Biomimetic photocatalysts for the conversion of aqueous- and gas-phase nitrogen species to molecular nitrogen via denitrification and ammonia oxidation

Abstract

Denitrification and anaerobic ammonium oxidation (anammox) are important biological processes of the nitrogen cycle that help to preserve the global ecosystem. However, indiscriminate development and global population growth result in the discharge of large amounts of nitrogen species (e.g., via the Haber–Bosch process), particularly nitrogen oxides and ammonia, which cannot be fully digested by microorganisms and therefore accumulate in soil and water. Photocatalysts can promote the conversion of nitrogen oxides and ammonia to molecular nitrogen under the action of photogenerated electrons and holes, thus mimicking denitrifying and anammox bacteria, respectively. Herein, we review the biomimetic photocatalysts and photoelectrochemical cells used to convert aqueous and airborne nitrogen species to molecular nitrogen and shed light on the charge transfer mechanism that should be selectively controlled to favor the formation of molecular nitrogen over that of nitrogen-containing intermediates and by-products. Last but not least, we discuss the outlooks and perspectives of solar-powered molecular nitrogen recovery and suggest guidelines for the design of high-performance denitrification/anammox bacteria-like photocatalysts.

---

Cheolwoo Park

Cheolwoo Park received a BS in Biochemical Engineering from Gangneung-Wonju National University in 2015 and is now pursuing a PhD in Energy Science under the supervision of Prof. Tae Kyu Ahn at Sungkyunkwan University (SKKU). He joined Sookmyung Women's University (advisor: Prof. Wooyul Kim) as a visiting researcher in 2016 and focuses on elucidating the reaction mechanisms of surface-modified metal oxide photocatalysts for energy conversion and environmental applications.

Hyelim Kwak

Hyelim Kwak received a BS in Chemical and Biological Engineering from Sookmyung Women's University (Seoul, Korea) in 2020 and is currently pursuing a master's degree under the supervision of Prof. Wooyul Kim at Sookmyung Women's University, focusing on photocatalysis for environmental and energy applications.

Gun-hee Moon

Gun-hee Moon received a BS in Chemical Engineering from Inha University (Incheon, Korea) in 2008, an MS degree in Environmental Engineering from POSTECH in 2011, and a PhD degree in Chemical Engineering from POSTECH (Pohang, Korea) in 2015. After a postdoctoral stay at the Max-Planck-Institut für Kohlenforschung (Germany, 2016–2020), he moved to the Korea Institute of Science and Technology (KIST), where he is currently employed as a senior scientist working on the design of photocatalysts, electrocatalysts, photoelectrochemical cells, and carbon materials for energy conversion and environmental applications.

Wooyul Kim

Wooyul Kim received a PhD in Environmental Engineering (advisor: Prof. Wonyong Choi) from POSTECH (Pohang, Korea) in 2012 and worked at the Lawrence Berkeley National Laboratory (USA, 2012–2016) as a postdoctoral fellow (advisor: Dr Heinz Frei). During his PhD period, he joined Osaka University (advisor: Prof. Tetsuro Majima) as a visiting researcher in 2009, 2010, and 2012. He joined the faculty of the Department of Chemical and Biological Engineering at Sookmyung Women's University (Seoul, Korea) as an assistant professor (2016) and was promoted to associate professor in 2021. His current research focuses on revealing the structural identity of key intermediates and their roles in the catalytic cycle (i.e., their kinetic relevancy) to overcome the kinetic barriers for photo (or electro) catalysis. At an early stage of his independent career within the fields of photo(electro) catalysis for energy and environmental applications, he was honored as a winner of the 2020 Energy & Environmental Science lectureship award (Royal Society of Chemistry).

---

1. Introduction

Unlike ammonia (NH₃), nitrite (NO₂⁻), and nitrate (NO₃⁻), molecular nitrogen (N₂), which constitutes ∼80% of the atmosphere by volume, is not chemically and biologically reactive and cannot therefore be utilized by plants for the synthesis of proteins to be supplied to animals and humans.^1,2 The conversion of N₂ into more reactive forms mainly occurs through microbially facilitated processes such as nitrogen fixation and nitrification. The fixed reactive forms are absorbed, used, released in the form of excrements and dead bodies, converted back to the biologically inert form (i.e., N₂) through denitrification and anammox, and eventually returned to the atmosphere.¹ This biological nitrogen cycle preserved the balance between nitrogen fixation and denitrification for thousands of years.² In the 20^th century, however, human activities began to strongly impact the modern nitrogen cycle (established some 2.5 billion years ago), which is likely to result in the establishment of a new steady state in several decades.

Over the past century, the development of industrial processes and the rapid increase in fossil fuel use to satisfy the growing global demand for food and energy have drastically disrupted the nitrogen cycle.^3–9 In particular, the Haber–Bosch process offers a way to synthetically fix nitrogen in the form of ammonia for the mass production of synthetic fertilizers,³ thus enabling abundant food production along with rapid world population growth. In the last 50 years, the consumption of fertilizers and fossil fuels has increased more than six-fold^4,5 and three-fold,⁶ respectively. The projected global population growth is expected to result in elevated fertilizer production and fossil fuel consumption. As both of these anthropogenic sources (i.e., the Haber–Bosch process and fossil fuel combustion) account for ∼45% of the annual fixed nitrogen production (Fig. 1a), the above increase will initiate a cascade of large-scale environmental impacts such as (i) the extensive eutrophication of terrestrial and aquatic systems, (ii) the increase in potent greenhouse gas (i.e., N₂O) inventory, and (iii) global acidification. The low (typically <40%) utilization efficiency of the nitrogen contained in fertilizers results in the nitrification-based conversion of large amounts of fertilizer (∼90% of NH₄⁺) to highly mobile NO₃⁻ ions, which can leach into aquatic systems such as rivers, lakes, and aquifers.² Moreover, besides producing N₂ as the main product, anaerobic denitrification also affords N₂O and thus significantly affects atmospheric N₂O levels.⁷ As N₂O reacts with the stratospheric ozone and is a potent greenhouse gas with global warming potential ∼300 times that of CO₂, denitrification contributes to climate change and stratospheric ozone depletion.⁷ In addition, the absorption of nitrogen compounds by agricultural soils results in their acidification and thus inhibits the activity of soil organisms and disturbs the ecosystem. Fertilizer nitrogen is easily converted into gaseous ammonia and therefore returns from the atmosphere to the watershed via precipitation as another reactive nitrogen form.⁸ Nitrogen oxides (NO_x) produced by fossil fuel combustion not only react with ammonia to produce fine dust and ozone and thus contribute to poor air quality, but also cause acid rain and, hence, soil and ocean acidification. The nitrogen cycle, the carbon cycle, and climate are known to exhibit numerous strong mutual interactions.⁹ The dramatic increase in atmospheric CO₂ levels (>30% above pre-industrial values) due to fossil fuel combustion and land use change is viewed as the primary cause of climate warming observed over the past century. The human activity-induced perturbations of global nitrogen and carbon cycles are in part related to each other, as exemplified by the possible interacting drivers of these cycles during the 21^st century (Fig. 1b).

Fig. 1 (a) Rates of nitrogen flux in the modern nitrogen cycle depend on the efficiency of transformations between reservoirs. Reprinted with permission from ref. 2. Copyright© 2010, American Association for the Advancement of Science. (b) The main anthropogenic drivers of nitrogen–carbon–climate interactions in the 21^st century. Reprinted with permission from ref. 9. Copyright© 2008, Nature Publishing Group.

From the perspective of nitrogen cycle management, the main objectives requiring special consideration are (i) the substantial decrease in nitrogen use,⁹ (ii) the direct up-cycling of used nitrogen to microbial protein,¹⁰ and (iii) the development of artificial denitrification processes powered by renewable energy. Among the various methods of decreasing nitrogen use, one can mention systematic crop rotation,¹⁰ optimization of fertilizer introduction timing and amount,¹¹ and the breeding/development of genetically engineered crops for increasing nitrogen use efficiency.¹² In view of the low efficiency of nitrogen utilization (e.g., agricultural nitrogen utilization efficiency = 40%, feed conversion efficiency = 15%, manure utilization efficiency = 50%),¹⁰ the direct up-cycling of used nitrogen to microbial protein has been proposed as an alternative to the formation of plant and meat proteins. As a renewable energy-powered direct denitrification process, the photocatalytic reduction of reactive nitrogen compounds to N₂ holds great promise since the advantages of photocatalysis compared with conventional catalysis, thermocatalysis, and electrocatalysis are that (i) it does not require energy-intensive processes (solar energy vs. heat or electricity), (ii) the operation is possible without the need for oxidants, reductants, or electrolytes, (iii) it is flexible for application in both aqueous and gas-phase reactions, and (iv) material cost is relatively cheap.^13–17 Herein, we introduce and discuss the most recent findings and advances in photocatalytic denitrification and ammonia oxidation processes, the ultimate goal of which is the conversion of reactive nitrogen compounds (e.g., ionic or gas-phase nitrogen oxides, ammonia) into inert N₂ on a scale comparable to that of anthropogenic nitrogen fixation. Most parts of this review deal with denitrification/anammox bacteria-like photocatalysts and the related mechanisms, which greatly affect activity and selectivity.

2. Photocatalysts and co-catalysts for denitrification

2.1. Reduction of ionic nitrogen oxides to N₂

The nitrogen cycle imbalance due to human activities (e.g., combustion, intensive fertilizer use, and agriculture) has resulted in increased nitrate levels in groundwater and other waters. Ionic nitrogen oxides including nitrate, nitrite, and NO⁻ are the prevalent contaminants in groundwater, causing eutrophication, water waste, and potentially health-threatening consequences such as cancer, birth defects, and cyanosis. Therefore, the release of ionic nitrogen oxides is strictly regulated, and their removal to secure water resources is a great challenge. Traditionally, the removal of NO₃⁻ from wastewater is achieved using reverse osmosis, ion exchange, and biological/catalytic treatment.^1,18,19

Among the various denitrification methods used to reduce ionic nitrogen oxides in aquatic environments, eco-friendly photocatalytic denitrification is the one most promising from the perspective of industrialization.^20–22 This photocatalytic reduction affords inert N₂ gas and mainly involves the reduction of NO₃⁻ to N₂via NO₂⁻.^20,23–27 To increase the overall efficiency and selectivity for N₂ formation, one needs to fully understand the underlying mechanisms, including those of the undesired nitrification (conversion of NO₂⁻ to NO₃⁻)^28,29 and the dissimilatory nitrate reduction to ammonium (DNRA).^30,31 In particular, these undesired reactions need to be precisely controlled (i.e., inhibited) to maximize N₂ formation selectivity. However, from the perspective of reactive nitrogen up-cycling, the highly selective production of ammonium (e.g., DNRA) could be useful.^32,33 In this section, we critically investigate the efforts made to enhance the selectivity for N₂ production in the photocatalytic denitrification of ionic nitrogen oxides and the efficiency of this process. In particular, we demonstrate the important roles of intrinsic photocatalyst properties, sacrificial agents, and specific reaction conditions.

2.1.1. Photocatalytic materials for ionic nitrogen oxide reduction to N₂. The discovery of the photocatalytic reduction of NO₃⁻ to NH₄⁺ in aqueous solutions (over Pt–TiO₂) in 1987 ³⁴ triggered the search for other denitrification photocatalysts. In nature, the corresponding reaction involves multiple-electron transfer and is primarily performed by bacteria such as Thiobacillus denitrificans (2NO₃⁻ + 10e⁻ + 12H⁺ → N₂ + 6H₂O), whereas photocatalytic denitrification predominantly occurs in a stepwise manner and involves the reduction of NO₃⁻ to NO₂⁻ and the subsequent reduction of NO₂⁻ to N₂ or NH₄⁺.^20,35–38 Each step of the nitrate-to-N₂ reduction has its own rate constant, which is largely determined by the interaction between the photocatalyst and the adsorbed reactants (e.g., NO₃⁻, NO₂⁻, NO, and N₂O) and reflects the ease of reactant adsorption on or product desorption from the photocatalyst.³⁹ Attempts to accelerate photocatalytic denitrification have resulted in the discovery and evaluation of numerous photocatalysts, among which TiO₂ is the one most frequently and thoroughly studied because of its durability, non-toxicity, long diffusion length of charge carriers, and other advantages. In particular, various methods of photocatalyst surface modification have been developed to accelerate photoconversion or alter the reaction mechanism and thus control selectivity. According to surface modifier type, these methods are classified into those relying on metal deposition,^40,41 inorganic adsorbates,^42,43 polymer coatings, dye sensitization,^44–47 impurity doping, charge transfer complexation,^48,49etc. Photocatalyst modification not only increases photoconversion efficiency but also affects selectivity by influencing the mechanism and kinetics of photocatalytic denitrification. Recent studies on photocatalysis by surface-modified semiconductors are summarized below.
Photocatalysts based on pristine TiO₂ and related (bi)metallic composites. Compared to TiO₂ powder (P25), TiO₂ nanotubes (TNTs) achieved a ∼50% higher NO₃⁻ conversion efficiency and a slightly elevated selectivity for N₂ formation, which was ascribed to their high specific surface area and abundant active sites.²¹ Given that bare TiO₂ exhibits poor selectivity for N₂ formation, TiO₂–metal composites (M/TiO₂) have been used for NO₃⁻ reduction in aquatic environments. The deposited metal changes the reactant adsorption properties and the charge carrier dynamics under illumination by modifying surface properties such as the space-charge region and charge density. In particular, metals deposited on TiO₂ promote its use as an electron sink (with the formation of a Schottky barrier potential) by decreasing the work function and, hence, increasing the electron affinity.²⁰ Sá et al. reported that the deposition of Cu, Fe, and Ag on TiO₂ increased its activity and selectivity for N₂, suggesting that the conversion efficiency and selectivity strongly depend on factors such as reaction temperature, hole scavenger presence/type, metal deposition method, and TiO₂ particle size. Among the various M/TiO₂ composites, Ag/TiO₂ showed the highest activity to meet the EU-stipulated levels for drinking water (Table 1).⁵⁰ The surface plasmon resonance effect of Ag⁰ and Au⁰ on TiO₂ extended the excitation range of this oxide semiconductor from UV to visible and promoted the separation of photogenerated charge carriers (Fig. 2a).⁵¹ Compared to other M/TiO₂ composites, Ag/TiO₂ exhibited high selectivity for the reduction of NO₃⁻ to N₂,^21,40,41 while Au/TiO₂ accelerated the formation of NH₄⁺.⁵¹ The non-noble metals/TiO₂ (Cu, Ni, Fe, Bi, Zn, etc.) composites also exhibited enhanced catalytic performance.^33,52–54 For instance, Fe/TiO₂ and Zn/TiO₂ exhibited better N₂ yield and selectivity than bare TiO₂.^54–56 Regarding Cu/TiO₂ and Ni/TiO₂, the formation of NO₂⁻ and NH₄⁺ was predominant, where the nitrate reduction was facilitated by the electrons accumulated on Cu or Ni but the interaction with intermediates inhibited the generation of N₂.^33,50,54 On the other hand, Cr/TiO₂ and Co/TiO₂ lowered the conversion of NO₃⁻, which was explained by the light shield effect at a specific wavelength and the fast charge recombination kinetics as well.^33,54 However, conventional metal-modified photocatalysts usually suffer from metal leaching, aggregation, and gradual deactivation and need to be substantially optimized in terms of N₂ formation selectivity.

