KNO₂ (Experimental section). Across the applied potentials from −0.4 to −1.4 V vs. RHE, HR-Fe₂O₃ exhibits the highest NH₄⁺ production rate and faradaic efficiency (FE_NH4⁺), followed by LR-Fe₂O₃, and lastly, pristine Fe₂O₃ (Fig. 1a and b). Specifically, HR-Fe₂O₃ achieves its optimal NH₄⁺ production performance at −1.0 V vs. RHE, with a production rate and FE_NH4⁺ of 154 nmol s⁻¹ cm⁻² and 93%, respectively. However, at more negative potentials beyond −1.0 V vs. RHE, while the NH₄⁺ production rate remains relatively high, FE_NH4⁺ begins to decrease, indicating increased competition from side reactions and reduced selectivity toward NH₄⁺.^39–41 The chronoamperometric i–t curves depicted in Fig. S1 demonstrate that HR-Fe₂O₃ maintains the highest partial current density for NH₄⁺ production of ∼90.3 mA cm⁻² among all samples at a NO₂RR applied potential of −1.0 V vs. RHE. As a control, to ensure that prolonged ERC treatment does not further change the catalytic activity, Fe₂O₃ was ERC-treated for 1 h (double reduction length) instead of 0.5 h at −2.0 V vs. RHE to create an HRR-Fe₂O₃ electrode, which was then tested for its NO₂RR activity under similar operating conditions. As a result, the HRR-Fe₂O₃ catalyst exhibits essentially identical performance to the HR-Fe₂O₃ electrode (Fig. S2), showing overlapping trends in both the NH₄⁺ production rate and FE_NH4⁺ across the applied potentials from −0.4 to 1.4 V vs. RHE, with only minor differences within experimental error bars. Hence, this work adopts a 0.5 h ERC pretreatment as the optimal pretreatment condition. Further, the analysis of side products (Fig. S3) reveals that HR-Fe₂O₃ generates the least amount of H₂ (22 nmol s⁻¹ cm⁻²), followed by LR-Fe₂O₃ (25 nmol s⁻¹ cm⁻²), with pristine Fe₂O₃ producing the most (28 nmol s⁻¹ cm⁻²). This suggests that ERC treatment helps suppress the hydrogen evolution reaction (HER), with more negative reduction treatment potentials leading to greater suppression of the HER.

Fig. 1 NO₂RR-to-NH₄⁺ production performance of Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃ electrodes across an applied potential range of −1.4 to −0.4 V vs. RHE: (a) NH₄⁺ production rate and (b) faradaic efficiency (FE_NH4⁺). (c) Partial current density for NH₄⁺ (j_NH4⁺) of all electrodes in 0.1 M KOH + 0.1 M KNO₂. (d(i)) Nyquist plot and its corresponding (d(ii)) equivalent circuit of Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃ electrodes.

Polarization curves (Fig. S4a–c) indicate that both LR-Fe₂O₃ and HR-Fe₂O₃ exhibit slight improvements in total current densities (j) compared to pristine Fe₂O₃ across tested potentials in electrolytes of 0.1 M KOH, 0.1 M KOH + 0.1 M KNO₃, and 0.1 M KOH + 0.1 M KNO₂. For instance, at −1.0 V vs. RHE in 0.1 M KOH + 0.1 M KNO₂ electrolyte (Fig. S4c), the j achieved by pristine Fe₂O₃ is −60.1 mA cm⁻², while LR-Fe₂O₃ and HR-Fe₂O₃ show improvements of ∼16% (−69.9 mA cm⁻²) and ∼22% (−73.0 mA cm⁻²), respectively. Similarly, in the electrolytes of 0.1 M KOH only (Fig. S4a) and 0.1 M KOH + 0.1 M KNO₃ (Fig. S4b), at −1.0 V vs. RHE, HR-Fe₂O₃ achieves the highest j of ∼−38.2 and −70.7 mA cm⁻², representing improvements of ∼15% and 13% over pristine Fe₂O₃ (−33.1 and −62.8 mA cm⁻², respectively). LR-Fe₂O₃ shows modest improvements of ∼10% (∼−36.4 mA cm⁻²) and ∼6.8% (−67.1 mA cm⁻²) over pristine Fe₂O₃ in the same electrolytes. These results suggest that ERC-conditioned Fe₂O₃ catalysts exhibit slightly enhanced intrinsic activity overall. However, the difference in j between LR-Fe₂O₃ and HR-Fe₂O₃ is minor. A more pronounced difference is observed in the partial current density for NH₄⁺ production (j_NH4⁺) under the electrolyte condition of 0.1 M KOH + 0.1 M KNO₂ (Fig. 1c). Compared to pristine Fe₂O₃, which achieves a j_NH4⁺ of ∼−49.1 mA cm⁻², LR-Fe₂O₃ and HR-Fe₂O₃ show significant improvements of ∼24% (−61.0 mA cm⁻²) and ∼37% (−67.4 mA cm⁻²), respectively. This indicates that the more negative ERC (HR) potential enhances selectivity toward NH₄⁺ production, with both treated catalysts showing a marked improvement in j_NH4⁺ over pristine Fe₂O₃. As a control, to compare the activity between the NO₂RR and the NO₃RR, pristine Fe₂O₃ and HR-Fe₂O₃ electrodes were also used to conduct the electrocatalytic NO₃RR in an electrolyte containing 0.1 M KOH and 0.1 M KNO₃. The results, presented in Fig. S5, clearly demonstrate that both electrodes produce greater NH₄⁺ and FE_NH4⁺ in the NO₂RR than in the NO₃RR, with HR-Fe₂O₃ consistently outperforming pristine Fe₂O₃ in nitrate and nitrite reduction. Specifically, for the NO₂RR, HR-Fe₂O₃ achieves a peak FE_NH4⁺ of 93% at −1.0 V vs. RHE, with an NH₄⁺ production rate of ∼154 nmol s⁻¹ cm⁻². This represents a 31% improvement in productivity and 15% increase in FE_NH4⁺ compared to pristine Fe₂O₃ (∼118 nmol s⁻¹ cm⁻² and 81% at −1.0 V vs. RHE). For the NO₃RR, HR-Fe₂O₃ also outperforms pristine Fe₂O₃, achieving an NH₄⁺ production rate of ∼101 nmol s⁻¹ cm⁻² at −1.0 V vs. RHE, corresponding to a 5% improvement over pristine Fe₂O₃ (∼96 nmol s⁻¹ cm⁻²). Importantly, the results indicate that the improvements in NH₄⁺ production are more pronounced in the NO₂RR than in the NO₃RR following the ERC treatment. This suggests that the ERC-treated Fe₂O₃ catalysts are more surface-active for the NO₂RR than for the NO₃RR, and that ERC treatment yields better performance enhancements for the NO₂RR than for the NO₃RR.

Electrochemical impedance spectroscopy (EIS) was employed to understand charge-transfer resistances during the NO₂RR process. The Nyquist plots (Fig. 1d) display a single semicircle for all electrodes, indicative of the charge transfer resistance at the electrode/electrolyte interface.^42,43 Notably, the radius of the semicircle decreases as the reduction treatment potential becomes more negative. Coupled with equivalent circuit modeling (Fig. 1d(ii)), the results reveal that HR-Fe₂O₃ exhibits a ∼35% lower charge transfer resistance compared to pristine Fe₂O₃ (1.4 kΩ vs. 2.2 kΩ), while LR-Fe₂O₃ shows a ∼21% reduction (1.8 kΩ vs. 2.2 kΩ). These results demonstrate that the electrode preconditioned at a more negative reduction potential has reduced charge transfer resistance, thereby benefiting NH₄⁺ production rates and selectivity during the NO₂RR.⁸ Additionally, the electrochemically active surface area (ECSA) of the catalysts was determined using cyclic voltammetry and double-layer capacitance methods (CDL) (Fig. S6 and S7).^44,45 The results (Table S1) indicate that the ECSA does not exhibit a significant difference (ranging from 5.1 to 6.7 cm²). Therefore, the changes in performance are likely influenced by other factors, such as modifications in the structure and/or surface chemistry (e.g., coordination environment and defects) as a result of ERC treatment.

To investigate the surface chemistry of the catalysts and understand how NH₄⁺ production performance is influenced by ERC treatments, X-ray photoelectron spectroscopy (XPS) measurements were performed on all as-prepared electrodes (pristine Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃). Fig. 2a presents the high-resolution Fe 2p XPS spectra, which show doublet peaks at ∼724.1 and 710.8 eV corresponding to Fe 2p_1/2 and Fe 2p_3/2, respectively.⁴⁶ These are typical binding energies for Fe³⁺ indicating Fe₂O₃.⁴⁶ In addition, a satellite peak at ∼720.0 eV, characteristic of α-Fe₂O₃,⁴⁶ is observed. Notably, a positive shift of ∼0.4 to 0.7 eV in the Fe 2p peaks is observed for LR-Fe₂O₃ and HR-Fe₂O₃ compared to pristine Fe₂O₃, respectively, suggesting possible changes in the coordination environment and/or induction of strain following ERC.^47–50 The fitted Fe 2p spectra (Fig. S8a–c) demonstrate that increasing the ERC potential (more negative) results in a further reduction of Fe species. Specifically, when the electrode is preconditioned at a low reduction (LR) potential (−1.0 V vs. RHE), 19.2% of Fe₂O₃ in LR-Fe₂O₃ is reduced to Fe and FeO, accounting for 2.6% and 16.6% (Fig. S8b), respectively. At a more negative (high reduction, HR) treatment potential (−2.0 V vs. RHE), HR-Fe₂O₃ experiences a slightly larger portion of Fe₂O₃ (23.5%) being reduced to Fe and FeO, comprising 2.8% and 20.7% (Fig. S8c), respectively. The emergence of these FeO and metallic Fe on the catalyst surface shows the tuneable FeO/Fe/Fe₂O₃ components with different ERC treatment potentials, and these reconstructed metal and metal oxide species likely provide complementary active sites for NO₂RR activity, as reported in prior NO₃RR studies.^29,30,51 To further understand the Fe oxidation states, ex situ and in situ X-ray absorption near-edge structure (XANES) spectra of LR-Fe₂O₃ and HR-Fe₂O₃ catalysts were evaluated (Fig. S9). The XANES spectra were compared with the reference compound spectra (metallic Fe and Fe₂O₃). The results clearly show that the absorption edge positions and white-line intensities of both catalysts lie between those of metallic Fe and Fe₂O₃ references, indicating that both LR-Fe₂O₃ and HR-Fe₂O₃ catalysts possessed mixed oxidation states between metallic Fe and Fe₂O₃, corroborating the results from XPS analyses. Furthermore, under experimental conditions of open-circuit potential and −1.0 V vs. RHE in 0.1 M KOH with 0.1 M KNO₂ electrolyte, the in situ XANES spectra remain consistent with the ex situ observations, confirming that these mixed-valence Fe species are maintained during operation. In addition, in situ PD spectra (Fig. S10) were collected under the same conditions as the in situ XANES measurements to further substantiate the coexistence of Fe and Fe oxide species in LR-Fe₂O₃ and HR-Fe₂O₃. The LR-Fe₂O₃ catalyst exhibits diffraction features at 21.3°, 24.8°, and 35.4°, assignable to Fe(011) (#96-230-0201), Fe₂O₃(024) (#96-210-1168), and FeO(113) (#96-900-8637) throughout the 30 min reaction, respectively, while an additional FeO(222) (#96-900-8637) reflection at 36.3° emerged during electrolysis. Similarly, the HR-Fe₂O₃ catalysts show characteristic reflections at 21.3°, 31.5°, 24.8°, 26.7°, and 35.4°, respectively, corresponding to Fe(011, 002) (#96-230-0201), Fe₂O₃(024, 116) (#96-210-1168), and FeO(113) (#96-900-8637) under reaction conditions. The results provide further evidence for the Fe and Fe oxide phases under reaction conditions. When correlating these findings with performance tests, higher proportions of these reduced Fe and Fe oxide species in LR-Fe₂O₃ and HR-Fe₂O₃ explain their improved NH₄⁺ production rates compared to pristine Fe₂O₃. The O 1s XPS spectra and corresponding curve fittings (Fig. S11a–c) show a clear decrease in the proportion of lattice oxygen (peak at ∼530 eV)⁴⁶ after ERC, which is consistent with the reduction of Fe₂O₃ evidenced by Fe 2p XPS spectra. Specifically, LR-Fe₂O₃ consists of 39.7% lattice oxygen, while HR-Fe₂O₃ comprises 34.9%, both markedly lower than pristine Fe₂O₃ (55.8%). Two other peaks at ∼531.0 and 532.5 eV are attributed to surface oxygen passivated with hydrogen and adsorbed oxygen (water), respectively.^52,53 The content of surface oxygen passivated with hydrogen increases from 35.4% in pristine Fe₂O₃ to 43.5% in LR-Fe₂O₃ and further to 47.5% in HR-Fe₂O₃ Similarly, adsorbed oxygen (water) increases from 8.8% in pristine Fe₂O₃ to 16.8% in LR-Fe₂O₃ and 17.6% in HR-Fe₂O₃. In fact, the literature reports a direct correlation between these two oxygen species (surface oxygen passivated with hydrogen and adsorbed oxygen) and water dissociation on the catalyst surface,^54–56 suggesting that their elevated presence in ERC-treated samples increases water adsorption and subsequent dissociation. This process is crucial for supplying hydrogen during the NO₂RR, facilitating the hydrogenation of reaction intermediates and the eventual formation of NH₄⁺ production (vide infra).

Fig. 2 (a) High-resolution Fe 2p XPS spectra of Fe₂O₃, LR-Fe₂O₃ and HR-Fe₂O₃. (b) Normalized XRD patterns (intensity normalized to the highest peak) of pristine Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃, with (i) and (ii) showing excerpts for patterns in diffraction peak ranges of 30°–70° and 32°–37°, respectively, compared with the full range (30°–90°) patterns in Fig. S12. (c) HR-TEM images of HR-Fe₂O₃ at (i) a magnification of 10 nm, with (ii) and (iii) showing enlarged views of specific areas indicated in (i). (d) Normalized Raman spectra of pristine Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃.

To investigate the bulk characteristics and crystal structure of the ERC-treated catalysts at different potentials, X-ray diffraction (XRD) measurements were conducted. The XRD patterns depicted in Fig. S12 and 2b show that pristine Fe₂O₃ displays peaks at ∼33.0°, 35.5°, 40.7°, 49.4°, 62.3°, and 63.9°, corresponding to the (104), (110), (113), (024), (214), and (300) planes of α-Fe₂O₃, respectively (JCPDS collection code: 01-080-0597). The post-conditioned electrodes, LR-Fe₂O₃ and HR-Fe₂O₃, exhibit similar peaks, confirming the retention of the α-Fe₂O₃ structure. However, a new split from the (110) peak at ∼35.0° is observed to appear in both conditioned samples (Fig. 2b(ii)). This split peak, corresponding to the (101) plane of α-Fe₂O₃, indicates a partial phase shift in the crystal structure, likely caused by non-uniform distortion within the cell,^57–62 while the lack of the (101) peak in pristine Fe₂O₃ indicates the absence of this phase shift. For samples after preconditioning reduction treatments, the I₁₁₀/I₁₀₁ ratio decreases with increasing (more negative) reduction potential: HR-Fe₂O₃ exhibits a lower I₁₁₀/I₁₀₁ of 1.47 compared to LR-Fe₂O₃ (1.65). This trend indicates enhanced distortion and greater phase transformation at more negative reduction treatment potentials. Such cell distortion and multiphasic phenomena have been reported in the literature as a result of strong lattice strains induced in the catalysts.^60,62 Furthermore, previous studies have shown that the reduction of surface species (in this work, Fe³⁺ to Fe²⁺/Fe) can induce significant lattice strain due to localized changes in the coordination environment.^63,64 This strain, in turn, promotes the formation of multiphasic phenomena, as observed in XRD patterns in Fig. 2b(ii). To confirm this, strain analysis using the Williamson–Hall (WH) equation was performed.^58,65 The WH plots in Fig. S13 estimate lattice strains of 6.0%, 26.3%, and 28.0% for Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃, respectively. Moreover, strain directly correlated with the d-band structure of the catalyst, modulating the adsorption energies of reaction intermediates.^33,34 Specifically, the increased strain can lead to the d-band center shift closer to the Fermi level,^38,66,67 which is responsible for the improved surface adsorption capabilities^4,68–71 and explains the improved water dissociation observed in HR-Fe₂O₃ and LR-Fe₂O₃ as compared to pristine Fe₂O₃ as evidenced by the XPS O 1s spectra. Furthermore, the main characteristic peaks of the catalysts, corresponding to the (104), (110) and (101) planes, were used to calculate their crystallite size (calculated via the Scherrer equation) and lattice spacing, as shown in Table S2. The results reveal no obvious trend or significant change in crystallite size and lattice spacing with different ERC potentials. Combining XPS and XRD findings confirms that the more negative ERC potentials lead to further Fe species reduction and thus stronger lattice strain formation, ultimately tuning the d-band center shift and leading to improved surface adsorption capabilities for protonation and NO₂⁻ adsorption processes.

The morphology of the Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃ electrodes were investigated using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) (Fig. S14a and b, respectively). The obtained SEM images of the catalysts (Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃) were statistically analyzed to determine the particle size distribution (Fig. S15). Specifically, the pristine Fe₂O₃ catalyst displays nanoparticles (60–100 nm) that agglomerate into large clustered particles. After ERC, both LR-Fe₂O₃ and HR-Fe₂O₃ have a morphology consisting of a mixture of nanoparticles and flake-like aggregates, while LR-Fe₂O₃ shows a dominant particle size between 60 and 100 nm and HR-Fe₂O₃ shows a relatively smaller particle size range from 40 to 70 nm. The morphological changes are consistent with the structure distortion generated during the surface reconstruction.^72,73 In this study, the surface reconstruction due to lattice strain and distortion (as evidenced by the XRD measurement above) during the ECR treatment is likely a key driver of the observed morphological changes in LR-Fe₂O₃ and HR-Fe₂O₃ electrodes. Notably, no significant change in particle size is observed, despite the catalysts being preconditioned at different reduction potentials, which is consistent with the XRD and ECSA analyses. High-resolution TEM (HR-TEM) images (Fig. 2c and S16) show that all samples demonstrate lattice spacings of 0.27 and 0.25 nm corresponding to the (104) and (110) planes of α-Fe₂O₃, respectively.^46,48 Additionally, both LR-Fe₂O₃ and HR-Fe₂O₃ display a lattice spacing of 0.26 nm, which corresponds to the (101) plane of α-Fe₂O₃, corroborating the dominant bulk α-Fe₂O₃ phase from XRD analyses.