Table 1 Materials used for the photocatalytic reduction of ionic nitrogen oxides

No.	Photocatalyst	Co-catalyst	Light	Initial conc.	Catalyst loading (g L^−1)	Sacrificial reagent	NO3^− conversion (%)	N2 selectivity (%)	By-products	Ref.
1	TiO2	—	Medium-pressure Hg lamp, 150 W	1 mM	2.5	Oxalic acid	15	—	NH4^+	82
2	TiO2	—	Medium-pressure Hg lamp, 400 W	10 mM	10	Oxalic acid	9.8	56.5	NO2^−, NH4^+	33
3	TiO2	—	Medium-pressure Hg lamp, 150 W	0.8 mM	0.45	Oxalic acid	90.1	55.4	NO2^−, NH4^+	68
4	TiO2	—	High-pressure Hg lamp, 100 W	0.8 mM	0.38	Formic acid	48.5	38.1	NO2^−, NH4^+	80
5						KI	25.5	18
6	TiO2	—	High-pressure Hg lamp, 300 W	1.6 mM	1	Formic acid	26.8	72.4	NO2^−, NH4^+	81
7	TiO2	—	High-pressure Hg lamp, 150 W	1.6 mM	1	Formic acid	35.8	87.7	NO2^−, NH4^+	21
8	TiO2 (TNTs)						53.3	89.5	NO2^−, NH4^+
9	TiO2	Au	400 W lamp	1.6 mM	0.21	Oxalic acid	44	—	NH4^+	83
10	TiO2	Cu	High-pressure Hg lamp, 110 W	100 ppm	0.38	Formic acid	100	63	NH4^+	50
11	TiO2	Ag	High-pressure Hg lamp, 300 W	1.6 mM	1	Formic acid	99.6	88.4	NO2^−, NH4^+	81
12	TiO2	Ag2O					97.5	82.9	NO2^−, NH4^+
13	TiO2 (TNTs)	AgCl	High-pressure Hg lamp, 150 W	1.6 mM	1	Formic acid	94.5	92.9	NO2^−, NH4^+	21
14	TiO2	Pt–Cu	High-pressure Hg lamp, 250 W	60 mg L^−1	1	Benzene	59	∼89	NO2^−, NH4^+	64
15	TiO2	Pd–Cu	High-pressure Hg lamp, 400 W	0.05 mM	100	Formic acid	56	98	NO2^−	65
16	TiO2	Pd–Cu	Medium-pressure Hg lamp, 150 W	1.6 mM	0.52	Formic acid 0.08 M	84	83	NO2^−, NH4^+	66
17	LiNbO3	—	High-pressure Hg lamp, 100 W	0.8 mM	0.38	Formic acid	98.4	95.8	NO2^−, NH4^+	80
18		—	High-pressure Hg lamp, 100 W	0.8 mM	0.38	KI	96.2	93	NO2^−, NH4^+
19	LiNbO3	—	UV lamp	10 mg L^−1	Membrane	Formic acid	81.82	98.04	NO2^−, NH4^+	22
20	LiNbO3	—	High-pressure Hg lamp, 100 W	0.8 mM	0.4	Formic acid	60.5	57.21	NO2^−, NH4^+	76
21		Fe					86.69	85.71	NO2^−, NH4^+
22	CuInS2	0.75 wt% Pt–0.75 wt% Ru	Hg lamp, 125 W	7.2 mg L^−1	0.5	Sodium oxalate	100	80.2	NO2^−	77
23			Xe lamp, 300 W (400 nm cut-off)				100	56.1	NO2^−
24	FeTiO3		Medium-pressure Hg lamp, 150 W	0.8 mM	0.45	Oxalic acid	100	93	NO2^−	68
25	GdCrO3	1 wt% Pd	High-pressure Hg lamp, 500 W	0.8 mM	0.5	Formic acid 0.4 mM	98.7	100	—	67
26		1 wt% Ag					85.1	83.2	NO2^−, NH4^+
27		1 wt% Cu					81.9	78.8	NO2^−, NH4^+
28		—					79.3	81.4	NO2^−, NH4^+
29	CuFe0.7Cr0.3S2	0.75 wt% Pd	Hg lamp, 500 W	1.6 mM	1	Sodium oxalate	100	59	NO2^−	78
		3 wt% Au
30	KTaO3	1 wt% Ni	High-pressure Hg lamp, 450 W	10 mM	2.5	—	97	44	H2, NO2^−, NH4^+	69

---

Fig. 2 Photocatalytic nitrate reduction promoted by (a) Au/TiO₂ and (b) Pd–Cu/TiO₂. Reprinted with permission from ref. 51 and 66. Copyright© 2011 and 2014, Elsevier. (c) Comparison of nitrate reduction promoted by LiNbO₃ and P25 TiO₂. Reprinted with permission from ref. 80. Copyright© 2016, American Chemical Society.

In bimetallic composites, the metals act as promoters and selectors. The promoter metal (e.g., Cu, Sn, In)^57–59 initiates the rate-limiting step of the NO₃⁻ to NO₂⁻ conversion, while the selector metal (e.g., Pd, Pt, Rh)^60,61 further reduces NO₂⁻ to NH₄⁺ and/or N₂. Among the available metal combinations, Pd–Cu is widely accepted as the most active and selective one for electrocatalytic NO₃⁻ reduction, the mechanism of which has been revealed by conventional electrochemical analysis and density functional theory (DFT) calculations.^45,46 The increased H₂ amount resulting from the elevated Pd loading and H₂ flow rate promoted the reduction of Cu^II to Cu⁰ and thus facilitated NO₃⁻ removal, while the high N : H ratio on the active Pd sites increased the selectivity for N₂.⁶² Bimetallic electrocatalysts have been widely deposited on photocatalysts for photocatalytic applications.⁶³ Precious metal (e.g., Pt, Pd)–Cu combinations are among those offering the highest activity and selectivity for catalytic NO₃⁻ reduction (Table 1). Notably, NO₃⁻ was mainly converted to ammonia (over Pt/TiO₂) or NO₂⁻ (over Cu/TiO₂), whereas Pt–Cu/TiO₂ catalysts exhibited a considerable selectivity for N₂ formation in photocatalytic NO₃⁻ reduction.⁶⁴ The fact that N₂ formation was observed for Pd/TiO₂ and Pd–Cu/TiO₂ systems but was negligible for the Cu–TiO₂ system means that Pd is indispensable for the photocatalytic reduction of NO₂⁻ to N₂.^65,66 Likewise, in bimetallic composites, electrons transferred from TiO₂ to promoter metal sites reduce NO₃⁻ to NO₂⁻, with the subsequent reduction of NO₂⁻ to N₂ occurring at selector metal sites. The adsorption of protons on the selector metal surface significantly affects the overall selectivity for N₂ (Fig. 2b). Hence, the metal deposited on TiO₂ controls the reaction path and, hence, the conversion efficiency and selectivity for N₂ or NH₄⁺, which implies that the optimization of the promoter-to-selector metal ratio is crucial for realizing selective N₂ formation. Finally, the presence of sacrificial electron donors and the occurrence of competitive reactions (e.g., H₂ production) present additional challenges.

New photocatalyst types. Much effort has been directed at the optimization of perovskite-based photocatalysts, as the unique properties of perovskites (e.g., chemical and optical stability, tunable bandgap and crystal structure, and long charge carrier lifetime) allow one to readily alter the dynamics of photogenerated charge carriers and the overall photocatalysis mechanism. Pd/GdCrO₃ exhibited faster nitrate reduction and higher selectivity for N₂ due to the negative conduction band energy level and the co-catalyst effect of Pd,⁶⁷ while FeTiO₃ was characterized by negligible NH₄⁺ formation (i.e., it exhibited a remarkably high selectivity for N₂) without the need for complex and expensive catalysts.⁶⁸ KTaO₃ effectively promoted the photocatalytic reduction of NO₃⁻ to NO₂⁻, N₂, and NH₃ under UV light irradiation even in the absence of co-catalysts or reducing reagents such as organic compounds.⁶⁹

Layered double hydroxides (LDHs) with hydrotalcite-like structures are some of the interesting materials due to their unique properties such as anions intercalated in 2D interlayer spaces, a bunch of surface hydroxyl groups, flexibility to change elements, and swelling nature, where divalent (e.g., Mg, Co, Ni, Cu, and Zn) and trivalent (e.g., Al, Cr, Ni, and Ga) metal cations are combined.^70,71 In particular, a high specific surface area, excellent electrical conductivity, high mobility of charge carriers, and high chemical stability make it possible to apply them in various photocatalytic reactions.^72–74 Therefore, it was reported that the MgAl-LDH used for NO₃⁻ reduction enhanced the selectivity to N₂ without any sacrificial reagent, which was ascribed to both attraction of NO₃⁻ ions near the photocatalyst surface and restriction of charge carrier recombination.⁷⁵

LiNbO₃ is a nonlinear optical material with high potential for NO₃⁻ removal,^22,60,76 offering spontaneous polarization screening by either free electrons and holes or ions/molecules adsorbed on the surface. The second harmonic generation effects of nonlinear optical materials facilitate the generation of electrons and inhibit the recombination of charge carriers to enhance the efficiency and stability of NO₃⁻ reduction. The superior (compared to that of bare TiO₂) activity of LiNbO₃ was attributed to the photocatalytic reduction of nitrate through direct heterogeneous interactions with electrons at the conduction band of this material, whereas in the conventional photocatalysis mechanism, nitrate is mainly reduced by CO₂˙⁻ generated from holes at the valence band (Fig. 2c). In an effort to develop a systematic and durable industrial-scale process, LiNbO₃ was applied to a membrane platform,²² which offered the inherent benefits of high separation performance and antifouling properties compared to common ultrafiltration membranes. In addition, LiNbO₃ has been successfully applied to membrane materials without significant photocatalytic activity inhibition (Table 1). Fe–LiNbO₃ exhibited an enhanced selectivity for N₂ formation as well as a high NO₃⁻ conversion efficiency,⁷⁶ which was ascribed to the increase in the specific surface area and the number of Lewis-acidic sites upon doping.

As wide-bandgap semiconductors (e.g., TiO₂, FeTiO₃, GdCrO₃, and KTaO₃) are intrinsic UV-light-driven photocatalysts, a more effective strategy would be to develop narrow-bandgap photocatalysts and thus utilize the whole solar spectrum. As a result, various chalcogenide materials (e.g., CuInS₂ ⁷⁷ and CuFe_0.7Cr_0.3S₂ ⁷⁸) have been developed. In particular, CuInS₂ has a narrow bandgap of 1.45 eV and an insufficient conduction band potential for H₂ production, thus preventing the over-reduction of nitrate to ammonia.

2.1.2. Insights into the mechanism of ionic nitrogen oxide reduction. The efficiency and selectivity of artificial solar denitrification systems can be increased by suppressing undesired reactions, which mainly correspond to DNRA (i.e., ammonification) and the re-oxidation of NO₂⁻ to NO₃⁻. In turn, a deep understanding of the overall mechanism is required to precisely control competitive reactions and thus selectively convert ionic nitrogen species to N₂ while maintaining sufficient catalytic rates for keeping up with the photon flux at maximum solar intensity. DNRA is widely known as the anaerobic microbial pathway of the natural nitrogen cycle. The important implication of DNRA in denitrification is the production of NH₄⁺ from NO₃⁻, which is the major side reaction for the reduction of NO₃⁻ to N₂. The NO₂⁻ ion produced by the reduction of NO₃⁻ (NO₃⁻+ 2e⁻ + 2H⁺ → NO₂⁻ + H₂O) can undergo ammonification (NO₂⁻ + 6e⁻ + 8H⁺ → NH₄⁺ + 2H₂O) or denitrification (NO₂⁻ + 6e⁻ + 8H⁺ → N₂ + 4H₂O), both of which are six-electron reductions. The preference for a particular NO₂⁻ reduction pathway is strongly affected by the environment, particularly by the ratio of N species to H atoms (N : H ratio) at active sites. The injection of excess H₂ increases the surface concentration of H⁺ at active sites because of the low-energy-barrier dissociative adsorption of H₂ and thus increases the selectivity for NH₄⁺ formation (Fig. 3a). Alternatively, the N : H ratio can be changed by controlling the steady-state concentration of NO₂⁻ to increase the probability of binding of two nitrogen species to form N₂ (Fig. 3b). The fine-tuning of the N : H ratio at surface active sites is required to suppress DNRA and thus selectively convert NO₂⁻ to N₂.

Fig. 3 Dependence of product (N₂ and NH₃) selectivity during NO₂⁻ reduction over Pd/TiO₂ on (a) H₂ flow rate and (b) NO₂⁻ concentration. Reprinted with permission from ref. 60. Copyright© 2014, American Chemical Society.

In the case of efficient nitrate conversion, photocatalytic nitrite oxidation, which is hard to detect during NO₃⁻ reduction, should be considered for low-efficiency nitrate reduction. Even if nitrification and denitrification occur simultaneously, it is difficult to identify the main factors of nitrification because of the same initial reactant and product. The formation of NO₃⁻ indicates that the oxidation of NO₂⁻ by holes occurs even in the presence of a hole scavenger in aqueous photocatalyst suspensions (Fig. 4). The low NO₃⁻ conversion efficiency is due to the low rate constant of NO₃⁻ reduction and the high rate constant of NO₂⁻ oxidation.

Fig. 4 (a) Schematic mechanism of Pd/TiO₂ operation and the combination of two photocatalytic systems for the reduction of NO₂⁻ to N₂ and NO₃⁻. (b) Time-dependent conversion of NO₂⁻ and the formation of N₂, H₂, and NO₃⁻ in suspensions of Pd–TiO₂ in aqueous sodium oxalate (diamonds: NO₂⁻, circles: N₂, squares: NO₃⁻, triangles: H₂). Reprinted with permission from ref. 29. Copyright© 2012, Royal Society of Chemistry.