To further understand the structure of the catalysts, Raman spectroscopy measurements were performed. The Raman spectra in Fig. 2d for pristine Fe₂O₃ show seven peaks at ∼225, 244, 292, 410, 499, 612 and 661 cm⁻¹, corresponding to α-Fe₂O₃.^74–77 For both LR-Fe₂O₃ and HR-Fe₂O₃, a negative shift in the α-Fe₂O₃ peaks is observed, with HR-Fe₂O₃ exhibiting a more pronounced shift than LR-Fe₂O₃. This negative shift in Raman wavenumbers is likely due to changes in the electronic structure, coordination environment and/or lattice effects.^78,79 Thus, the more pronounced shift observed in HR-Fe₂O₃ compared to LR-Fe₂O₃ aligns with previous characterization results, which show that higher lattice strain in HR-Fe₂O₃ resulted from greater surface Fe species reduction. Besides, significant broadening of the Raman peaks observed in both LR-Fe₂O₃ and HR-Fe₂O₃ further supports the introduction of lattice strain following ERC,⁵⁷ consistent with XRD and HR-TEM characterization results. Hence, the observed negative shift and peak broadening in the Raman spectra of LR-Fe₂O₃ and HR-Fe₂O₃ can be attributed to structural disorders from non-equivalent sites in the partly reduced Fe species (Fe²⁺/Fe) structure and/or the built strain induced within the catalysts.^57,77 Nonetheless, it can be deduced that the Fe/FeO/Fe₂O₃ components were tuned in the post-conditioned catalysts following ERC treatment. These findings collectively indicate that more negative ERC potentials lead to greater Fe³⁺ reduction and more pronounced strain. As a result, HR-Fe₂O₃ demonstrates the best NH₄⁺ production performance from the NO₂RR due to its pronounced strain in the distorted structure built during the surface Fe₂O₃ reduction.

Overall, the characterization shows that ERC creates structural disorders in the catalysts by partially reducing Fe₂O₃ to FeO and/or metallic Fe, tuning Fe/FeO/Fe₂O₃ components and inducing lattice strain in the catalysts. The extent of Fe₂O₃ species reduction and strain induction increases with more negative ERC treatment potentials, with HR-Fe₂O₃ exhibiting the most pronounced of these characteristics compared to LR-Fe₂O₃ and pristine Fe₂O₃. Given that previous studies have collectively demonstrated the limited contribution of individual Fe or FeO species to enhanced NO_xRR activity,^30,51,80 the improved NO₂RR-to-NH₄⁺ performance observed in this work with HR-Fe₂O₃ and LR-Fe₂O₃ is attributed to the synergistic effects of the tuned FeO/Fe/Fe₂O₃ components and lattice strain. Notably, HR-Fe₂O₃ achieves the best NO₂RR activity with optimized FeO/Fe/Fe₂O₃ components and pronounced strain, delivering the most optimal NH₄⁺ production rate (154 nmol s⁻¹ cm⁻²) and FE_NH4⁺ (93%) at −1.0 V vs. RHE.

To gain further mechanistic insight into the structural change of the active sites in the catalysts during the ERC, in situ Raman measurements were performed. The Raman spectra (Fig. 3a and b) reveal that the peak at ∼408 cm⁻¹, attributed to α-Fe₂O₃,^74–77 remains until the end of the ERC treatment for all samples. However, HR-Fe₂O₃ exhibits a greater reduction in the intensity of this peak (∼80%) compared to LR-Fe₂O₃ (∼59%), indicating a more substantial reduction of Fe₂O₃ at more negative ERC potentials. While the low signal intensity of other α-Fe₂O₃ peaks in the HR-Fe₂O₃ spectra, likely caused by electrolyte blockage and/or high amounts of adsorbed species on the catalyst surface,⁸¹ makes these peaks less visible, the predominant presence of Fe₂O₃ in HR-Fe₂O₃ is supported by the persistence of the α-Fe₂O₃ peak at ∼408 cm⁻¹. Throughout the ERC, both LR-Fe₂O₃ and HR-Fe₂O₃ are observed to exhibit peak shifts, peak broadening, and the emergence of new shoulder peaks (marked with black arrows in the graphs), all indicative of non-uniform strain formation in the catalysts.^57,79,82 This non-uniform strain has been reported to significantly influence catalyst performance by altering the electronic structure and shifting the d-band center closer to the Fermi level, which is a well-recognized indicator for improving the adsorption and activation abilities of the catalysts toward reaction intermediates involved in reactions.^{4,33,34,68–71} Comparatively, HR-Fe₂O₃ displays these changes more prominently and earlier than LR-Fe₂O₃, suggesting a higher susceptibility to strain induction at more negative ERC potentials. This may explain why HR-Fe₂O₃ exhibits the greatest NH₄⁺ production performance during the NO₂RR. Unlike ex situ Raman and XPS analyses, in situ Raman spectra do not consistently show a negative shift to lower wavenumbers or the FeO peak at approximately 580 cm⁻¹. This discrepancy suggests that the FeO observed in ex situ analyses may not be attributed to the residual oxide after surface reduction, which is possibly attributed to the reconstructed FeO during the reoxidation when exposed to air after the reaction.⁸³ Thus, in situ Raman provides a clear view of the structural changes occurring within the catalysts at different ERC potentials. These results demonstrate that HR-Fe₂O₃, preconditioned at a more negative ERC potential, possesses more structural disorders through more obvious reduction of Fe₂O₃ to Fe²⁺/Fe species while building lattice strain, collectively leading to enhanced NH₄⁺ production performance from NO₂RR.

Fig. 3 In situ Raman spectra of (a) LR-Fe₂O₃ and (b) HR-Fe₂O₃ during 30 min ERC treatment under the electrolyte condition of 0.1 M KOH + 0.1 M KNO₂. Black arrows mark the emergence of peak broadening and new shoulder peaks. (c) In situ FTIR spectra of HR-Fe₂O₃ at a NO₂RR applied potential of −1.0 V vs. RHE for 15 min under the electrolyte condition of 0.1 M KOH + 0.1 M KNO₂. (d) DFT calculations of NO₂⁻ adsorption energy on simulated surfaces representing pristine (pristine slab) and defective (oxygen vacancy slab) hematite (α-Fe₂O₃) (104) surfaces.

Further, in situ Fourier transform infrared (FTIR) spectroscopy measurements were conducted under electrochemical NO₂RR conditions with electrolyte (0.1 M KOH + 0.1 M KNO₂) at −1.0 V vs. RHE for 15 min to probe the reaction intermediate species and elucidate the reaction pathways undertaken by these catalysts for producing NH₄⁺. The in situ FTIR spectra, shown in Fig. 3c and S17, reveal peaks corresponding to various intermediates, i.e., NH₂OH* (1215 cm⁻¹), NO₂* (1362 cm⁻¹), NH₂* (1563 cm⁻¹), NH₃* (1767 cm⁻¹), NO* (1971 cm⁻¹), NH₄* (3142 cm⁻¹), and H₂O* (3500 cm⁻¹).^44,84–92 Therefore, in agreement with the literature,^3,6,8,44 this in situ FTIR measurement validates the reaction mechanism for the eR-Fe₂O₃ catalysts. Specifically, ionic NO₂⁻ is adsorbed onto the catalyst surface to form NO₂*, the N–O bond of which is then cleaved to form NO*. The formed NO* is hydrogenated subsequently to produce HNO*, H₂NO*, NH₂OH*, NH₂*, NH₃*, and NH₄*, which ultimately desorbs to form NH₄⁺. Moreover, it is observed that both NO₂* and NO* peaks (at ∼1362 and 1971 cm⁻¹, respectively) exhibit a production trend in pristine Fe₂O₃, but a consumption trend in LR-Fe₂O₃ and HR-Fe₂O₃, suggesting enhanced conversion of these intermediates over the post-conditioned catalysts. When comparing LR-Fe₂O₃ and HR-Fe₂O₃, HR-Fe₂O₃ shows boosted consumption rates of NO* and NO₂* during the in situ experiment, benefiting from the structural reconstruction under the more negative ERC treatment potential as discussed above. Additionally, all samples display a production trend for NH₂OH*, NH₂*, NH₃*, and NH₄* peaks. Notably, HR-Fe₂O₃ demonstrated the most significant increase in NH₂OH* peak (at ∼1215 cm⁻¹) intensity compared to pristine Fe₂O₃ and LR-Fe₂O₃. These results collectively suggest that ERC treatment promotes effective adsorption and faster consumption of NO₂RR intermediates (NO₂* and NO*), leading to improved hydrogenation pathways and ultimately higher NH₄⁺ productivity. This demonstrates that the effects of ERC-induced structural disorders, i.e., optimized Fe³⁺/Fe²⁺/Fe components and lattice strain, collectively drive superior NO₂RR-to-NH₄⁺ performance.

To gain insight into the enhanced NO₂RR performance, density functional theory (DFT) calculations were performed on pristine and defective hematite (α-Fe₂O₃) (104) surfaces. The (104) surface was selected as it represents the most prominent surface orientation in pristine hematite before ERC. Given that ERC reduces the surface oxide and introduces structural disorders, the (104) surface with oxygen vacancies was investigated to study the defective surface effects on NO₂⁻ adsorption and water activation processes. The calculations reveal that NO₂⁻ exhibits stronger adsorption on the coordinatively unsaturated Fe site adjacent to the vacancy (−0.87 eV) compared to the pristine surface (−0.57 eV) (Fig. 3d). This correlates with the experimentally observed enhanced NO₂⁻ conversion rates in ERC-treated catalysts. Along with enhanced nitrite binding, oxygen vacancies act as bifunctional active sites that fundamentally alter water interactions. While water dissociation on the pristine surface is thermoneutral (−0.04 eV), vacancy formation renders this process exothermic (−0.21 eV). The preferred water binding mode shifts from O-down on pristine surfaces (−1.25 eV) to a tilted configuration with hydrogen atoms oriented closer to the surface on the vacancy slab (−1.55 eV), facilitating water dissociation and creating a favorable environment for proton-coupled electron transfer processes (Fig. S20). This enhanced water activation capability ensures a steady supply of surface protons (H*) required for the improved hydrogenation steps observed in the FTIR measurements, providing molecular-level understanding of why ERC treatment enhances NO₂RR performance.

Further, the NO₂RR performance of all samples (Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃) was evaluated over extended periods through a 15-cycle test at −1.0 V vs. RHE, with each cycle comprising 0.5 h of chronoamperometry. Cycling tests were selected over continuous long-term testing to mitigate issues related to electrolyte depletion and concentration drift over time, which occur due to the consumption and gradual depletion of NO₂⁻ species in a batch mode setup.^8,93,94 By mimicking this through a cycling test, it ensures reproducible stability evaluation and a more reliable assessment of catalyst performance under conditions closer to practical operation.^95–98 As evidenced by the steady chronoamperometric i–t curves and stable NH₄⁺ production rate shown in Fig. S18, all electrodes demonstrate high NO₂RR activity over prolonged periods, confirming their stability. Among them, the HR-Fe₂O₃ catalyst exhibits the highest NO₂RR performance, achieving an average NH₄⁺ production rate of 153.0 nmol s⁻¹ cm⁻² and FE_NH4⁺ of 93%, which is competitive with findings reported in the literature (Table S3). While the NH₄⁺ production rate and FE_NH4⁺ are critical metrics for evaluating catalyst performance, the HR-Fe₂O₃ catalyst also excels in other key performance metrics.⁸⁹ Specifically, it demonstrates an exceptionally high partial current density for NH₄⁺ production, reaching ∼96.5 mA cm⁻² at −1.0 V vs. RHE, positioning it among the most active NO₂RR catalysts for NH₄⁺ generation in the literature (Table S3). While it is also important to evaluate NO₂RR performance at lower concentrations and under conditions representative of real wastewater (NO_x concentrations ranging from 0.88 mM to 1.95 M),^99,100 the investigation extends to experiments with PAAE (plasma air activated electrolyte). This electrolyte solution is generated using a plasma system as discussed in previous studies^6,44,81 (Experimental section). Through this approach, it enables a decentralized, air-to-ammonia production pathway, independent of wastewater sources, thus broadening its applicability across a wide range of deployable scenarios. Building on previous studies, this work focuses on optimizing the production of NO₂⁻ over NO₃⁻ by manipulating various operating conditions. First, the pH of the reaction medium was altered by comparing neutral (deionized water) to alkaline conditions (0.1 M KOH). The results (Fig. S19a) indicate that the alkaline condition (0.1 M KOH) significantly favors the production of NO₂⁻ (9.6 mM) over NO₃⁻ (0.8 mM), whereas in neutral environments, the production trends reverse, favoring NO₃⁻ over NO₂⁻. Further, the production of NO_x via plasma was tested across different plasma voltages, as voltage is reported to fundamentally affect the concentration of NO_x production.^101,102 The results (Fig. S19b) indicate that the highest amounts of NO₂⁻ and total NO_x are produced at 200 V, with concentrations of 13.7 and 15.0 mM, respectively. NO_xRR performance testing in the optimally produced PAAE was conducted for 10 cycles (0.5 h chronoamperometry each cycle) within a commercial H-cell using pristine Fe₂O₃ and the best performing catalyst, HR-Fe₂O₃. The results further demonstrate the high stability of both Fe₂O₃ and HR-Fe₂O₃ across the 10-cycle operation at −1.0 V vs. RHE, evidenced by steady chronoamperometric i–t curves and consistent NH₄⁺ production rates (Fig. S20a and b, respectively). Additionally, the results suggest that HR-Fe₂O₃ outperforms pristine Fe₂O₃ even under low NO_x concentrations, achieving a stable NH₄⁺ production rate of 32.0 nmol s⁻¹ cm⁻² and a steady current density of ∼52 mA cm⁻² across 10-cycle tests at −1.0 V vs. RHE. These results are comparable to those reported in the literature (Table S3) for studies involving similar low concentrations of NO_x. Hence, conducting the NO_xRR using PAAE reveals the potential for a self-sustaining NH₄⁺ production system through the integration of a plasma-electrolyzer setup. Furthermore, the ability of HR-Fe₂O₃ to efficiently convert trace amounts of NO_x to NH₄⁺ in PAAE underscores its potential for real-world applications, particularly in treating waste streams from industries, power plants, and agriculture, where NO_x is commonly found at low concentrations.

3 Conclusions

In conclusion, this work addresses key knowledge gaps regarding ERC-treated Fe-based catalysts with structural disorders in enhancing NO₂RR activity for NH₄⁺ production in alkaline environments. A combination of ex situ and in situ characterization techniques collectively reveals that more negative ERC potentials enhance the reduction of surface Fe₂O₃ to FeO and metallic Fe, optimizing Fe³⁺/Fe²⁺/Fe components, while simultaneously inducing significant lattice strain. This strain alters the d-band structure, promoting the hydrogenation of NO₂RR intermediates and enhancing NH₄⁺ production. These findings underscore the synergistic effects of lattice strain engineering and Fe³⁺/Fe²⁺/Fe components in creating highly effective active sites for the NO₂RR, offering valuable insights for designing next-generation metal oxide catalysts. Additionally, this work addresses significant knowledge gaps in utilizing plasma technology to optimize NO_x feedstock generation for NO₂⁻ formation. Systematic optimization of plasma conditions (solution pH and reactor voltage) enhances NO₂⁻ generation up to 13.7 mM, thereby offering a more efficient and targeted feedstock for the NO₂RR. Overall, this work showcases that the direct reduction of NO₂⁻ to NH₄⁺ offers substantial advantages for green ammonia production, bypassing the most energy-intensive NO₃⁻-to-NO₂⁻ step and reducing overall energy costs. The enhanced hydrogenation capabilities of structurally engineered Fe₂O₃ catalysts in this work enable greater selectivity toward NH₄⁺, with minimal H₂ by-products, thereby improving sustainability. Moreover, the integration of this plasma-electrolyzer system enables a decentralized, self-sustaining pathway for transforming air-derived NO_x into ammonia-related products on demand.

4 Experimental

4.1 Materials

All chemical reagents and solvents utilized in this work were used as received and without any further purification. Deionized water (resistivity 18.2 MΩ cm⁻¹) was used throughout all experiments.

4.2 Catalyst synthesis

Fe₂O₃ catalyst powders were prepared through a precipitation method followed by calcination. First, 2 g of iron(III) nitrate nonahydrate (Sigma-Aldrich, ≥99.95%) was dissolved in 15 mL of deionized water and stirred for 0.5 h to achieve a homogeneous solution. This solution was then gradually added dropwise to 100 mL of a 1 mol L⁻¹ Na₂CO₃ (Chem-Supply, analytical grade, 99.5%) solution maintained at 60 °C. The mixture was aged and stirred continuously for 3 h. The resulting slurry was centrifuged and washed with ethanol (Chem-Supply, analytical grade, 99.5%) and water (3–5 times each) to remove impurities and then dried overnight at 100 °C. The black crystals obtained were ground into powders and calcined in flowing air (Coregas, purity 100%) at 400 °C for 2 h, yielding brown Fe₂O₃ powders.

To prepare pristine Fe₂O₃ working electrodes, these Fe₂O₃ powders were spray-cast onto carbon fibre paper. Specifically, 3 mg of Fe₂O₃ powder was dispersed in a solution containing 1 mL deionized water, 1 mL ethanol, and 30 µL of a Nafion 117 solution (Sigma-Aldrich, ∼5% in a mixture of lower aliphatic alcohols and water). The mixture was sonicated for 10–15 min to ensure homogeneity. One-third of this mixture was sprayed onto carbon fibre paper placed on a hot plate at 110 °C, evenly covering an area of 1 cm × 6 cm. After cooling, the sprayed carbon fibre paper was cut into six pristine Fe₂O₃ electrodes.

To modify the surface properties of Fe₂O₃, the ERC strategy was employed, with the ERC treated Fe₂O₃ electrodes referred to as eR-Fe₂O₃. Specifically, prior to electrochemical testing, the Fe₂O₃ electrodes were pre-treated at reduction potentials of −1.0 V and −2.0 V vs. RHE for 0.5 h. The Fe₂O₃ electrodes conditioned at −1.0 V are referred to as Low-Reduction Fe₂O₃ (LR-Fe₂O₃), while those treated at −2.0 V are named High-Reduction Fe₂O₃ (HR-Fe₂O₃). This pre-treatment was conducted in a commercial three-electrode H-cell containing 0.1 M KOH (Sigma-Aldrich, ACS reagent grade, 90% flakes) and 0.1 M KNO₂ (Sigma-Aldrich, ACS reagent grade, ≥ 96.0%) as the catholyte, similar to the setup used for electrochemical NO₂RR-to-NH₄⁺ activity testing.

4.3 Electrochemical experiments

All chemical reagents and solvents utilized in this work were used as received and without any further purification. Deionized water (resistivity: 18.2 MΩ cm⁻¹) was used throughout all experiments. The electrocatalytic NO₂RR-to-NH₄⁺ was carried out using a commercial H-cell setup. A three-electrode system was employed, comprising the eR-Fe₂O₃ catalyst as the working electrode (cathode), a Pt plate as the counter electrode (anode), and a Hg/HgO reference electrode. The anolyte was 50 mL of 0.1 M KOH solution, whereas the catholyte was 50 mL of 0.1 M KOH with 0.1 M KNO₂. An anion exchange membrane (AEM) (Dioxide Materials, X37-50 Grade RT) was used to separate the electrolytes while allowing ions to pass through. PAAE was also used as the catholyte to evaluate the NO₂RR performance of the catalysts at lower concentrations and under conditions representative of real wastewater.^99,100 All electrochemical NO₂RR processes were performed using an Autolab Potentiostat (Nova 2 Metrohm).