The minimal loss of photogenerated electron–hole pairs offers flexibility for maximizing photocatalytic efficiency by adding sacrificial hole or electron scavengers to restrict charge carrier recombination. Sacrificial electron donors (i.e., hole scavengers) were used for photocatalytic denitrification in aqueous media to enable the efficient reduction of ionic nitrogen species⁵ and were shown to affect reactant–photocatalyst interactions. Sacrificial reagents act not only as efficient hole scavengers but also as precursors of active radicals for ionic nitrogen oxide reduction. Sacrificial reagents were demonstrated to promote the efficient removal of holes and thus reduce charge carrier recombination while being oxidized³⁴ to afford strongly reducing (−1.81 vs. SHE) carboxyl radicals (CO₂˙⁻), which also resulted in activity enhancement.⁷⁹ The most common sacrificial reagents are organic compounds such as formic acid,^21,80,81 oxalic acid,^68,82,83 humic acid,⁸⁴ and methanol.³⁴ Among them, formic acid is the best hole scavenger for NO₃⁻ reduction, as its simple structure results in the exclusive formation of the strongly reducing CO₂˙⁻, while the release of protons promotes efficient N₂-selective reduction. Oxalic acid is the second most used hole scavenger, featuring a higher selectivity for NH₄⁺ formation than formic acid (Table 1). The dependence of selectivity on hole scavenger type is attributed to the reduction ability of the reactant and intermediates. Simple carboxyl compounds (formic acid, sodium formate, etc.) are oxidized to afford abundant CO₂˙⁻ radicals and therefore allow for more efficient conversion than other organic hole scavengers. On the other hand, oxalic acid remarkably enhances the conversion of nitrite to N₂ while exhibiting a modest hole scavenging ability (Fig. 5a and b).^21,85 The understanding of the complicated interactions between various intermediates and sacrificial reagents is particularly challenging in the case of photocatalytic denitrification. In addition, the presence of additional reagents such as SO₄²⁻, H₂PO₄⁻, F⁻, Cl⁻, and HCO₃⁻ increases system complexity and therefore leads to hole blocking and, as a consequence, inhibits hole scavenger oxidation by promoting surface anionization (Fig. 5c).⁸⁶

Fig. 5 (a) Effects of organic hole scavengers on the reduction of NO₃⁻ over AgCl/TiO₂ nanotubes. Reprinted with permission from ref. 21. Copyright© 2018, Royal Society of Chemistry. (b) Effects of organic hole scavengers on the reduction of NO₂⁻ over Ag/TiO₂. Reprinted with permission from ref. 85. Copyright© 2007, American Chemical Society. (c) Photocatalytic nitrate conversion efficiency and N₂ selectivity achieved over TiO₂ in the presence of different levels of SO₄²⁻, HCO₃⁻, H₂PO₄⁻, F⁻ and Cl⁻ (irradiation duration: 3 h, sacrificial agent: formic acid). Reprinted with permission from ref. 86. Copyright© 2020, Elsevier.

2.2. Denitrification of airborne NO_x and N₂O to N₂

Airborne nitrogen oxides (NO_x) including NO and NO₂ originate from anthropogenic (combustion of fossil fuels at high temperatures in automobile engines and power plants) and biogenic sources, significantly affecting human health and the environment.⁸⁷ N₂O is produced by human activities and natural processes and is both a greenhouse gas and an ozone-depleting substance.⁸⁸ Taken together, NO_x and N₂O significantly affect the environment via (i) ozone and smog generation due to the reaction of NO_x with volatile organic compounds (VOCs) upon irradiation with light, (ii) the acidification of water vapor (acid rain), (iii) excessive algal growth due to the dissolution of NO_x in water (eutrophication), (iv) climate change and ozone layer depletion, and (v) secondary fine dust formation through the combination of the above oxides with water vapor, O₃, ammonia, etc. Selective catalytic reduction (SCR)-based denitrification is a long-standing industrial process that is used to control air quality and is performed using different types of reactors and catalysts, depending on the application. Regarding the conventional deNO_x process, ammonia has been used as a reducing agent instead of hydrogen to improve process safety and reduce the operating temperature.⁸⁹ The conversion of NO_x to N₂ is described by Eley–Rideal and Langmuir–Hinshelwood models, which rely on the decomposition of intermediates generated through the reaction of surface-adsorbed activated NH₃ with free gaseous NO and NO adsorbed on basic catalyst sites, respectively. Alternatively, airborne NO_x can be removed via complete oxidation to NO₃⁻. Although this approach (nitrification) has received widespread attention for the development of smart cement and asphalt used in the construction of the state-of-the-art urban infrastructure,⁹⁰ the accumulation of NO₃⁻ on the surface of photocatalysts can result in their deactivation. As this review focuses on denitrification only, readers interested in solar-driven nitrification are advised to consult appropriate literature.

The photocatalytic reduction of NO_x and N₂O to N₂ offers the following advantages: (i) the use of water instead of the explosive H₂ and the toxic NH₃, (ii) operation at standard temperature and pressure, (iii) net zero carbon emission for operation under natural sunlight, and (iv) the availability of cheap and environmentally benign materials.^91–94 As electron–hole pairs photogenerated in semiconducting materials can be transformed to the strongly oxidizing reactive oxygen species (ROS), aerobic conditions favor nitrification, in which case O₂ acts as a good electron acceptor and a precursor of mobile hydroxyl radicals .⁹⁵ Under anaerobic conditions, the electrons can be transferred to NO_x or N₂O without interference from O₂ but can still be intercepted by water vapor (2H₂O + 2e⁻ → H₂ + 2OH⁻) (Fig. 6). Ideally, the residual holes in photocatalysts oxidize water vapor (2H₂O + 4h⁺ → O₂ + 4H⁺), otherwise, nitrification is driven by oxidation with holes or OH˙ (H₂O + h⁺ → OH˙ + H⁺). Along with selective charge transfer, other problems such as low solar light absorption, poor catalytic activity, need for noble metal–based co-catalysts, and lack of long-term durability should be overcome for practical applications. Herein, we summarize the strategies (e.g., structure and morphology control, co-catalyst loading, heteroatom doping, and hybridization with different types of materials) used to address the weaknesses of the photocatalytic reduction of NO_x and N₂O to N₂.

Fig. 6 Comparison of photocatalytic nitrification (left panel) and denitrification (right panel) mechanisms. The bottom right panel presents NH₃-mediated denitrification, the release of trapped oxygen atoms from NO (green box), and the oxidation of sacrificial gases (purple box; ED stands for carbon-containing electron donors such as CO, CH₄, C₃H₈, CH₃OH, and C₂H₅OH).

2.2.1. Photocatalysts for the reduction of airborne NO_x to N₂.
DeNO_x without additional electron donors. Water vapor and surface-trapped oxygen atoms formed by electron transfer to NO are ideal electron donors for the photocatalytic denitrification of NO_x; however, the kinetics of water oxidation is sluggish, and the control of selective electron transfer to NO_x is difficult. Therefore, whereas most studies focus on photocatalyst structure/surface modification and the operation of deNO_x in the presence of nitrogen- or carbon-containing electron donors (e.g., NH₃, CO, CH₄, C₃H₈, CH₃OH, and C₂H₅OH), very few processes have been performed without these donors. Wu et al. optimized the selectivity of NO reduction to N₂ by controlling oxygen vacancies in TiO₂ nanoparticles.⁹⁶ These vacancies were introduced by doping the TiO₂ lattice with Fe³⁺, and the substitution of Ti⁴⁺ by Fe³⁺ contributed to the charge compensation between negatively charged dopants and positively charged oxygen vacancies.⁹⁷ In air, the photocatalytic conversion of NO over pristine TiO₂ was not selective for N₂ (Fig. 7a and Table 2), whereas a significantly improved selectivity was observed for Fe-doped TiO₂ (Fig. 7b). The highest yield and selectivity for N₂ were obtained under anaerobic conditions, and the formation of NO₂ over Fe-doped TiO₂ was completely suppressed (Fig. 7c). Thus, only N₂ and O₂ were detected along with unreacted NO in the outlet (Fig. 7d). The proposed mechanism (Table 3) is believed to involve ROS (superoxide anion radicals and hydroxyl radicals) as nitrification agents. On the other hand, oxygen vacancies stabilized by Fe doping enhanced selective charge transfer to NO, which did not involve radical species, providing a clue to the design of highly effective denitrification photocatalysts.

Fig. 7 Photocatalytic conversion of NO under UV light irradiation in air over (a) pristine TiO₂ and (b) Fe-doped TiO₂. (c) Photocatalytic conversion of NO to N₂ over Fe-doped TiO₂ and the schematic mechanism of this conversion. The rapid decrease of NO in (a–c) is due to the formation of NO₃⁻via a reaction with superoxide anion radicals produced from adsorbed oxygen. (d) NO conversion ([NO] = 100 ppm) to N₂ and O₂ over pristine TiO₂ and Fe-doped TiO₂ in He. Reprinted with permission from ref. 96. Copyright© 2012, American Chemical Society.

Table 2 Performances of various materials as photocatalysts for the denitrification of airborne NO_x

No.	Photocatalyst	Target (Conc.)	Light	Temp.	Flow rate & GHSV	Carrier gas	Supplements	NO conversion (%)	N2 selectivity (%)	By-products	Ref.
1	TiO2	NO (1000 ppb)	UV	r.t.	1 L min^−1	Air	None	∼50	—	NO2, NO3^−	96
2	Fe-doped TiO2					Air		∼6	∼50	NO2, NO3^−
3						N2		∼4.5	∼100	NO2, NO3^−
4	g-C3N4	NO (600 ppb)	Visible (420 nm ≤)	r.t.	1 L min^−1	Air	None	∼38	—	NO2, undefined	98
5		NO (1500 ppb)				Ar		Almost negligible	—	—
6	g-C3N4 with carbon vacancies	NO (600 ppb)				Air		∼48	—	NO2, undefined
7		NO (1500 ppb)				Ar		∼34	∼66	NO2
8	Cu^+-ZSM-5	NO (2 Torr)	UV (280 nm <)	r.t.	—	—	None	∼2 (4 h)	∼100	—	99
9	Ti-MCM-41	NO (180 μmol Eg-cat^−1)	UV (240 nm <)	r.t.	—	—	None	∼1.1 (1 h)	∼75	N2O, undefined	103
10	JCR-TiO2 (anatase)	NO (1000 ppm)	UV	r.t.	GHSV 32 000 h^−1	2% O2, 98% Ar	NH3 (1000 ppm)	41	100	—	108
11	JCR-TiO2 (rutile)							53	100	—
12	JCR-TiO2 (anatase 91.3% + rutile 8.7%)							63	100	—
13	V2O5/TiO2 (1 wt%)	NO (1000 ppm)	UV	r.t.	GHSV 50 000 h^−1	2% O2, 98% Ar	NH3 (1000 ppm)	17.7	100	—	114
14	CrO6/TiO2 (1 wt%)							34.2	100	—
15	MnO/TiO2 (1 wt%)							12.1	100	—
16	Fe2O3/TiO2 (1 wt%)							29.6	100	—
17	CoO/TiO2 (1 wt%)							21.6	100	—
18	NiO/TiO2 (1 wt%)							27.0	100	—
19	CuO/TiO2 (1 wt%)							26.1	100	—
20	ZnO/TiO2 (1 wt%)							46.6	100	—
21	Y2O3/TiO2 (1 wt%)							47.0	100	—
22	ZrO2/TiO2 (1 wt%)							41.1	100	—
23	Nb2O5/TiO2 (1 wt%)							58.4	>99	—
24	MoO3/TiO2 (1 wt%)							60.2	>99	—
25	Ta2O3/TiO2 (1 wt%)							38.6	100	—
26	WO3/TiO2 (1 wt%)							63.6	>99	—
27	N3-dye TiO2	NO (1000 ppm)	Visible (420 nm ≤)	r.t.	GHSV 100 000 h^−1	2% O2, 98% He	NH3 (1000 ppm)	>99	>99	—	121
28	LaFe0.4Mn0.6O3/attapulgite	NO (1000 ppm)	UV	r.t.	GHSV 50 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼85	∼100	—	122
29	La0.7Ce0.3FeO3/attapulgite	NO (1000 ppm)	UV	r.t.	0.1 L min^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼80	—	—	123
30	LaFe0.5Ni0.5O3/palygorskite	NO (1000 ppm)	Visible (420 nm ≤)	200 °C	GHSV 50 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼92	∼98	—	124
31	La0.5Pr0.5CoO3/palygorskite	NO (1000 ppm)	Visible (420 nm ≤)	200 °C	GHSV 40 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼95	∼99	—	125
32	N-doped carbon quantum dot-modified PrFeO3/palygorskite	NO (1000 ppm)	Visible	150–200 °C	GHSV 50 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼93	100	—	127
33	Fe2O3/SmFeO3/palygorskite	NO (1000 ppm)	UV	<200 °C	GHSV 40 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼95	100	—	128
34	LaCoO3/appapulgite/rGO	NO (1000 ppm)	UV	100–150 °C	GHSV 50 000 h^−1	3% O2, 97% N2	NH3 (1000 ppm)	∼95	100	—	129
35	Ag/TiO2 (1 wt%)	NO (909 ppm)	UV	r.t.	5.5 cm^3 min^−1	Ar	CO (1818 ppm)	∼35 μmol h g^−1cat^−1	90	—	136
36	Ag/TiO2 (5 wt%)							∼10 μmol h g^−1cat^−1	100	—
37	TiO2	NO (3000 ppm)	UV	150 °C	—	5% O2, 3% H2O in N2	Carbon black	97	99	N2O	150

---

Table 3 Proposed mechanism for the conversion of NO on pristine and Fe-doped TiO₂.⁹⁶

Byproduct (i.e., NO2 and NO3^−) formation over TiO2
[Reductive pathway]	[Oxidative pathway]
Superoxide radical-mediated	Hydroxyl radical-mediated
a Numbers 1–4 denote the reaction pathway numbered in the scheme of Fig. 7c. and denote charged and neutral oxygen vacancies, respectively.
Ti(O2)ads + e^− → Ti(O2^−)ads	Ti–OH^− + h^+ → Ti–OH˙
Ti(O2^−)ads + NO(g) → Ti(NO3^−)ads	Ti–OH˙ + NO(g) → Ti–H + NO2(g)
Ti(O2)ads + Ti–OH^− + hv + 2NO(g) → Ti(NO3^−)ads + Ti–H + NO2(g)

---

Conversion of NO to N2 over Fe-doped TiO2
[Reductive pathway]	[Oxidative pathway]
(1)	(4)
2Osurf–N → 2Osurf + N2(g) (2)
2Osurf → O2(g) (3)
2NO(g) + 4hv → N2(g) + O2(g)

---

Likewise, according to Dong et al.,⁹⁸ the carbon vacancy tailoring of graphitic carbon nitride (g-C₃N₄) nanosheets (C_v-g-C₃N₄) significantly enhanced NO reduction under visible light irradiation (Fig. 8a). This reduction was faster in air than in argon irrespective of structure modification, as the ROS-mediated oxidation of NO to NO₂ is much more favorable under oxic conditions (Fig. 8b). While the photocatalytic removal of NO over g-C₃N₄ was almost prohibited under anaerobic conditions, C_v-g-C₃N₄ was characterized by a relatively high conversion of NO and high selectivity for N₂ formation (Fig. 8c), as in the case of Fe-doped TiO₂. In line with the enhanced light absorption of C_v-g-C₃N₄ and the restricted recombination of charge carriers therein, electron spin resonance (ESR) spectra suggested that surface adsorption sites stimulated the chemisorption of airborne NO (via the interaction between a carbon atom with an unpaired electron in g-C₃N₄ and a nitrogen atom with an unpaired electron in NO), with carbon vacancies acting as active centers to induce interactions with the NO oxygen (i.e., C_v–O–N). Thus, after the adsorption of NO on g-C₃N₄ and C_v-g-C₃N₄, the ESR signal intensity due to carbon atoms with unpaired electrons decreased and increased, respectively (Fig. 8d and e). Illumination of C_v-g-C₃N₄ with pre-adsorbed NO induced a peak shift and the appearance of two new peaks at 3535 and 3555 G, which indicated the change of defect sites and the decomposition of NO into atomic N and O, respectively (Fig. 8f).

Fig. 8 (a) Photocatalytic removal of NO ([NO] = 1500 ppb) in air and argon over g-C₃N₄ and C_v-g-C₃N₄ under UV light irradiation. Production of (b) NO₂ in air and (c) N₂ in argon over g-C₃N₄ and C_v-g-C₃N₄. ESR spectra of (d) g-C₃N₄ and (e) C_v-g-C₃N₄ before and after NO adsorption. (f) ESR spectra of C_v-g-C₃N₄ with adsorbed NO recorded in the dark and under UV light irradiation. Reprinted with permission from ref. 98. Copyright© 2017, Elsevier.

Prior to the strategies described above, Anpo et al. reported the UV light-promoted decomposition of NO into N₂ and O₂ over cation-exchanged ZSM-5 (Cu⁺, Ag⁺, and Pb²⁺),^99–101 vanadium silicate (VS)/ZSM-5,¹⁰² and Ti-MCM-41.¹⁰³ Herein, we do not discuss the characteristics of such materials and the corresponding kinetic analysis in detail because of the multitude of related reviews^104–106 but briefly overview the concept of transition metal ion-mediated electron transfer to NO. Cu⁺ and Ag⁺ immobilized in zeolites can be excited under illumination and transfer an electron to the π-antibonding orbital of NO while concomitantly accepting the electron of another NO molecule. Consequently, two contiguous N⋯O species adsorbed at metal ion sites are converted into N₂ and O₂. Moreover, the coordination and distribution of metal oxide species, e.g., four-fold tetrahedrally coordinated vanadium oxide species with a terminal oxovanadium group (V=O) in VS/ASM-5 and Ti oxide species with tetrahedrally coordinated Ti⁴⁺ in Ti-MCM-41, strongly affected NO removal activity and selectivity. Although several studies demonstrated the selective conversion of NO to N₂ in the absence of supplements, the related yields were quite low, and the formation of undesired products could not be avoided, which was ascribed to catalyst inactivation via product accumulation on active sites. Another way to overcome this issue is the utilization of NH₃ and carbon-containing compounds as sacrificial molecules.

DeNO_xvia NH₃(g) oxidation. The photo-assisted SCR (photo-SCR) of NO(g) with NH₃(g) is an effective way to return N₂ to the atmosphere (4NO + 4NH₃ + O₂ → 4N₂ + 6H₂O). Back in 1992, Cant et al. reported photo-SCR based on the reaction of NO with NH₃ over TiO₂ under UV light irradiation and used isotope labeling to elucidate the reaction pathway (4¹⁴NO + 4¹⁵NH₃ + O₂ → 4¹⁴N¹⁵N + 6H₂O).¹⁰⁷ In this system, NO removal was very slow, and N₂O and NO₃⁻ were still formed; moreover, the experiment was conducted under unrealistic conditions. Over the past 20 years, Tanaka's group systematically investigated the photo-SCR of NO, mainly focusing on the synthesis of TiO₂-based photocatalysts, reaction mechanism verification, and the design of flow-type photoreactors operable at high gas hourly space velocities (GHSVs).^108,109 From a mechanistic point of view, the photo-SCR of NO over TiO₂ photocatalysts is suggested to comprise five steps (Fig. 9a).^110,111 According to this mechanism, NH₃(g) is adsorbed on the Lewis-acidic sites of TiO₂ (step 1) and is then oxidized by a photogenerated hole (step 2), while the photogenerated electron is trapped by Ti⁴⁺. The adsorbed reacts with NO(g) through the Eley–Rideal mechanism (step 3), and the nitrosoamide (NH₂NO) formed as an intermediate decomposes into N₂ and H₂O (step 4). The remaining Ti³⁺ is oxidized by electron transfer to O₂(g) to regenerate Ti⁴⁺ (step 5). As NH₃ adsorption takes place on Lewis-acidic sites, surface acidity control is an effective way of improving photo-SCR performance, whereas non-acidic surface area and the crystal phase do not contribute to activity enhancement.¹¹² The rate-limiting step is affected by the concentration of O₂, corresponding to step 4 in the presence of excess O₂ and step 5 at O₂ contents of <2 vol%. Ji et al.¹¹³ suggested that (i) the adsorption of NH₃ and its direct oxidation by photogenerated holes was preferred to the dissociative adsorption of NH₃ on TiO₂, as proton-coupled hole transfer was energetically favored (NH₃ + h⁺ + O_2f⁻ → + O_2fH⁻; O_2f denotes a two-fold coordinated O atom in TiO₂) and the formation of NH₂NO proceeded via the Eley–Rideal mechanism, and (ii) the decomposition of NH₂NO into N₂ and O₂ was initiated by the transfer of H_b to the O atom, which was followed by either the transfer of H_a to O–H_b or the transfer of O–H_b to the surface Ti atom (Fig. 9b). Step 4 was calculated to have the highest energy barrier among other steps (i.e., it was the rate-limiting step), in line with the experimental result of Tanaka's group.

Fig. 9 (a) Mechanism of the photocatalytic reduction of NO by NH₃ over TiO₂. Reprinted with permission from ref. 111. Copyright© 2004, Elsevier. (b) Potential energy diagram for the decomposition of NH₂NO on the (101) surface of anatase TiO₂. Reprinted with permission from ref. 113. Copyright© 2014, American Chemical Society.

To realize high catalytic performance or to run the system under visible light, one should appropriately design or modify the photocatalysts. Herein, photocatalysts were classified as those based on TiO₂ or other materials. When TiO₂ was modified with transition metal (V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Y, Zr, Nb, Ta, and W) oxides, increased NO conversion was observed only for the more acidic ZnO, Y₂O₃, Nb₂O₅, MoO₃, and WO₃ (Fig. 10a and Table 2).¹¹⁴ The low activity observed for other oxides was ascribed to their non-photocatalytic nature or the instability of active sites. The highest activity of WO₃/TiO₂ was attributed to the facile decomposition of NH₂NO on the weakly Lewis-acidic sites of WO₃.¹¹⁵ The doping of Si into TiO₂ caused the formation of smaller crystals with a higher surface area and pore volume, and acidity was enhanced because of the increased concentration of surface hydroxyl groups.¹¹⁶ The morphology of TiO₂ was tailored by Ti foil anodization, and high-aspect-ratio TiO₂ nanotubes provided more sites for NH₃ adsorption than spherical TiO₂ (P25).¹¹⁷ Although TiO₂ does not absorb visible light, the adsorption of NH₃ on TiO₂ could generate an extra energy level in the bandgap via in situ doping to induce direct electron transfer from the N 2p orbital to the Ti 3d orbital under irradiation with visible light (λ ≥ 400 nm).¹¹⁸ This concept resembles that of ligand-to-metal charge transfer.¹¹⁹ Dye sensitization is an effective way to inject electrons from dye molecules into the conduction band of TiO₂ under visible light. Among the 15 dyes anchored on TiO₂, the Ru(2,2′-bipyridyl-4,4′-dicarboxylic acid)₂(NCS)₂ complex (N3-dye) showed the highest performance (Fig. 10b).^120,121 The remaining holes in the HOMOs of dye molecules activated NH₃, and N₂ was selectively formed by the reaction between NO₂⁻ and . Consequently, the complete conversion of NO and a 100% selectivity for N₂ were achieved at a high GHSV of 100 000 h⁻¹ under 30 min irradiation with visible light (Table 2). One of the serious problems of dye-sensitized systems in aqueous media is the detachment of dye molecules from TiO₂ and the dependence of charge transfer on the complexation between functional groups. However, the occurrence of the reaction at the gas–solid interface allows dye desorption to be ignored. Therefore, numerous dyes are available for dye-sensitized SCR.

Fig. 10 (a) Photo-SCR of NO over various metal oxide (1.0 wt%)-promoted TiO₂ (GHSV: 50 000 h⁻¹). (b) Photo-SCR of NO over dye-modified TiO₂ under visible light irradiation (dye loading: 12.5 μmol g⁻¹, GHSV: 100 000 h⁻¹). (1) N3 dye, (2) Rose Bengal, (3) eosin Y, (4) Ru(bpy)₃Cl₂, (5) rhodamine B, (6) coumarin 343, (7) TCPP, (8) methylene blue, (9) Zn phthalocyanine, (10) Congo Red, (11) phthalocyanine, (12) RhCl₃, (13) Indigo Carmine, (14) Cu phthalocyanine, and (15) carmine dyes. Reprinted with permission from ref. 109 and 121. Copyright© 2016 and 2015, Wiley.

Yao's group designed diverse types of photocatalytic systems for the photo-SCR of NO, mainly those relying on (i) cascadal electron transfer [LaFe_1−xMn_xO₃/palygorskite,¹²² La_1−xCe_xFeO₃/palygorskite,¹²³ LaFe_1−xNi_xO₃/palygorskite,¹²⁴ La_1−xPr_xCoO₃/palygorskite,¹²⁵ CaTi_1−xMn_xO_3−δ¹²⁶], (ii) Z-scheme electron transfer [N-doped carbon quantum dots/PrFeO₃,¹²⁷ Fe₂O₃/SmFeO₃/palygorskite,¹²⁸ LaCoO₃/palygorskite/reduced graphene oxide,¹²⁹ Pr_1−xCe_xFeO₃/palygorskite,¹³⁰ CeVO₄/modified palygorskite¹³¹], and (iii) up-conversion (near-infrared light → UV and visible light) [CeO₂/Pr³⁺/palygorskite¹³² and CeO₂/palygorskite¹³³] (Fig. 11 and Table 2). The metal ion content and hetero-element doping in mixed oxides altered the photocatalyst’s physical properties such as particle size, electronic band structure, surface acidity, and charge trapping sites, and the supports (palygorskite) were shown to prevent nanoparticle agglomeration and provide sites for NH₃ adsorption. High NO conversion and the selective formation of N₂ were achieved, and the mechanism of NO conversion to N₂ was the same as that reported by Tanaka's group despite the difference in electron transfer pathways proposed. The Ag nano- and sub-nano-clusters incorporated in zeolites also promoted photo-assisted SCR under visible light irradiation (λ ≥ 390 nm), with activity determined by the reaction temperature (room temperature vs. 150 °C).¹³⁴ Ag_n^δ+ clusters were utilized as sensitizers because of their surface plasmon resonance and they favored the decomposition of NH₂NO to N₂ at 150 °C as opposed to the further oxidation of to NO and then to NO₂ by singlet oxygen at room temperature.

Fig. 11 Schematic diagrams of (a) cascadal electron transfer, (b) Z-scheme electron transfer, and (c) electron transfer in composite materials for the photocatalytic oxidation of NH₃. Reprinted with permission from ref. 124 and 133. Copyright© 2018 and 2020, Elsevier. Reprinted with permission from ref. 127. Copyright© 2018, American Chemical Society.

DeNO_xvia the oxidation of carbon-containing compounds. NH₃ can be replaced by CH₄, C₃H₈, C₄H₁₀, CO, CO(NH₂)₂, C₂H₅OH, carbon black, etc. for the photo-SCR of NO to N₂ over TiO₂-based catalysts under aerobic conditions. Bowering et al. tested the photocatalytic conversion of NO to N₂ over TiO₂ (P25) in the presence of CO as a reducing agent under UV light irradiation and showed that the reaction did not follow the Eley–Rideal mechanism but was rather driven by the adsorbed CO and NO (i.e., (i) CO_ads + O_ads → CO₂(g); (ii) CO_ads + 2NO_ads → N₂O_ads + CO_2ads; (iii) CO_ads + N₂O_ads → N₂(g) + CO_2ads).¹³⁵ Therefore, the rate of NO conversion was the highest under CO-free conditions owing to the absence of competitive adsorption, whereas the highest selectivity for N₂ was achieved in the presence of CO even though NO conversion was reduced. In the latter case, the surface hydroxyl groups significantly influenced selectivity (more hydroxyl groups led to better performance), and the loading of Ag on TiO₂ markedly enhanced selectivity while reducing NO conversion, as Ag clusters acted as centers for electron–hole pair recombination.¹³⁶ The electron defects (Ti³⁺, F⁺, and F centers) intentionally introduced on TiO₂ by partial reduction could reduce NO under visible light irradiation, and the presence of CO as a regenerator of donor centers increased the selectivity for N₂.¹³⁷ Although extra experiments were also performed in the presence of hydrocarbons such as C₂H₆, C₂H₄, C₃H₈, propylene, C₄H₁₀, benzene, toluene, ethylbenzene, and o-xylene, the research goal was not the selective conversion of NO to N₂ but the utilization of NO as an oxidant for the removal of VOC.^138,139

Wu's group employed photo-SCR for denitrification in the presence of saturated hydrocarbons including CH₄,^140,141 C₃H₈,^142–144 and C₄H₁₀ ^145–147 as reducing agents, focusing on the synthesis of TiO₂ and its structure/surface modification in Pd/TiO₂, PtO_xPdO_y/TiO₂, PdO/TiO₂, Ag/TiO₂, Cu/TiO₂, Pt/TiO₂, and TiO₂ nanosheets. In the absence of co-catalysts on TiO₂, the electron-donating behavior of hydrocarbons was not effectively utilized, unlike in the case of NH₃. However, the temperature and the presence of moisture and oxygen were important for controlling NO conversion and selectivity for N₂. For example, when PtO_xPdO_y/TiO₂ was tested at temperatures of 25, 70, and 120 °C, the best performance was observed at the highest temperature when either oxygen or water vapor was present, as these conditions helped to avoid the accumulation of nitrate and the desorption of water vapor from active sites, respectively. On the other hand, the opposite trend was observed under vapor- and oxygen-free conditions because of the poor adsorption of C₃H₈ and NO (competitive adsorption as in the case of CO) on the catalyst surface at high temperature. Without PtO_xPdO_y catalysts, NO oxidation was dominant, and therefore, nitrate was formed as the major product, with its accumulation on the surface resulting in a decrease of activity with reaction time.