Chronoamperometry tests were employed to identify the optimal potential for maximum NH₄⁺ production efficiency. Each test was conducted for a duration of 15 min and repeated three times to obtain average values. All potentials were converted to the RHE using the following equations:^103–105

E_RHE (V) = E_Hg/HgO (V) + 0.098 + 0.059 × pH

Long-term stability was assessed at the optimal applied potential under conditions similar to the chronoamperometry tests. The stability tests consisted of 15 cycles at the optimal potential, with each cycle lasting 30 min. LSV measurements were conducted using the same H-cell setup under three different catholyte conditions: 0.1 M KOH, 0.1 M KOH + 0.1 M KNO₃, and 0.1 M KOH + 0.1 M KNO₂. EIS experiments were performed under similar conditions to the chronoamperometry tests, with a frequency range from 1 MHz to 0.01 Hz. The measurements were taken at a potential of 0.615 V vs. RHE.

The ECSAs of the catalysts were estimated using the CDL method. Briefly, cyclic voltammetry (CV) scans were recorded within a potential window of ±0.05 V around each catalyst's onset potential, as identified from the LSV curves. The scan rates used were 100, 200, 300, 400, and 500 mV s⁻¹. The differences between the cathodic and anodic currents at the midpoint potential were plotted against the scan rates. The slope of the linear fit to this plot corresponds to the CDL. The ECSA was then calculated by dividing the CDL of the catalyst by the specific capacitance of a reference blank glassy carbon electrode (40 µF cm⁻²) and multiplying by the geometric surface area of the electrode.⁴⁵

4.4 Products analysis

The concentration of NH₄⁺ produced in the catholyte was determined using the indophenol blue test. In this method, 0.5 mL of the catholyte sample was transferred into a 1.5 mL sample tube. To this, 0.4 mL of a 1 M NaOH solution (Sigma-Aldrich, semiconductor grade, 99.99%) containing 5 wt% salicylic acid (Sigma-Aldrich, ACS grade, 99.0%) and 5 wt% sodium citrate (Sigma-Aldrich, ACS grade, 99.0%) was added, followed by 0.1 mL of a 0.05 M sodium hypochlorite solution (Sigma-Aldrich, ACS grade, 10–15%) and 30 µL of a 1 wt% sodium nitroferricyanide solution (Sigma-Aldrich, ACS grade, 99%). The mixture was thoroughly mixed and incubated in the dark at room temperature for 1 h. If the NH₄⁺ concentration was excessively high, the catholyte was appropriately diluted prior to analysis. After incubation, the amount of NH₄⁺ was quantified using ultraviolet-visible (UV-vis) spectroscopy (Shimadzu, UV-360). Absorbance measurements were taken over wavelengths ranging from 550 to 800 nm. A calibration curve was constructed using the peak absorbance values (Fig. S21). The performance of the catalysts in generating NH₄⁺ was assessed by calculating the production rate (Rate_NH4⁺) and faradaic efficiency (FE_NH4⁺) using the following equations:^106,107

where V is the volume of the catholyte, t is the time taken for the NO₂RR, S is the effective area of the working electrode, n is the desired electrons to synthesise one NH₄⁺ molecule (n = 6), F is the Faraday constant (F = 96 485.33 A s mol⁻¹) and Q is the overall electrical energy consumed by electrodes throughout the duration of the electrochemical NO₂RR process.

In addition, ion chromatography (IC) was employed to quantify the concentrations of NH₄⁺, NO₂⁻, and NO₃⁻ and validate the calorimetric methods. The analyses were performed using an AQ400 ion chromatograph from Seal Analytical. To measure the amount of H₂ produced in the cathodic chamber, the solution was purged with Ar (Coregas, >99.95% purity) gas for approximately 15 min before conducting performance tests. A Shimadzu GC-2010 gas chromatograph (GC) equipped with a thermal conductivity detector (TCD) and a Supelco Carboxen 1010 column was used for analysis. A calibration curve (Fig. S22) was generated using known concentrations of H₂ to ensure accurate quantification.

4.5 Physical characterization

For the study of the surface morphology, crystal size, and surface element distribution of the catalysts, a Nano SEM 230 scanning electron microscope (SEM) and a FEI Tecnai G2 20 transmission electron microscope (TEM) were utilized. An accelerating voltage of 200 kV was applied for the TEM measurement. To prepare samples for TEM analysis, the catalyst nanoparticles were first scrapped off from the electrode substrate. They were then dispersed in ethanol by gentle ultrasonic agitation for a few minutes. Using a micropipette, a small droplet of the resulting dispersed suspension was drop-cast onto a gold TEM grid. Finally, the samples were dried under ambient conditions before being used for TEM measurement. X-ray diffraction (XRD) studies were performed using an X'Pert PRO diffractometer with Cu Kα radiation (λ = 1.54060 Å). The scanning range spanned from 10° to 90° with a step size of 0.013°, aiming to identify the phase compositions and crystallinity of the catalysts. Data collected from the XRD experiments were analyzed using HighScore Plus software.

The surface chemical states of the catalysts were investigated using a Thermo ESCALAB 250Xi X-ray photoelectron spectrometer (XPS). A monochromatic Al Kα source with an excitation energy of 1486.68 eV was employed. The instrument operated at a typical power of 120 W, with an average spot size of 500 µm in diameter. Survey scans were recorded with a pass energy of 100 eV, while detailed regional scans used a pass energy of 20 eV. Calibration of the C 1s XPS spectra was performed by setting the graphitic C–C peak position to 284.8 eV. Raman spectra were acquired at room temperature using either an inVia 2 or an inVia Qontor Raman spectrometer. The inVia 2 instrument utilized a 532 nm wavelength diode laser with 50× magnification, while the inVia Qontor used a 514 nm diode laser, also at 50× magnification. These measurements provided insights into the surface chemistry and composition of the catalysts. Additionally, in situ Raman measurements were conducted under applied potentials corresponding to the electrochemical reactions being studied, maintaining similar operational parameters.

In situ Fourier transform infrared (FTIR) measurements were performed at the infrared beamline of the Australian Synchrotron. The experimental setup involved a three-electrode electrochemical cell equipped with a ZnSe crystal serving as the infrared transmission window. An Ag/AgCl electrode acted as the reference electrode, while a Pt wire served as the counter electrode. The working electrode (i.e., pristine Fe₂O₃, LR-Fe₂O₃, and HR-Fe₂O₃) was connected to a potentiostat to apply the required potential for the electrochemical reaction under investigation. FTIR spectra were continuously collected during the experiments in reflection mode at a resolution of 4 cm⁻¹. To enhance the signal-to-noise ratio, each spectrum was the result of accumulating 512 scans. Background spectra were obtained using the same setup at open-circuit voltage before commencing the tests.

In situ Fe K-edge XANES measurement was carried out at the XAS beamline of the Australian Synchrotron Center. The data were collected in fluorescence mode using a solid-state 18-element detector. The in situ cell was mounted on the beamline sample stage, with the electrode oriented at a 45° angle relative to the fluorescence detector.¹⁰⁸ The incident beam intensity was monitored using an OKEN ionization chamber, and all measurements were performed at ambient temperature.

In situ powder diffraction (PD) experiments were carried out using a custom-designed three-electrode electrochemical cell, equipped with a Kapton window (2 cm diameter) to enable X-ray irradiation.¹⁰⁹ For the measurements, the cell was configured with an Ag/AgCl electrode as the reference, a Pt wire as the counter electrode, and the prepared catalyst as the working electrode.

4.6 Plasma system for PAAE generation

An in-house, custom-built plasma system (as described in the work by Sun et al.⁶), comprising a plasma generator (‘Leap 100’ from PlasmaLeap Technologies) and a plasma reactor generating PAAE (plasma air activated electrolyte), was set up as illustrated in Fig. S23. The plasma reactor was filled with 250 mL of 0.1 M KOH. Air (Coregas, purity 100%) was fed into the plasma system at a constant flow rate of 20 L min⁻¹. The plasma system reaction was operated under constant plasma generator operating parameters (i.e., voltage, duty, discharge frequency, and resonance frequency) for 2 h, with the solution in the reactor continuously stirred at a constant speed of 350 rpm. After 2 h, the plasma activated solution, PAAE rich in NO_x species, including NO₂⁻ and NO₃⁻, was cooled down to room temperature before being used for subsequent electrochemical measurements.

4.7 Computational methods

All spin-polarized calculations were implemented using the Vienna Ab initio Simulation Package (VASP) with the Projector Augmented Wave (PAW) method, interfaced through the Atomic Simulation Environment (ASE).^110–115 Electronic wavefunctions were expanded in a plane wave basis set with a 500 eV cutoff energy using gamma-point calculations. Electronic self-consistency was achieved with a convergence criterion of 10⁻⁶ eV, while ionic relaxations were terminated when forces on all unconstrained atoms fell below 0.05 eV Å⁻¹. The Perdew–Burke–Ernzerhof (PBE) functional was employed to account for exchange–correlation effects within the generalized gradient approximation (GGA) framework. To accurately account for the strongly correlated Fe 3d electrons, the DFT+U approach was implemented with a Hubbard U parameter of 4.2 eV for Fe atoms, following Dudarev's formalism. To model hematite surfaces a five-layer α-Fe₂O₃(104) slab with a 15 Å vacuum gap in the z-direction was constructed (Fig. S24). The top three atomic layers were fully relaxed to interact with adsorbates, while the bottom two layers were constrained to maintain bulk-like properties. The charged NO₂⁻ species was avoided as a reference in our DFT calculation. A thermodynamic cycle method was instead used to find the reference energy for NO₂⁻. This approach uses the deprotonation of HNO₂ as a reference.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: Fig. S1–S24 and Tables S1–S3. See DOI: https://doi.org/10.1039/d5ta09764a.