The utilization of urea, C₂H₅OH, and carbon black as reducing agents was also possible for the photocatalytic denitrification of NO to N₂. In the case of TiO₂ and urea co-supported carbon fiber, TiO₂ and urea promoted the formation of NO₂ and the sequential reduction of NO₂ to N₂ at room temperature, respectively, with moisture accelerating the desorption of NO₂ from TiO₂.¹⁴⁸ In the case of Au/TiO₂ + ethanol, the adsorption and dissociation of ethanol on Au particles or at the Au/TiO₂ interface initiated the reaction at room temperature when C₂H₅O_ads accepted a photogenerated electron. This reaction was also promoted by the by-products (CH₃CHO, H₂, CO, and CH₄) formed by ethanol decomposition.¹⁴⁹ Last, the reduction of NO to N₂ was conducted over TiO₂ along with the photocatalytic oxidation of carbon black to CO₂ in the presence of O₂ and moisture at 150 °C, providing the possibility of utilizing solid materials as reductants (Table 2).¹⁵⁰

2.2.2. Photocatalytic removal of N₂O(g) to N₂. Several decades ago, the photocatalytic denitrification of N₂O to N₂ was investigated on ZnO at 371–431 °C under UV light irradiation, which induced the decomposition of N₂O via a combination of thermocatalysis and photocatalysis.¹⁵¹ The thermocatalytic reaction followed first-order kinetics, while the photocatalytic reaction kinetics was more complicated and governed by N₂O⁻ that was formed as an intermediate through electron transfer from ZnO to N₂O. The mechanism proposed for n-type metal oxides in the dark at 20 °C comprises five steps and features step 2 as the rate-limiting step and O_ads⁻⋯MO⁺(s) as the dominant species because of the fast dissociation of N₂O_ads⁻ (Fig. 12a).¹⁵² Therefore, the N₂ formation rate sharply increased and then saturated within the initial reaction stage over ZnO, in which case the kinetics was much faster than in the case of the photocatalytic reaction (Fig. 12b). Although the quantum efficiency of photo-assisted dissociation was small, an additional increase in N₂ generation clearly appeared with time, and the reaction pathway (i.e., N₂O_ads + (e⁻ + h⁺) → → N₂(g) + O_ads) was specified by the migration of photogenerated charge carriers to the surface of ZnO. Electron paramagnetic resonance spectroscopy revealed that electrons are transferred to N₂O_ads (N₂O_ads + e⁻ → N₂(g) + O_ads⁻), and O_ads⁻ might be localized at oxide ion vacancies via migration, while holes can be trapped at oxide ions via migration through the lattice.¹⁵³ Anpo et al. detected hyperfine splitting caused by one nitrogen atom at 77 K, which indicated the formation of either N₂O⁻ or N₂O₂⁻ (N₂O + O⁻ → N₂O₂⁻) on TiO₂ supported by porous Vycor glass.^154,155

Fig. 12 (a) Mechanism of the conversion of N₂O to N₂ and O₂ over a metal oxide surface. (b) Time-dependent production of N₂ (V_N) from N₂O (10 Torr) over an activated ZnO surface at 20 °C under (i) dark and light on/off conditions indicated by arrows and (ii) continuous illumination. Reprinted with permission from ref. 152. Copyright© 1971, American Chemical Society. (c) Relative energy diagram for the photocatalytic decomposition of N₂O on perfect anatase (001) facets. Reprinted with permission from ref. 173. Copyright© 2018, Royal Society of Chemistry. (d) Reaction diagram for the conversion of N₂O to various adsorption and decomposition products (N₂O*, , O*, , N₂(g) and O₂(g); * designates adsorbed species). (1–3) Dissociative adsorption of N₂O on Ir(111), (4) formation of surface peroxides, and (5) O₂ desorption from Ir(111). Reprinted with permission from ref. 177. Copyright© 2019, American Chemical Society.

Kudo et al. reported the denitrification of N₂O over metal (Pt, Ag, and Cu)-loaded TiO₂ in the presence of electron donors (water or/and methanol vapor) under UV light.^156,157 Pt promoted the separation of electron–hole pairs, the dissociation of N₂O, and the supply of adsorbed hydrogen atoms (i.e., H⁺ + e⁻ → H; N₂O + 2H → N₂ + H₂O; N₂O⁻ + H → N₂ + OH⁻), while water was oxidized on TiO₂ (4OH⁻ + 4h⁺ → O₂ + 2H₂O). In the presence of both water and CH₃OH vapor, the photocatalytic activity of Pt/TiO₂ for N₂O reduction was almost negligible, as the photogenerated electrons were selectively transferred to water to produce H₂. Ag- and Cu-loaded TiO₂ promoted denitrification, probably because of the facile dissociation of N₂O_ads on Au and Cu surfaces as well as the relatively high kinetic barrier for the reduction of water by electrons. Sano et al. further probed the removal of N₂O over Ag/TiO₂ in the presence of CH₃OH vapor and showed that the photocatalytic performance was affected by the oxidation state of Ag.¹⁵⁸ In particular, partially reduced Ag₂O prepared by photodeposition was more active than metallic Ag, which was ascribed to Ag⁺-mediated charge transfer (N₂O_ads + e⁻ → N₂O⁻_ads; N₂O⁻_ads + Ag⁺→ N₂ + Ag–O; 3Ag–O + CH₃OH_ads + 3h⁺ → 3Ag⁺ + CO₂ + 2H₂O).

Metal ions (Cu⁺, Ag⁺, Pb²⁺, and Pr³⁺) were immobilized on the surface of metal oxides (SiO₂, Al₂O₃, and SiO₂/Al₂O₃) or incorporated inside ZSM-5 zeolite pores to promote the photo-assisted removal of N₂O. Cu⁺-anchored metal oxides were prepared by an ion-exchange method with thermovacuum treatment, and linear two-coordinate and planar three-coordinate Cu⁺ ions were observed on SiO₂/Al₂O₃ and Al₂O₃ or SiO₂, respectively.¹⁵⁹ In this case, Cu⁺ was assumed to undergo a (3d¹⁰),¹S₀ → (3d)⁹(4s)¹,¹D₂ electronic transition under UV light irradiation, and electron transfer from the photo-excited Cu⁺ to N₂O initiated denitrification. Photocatalytic activity was affected by the coordination geometry (linear or planar) and the aggregation state (isolated Cu⁺ monomer or Cu⁺–Cu⁺ dimer) of Cu⁺ and was the highest for the isolated linearly coordinated Cu⁺ monomer owing to the long lifetime of charge carriers coupled with the low accumulation of O_ads due to O₂ release. As Cu⁺ was incorporated in ZSM-5 and Y zeolite cavities, the type of Cu⁺ species depended on the degassing temperature, which suggested that the excited state of the Cu⁺–Cu⁺ dimer was an effective N₂O quencher.^160–162 In the case of Ag⁺-exchanged ZSM5, UV light irradiation induced the 4d¹⁰ → 4d⁹5s¹ transition of two-coordinate isolated Ag⁺ ions, and the complexation of Ag⁺ with N₂O provided a channel for electron transfer from the excited Ag⁺ to the antibonding molecular orbital of N₂O.¹⁶³ For Pb²⁺-exchanged and Pr³⁺-supported catalysts, the reaction mechanisms were similar to those observed for Cu⁺- and Ag⁺-exchanged ones.^164,165

Kočí's group reported diverse photocatalysts for the decomposition of N₂O under UV light irradiation, e.g., ZnS/montmorillonite,¹⁶⁶ cordierite/steatite/CeO₂,¹⁶⁷ TiO₂/C₃N₄,¹⁶⁸ WO₃/C₃N₄,¹⁶⁹ ZnO/C₃N₄,¹⁷⁰ BiOIO₃/C₃N₄,¹⁷¹ and BiVO₄/C₃N₄.¹⁷² Among them, binary photocatalyst combinations helped to inhibit charge carrier recombination and therefore exhibited enhanced photocatalytic denitrification performances. Although some of these photocatalysts exhibited visible-light activity, all experiments were carried out under UV light. Moreover, the physicochemical interactions between the catalyst surface and N₂O were not deeply investigated. To bridge this gap, Liu's group used DFT calculations to model the decomposition of N₂O on TiO₂,¹⁷³ CeO₂,¹⁷⁴ BiVO₄,^175,176 BiMoO₆,¹⁷⁶ and Bi₂WO₆,¹⁷⁶ obtaining results well correlated with experimental findings. For example, in the case of TiO₂, the photogenerated electrons did not affect N₂O adsorption, but the presence of oxygen vacancies or excited electrons promoted the N₂O decomposition reaction. The surface-trapped electrons at five-coordinate Ti (Ti_5c⁴⁺ + e⁻ → Ti_5c³⁺) centers could act as active sites for N–O bond cleavage, with the reaction pathway depending on the adsorption geometry, i.e., on whether N₂O (O=N⁺=N⁻ ↔ ⁻O–N⁺≡N) was adsorbed on TiO₂via the oxygen or the nitrogen end. In the case of decomposition on perfect anatase (001) facets, the N₂O adsorbed on Ti³⁺via the oxygen end possessed an exothermic energy of 0.17 eV, and the O–N bond cleavage by the transfer of excited electrons from Ti³⁺ to N₂O featured an exothermicity of 0.29 eV and produced N₂ (Fig. 12c). On the other hand, the N₂O adsorbed on Ti³⁺via the nitrogen end formed an intermediate bridging configuration (with a binding energy of 0.19 eV), and the N–O bond cleavage was characterized by an enthalpy change of −0.40 eV. Finally, N₂ release from TiO₂ was an endothermic (by 0.13 eV) process. The removal of O⁻ was ascribed to O⁻ discharge followed by recombination with another O atom, which proceeded via hole transfer to O⁻ and could decrease the energy barrier for O₂ production.

An Al–Ir plasmonic antenna reactor combining plasmonic metallic antenna nanoparticles (Al nanocrystals) with nearby catalytic reactors (Ir nanoparticles) was designed for the photocatalytic conversion of N₂O to N₂ and O₂.¹⁷⁷ At high GHSVs (≥80 000 h⁻¹), the conversion efficiency reached 10%, and N₂ and O₂ were the only products formed. The apparent activation energy was maintained irrespective of illumination, which suggested that photothermal heating rather than hot carriers generated by the plasmon effect was responsible for N₂O decomposition. As depicted in Fig. 12d, the pre-adsorption of N₂O on Ir (step 1) and the dissociation of N₂O (step 2) are not necessary because of the high exothermicity of the dissociative adsorption of N₂O(g) into and O* at high operating temperatures (step 3). For fully saturated O*, the direct interaction between N₂O(g) and O* can be driven by the Eley–Rideal mechanism to produce surface peroxide (step 4, moderately endothermic). Finally, the reaction is completed by the highly endothermic desorption of surface peroxide (; step 5), which was assumed to be the rate-limiting step for the overall N₂O decomposition on Ir(111).

3. Oxidation of ammonia to N₂ under aerobic and anaerobic conditions

Ammonia is one of the most valuable chemicals in agricultural and other industries, and has recently received much attention as a hydrogen carrier.¹⁷⁸ As the manufacture of NH₃ by the Haber–Bosch process is highly energy-intensive and consumes H₂ that is mainly derived from fossil fuels, the economically feasible utilization of NH₃ as a hydrogen carrier requires the development of highly active catalysts for the production of NH₃ and its decomposition to H₂ under mild conditions.^179,180 Therefore, much effort has been directed at the establishment of new methods of ambient-condition N₂ fixation, particularly those using renewable energy resources.¹⁸¹ Among these methods, the photocatalytic reduction of N₂ to NH₃ holds great promise, as the electrons and hydrogen are provided by sunlight and water, respectively, although the cleavage of the N≡N bond in N₂ at standard temperature and pressure is a big challenge because of the low solubility of this gas.¹⁸² From an environmental perspective, NH₃ is not a useful chemical but a pollutant because of its high toxicity, corrosivity, odor, etc., and should therefore be effectively removed from air and water. The expansion of agricultural infrastructure to satisfy the increasing global food demand is facilitating the release of NH₃ (from fertilizers, livestock excretions, etc.) to the atmosphere and water bodies.¹⁸³ Most studies on photocatalysis target the oxidation of NH₃ to N₂ or NO_x (2NH₃ + 1.5O₂ → N₂ + 3H₂O, Δ_rG⁰₂₉₈ = −652.41 kJ mol⁻¹; 2NH₃ + 4O₂ → 2HNO₃ + 2H₂O, Δ_rG⁰₂₉₈ = −585.4 kJ mol⁻¹), with comprehensive catalysts and relevant reaction mechanisms summarized in recent reviews.^184,185 Herein, we briefly describe the photocatalytic decomposition of NH₃ on TiO₂ and present several examples of relatively high performance for the selective conversion of NH₃ to N₂ at room temperature.

Fig. 13a presents the mechanism of the photocatalytic oxidation of gas-phase NH₃ on Pt/TiO₂ in the presence/absence of water vapor under anaerobic conditions.¹⁸⁶ Initially, NH₃ is adsorbed on both Lewis- and Bronsted-acidic sites of TiO₂ (mainly hydroxyl groups), and the reaction is initiated by the charge carriers generated under UV light irradiation. The electrons migrate to Pt nanoparticles to reduce protons and thus produce H₂. The oxidation of adsorbed NH₃ occurs via hole transfer, and the coupling of two amide radicals produces N₂H₄, which can be subsequently converted into H₂ and N₂H₂. Finally, N₂H₂ self-decomposes into N₂ and H₂ or disproportionates into N₂ and N₂H₄. As this process does not involve the formation of NO_x, the H₂ : N₂ molar ratio was recorded as 2.9, which was close to the theoretical value of 3.0 for the decomposition of NH₃ to N₂ and H₂. Although the hole-mediated oxidation of to N through NH to release N₂ is also possible, it is energetically unfavorable because of its higher net activation energy.¹⁶⁰ Under dry conditions, the accumulation of NH₄⁺ ions on TiO₂ promotes catalyst deactivation, as these ions cannot easily migrate to Pt nanoparticles in the absence of water (Fig. 13b).

Fig. 13 Proposed mechanism of the photocatalytic decomposition of NH₃ on Pt/TiO₂ in the (a) presence and (b) absence of water. Reprinted with permission from ref. 186. Copyright© 2012, American Chemical Society.

When TiO₂ is used under aerobic and humid conditions, various nitrogen-containing species (e.g., NO, NO₂, NO₂⁻, NO₃, NO₃⁻, N₂O, HONO, and N₂H₄) might be involved as intermediates or produced as by-products during NH₃ oxidation,^188–194 which complicates the selective production of N₂. When the experiment was carried out by irradiating TiO₂ in a flow tube with a stream of NH₃-containing air, HONO was formed as an intermediate.¹⁹⁰ The production of HONO was negligible in the absence of O₂ and exhibited a volcano-type dependence on the concentration of NH₃. The increase in [HONO] was ascribed to the photoreduction of NO₂ (NH₃ → NO₂ → HONO), while the decrease in [HONO] at higher NH₃ concentrations was ascribed to the saturation of surface-active sites according to the Langmuir–Hinshelwood model and the reaction with NH₃ (NH₃ + HONO → NH₄NO₂ → N₂ + 2H₂O). In a similar manner, [HONO] exhibited a volcano-like dependence on the relative humidity of the gas flow. Water accelerated the formation of HONO at low humidity, although excess water could occupy the pores of TiO₂, hinder the access of NH₃ to active sites, and facilitate the quenching of OH˙ to decrease [HONO]. In this experiment, NO_x was formed as the major by-product. Instead of probing the complete conversion of NH₃ to N₂, almost all studies investigated the photocatalytic abatement of NH₃ without analyzing the composition of the final products.