Acknowledgements

The work was supported by the Australian Research Council (ARC) Training Centre for Global Hydrogen Economy (IC200100023). The authors acknowledge research funding provided by the ARC (DE230100789 and DE230101396). R. D. acknowledges funding from the UNSW Scientia Fellowship. Z. M. acknowledges the UNSW-Tsinghua Collaborative Research Seed Program 2023 and the fellowship program by the International Hydrogen Research Collaboration Program funded by the Commonwealth Scientific and Industrial Research Organisation (CSIRO). Z. M., R. D., R. A., and E. L. acknowledge funding support from ARENA Hydrogen R&D (TRAC 2023) Pro-1029. The authors also acknowledge the Infrared Microscopy/ANSTO Grant (AS241/IRM/21262) for the in situ FTIR beamtime, the XAS/ANSTO Grant (AS261/XAS/24311a) for the in situ/ex situ XANES, and the PD/ANSTO Grant at the Australian Synchrotron.

References

1. G. Eliseev, in 2024 Annual Conference (Fertecon/S&P Global Commodity Insights), S&P Global Commodity Insights (Fertecon): Conference Presentation, 2024, pp. 1–17.
2. H. Wang, N. Lin and M. Arzumanyan, Gases, 2024, 4, 224–235,  DOI:10.3390/gases4030013.
3. Y. Wang, W. Zhou, R. Jia, Y. Yu and B. Zhang, Angew. Chem., Int. Ed., 2020, 59(13), 5350–5354,  DOI:10.1002/anie.201915992.
4. Y. Wang, A. Xu, Z. Wang, L. Huang, J. Li, F. Li, J. Wicks, M. Luo, D.-H. Nam, C.-S. Tan, Y. Ding, J. Wu, Y. Lum, C.-T. Dinh, D. Sinton, G. Zheng and E. H. Sargent, J. Am. Chem. Soc., 2020, 142(12), 5702–5708,  DOI:10.1021/jacs.9b13347.
5. Y. Guo, R. Zhang, S. Zhang, Y. Zhao, Q. Yang, Z. Huang, B. Dong and C. Zhi, Energy Environ. Sci., 2021, 14(7), 3938–3944,  10.1039/d1ee00806d.
6. J. Sun, D. Alam, R. Daiyan, H. Masood, T. Zhang, R. Zhou, P. J. Cullen, E. C. Lovell, A. Jalili and R. Amal, Energy Environ. Sci., 2021, 14(2), 865–872,  10.1039/d0ee03769a.
7. O. Serpell, Z. Hsain, A. Chu and W. Johnsen, Ammonia's Role in a Net-Zero Hydrogen Economy, 2023.
8. R. Daiyan, T. Tran-Phu, P. Kumar, K. Iputera, Z. Tong, J. Leverett, M. H. A. Khan, A. Asghar Esmailpour, A. Jalili, M. Lim, A. Tricoli, R.-S. Liu, X. Lu, E. Lovell and R. Amal, Energy Environ. Sci., 2021, 14(6), 3588–3598,  10.1039/d1ee00594d.
9. C. Smith, A. K. Hill and L. Torrente-Murciano, Energy Environ. Sci., 2020, 13(2), 331–344,  10.1039/c9ee02873k.
10. D. Ye and S. C. E. Tsang, Nat. Synth., 2023, 2(7), 612–623,  DOI:10.1038/s44160-023-00321-7.
11. M. L. Carreon, J. Phys. D: Appl. Phys., 2019, 52(48), 483001,  DOI:10.1088/1361-6463/ab3b2c.
12. X. Bian, Y. Zhao, S. Zhang, D. Li, R. Shi, C. Zhou, L. Z. Wu and T. Zhang, ACS Mater. Lett., 2021, 3(11), 1521–1527,  DOI:10.1021/acsmaterialslett.1c00504.
13. B. Lin, T. Wiesner and M. Malmali, ACS Sustain. Chem. Eng., 2020, 8(41), 15517–15531,  DOI:10.1021/acssuschemeng.0c04313.
14. R. Hawtof, S. Ghosh, E. Guarr, C. Xu, R. M. Sankaran and J. N. Renner, Asian J. Chem., 2019, 31(2), 5778–5789,  DOI:10.1126/sciadv.aat5778.
15. Y. He, Z. Chen, Z. Li, G. Niu and J. Tang, Front. Optoelectron., 2018, 11(1), 92–96,  DOI:10.1007/s12200-018-0807-z.
16. S. Giddey, S. P. S. Badwal, C. Munnings and M. Dolan, ACS Sustain. Chem. Eng., 2017, 5(11), 10231–10239,  DOI:10.1021/acssuschemeng.7b02219.
17. S. Kumari, S. Pishgar, M. E. Schwarting, W. F. Paxton and J. M. Spurgeon, Chem. Commun., 2018, 54(95), 13347–13350,  10.1039/c8cc07869f.
18. H. L. Du, M. Chatti, R. Y. Hodgetts, P. V. Cherepanov, C. K. Nguyen, K. Matuszek, D. R. MacFarlane and A. N. Simonov, Nature, 2022, 609(7928), 722–727,  DOI:10.1038/s41586-022-05108-y.
19. H. Chen, D. Yuan, A. Wu, X. Lin and X. Li, Waste Disposal Sustainable Energy, 2021, 3, 201–217,  DOI:10.1007/s42768-021-00074-z.
20. L. Li, C. Tang, X. Cui, Y. Zheng, X. Wang, H. Xu, S. Zhang, T. Shao, K. Davey and S.-Z. Qiao, Angew. Chem., 2021, 133(25), 14250–14256,  DOI:10.1002/anie.202104394.
21. W. Li, S. Zhang, J. Ding, J. Liu, Z. Wang, H. Zhang, J. Ding, L. Chen and C. Liang, ACS Sustain. Chem. Eng., 2023, 11(3), 1168–1177,  DOI:10.1021/acssuschemeng.2c06525.
22. Z. Zhang, N. Zhang, J. Zhang, B. Deng, Z. Cao, Z. Wang, G. Wei, Q. Zhang, R. Jia, P. Xiang and S. Xia, Chem. Eng. J., 2024, 483, 148952,  DOI:10.1016/j.cej.2024.148952.
23. S. Yin, R. Cao, Y. Han, J. Shang, J. Zhang, W. Jiang and G. Liu, J. Energy Chem., 2024, 96, 642–668,  DOI:10.1016/j.jechem.2024.05.023.
24. X. Zou, J. Xie, C. Wang, G. Jiang, K. Tang and C. Chen, Chin. Chem. Lett., 2023, 34(6), 107908,  DOI:10.1016/j.cclet.2022.107908.
25. J.-Y. Fang, J.-L. Fan, S.-B. Liu, S.-P. Sun and Y.-Y. Lou, Materials, 2023, 16(11), 4000,  DOI:10.3390/ma16114000.
26. L. Xie, Y. Wang, Q. Kong and R. Cao, ChemCatChem, 2024, 16(21), e202400956,  DOI:10.1002/cctc.202400956.
27. Z. Yu, M. Gu, Y. Wang, H. Li, Y. Chen and L. Wei, Adv. Energy Sustainability Res., 2024, 5(5), 2300284,  DOI:10.1002/aesr.202300284.
28. J. Liang, Z. Li, L. Zhang, X. He, Y. Luo, D. Zheng, Y. Wang, T. Li, H. Yan, B. Ying, S. Sun, Q. Liu, M. S. Hamdy, B. Tang and X. Sun, Chem, 2023, 9(7), 1768–1827,  DOI:10.1016/j.chempr.2023.05.037.
29. T. Li, C. Tang, H. Guo, H. Wu, C. Duan, H. Wang, F. Zhang, Y. Cao, G. Yang and Y. Zhou, ACS Appl. Mater. Interfaces, 2022, 14(44), 49765–49773,  DOI:10.1021/acsami.2c14215.
30. T. Li, C. Tang, H. Guo, J. Yang, F. Zhang, G. Yang and Y. Zhou, Chem. Eng. J., 2024, 485, 149560,  DOI:10.1016/j.cej.2024.149560.
31. Q. Liu, Q. Liu, L. Xie, L. Yue, T. Li, Y. Luo, N. Li, B. Tang, L. Yu and X. Sun, Chem. Commun., 2022, 58(33), 5160–5163,  10.1039/d2cc00611a.
32. G.-F. Chen, Y. Yuan, H. Jiang, S.-Y. Ren, L.-X. Ding, L. Ma, T. Wu, J. Lu and H. Wang, Nat. Energy, 2020, 5(8), 605–613,  DOI:10.1038/s41560-020-0654-1.
33. Z. Hou, C. Cui, Y. Li, Y. Gao, D. Zhu, Y. Gu, G. Pan, Y. Zhu and T. Zhang, Adv. Mater., 2023, 35(39), 2209876,  DOI:10.1002/adma.202209876.
34. A. Herklotz, D. Lee, E. J. Guo, T. L. Meyer, J. R. Petrie and H. N. Lee, J. Phys.: Condens. Matter, 2017, 29(49), 493001,  DOI:10.1088/1361-648X/aa949b.
35. L. Fang, S. Wang, C. Song, S. Lu, X. Yang, X. Qi and H. Liu, Chem. Eng. J., 2022, 446, 137341,  DOI:10.1016/j.cej.2022.137341.