As mentioned earlier, the selective conversion of gaseous NH₃ to N₂ under aerobic conditions is challenging. From a practical viewpoint, operation under anaerobic conditions does not make sense, as the photocatalytic process is designed to remove few-ppm-level NH₃ from air. In this regard, an anammox-like process aims to completely remove nitrogen species from aqueous systems (mainly NH₃-containing wastewater) or use concentrated NH₃ solutions as hydrogen carriers to provide H₂ for fuel cells and should be more feasible owing to the ease of inert atmosphere generation via N₂ or Ar purging. The protonation of NH₃ (pK_a ≈ 9.25) and the positive change of TiO₂ surface charge (pH_zpc 6–7 for P25) in acidic and neutral media cause electrostatic repulsion (i.e., NH₄⁺ ↔ >Ti–OH₂⁺), which hinders the adsorption of NH₄⁺ and inhibits the photocatalytic reaction.¹⁹⁵ Moreover, whereas NH₄⁺ is stable against attack by OH˙, neutral NH₃ is degraded by OH˙ under photocatalytic conditions.^196–198 Therefore, high photocatalytic performance was achieved at pH 10–11, whereas an activity decrease was observed at higher pH, probably because of the low solubility of NH₃ under these conditions.

The use of metal nanoparticles as co-catalysts offers a simple way to increase the yield and selectivity of photocatalytic processes, prolong charge carrier lifetime, and provide catalytically active sites. Among the various metal nanoparticles used in conjunction with TiO₂, Pt nanoparticles exhibited an outstanding performance for the decomposition of NH₃ into N₂ and H₂ under both oxic and anoxic conditions.^199–201 Based on the calculated adsorption energies, Pt (−394 kJ mol⁻¹) has a moderate atomic nitrogen affinity for N₂ formation among the tested metals (e.g., Ag (−156 kJ mol⁻¹), Au (−162 kJ mol⁻¹), Rh (−448 kJ mol⁻¹), Ru (−525 kJ mol⁻¹)).²⁰² In comparison with bare TiO₂, which generated only nitrite and nitrate as end-products under air-saturated conditions, the loading of Pt (0.2 wt%) accelerated the reaction kinetics and promoted the evolution of N₂ to reduce the total N content in the NH₃ solution.¹⁷⁰ Interestingly, the presence of O₂ had little influence on the kinetics over Pt/TiO₂, for which the efficiency of the NH₃ to N₂ conversion after 2 h irradiation equaled 65–70% in both air and N₂. Pt nanoparticles on TiO₂ probably stabilized NH_x species generated as intermediates by OH˙-mediated chain reactions. When O₂ was replaced by N₂O, more OH radicals were formed through the reductive dissociation of N₂O on Pt to increase the efficiency of the NH₃ to N₂ conversion to 80%. The photocatalytic conversion of NH₃ and the selectivity for N₂ simultaneously increased with the increase in the loading of Pt on TNTs under oxic conditions. In particular, an ammonia conversion of 100% (for [NH₃]_i = 20 ppm) and a selectivity of 87.5% were achieved after 3 h irradiation for Pt/TNTs (25 wt% Pt).²⁰⁰ Although the reductive dissociation of NH₃ on Pt and the overoxidation of NH₃ on TNTs might be responsible for the formation of N₂ and NO_x ions, respectively, it is still unclear whether the reductive dissociation of NH₃ is energetically favorable or not, and the function of nitrogen hydrogen radicals on Pt as electron/hole recombination centers remains to be explored.

Under anaerobic conditions, the H₂ : N₂ molar ratio achieved at alkaline pH using metallized photocatalysts (Pt/TiO₂, Pt/Fe-doped TiO₂, Ni/TiO₂, Pt_0.9Au_0.1/TiO₂, and Ru/ZnS) was close to the theoretical molar ratio (3 : 1). The main advantage of the anammox-like process is its ability to achieve both the complete removal of NH₃ from wastewater and the recovery of H₂ as a fuel for fuel cells at room temperature under sunlight. For example, in a highly concentrated solution (0.59 M), NH₃ was decomposed at pH 10–12 over Pt/TiO₂ (0.5 wt% Pt) to afford H₂ and N₂ in a 3 : 1 molar ratio, and the catalyst performance was governed by Pt loading, pH, photocatalyst type, and co-catalyst type. Despite the lack of supporting evidence, Pt was assumed to provide active sites for the reduction of protons to H₂, while the oxidation of NH₃ occurred on TiO₂. When Pt/Fe-doped TiO₂ (0.5 wt% Pt and 1.0 wt% Fe) was tested in 0.59 M NH₃ under UV light irradiation, a 3 : 1 (mol/mol) H₂ : N₂ ratio was recorded.²⁰³ The higher H₂ yield of Pt/Fe-doped TiO₂ (27 μmol mg_cat mol⁻¹) than that of Pt/TiO₂ (18 μmol mg_cat mol⁻¹) was due to the better absorption of visible light in the former case. Except for the case of Ni/TiO₂, the loading of non-noble-metals (V, Cr, Mn, Fe, Co, and Cu) on TiO₂ slightly decreased the H₂ yield, whereas the H₂ : N₂ molar ratio of 3 : 1 was maintained (0.59 M NH₃).¹⁸⁸ As seen in Fig. 14a, the amounts of N₂ and H₂ produced over Ni/TiO₂ (0.5 wt% Ni) linearly increased with increasing irradiation time, and the reaction completely stopped in the dark. Isotope labeling experiments performed with D₂O revealed that no D₂ and HD were produced, i.e., the hydrogen in H₂ stemmed from NH₃ and not from water (Fig. 14b). This result indicates that the photodecomposition of NH₃ occurred on the interface between metal nanoparticles and TiO₂ and involved the direct migration of H˙ (formed by the hole-mediated reaction of NH₃) to Pt. DFT calculations indicated the existence of two possible pathways for TiO₂-based NH₃ decomposition, namely (i) 2NH_{3, ads} → + H₂(g) → H₂N–NH₂ + H₂(g) → ˙N=N˙ + 3H₂(g) → N₂(g) + 3H₂(g) and (ii) NH_{3, ads} + NH₃ → + H˙ + NH₃ → NH₂–NH₃ + H˙ → H₂N–NH₂ + H₂(g) → ˙N=N˙ + 3H₂(g) → N₂(g) + 3H₂(g). The formation of NH₂–NH₂ was probably assisted by metallic Ni. The loading of bimetallic alloy nanoparticles on TiO₂ is also a good way to enhance the photocatalytic activity of monometallic nanoparticle/TiO₂ hybrids, with the highest activity obtained for Pt_0.9Au_0.1 under UV light irradiation (Fig. 14c).²⁰⁴ As depicted in Fig. 14d, charge separation efficiency is determined by the Schottky barrier (φ_B; φ_B = metal work function (W) − electron affinity of the TiO₂ conduction band (χ)). The introduction of Au into Pt reduces φ_B, which was calculated as 1.84, 1.62, and 0.97 eV for Pt/TiO₂, Pt_0.9Au_0.1/TiO₂, and Au/TiO₂, respectively. Overly high and low φ_B values suppress electron separation and promote reverse electron transfer, thus decelerating photocatalytic reactions. The decomposition of NH₃ into N₂ and H₂ was also carried out using other photocatalysts such as RuO₂–NiO–SrTiO₃,²⁰⁵ ZnO,²⁰⁶ and Ru/ZnS,²⁰⁷ the activities of which were much lower than that of Pt/TiO₂. The visible light-induced decomposition of NH₃ into N₂ and H₂ was also attempted in a dye-sensitized system comprising a homogeneous tris(bipyridine)ruthenium(II) (Ru(bpy)₃²⁺) dye, methyl viologen as an electron mediator, and O₂ as an electron acceptor.²⁰⁸ Under visible light irradiation, Ru(bpy)₃³⁺ oxidized NH₃ and was converted to the original state, Ru(bpy)₃²⁺.

Fig. 14 (a) Time profiles of H₂ and N₂ production by the photodecomposition of NH₃ over Ni/TiO₂ (1.0 wt%) in the dark and under illumination. (b) Time profiles of gas-phase product yields for the photocatalytic decomposition of NH₃ over Ni/TiO₂ in D₂O. Reprinted with permission from ref. 187. Copyright© 2017, Elsevier. (c) Amounts of H₂ and N₂ evolved during the 6 h photocatalytic decomposition of NH₃ on Pt_0.9M_0.1/TiO₂ (M: Au, Pd, Cu, Ni, and Ag, total metal loading on TiO₂: 0.1 mol%). (d) Schematic electronic structure of a metal/semiconductor interface. E_vac, E_F, W, φ_B, and χ denote the vacuum level, Fermi energy level, metal work function, Schottky barrier, and the electron affinity of the semiconductor conduction band, respectively. Reprinted with permission from ref. 204. Copyright© 2020, American Chemical Society.

4. Photoelectrochemical denitrification and ammonia oxidation

Only a few studies deal with photoelectrochemical denitrification and ammonia oxidation, focusing on the recovery of N₂ from nitrogen species. The photoelectrochemical denitrification of nitrite (1 mM NaNO₂ at pH 7) was first achieved in 1999 using a three-electrode system (counter electrode (CE) = Pt wire, reference electrode (RE) = saturated calomel electrode (SCE), and working electrode (WE) = roughened Ag electrode) in 0.1 M Na₂SO₄ as an electrolyte under laser irradiation (362, 413, 457, 476, 488, 496, 514, and 647 nm) and a nitrogen atmosphere.²⁰⁹ Denitrification was initiated by the excitation of Ag via plasmon resonance and the electrochemical current generated at −1.0 V_SCE. Notably, irradiation brought about not only an increase in cathodic current but also a positive shift of the onset potential. The quantum efficiency was estimated as 0.04% without the analysis of real-time NO₂⁻, NO₃⁻, N₂, and NH₃ concentrations. Nitrate reduction was believed to involve the electrochemical nitrate to nitrite conversion followed by the photoelectrochemical reduction of nitrite to NH₃ and N₂. The photoelectrochemical nitrate to nitrite conversion was also observed for Ag nanopyramids (1 M NaNO₃ at pH 5.7 under Ar), in which case the plasmon resonance of Ag resulted in an almost 100% faradaic efficiency at −1.0 V_RHE.²¹⁰

The photoelectrochemical nitrate to nitrite reduction was also promoted by semiconducting photocathode materials such as p-GaInP₂, nanoporous p-Si, and CuI/PbI₂. In the case of p-GaInP₂, data were collected in a three-electrode system (CE = Pt black, RE = Ag/AgCl, and WE = p-GaInP₂) in 0.1 M HNO₃ + 0.5 M NH₄NO₃ as an electrolyte (pH 1) under simulated solar light at an air mass (AM) of 1.5 G.²¹¹ The faradaic efficiency of nitrate reduction was calculated as 80%, and the incident-photon-to-current efficiency (IPCE) at −1.0 V_Ag/AgCl was recorded as 100, 60, and 5% under excitation at 400, 580, and 610 nm, respectively. As a close to zero current was obtained in the dark, illumination was concluded to stimulate the rate-limiting step, and the catalytically active sites were assumed to be Ga and/or In. For nanoporous p-Si under similar conditions, the faradaic efficiency of nitrate reduction at −0.6 V_Ag/AgCl equaled 65%, and no NH₃ and N₂ were observed.²¹² In the case of CuI–PbI₂, the faradaic efficiency of nitrate reduction in 0.1 M NaNO₃ exceeded 52%, and the IPCE at 400 nm was around 15%.²¹³ The bubbles evolved on the photoelectrode surface probably contained N₂ rather than H₂, as no H₂ signal was observed by gas chromatography. Interestingly, isotope labeling experiments performed in Ar-saturated 0.1 M Na¹⁵NO₃ solution (98% ¹⁵N) revealed that the generation of NH₃ was due to an external contamination and not nitrate reduction.

n-type semiconductors can be used as photoanodes for the water oxidation-induced conversion of nitrate to N₂. In the presence of NH₃ as an electron donor (i.e., using the same concept as that discussed in Section 2.2.1, the photo-SCR deNO_x), ammonia oxidation and denitrification simultaneously occurred over TiO₂ and Pt black, respectively, in the absence of a bias voltage under UV light irradiation (1 mM NH₃, 100 mM KNO₃) (Fig. 15a).²¹⁴ When a mixture of pig urine/wash water (1/4) containing NH₄⁺, NO₃⁻, and NO₂⁻ was tested under aerobic conditions, the following concentration decreases were observed after 24 h: NH₄⁺ (2580 → 166 ppm), NO₃⁻ (18.6 → 17.0 ppm), and NO₂⁻ (4.84 → 3.17 ppm). The imbalance in the removal of NO_x⁻ and NH₄⁺ was ascribed to the competitive reduction of oxygen to H₂O. Similarly, in a biophotochemical cell, H₂O or biorefractory organics were oxidized at the photoanode, while denitrification proceeded at the biocathode.^215,216 The biocathode was prepared using activated sludge as an inoculum and was separated from the TiO₂ photoanode by a cation exchange membrane. As seen in Fig. 15b, the concentration of nitrate continuously decreased under illumination, whereas the abiotic cathode did not show any activity. Indeed, NO₂⁻ and N₂O were formed as intermediates, but the concentration of these intermediates and NO₃⁻ decreased to zero after 30 h (Fig. 15c). NH₄⁺ ions were always present at levels below the detection limit, which indicated that nitrate was selectively converted to N₂. The faradaic efficiency of the cathode was estimated as 97%, and a small number of electrons was assumed to be consumed by microbial growth.

Fig. 15 (a) Photoelectrochemical denitrification of NO₃⁻ to N₂ over a TiO₂ photoanode connected to a Pt cathode in the presence of NH₃ and H₂O as electron donors under Ar. Reprinted with permission from ref. 214. Copyright© 2009, Royal Society of Chemistry. (b) Decrease of nitrate level under a photo-generated current during on–off intermittent illumination and (c) change of nitrogen oxide levels with time. Reprinted with permission from ref. 215. Copyright© 2017, American Chemical Society. (d) Proposed mechanism for the oxidation of NH₃ over a CuO/Co₃O₄ photocathode in the presence of peroxydisulfate. Reprinted with permission from ref. 219. Copyright© 2020, Elsevier.

The photoelectrochemical oxidation of ammonia to N₂ can be accomplished using the anodic or cathodic reaction to control photogenerated holes or radical species (hydroxyl or sulfate radicals) activated by electrons, respectively.^217–220 In the employed system (CE = Pt wire, RE = Ag/AgCl, and WE = TiO₂ photoanode in 10 M NH₃ + 0.1 M KNO₃ at pH 14.1), the H₂ : N₂ molar ratio equaled 3.08 under short-circuit conditions after 2 h irradiation.¹⁸⁸ The holes in TiO₂ oxidized NH₃ to N₂, and the electrons transferred to Pt reduced water to H₂. The OH˙ and SO₄˙⁻ species generated by the activation of peroxydisulfate (S₂O₈²⁻ + e⁻ → SO₄˙⁻ + SO₄²⁻ and SO₄˙⁻ + H₂O → H⁺ + OH˙ + SO₄²⁻) at the CuO/Co₃O₄ photocathode oxidized NH₃ to N₂ (Fig. 15d).²¹⁹ The removal of 96.1% NH₃ (100 ppm) was achieved under visible light irradiation, and the reactive sites were identified as Co and Cu species.

5. Summary and outlook

There are various methods of decreasing the levels of reactive nitrogen compounds, which can include systematic crop rotation, optimization of the timing and amount of fertilizer input, the breeding or development of genetically engineered varieties of crops for increased nitrogen utilization efficiency, direct up-cycling of used nitrogen to microbial protein, and the development of artificial denitrification/ammonia oxidation processes powered by renewable energy. Among them, the solar-powered photocatalytic and photoelectrochemical approaches are promising and future-oriented ways to treat aqueous and airborne NO_x, N₂O, and NH₃ because of their economically feasible and environmentally benign nature. However, the low efficiency and selectivity to N₂ and the scale-up problems are still a bottleneck for practical applications. Basically, the efficiency of artificial solar-powered denitrification/ammonia oxidation can be determined by (i) the absorbance of photocatalysts, (ii) the electronic structure of semiconducting materials, (iii) the recombination of charge carriers (i.e., the lifetime of photogenerated electrons and holes), (iv) the control of surface properties to suppress undesired reactions such as ammonification and the re-oxidation of intermediates and products, and (v) the surroundings (e.g., the presence/absence of oxygen, different types of electron donors, pH, humidity, etc.). Due to the different reaction pathways, the development of new photocatalysts and the systematic design of reactors have to be different depending on the treatment of aqueous and gas phase NO_x, N₂O, and NH₃. In an aqueous system, the bimetallic nanoparticles loaded on TiO₂ (e.g., Pd–Cu and Pt–Cu) showed a high performance of NO₃⁻ conversion selective to N₂, which gives a hint that the separation of the catalytic sites, the reduction of NO₃⁻ to NO₂⁻ and the further reduction of NO₂⁻ to N₂, is important to increase the selectivity. Precise control of the N : H ratio at surface active sites of catalysts including Lewis acid sites and defect sites is particularly required to suppress ammonification and thus selectively convert NO₂⁻ to harmless N₂. The reason why it is hard to reach a high conversion efficiency of NO₃⁻ or NO₂⁻ as well as a high selectivity to N₂ is due to the competitive charge transfer to H⁺, O₂, and H₂O and the re-oxidation of intermediates and by-products. In the case of the photocatalytic treatment of airborne NO_x and N₂O to N₂, it has been thoroughly investigated during the last three decades, offering the advantages of using water instead of the explosive H₂ and toxic NH₃, operation at standard temperature and pressure, a net zero carbon emission in the case of operation under natural sunlight, and the availability of cheap and environmentally benign materials. Although various strategies such as structure and morphology control, co-catalyst loading, heteroatom doping, and hybridization with different types of materials have been developed to overcome the bottleneck of the conversion of NO_x and N₂O to N₂, the problems of insufficiently selective charge transfer, low solar light absorption, poor catalytic activity, need for noble metals as co-catalysts, and the lack of long-term durability need to be addressed to minimize the environmental impact of airborne NO_x and N₂O. In order to increase the possibility of commercialization, the addition of sacrificial hole scavengers should be necessary, where organic pollutants, particularly persistent organic pollutants (POPs), and ammonia (or carbon monoxide) are good candidates for wastewater and polluted air treatment, respectively, to compensate the operation cost.

Although ammonia is a very important feedstock, its high toxicity, corrosivity, and noxious odor make it a pollutant from the perspective of the environmental management of the nitrogen cycle. Given the large annual production of ammonia via the Haber–Bosch process and the low nitrogen use efficiency of ammonia-based fertilizers, ammonia should be effectively removed from air and water on a comparable scale. Recently, considering ammonia as a hydrogen carrier, the development of highly active catalysts for the decomposition of ammonia to H₂ and N₂ under mild conditions is highly desired. Although the conventional photocatalytic processes have focused on ammonia abatement, directing the production of nitrate instead of N₂, with a future-oriented point of view, the photocatalytic recovery of H₂ from concentrated ammonia solution seems quite promising for fuel cell applications. The oxidation of adsorbed NH₃ occurring via photogenerated holes does not involve the formation of NO_x, but the H₂ : N₂ molar ratio becomes close to the theoretical value of 3 : 1 through the decomposition of NH₃ to N₂ and H₂, which is driven at standard temperature and pressure under illumination.

A photoelectrochemical cell can selectively control the reduction and oxidation reaction of nitrogen-containing species, in which the photogenerated electron–hole pairs are easily separated and consequently participate in the denitrification and anammox upon applying extra bias. To date, very few studies have been reported, in particular targeting the removal of toxic nitrite and ammonia from wastewater; however, the application should be more suitable for anammox in order to secure H₂ from concentrated ammonia. Contrary to photocatalysis, it does not need to separate N₂ and H₂ because the oxidation and the reduction are proceeded in the anode and cathode, respectively, which is comparted by a membrane. Indeed, the addition of an electrolyte is unnecessary in that the pH of concentrated ammonia solution (>12) is conductive enough to transport ions in the electrolyte. The in-depth investigation and successful development of photoelectrochemical ammonia oxidation systems will enable a counterpart of (photo)electrocatalytic nitrogen fixation, in other words the combination of the production and the utilization of ammonia as a hydrogen carrier.

It is time to take this issue seriously and think about it, and photocatalysis is the greenest way to restore the nitrogen cycle with a future-oriented technology. In order to go one step further to commercialization, the following can be considered: (i) the development of new materials to overcome the intrinsic problems of photocatalysts, (ii) the control of composition, morphology, and size of catalysts (e.g., high entropy alloy, single atom catalyst, etc.), (iii) the systematic modification of photocatalysts including hybridization such as ternary and quaternary composites, co-doping, anchoring homogeneous sensitizers or promoters, selective surface passivation, etc., (iv) the separation of catalytic sites by the control of the boundary between the catalysts and supporter or by a Janus structure, (v) the precise control of the micro-environment on the catalysts or electrodes, (vi) the finding of suitable POPs and greenhouse gases that cannot be removed by conventional treatment or typical AOPs, (vii) the design of a photo-reactor and its scale-up, (viii) the combination with other processes such as the pretreatment or final treatment through biological processes, and (ix) in situ analysis (e.g., time-resolved surface enhanced infrared absorption/Raman spectroscopy) to unveil the real-time charge transfer and the formation of intermediates for the optimization of desired reactions.

Author contributions

Cheolwoo Park: investigation, visualization, writing – original draft. Hyelim Kwak: investigation, visualization, writing – original draft. Gun-hee Moon: conceptualization, supervision, writing – original draft. Wooyul Kim: conceptualization, supervision, writing – review & editing.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was financially supported by the Basic Science Research Program (NRF-2019R1C1C1006833) funded by the Korea government (MSIT) through the National Research Foundation of Korea (NRF), the Ecological Imitation-Based Environmental Pollution Management Technology Development Project funded by the Korea government (MOE) through KEITI (No. 2019002790008), the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2020M3H4A3106354), and the KIST internal project (3E311191) funded by the Korea Institute of Science and Technology (KIST).