36. Y. Qin, T. Yu, S. Deng, X.-Y. Zhou, D. Lin, Q. Zhang, Z. Jin, D. Zhang, Y.-B. He, H.-J. Qiu, L. He, F. Kang, K. Li and T.-Y. Zhang, Nat. Commun., 2022, 13(1), 3784,  DOI:10.1038/s41467-022-31468-0.
37. J. Li, G. Zhan, J. Yang, F. Quan, C. Mao, Y. Liu, B. Wang, F. Lei, L. Li, A. W. M. Chan, L. Xu, Y. Shi, Y. Du, W. Hao, P. K. Wong, J. Wang, S.-X. Dou, L. Zhang and J. C. Yu, J. Am. Chem. Soc., 2020, 142(15), 7036–7046,  DOI:10.1021/jacs.0c00418.
38. S. Liu, W. Miao, K. Ma, H. Teng, X. Zhang, J. Li, W. Li, X. Cui and L. Jiang, Appl. Catal., B: Environ. Energy, 2024, 350, 123919,  DOI:10.1016/j.apcatb.2024.123919.
39. J. Wu and Y.-X. Yu, Catal. Sci. Technol., 2021, 11(21), 7160–7170,  10.1039/d1cy01217g.
40. X. Zhang, Q. Zhang, Y. Peng, X. Ma and G. Fan, Int. J. Hydrogen Energy, 2022, 47(12), 7793–7801,  DOI:10.1016/j.ijhydene.2021.12.122.
41. F.-Y. Chen, Z.-Y. Wu, S. Gupta, D. J. Rivera, S. V. Lambeets, S. Pecaut, J. Y. T. Kim, P. Zhu, Y. Z. Finfrock, D. M. Meira, G. King, G. Gao, W. Xu, D. A. Cullen, H. Zhou, Y. Han, D. E. Perea, C. L. Muhich and H. Wang, Nat. Nanotechnol., 2022, 17(7), 759–767,  DOI:10.1038/s41565-022-01121-4.
42. T. Duan, Y. Chen, Q. Wen and Y. Duan, RSC Adv., 2015, 5(25), 19601–19612,  10.1039/c5ra01876e.
43. T. Rajaramanan, M. Natarajan, P. Ravirajan, M. Senthilnanthanan and D. Velauthapillai, Energies, 2020, 13(7), 1532,  DOI:10.3390/en13071532.
44. M. Lim, Z. Ma, G. O'Connell, J. A. Yuwono, P. Kumar, R. Jalili, R. Amal, R. Daiyan and E. C. Lovell, Small, 2024, 20(33), 2401333,  DOI:10.1002/smll.202401333.
45. Z. Ma, C. Tsounis, C. Y. Toe, P. V. Kumar, B. Subhash, S. Xi, H. Y. Yang, S. Zhou, Z. Lin, K. H. Wu, R. J. Wong, L. Thomsen, N. M. Bedford, X. Lu, Y. H. Ng, Z. Han and R. Amal, ACS Catal., 2022, 12(9), 4792–4805,  DOI:10.1021/acscatal.1c05431.
46. P. X. Li, L.-M. Qu, C.-H. Zhang, X.-B. Ren, H.-X. Wang, J.-L. Zhang, Y.-W. Mu and B.-L. Lü, J. Fuel Chem. Technol., 2021, 49(10), 1558–1566,  DOI:10.1016/s1872-5813(21)60154-8.
47. H. Chen, Y. Ma, Y. Han, X. Mao, Y. Hu, X. Zhao, Q. Dong, B. Wen, A. Du, X. Wang, X. Lyu and Y. Jia, Small, 2024, 20(26), 2309689,  DOI:10.1002/smll.202309689.
48. Y. Y. Fu, R. M. Wang, J. Xu, J. Chen, Y. Yan, A. V. Narlikar and H. Zhang, Chem. Phys. Lett., 2003, 379(3–4), 373–379,  DOI:10.1016/j.cplett.2003.08.061.
49. D. J. Morgan, X-Ray Photoelectron Spectroscopy (XPS): An Introduction, 2014, https://sites.cardiff.ac.uk/xpsaccess/files/2014/07/AccessXPS_Primer_Paper.pdf.
50. X. Liang, X. Wang, X. Zhang, S. Lin, M. Ji, Q. Liu and M. Wang, ACS Catal., 2024, 14(7), 4648–4655,  DOI:10.1021/acscatal.4c00016.
51. C. Liu, G. Zhang, W. Zhang, Z. Gu and G. Zhu, Proc. Natl. Acad. Sci. U. S. A., 2023, 120(3), e2209979120,  DOI:10.1073/pnas.
52. T. J. Frankcombe and Y. Liu, Chem. Mater., 2023, 35(14), 5468–5474,  DOI:10.1021/acs.chemmater.3c00801.
53. M. C. Biesinger, B. P. Payne, A. P. Grosvenor, L. W. M. Lau, A. R. Gerson and R. S. C. Smart, Appl. Surf. Sci., 2011, 257(7), 2717–2730,  DOI:10.1016/j.apsusc.2010.10.051.
54. J. Bernal Romero, O. Chavez, C. Rodriguez, S. Gonzalez, J. Fregoso, V. Carbajal, G. Clavel, O. Hudak, J. Cheng, J. Tandoc, J. Brown and L. Gao, MRS Adv., 2024, 9(15), 1176–1182,  DOI:10.1557/s43580-024-00866-5.
55. R. Rahul, R. K. Singh, B. Bera, R. Devivaraprasad and M. Neergat, Phys. Chem. Chem. Phys., 2015, 17(23), 15146–15155,  10.1039/c5cp00692a.
56. V. G. Antunes, C. A. Figueroa and F. Alvarez, Langmuir, 2022, 38(6), 2109–2116,  DOI:10.1021/acs.langmuir.1c03189.
57. Y. Kim, M. Watanabe, J. Matsuda, J. T. Song, A. Takagaki, A. Staykov and T. Ishihara, Appl. Catal., B, 2020, 278, 119292,  DOI:10.1016/j.apcatb.2020.119292.
58. B. Nasiri-Tabrizi, J. Adv. Ceram., 2014, 3(1), 31–42,  DOI:10.1007/s40145-014-0090-4.
59. F. Craciun, C. Galassi and R. Birjega, J. Appl. Phys., 2012, 112(12), 124106,  DOI:10.1063/1.4770326.
60. J. Li, X. L. Wu, D. S. Hu, Y. M. Yang, T. Qiu and J. C. Shen, Solid State Commun., 2004, 131(1), 21–25,  DOI:10.1016/j.ssc.2004.04.026.
61. G. Perenlei, P. C. Talbot, W. N. Martens, J. Riches and J. A. Alarco, RSC Adv., 2017, 7(26), 15632–15643,  10.1039/c6ra27633d.
62. Y. Sun, Y. Ren, Y. Liu, J. Wen, J. S. Okasinski and D. J. Miller, Nat. Commun., 2012, 3, 6–11,  DOI:10.1038/ncomms1963.
63. Q. Lei, L. Huang, J. Yin, B. Davaasuren, Y. Yuan, X. Dong, Z. P. Wu, X. Wang, K. X. Yao, X. Lu and Y. Han, Nat. Commun., 2022, 13(1), 4857,  DOI:10.1038/s41467-022-32601-9.
64. Q. Fan, P. Yan, F. Liu, Z. Xu, P. Liang, X. Cao, C. Ye, M. Liu, L. Zhao, S. Ren, H. Miao, X. Zhang, Z. Yang, X. Ding, J. Yang, C. Kong and Y. Wu, Sci. Bull., 2024, 69(18), 2881–2891,  DOI:10.1016/j.scib.2024.06.031.
65. F. Zaza, G. Orio, E. Serra, F. Caprioli and M. Pasquali, AIP Conf. Proc., 2015, 1667, 020004,  DOI:10.1063/1.4922560.
66. M. Biswas, D. Misra and T. K. Kundu, Eur. Phys. J. B, 2023, 96(6), 74,  DOI:10.1140/epjb/s10051-023-00547-6.
67. J. Duan, Y. Zhao, Y. Wu, Y. Liu, J. Chen, R. Yang, J. Huang, C. Luo, M. Wu, X. Zheng, P. Li, X. Jiang, J. Guan and T. Zhai, Chem. Sci., 2024, 15, 19546–19555,  10.1039/d4sc05873a.
68. W. Wen, P. Yan, W. Sun, Y. Zhou and X.-Y. Yu, Adv. Funct. Mater., 2023, 33(6), 2212236,  DOI:10.1002/adfm.202212236.
69. R. Zhang, S. Shang, F. Wang and K. Chu, Dalton Trans., 2024, 53(8), 3470–3475,  10.1039/d4dt00043a.
70. K. Huang, K. Tang, M. Wang, Y. Wang, T. Jiang and M. Wu, Adv. Funct. Mater., 2024, 34(24), 2315324,  DOI:10.1002/adfm.202315324.
71. J. Geng and S. Ji, Nano Res., 2024, 17(6), 4898–4907,  DOI:10.1007/s12274-024-6480-1.
72. C. Zhao, G. Tan, W. Yang, C. Xu, T. Liu, Y. Su, H. Ren and A. Xia, Sci. Rep., 2016, 6(1), 38603,  DOI:10.1038/srep38603.
73. P. Platt, R. Mella, W. DeMaio, M. Preuss and M. R. Wenman, Comput. Mater. Sci., 2017, 140, 322–333,  DOI:10.1016/j.commatsci.2017.09.001.
74. H. Cao, G. Wang, L. Zhang, Y. Liang, S. Zhang and X. Zhang, ChemPhysChem, 2006, 7(9), 1897–1901,  DOI:10.1002/cphc.200600130.
75. I. Chourpa, L. Douziech-Eyrolles, L. Ngaboni-Okassa, J. F. Fouquenet, S. Cohen-Jonathan, M. Soucé, H. Marchais and P. Dubois, Analyst, 2005, 130(10), 1395–1403,  10.1039/b419004a.
76. N. D. Phu, D. T. Ngo, L. H. Hoang, N. H. Luong, N. Chau and N. H. Hai, J. Phys. D: Appl. Phys., 2011, 44(34), 345002,  DOI:10.1088/0022-3727/44/34/345002.
77. D. L. A. De Faria, S. Venâncio Silva and M. T. De Oliveira, J. Raman Spectrosc., 1997, 28(11), 873–878,  DOI:10.1002/(sici)1097-4555(199711)28:11<873::aid-jrs177>3.0.co;2-b.
78. G. M. Bhalerao, M. K. Singh, A. K. Sinha and H. Ghosh, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 86(12), 125419,  DOI:10.1103/physrevb.86.125419.
79. S. Y. Davydov and O. V. Posrednik, Phys. Solid State, 2021, 63(4), 530–533,  DOI:10.1134/s1063783421040053.
80. H. Xiang, W. Liu, L. Su, S. Chen, Y. Han, C. Zhu, S. Wang, C. Tan and L. Zhang, Sep. Purif. Technol., 2024, 330, 125357,  DOI:10.1016/j.seppur.2023.125357.
81. M. Lim, J. Sun, Z. Ma, R. Jalili, R. Daiyan, E. C. Lovell and R. Amal, ACS Appl. Nano Mater., 2023, 6(6), 4936–4945,  DOI:10.1021/acsanm.3c00681.
82. M. K. Zhu, P. X. Lu, Y. D. Hou, X. M. Song, H. Wang and H. Yan, J. Am. Ceram. Soc., 2006, 89(12), 3739–3744,  DOI:10.1111/j.1551-2916.2006.01281.x.
83. Z. Yuan and H. Y. Sohn, ISIJ Int., 2014, 54(6), 1235–1243,  DOI:10.2355/isijinternational.54.1235.
84. W. Fu, Z. Hu, Y. Zheng, P. Su, Q. Zhang, Y. Jiao and M. Zhou, Chem. Eng. J., 2022, 433, 133680,  DOI:10.1016/j.cej.2021.133680.
85. J.-Y. Fang, Q.-Z. Zheng, Y.-Y. Lou, K.-M. Zhao, S.-N. Hu, G. Li, O. Akdim, X.-Y. Huang and S.-G. Sun, Nat. Commun., 2022, 13(1), 7899,  DOI:10.1038/s41467-022-35533-6.
86. K. Chu, Y. Luo, P. Shen, X. Li, Q. Li and Y. Guo, Adv. Energy Mater., 2022, 12(3), 2103022,  DOI:10.1002/aenm.202103022.
87. A. I. Barbosa, S. A. Costa Lima and S. Reis, Molecules, 2019, 24(2), 346,  DOI:10.3390/molecules24020346.
88. A. Al-Abduly and P. Christensen, Plasma Sources Sci. Technol., 2015, 24(6), 065006,  DOI:10.1088/0963-0252/24/6/065006.
89. S. Han, H. Li, T. Li, F. Chen, R. Yang, Y. Yu and B. Zhang, Nat. Catal., 2023, 6(5), 402–414,  DOI:10.1038/s41929-023-00951-2.
90. N. Alenazi, M. Hussein, K. Alamry and A. Asiri, C, 2018, 4(2), 30,  DOI:10.3390/c4020030.
91. A. I. Karelin, R. R. Kayumov and Y. A. Dobrovolsky, Spectrochim. Acta, Part A, 2019, 215, 381–388,  DOI:10.1016/j.saa.2019.03.007.
92. S. K. Verma and M. K. Deb, Anal. Chim. Acta, 2007, 582(2), 382–389,  DOI:10.1016/j.aca.2006.09.020.
93. T. Søndergaard, L. N. Cleemann, L. Zhong, H. Becker, T. Steenberg, H. A. Hjuler, L. Seerup, Q. Li and J. O. Jensen, Electrocatalysis, 2018, 9(3), 302–313,  DOI:10.1007/s12678-017-0427-1.
94. S. Katipamula, N. M. White and K. M. Waldie, Chem Catal., 2023, 3(3), 100561,  DOI:10.1016/j.checat.2023.100561.
95. X. Zhang, X. Liu, Z.-F. Huang, L. Guo, L. Gan, S. Zhang, M. Ajmal, L. Pan, C. Shi, X. Zhang, G. Yang and J.-J. Zou, ACS Catal., 2023, 13(22), 14670–14679,  DOI:10.1021/acscatal.3c04541.
96. E. Murphy, B. Sun, M. Rüscher, Y. Liu, W. Zang, S. Guo, Y. H. Chen, U. Hejral, Y. Huang, A. Ly, I. V. Zenyuk, X. Pan, J. Timoshenko, B. R. Cuenya, E. D. Spoerke and P. Atanassov, Adv. Mater., 2024, 36(27), 2401133,  DOI:10.1002/adma.202401133.
97. Y. Wang, S. Xia, R. Cai, J. Zhang, J. Wang, C. Yu, J. Cui, Y. Zhang, J. Wu, S. Yang, H. H. Tan and Y. Wu, Small, 2023, 19(26), 2302295,  DOI:10.1002/smll.202302295.
98. S. R. Choi, M. Lim, D. Y. Kim, W. Y. An, S. W. Lee, S. Choi, S. J. Bae, S. D. Yim and J. Y. Park, Int. J. Hydrogen Energy, 2022, 47(39), 17379–17392,  DOI:10.1016/j.ijhydene.2022.03.222.
99. W. Luo, Z. Guo, L. Ye, S. Wu, Y. Jiang, P. Xu, H. Wang, J. Qian, X. Zhou, H. Tang, Y. Ge, J. Guan, Z. Yang and H. Nie, Small, 2024, 20(30), 2311336,  DOI:10.1002/smll.202311336.
100. Y. Li, C. Wang, L. Yang, W. Ge, J. Shen, Y. Zhu and C. Li, Adv. Energy Mater., 2024, 14(7), 2303863,  DOI:10.1002/aenm.202303863.
101. B. M. Penetrante, M. C. Hsiao, B. T. Merritt and G. E. Vogtlin, Fundamental limits on NOx reduction by plasma, 1997, vol. 106,  DOI:10.4271/971715.
102. C. R. McLarnon and B. M. Penetrante, Effect of reactor design on the plasma treatment of NOx, 1998,  DOI:10.4271/982434.
103. Z. Ding, K. Wang, Z. Mai, G. He, Z. Liu and Z. Tang, Int. J. Hydrogen Energy, 2019, 44(45), 24680–24689,  DOI:10.1016/j.ijhydene.2019.07.244.
104. M. Wang, S. Ji, H. Wang, X. Wang, V. Linkov, X. Ma and R. Wang, J. Alloys Compd., 2023, 948, 169713,  DOI:10.1016/j.jallcom.2023.169713.
105. H. Wang, M. Wang and J. Wang, Curr. Chin. Sci., 2023, 3(3), 204–212,  DOI:10.2174/2210298103666221102092735.
106. F. Lei, K. Li, M. Yang, J. Yu, M. Xu, Y. Zhang, J. Xie, P. Hao, G. Cui and B. Tang, Inorg. Chem. Front., 2022, 9(11), 2734–2740,  10.1039/d2qi00489e.
107. X. Fan, D. Zhao, Z. Deng, L. Zhang, J. Li, Z. Li, S. Sun, Y. Luo, D. Zheng, Y. Wang, B. Ying, J. Zhang, A. A. Alshehri, Y. Lin, C. Tang, X. Sun and Y. Zheng, Small, 2023, 19(17), 2208036,  DOI:10.1002/smll.202208036.
108. P. Ramadhany, T. Trần-Phú, J. A. Yuwono, R. K. Hocking, Z. Ma, X. M. C. Ta, P. Kumar, D. Gunawan, B. Johannessen, A. Tricoli, A. N. Simonov, R. Amal and R. Daiyan, Nat. Commun., 2026, 17, 1776 , https://www.nature.com/articles/s41467-026-68481-6.
109. Z. Ma, J. Leverett, J. A. Yuwono, J. Pan, S. Zhou, D. Zhang, M. Zhang, B. Xie, L. Peng, M. H. A. Khan, S. L. Londono, P. V. Kumar, E. Lovell, R. Daiyan and R. Amal, Chem. Eng. J., 2025, 524, 168996,  DOI:10.1016/j.cej.2025.168996.
110. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50(24), 17953–17979,  DOI:10.1103/physrevb.50.17953.
111. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59(3), 1758–1775,  DOI:10.1103/physrevb.59.1758.
112. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54(16), 11169–11186,  DOI:10.1103/physrevb.54.11169.
113. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6(1), 15–50,  DOI:10.1016/0927-0256(96)00008-0.
114. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77(18), 3865–3868,  DOI:10.1103/physrevlett.77.3865.
115. L. J. Cardenas-Flechas, J. J. Barba-Ortega and M. R. Joya, Ceramica, 2022, 68(385), 52–59,  DOI:10.1590/0366-69132022683853152.

---