Notes and references

1. A. Bernhard, Nat. Educ. Knowl., 2010, 2, 1–8.
2. D. E. Canfield, A. N. Glazer and P. G. Falkowski, Science, 2010, 330, 192–196.
3. J. N. Galloway, A. M. Leach, A. Bleeker and J. W. Erisman, Philos. Trans. R. Soc., B, 2013, 368, 20130120.
4. M. A. Sutton, C. M. Howard, T. K. Adhya, E. Baker, J. Baron, A. Basir, W. Brownlie, C. Cordovil, W. de Vries, V. Eory, R. Green, H. Harmens, K. W. Hicks, R. Jeffery, D. Kanter, L. Lassaletta, A. Leip, C. Masso, T. H. Misselbrook, E. Nemitz, S. P. Nissanka, O. Oenema, S. Patra, M. Pradhan, J. Ometto, R. Purvaja, N. Raghuram, R. Ramesh, N. Read, D. S. Reay, E. Rowe, A. SanzCobena, S. Sharma, K. R. Sharp, U. Skiba, J. U. Smith, I. van der Beck, M. Vieno, and H. J. M. van Grinsven, Nitrogen - Grasping the Challenge. A Manifesto for Science-In-Action through the International Nitrogen Management System. Summary Report, Ecology & Hydrology, Edinburgh, UK, 2019.
5. Food and Agriculture Organization, FAO Statistical Databases, 2006, Rome, available at http://faostat.fao.org/default.aspx.
6. International Energy Agency, IEA Statistical Databases, 2015, Paris, available at http://iea.org/data-and-statistics.
7. Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories, ed. J. T. Houghton, L. G. Meria Filho, K. Lim, I. Trennton, I. Mamaty, Y. Bonduki, D. J. Griggs and B. A. Callander, IPCC, OECD, IEA, 1996.
8. D. Laffoley and J. M. Baxter, Ocean Deoxygenation: Everyone’s Problem, IUCN, Global Marine and Polar Programme, 2019.
9. N. Gruber and J. N. Galloway, Nature, 2008, 451, 293–296.
10. S. Matassa, D. J. Batstone, T. Hülsen, J. Schnoor and W. Verstraete, Environ. Sci. Technol., 2015, 49, 5247–5254.
11. S. P. Seitzinger, C. Kroeze, A. F. Bouwman, N. Caraco, F. Dentener and R. V. Styles, Estuaries, 2002, 25, 640–655.
12. M. B. Peoples, J. Brockwell, D. F. Herridge, I. J. Rochester, B. J. R. Alves, S. Urquiaga, R. M. Boddey, F. D. Dakora, S. Bhattarai, S. L. Maskey, C. Sampet, B. Rerkasem, D. F. Khan, H. Hauggaard-Nielsen and E. S. Jensen, Symbiosis, 2009, 48, 1–17.
13. Y. Lan, J. Chen, H. Zhang, W.-X. Zhang and J. Yang, J. Mater. Chem. A, 2020, 8, 15853–15863.
14. W. Hong, L. Su, J. Wang, M. Jiang, Y. Ma and J. Yang, Chem. Commun., 2020, 56, 14685–14688.
15. F. Ni, Y. Ma, J. Chen, W. Luo and J. Yanga, Chin. Chem. Lett., 2021, 32, 2073–2078.
16. L. Su, D. Han, G. Zhu, H. Xu, W. Luo, L. Wang, W. Jiang, A. Dong and J. Yang, Nano Lett., 2019, 19, 5423–5430.
17. H. Xu, J. Wu, W. Luo, Q. Li, W. Zhang and J. Yang, Small, 2020, 16, 2001775.
18. D. R. Keeney, R. L. Chen and D. A. Graetz, Nature, 1971, 233, 66–67.
19. A. Kapoor and T. Viraraghavan, J. Environ. Eng., 1997, 123, 371–380.
20. H. O. N. Tugaoen, S. Garcia-Segura, K. Hristovski and P. Westerhoff, Sci. Total Environ., 2017, 599–600, 1524–1551.
21. Z. Geng, Z. Chen, Z. Li, X. Qi, X. Yang, W. Fan, Y. Guo, L. Zhang and M. Huo, Dalton Trans., 2018, 47, 11104–11112.
22. H. Xu, Y. Li, M. Ding, W. Chen, K. Wang and C. Lu, ACS Sustainable Chem. Eng., 2018, 6, 7042–7051.
23. S. Roy, J. Phys. Chem. C, 2020, 124, 28345–28358.
24. H. Zhang, Z. Liu, Y. Li, C. Zhang, Y. Wang, W. Zhang, L. Wang, L. Niu, P. Wang and C. Wang, Appl. Surf. Sci., 2020, 503, 144092.
25. J. E. Silveira, A. R. Ribeiro, J. Carbajo, G. Pliego, J. A. Zazo and J. A. Casas, Water Res., 2021, 200, 117250.
26. L. Wang, W. Fu, Y. Zhuge, J. Wang, F. Yao, W. Zhong and X. Ge, Chemosphere, 2021, 278, 130298.
27. T. Caswell, M. W. Dlamini, P. J. Miedziak, S. Pattisson, P. R. Davies, S. H. Taylor and G. J. Hutchings, Catal. Sci. Technol., 2020, 10, 2082–2091.
28. H. Kominami, K. Kitsui, Y. Ishiyama and K. Hashimoto, RSC Adv., 2014, 4, 51576–51579.
29. H. Gekko, K. Hashimoto and H. Kominami, Phys. Chem. Chem. Phys., 2012, 14, 7965–7970.
30. W. L. Silver, D. J. Herman and M. K. Firestone, Ecology, 2001, 82, 2410–2416.
31. E. M. Sander, B. Virdis and S. Freguia, RSC Adv., 2015, 5, 86572–86577.
32. H. Hirakawa, M. Hashimoto, Y. Shiraishi and T. Hirai, ACS Catal., 2017, 7, 3713–3720.
33. H. Kominami, A. Furusho, S.-y. Murakami, H. Inoue, Y. Kera and B. Ohtani, Catal. Lett., 2001, 76, 31–34.
34. K. Akihiko, D. Kazunari, M. Ken-ichi and O. Takaharu, Chem. Lett., 1987, 16, 1019–1022.
35. M. Duca and M. T. M. Koper, Energy Environ. Sci., 2012, 5, 9726–9742.
36. Z. Hou, J. Chu, C. Liu, J. Wang, A. Li, T. Lin and C. P. François-Xavier, Chem. Eng. J., 2021, 415, 2082–2091.
37. S. Lee, S. Kim, C. Park, W. Kim, S. Ryu and W. Choi, Energy Environ. Sci., 2021 10.1039/D1EE01342D.
38. P. Li, Z. Jin, Z. Fanga and G. Yu, Energy Environ. Sci., 2021, 14, 3522–3531.
39. M. Shand and J. A. Anderson, Catal. Sci. Technol., 2013, 3, 879–899.
40. H.-i. Kim, K. Kim, S. Park, W. Kim, S. Kim and J. Kim, Sep. Purif. Technol., 2019, 209, 580–587.
41. L. Lei, W. Wang, C. Wang, H. Fan, A. K. Yadav, N. Hu, Q. Zhong and P. Müller-Buschbaum, J. Mater. Chem. A, 2020, 8, 23812–23819.
42. Q. Liu, S. Wang, Q. Ren, T. Li, G. Tu, S. Zhong, Y. Zhao and S. Bai, J. Mater. Chem. A, 2021, 9, 1552–1562.
43. X. Li, H. Jiang, C. Ma, Z. Zhu, X. Song, X. Li, H. Wang, P. Huo and X. Chen, J. Mater. Chem. A, 2020, 8, 18707–18714.
44. S. Lee, S. Kim, C. Park, G.-h. Moon, H.-J. Son, J.-O. Baeg, W. Kim and W. Choi, ACS Sustainable Chem. Eng., 2020, 8, 3709–3717.
45. H. Lee, S. Jee, R. Kim, H.-T. Bui, B. Kim, J.-K. Kim, K. Park, W. Choi, W. Kim and K. Choi, Energy Environ. Sci., 2020, 13, 519–526.
46. E.-T. Yun, H.-Y. Yoo, W. Kim, H.-E. Kim, G. Kang, H. Lee, S. Lee, T. Park, C. Lee, J.-H. Kim and J. Lee, Appl. Catal., B, 2017, 203, 475–484.
47. Y. Zhang, W. Hu, D. Wang, B. J. Reinhart and J. Huang, J. Mater. Chem. A, 2021, 9, 6180–6187.
48. S. Weon, E. Choi, H. Kim, J. Kim, H.-J. Park, S.-m. Kim, W. Kim and W. Choi, Environ. Sci. Technol., 2018, 52, 9330–9340.
49. Q. Guo, F. Liang, Z. Sun, Y. Wang, X.-B. Li, S.-G. Xia, Z. C. Zhang, L. Huang and L.-Z. Wu, J. Mater. Chem. A, 2020, 8, 22601–22606.
50. J. Sá, C. A. Agüera, S. Gross and J. A. Anderson, Appl. Catal., B, 2009, 85, 192–200.
51. J. A. Anderson, Catal. Today, 2011, 175, 316–321.
52. W. Gao, R. Jin, J. Chen, X. Guan, H. Zeng, F. Zhang and N. Guan, Catal. Today, 2004, 90, 331–336.
53. S. Rengaraj and X. Z. Li, Enhanced photocatalytic reduction reaction over Bi3+-TiO2 nanoparticles in presence of formic acid as a hole scavenger, Chemosphere, 2007, 66, 930–9383.
54. D. D. B. Luiz, S. L. F. Andersen, C. Berger, H. J. José and R. D. F. P. M. Moreira, J. Photochem. Photobiol. A, 2012, 246, 36–44.
55. J. R. Pan, C. Huang, W. Hsieh and B. Wu, Sep. Purif. Technol., 2012, 84, 52–55.
56. K. Kobwittaya and S. Sirivithayapakorn, APCBEE Proc., 2014, 10, 321–325.
57. S.-E. Bae, K. L. Stewart and A. A. Gewirth, J. Am. Chem. Soc., 2007, 129, 10171–10180.
58. I. Sanjuán, L. García-Cruz, J. Solla-Gullón, E. Expósito and V. Montiel, Electrochim. Acta, 2020, 340, 135914.
59. S. Hamid, M. A. Kumar, J.-I. Han, H. Kim and W. Lee, Green Chem., 2017, 19, 853–866.
60. H. Shin, S. Jung, S. Bae, W. Lee and H. Kim, Environ. Sci. Technol., 2014, 48, 12768–12774.
61. T. Gu, W. Teng, N. Bai, Z. Chen, J. Fan, W.-x. Zhang and D. Zhao, J. Mater. Chem. A, 2020, 8, 9545–9553.
62. S. Jung, S. Bae and W. Lee, Environ. Sci. Technol., 2014, 48, 9651–9658.
63. A. J. Calandra, C. Tamayo, J. Herrera and A. J. Arvía, Electrochim. Acta, 1972, 17, 2035–2053.
64. L. Li, Z. Xu, F. Liu, Y. Shao, J. Wang, H. Wan and S. Zheng, J. Photochem. Photobiol., A, 2010, 212, 113–121.
65. H. Kominami, T. Nakaseko, Y. Shimada, A. Furusho, H. Inoue, S.-y. Murakami, Y. Kera and B. Ohtani, Chem. Commun., 2005, 2933–2935.
66. O. S. G. P. Soares, M. F. R. Pereira, J. J. M. Órfão, J. L. Faria and C. G. Silva, Chem. Eng. J., 2014, 251, 123–130.
67. Z. Hou, F. Chen, J. Wang, C. P. François-Xavier and T. Wintgens, Appl. Catal., B, 2018, 232, 124–134.
68. J. A. Zazo, P. García-Muñoz, G. Pliego, J. E. Silveira, P. Jaffe and J. A. Casas, Appl. Catal., B, 2020, 273, 118930.
69. H. Kato and A. Kudo, Phys. Chem. Chem. Phys., 2002, 4, 2833–2838.
70. L. Mohapatra and K. Parida, J. Mater. Chem. A, 2016, 4, 10744–10766.
71. N. Dewangan, W. M. Hui, S. Jayaprakash, A.-R. Bawah, A. J. Poerjoto, T. Jie, A. Jangam, K. Hidajat, S. Kawi, Catal. Today, 356, 490–513.
72. S. Zhang, Y. Zhao, R. Shi, C. Zhou, G. I. N. Waterhouse, L.-Z. Wu, C.-H. Tung and T. Zhang, Adv. Energy Mater., 2020, 10, 1901973.
73. D. P. Sahoo, K. K. Das, S. Patnaik and K. Parida, Inorg. Chem. Front., 2020, 7, 3695–3717.
74. Y. Zhao, G. Chen, T. Bian, C. Zhou, G. I. N. Waterhouse, L.-Z. Wu, C.-H. Tung, L. J. Smith, D. O'Hare and T. Zhang, Adv. Mater., 2015, 27, 7824–7831.
75. M. Adachi and A. Kudo, Chem. Lett., 2012, 41, 1007–1008.
76. X. Li, S. Wang, H. An, G. Dong, J. Feng, T. Wei, Y. Ren and J. Ma, Appl. Surf. Sci., 2021, 539, 148257.
77. M. Yue, R. Wang, B. Ma, R. Cong, W. Gao and T. Yang, Catal. Sci. Technol., 2016, 6, 8300–8308.
78. R. Wang, M. Yue, R. Cong, W. Gao and T. Yang, J. Alloys Compd., 2015, 651, 731–736.
79. B. Wang, B. An, Z. Su, L. Li and Y. Liu, Chemosphere, 2021, 269, 128754.
80. G. Liu, S. You, M. Ma, H. Huang and N. Ren, Environ. Sci. Technol., 2016, 50, 11218–11225.
81. H.-T. Ren, S.-Y. Jia, J.-J. Zou, S.-H. Wu and X. Han, Appl. Catal., B, 2015, 176–177, 53–61.
82. Y. Li and F. Wasgestian, J. Photochem. Photobiol., A, 1998, 112, 255–259.
83. J. A. Anderson, Catal. Today, 2012, 181, 171–176.
84. B. Bems, F. C. Jentoft and R. Schlögl, Appl. Catal., B, 1999, 20, 155–163.
85. F. Zhang, Y. Pi, J. Cui, Y. Yang, X. Zhang and N. Guan, J. Phys. Chem. C, 2007, 111, 3756–3761.
86. J. M. A. Freire, M. A. F. Matos, D. S. Abreu, H. Becker, I. C. N. Diógenes, A. Valentini and E. Longhinotti, J. Environ. Chem. Eng., 2020, 8, 103844.
87. A. Oita, A. Malik, K. Kanemoto, A. Geschke, S. Nishijima and M. Lenzen, Nat. Geosci., 2016, 9, 111–115.
88. D. S. Reay, E. A. Davidson, K. A. Smith, P. Smith, J. M. Melillo, F. Dentener and P. J. Crutzen, Nat. Clim. Change, 2012, 2, 410–416.
89. G. Busca, L. Lietti, G. Ramis and F. Berti, Appl. Catal., B, 1998, 18, 1–36.
90. L. Yang, A. Hakki, F. Wang and D. E. Macphee, Appl. Catal., B, 2018, 222, 200–208.
91. Z. Gu, B. Zhang, Y. Asakura, S. Tsukuda, H. Kato, M. Kakihana and S. Yin, Appl. Surf. Sci., 2020, 521, 146213.
92. V. Khanal, N. O. Balayeva, C. Günnemann, Z. Mamiyev, R. Dillert, D. W. Bahnemann and V. R. Subramanian, Appl. Catal., B, 2021, 291, 119974.
93. G. Liu, H. Xia, W. Zhang, L. Song, Q. Chen and Y. Niu, J. Hazard. Mater., 2021, 418, 126337.
94. J. V. C. d. Carmo, C. L. Lima, G. Mota, A. M. S. Santos, L. N. Costa, A. Ghosh, B. C. Viana, M. Silva, J. M. Soares, S. Tehuacanero-Cuapa, R. Lang, A. C. Oliveira, E. Rodríguez-Castellón and E. Rodríguez-Aguado, Materials, 2021, 14, 2181.
95. W. Kim, T. Tachikawa, G.-h. Moon, T. Majima and W. Choi, Angew. Chem., Int. Ed., 2014, 53, 14036–14041.
96. Q. Wu and R. van de Krol, J. Am. Chem. Soc., 2012, 134, 9369–9375.
97. Q. Wu, Q. Zheng and R. van de Krol, J. Phys. Chem. C, 2012, 116, 7219–7226.
98. G. Dong, D. L. Jacobs, L. Zang and C. Wang, Appl. Catal., B, 2017, 218, 515–524.
99. M. Anpo, Y. Shioya, H. Yamashita, E. Giamello, C. Morterra, M. Che, H. H. Patterson, S. Webber and S. Ouellette, J. Phys. Chem., 1994, 98, 5744–5750.
100. M. Matsuoka, E. Matsuda, K. Tsuji, H. Yamashita and M. Anpo, J. Mol. Catal. A: Chem., 1996, 107, 399–403.
101. W.-S. Ju, M. Matsuoka and M. Anpo, Catal. Lett., 2001, 71, 91–93.
102. M. Anpo, S. G. Zhang, H. Mishima, M. Matsuoka and H. Yamashita, Catal. Today, 1997, 39, 159–168.
103. Y. Hu, G. Martra, J. Zhang, S. Higashimoto, S. Coluccia and M. Anpo, J. Phys. Chem. B, 2006, 110, 1680–1685.
104. M. Matsuoka and M. Anpo, in Catalysis by Unique Metal Ion Structures in Solid Matrices, ed. G. Centi, B. Wichterlová and A. T. Bell, Springer, Dordrecht, 2001, ch. Photocatalysis of Cations Incorporated within Zeolites, vol. 13, pp. 249–262.
105. M. Anpo, M. Matsuoka, K. Hanou, H. Mishima, H. Yamashita and H. H. Patterson, Coord. Chem. Rev., 1998, 171, 175–184.
106. M. Anpo and J. M. Thomas, Chem. Commun., 2006, 3273–3278.
107. N. W. Cant and J. R. Cole, J. Catal., 1992, 134, 317–330.
108. T. Shishido, K. Teramura and T. Tanaka, Catal. Sci. Technol., 2011, 1, 541–551.
109. A. Yamamoto, K. Teramura and T. Tanaka, Chem. Rec., 2016, 16, 2268–2277.
110. K. Teramura, T. Tanaka and T. Funabiki, Langmuir, 2003, 19, 1209–1214.
111. K. Teramura, T. Tanaka, S. Yamazoe, K. Arakaki and T. Funabiki, Appl. Catal., B, 2004, 53, 29–36.
112. S. Yamazoe, T. Okumura, K. Teramura and T. Tanaka, Catal. Today, 2006, 111, 266–270.
113. Y. Ji and Y. Luo, J. Phys. Chem. C, 2014, 118, 6359–6364.
114. S. Yamazoe, Y. Masutani, T. Shishido and T. Tanaka, Res. Chem. Intermed., 2008, 34, 487–494.
115. S. Yamazoe, Y. Masutani, K. Teramura, Y. Hitomi, T. Shishido and T. Tanaka, Appl. Catal., B, 2008, 83, 123–130.
116. R. Jin, Z. Wu, Y. Liu, B. Jiang and H. Wang, J. Hazard. Mater., 2009, 161, 42–48.
117. Y.-C. Chou and Y. Ku, Chem. Eng. J., 2013, 225, 734–743.
118. S. Yamazoe, K. Teramura, Y. Hitomi, T. Shishido and T. Tanaka, J. Phys. Chem. C, 2007, 111, 14189–14197.
119. G. Zhang, G. Kim and W. Choi, Energy Environ. Sci., 2014, 7, 954–966.
120. A. Yamamoto, Y. Mizuno, K. Teramura, S. Hosokawa, T. Shishido and T. Tanaka, Catal. Sci. Technol., 2015, 5, 556–561.
121. A. Yamamoto, K. Teramura, S. Hosokawa, T. Shishido and T. Tanaka, ChemCatChem, 2015, 7, 1818–1825.
122. X. Li, X. Yan, S. Zuo, X. Lu, S. Luo, Z. Li, C. Yao and C. Ni, Chem. Eng. J., 2017, 320, 211–221.
123. H. Zhang, X. Li, Y. Hui, L. Yu, Q. Xia, S. Luo and C. Yao, J. Mater. Sci.: Mater. Electron., 2017, 28, 9371–9377.
124. X. Li, H. Shi, W. Zhu, S. Zuo, X. Lu, S. Luo, Z. Li, C. Yao and Y. Chen, Appl. Catal., B, 2018, 231, 92–100.
125. K. Wei, X. Yan, S. Zuo, W. Zhu, F. Wu, X. Li, C. Yao and X. Liu, Clays Clay Miner., 2019, 67, 348–356.
126. Z. Zhang, H. Lü, X. Li, X. Li, S. Ran, Z. Chen, Y. Yang, X. Wu and L. Li, ACS Sustainable Chem. Eng., 2019, 7, 10299–10309.
127. X. Li, H. Shi, X. Yan, S. Zuo, Y. Zhang, T. Wang, S. Luo, C. Yao and C. Ni, ACS Sustainable Chem. Eng., 2018, 6, 10616–10627.
128. X. Li, H. Shi, T. Wang, Y. Zhang, X. Lu, S. Zuo, Z. Li and C. Yao, J. Taiwan Inst. Chem. Eng., 2018, 89, 119–128.
129. X. Li, X. Yan, X. Lu, S. Zuo, Z. Li, C. Yao and C. Ni, J. Catal., 2018, 357, 59–68.
130. X. Li, H. Shi, X. Yan, S. Zuo, Y. Zhang, Q. Chen, C. Yao and C. Ni, J. Catal., 2019, 369, 190–200.
131. X. Li, Z. Wang, H. Shi, D. Dai, S. Zuo, C. Yao and C. Ni, J. Hazard. Mater., 2020, 386, 121977.
132. X. Li, H. Zhang, H. Lü, S. Zuo, Y. Zhang and C. Yao, Environ. Sci. Pollut. Res., 2019, 26, 12842–12850.
133. X. Li, Z. Wang, X. Chu, B. Gao, S. Zuo, W. Liu and C. Yao, Appl. Clay Sci., 2020, 199, 105871.
134. H. I. Hamoud, M. Lafjah, F. Douma, O. I. Lebedev, F. Djafri, V. Valchev, M. Daturi and M. El-Roz, Sol. Energy, 2019, 189, 244–253.
135. N. Bowering, G. S. Walker and P. G. Harrison, Appl. Catal., B, 2006, 62, 208–216.
136. N. Bowering, D. Croston, P. G. Harrison and G. S. Walker, Int. J. Photoenergy, 2007, 2007, 090752.
137. A. A. Lisachenko, R. V. Mikhailov, L. L. Basov, B. N. Shelimov and M. Che, J. Phys. Chem. C, 2007, 111, 14440–14447.
138. C. H. Ao, S. C. Lee, C. L. Mak and L. Y. Chan, Appl. Catal., B, 2003, 42, 119–129.
139. S. Poulston, M. V. Twigg and A. P. Walker, Appl. Catal., B, 2009, 89, 335–341.
140. Y.-T. Wu, Y.-H. Yu, V.-H. Nguyen and J. C. S. Wu, Res. Chem. Intermed., 2015, 41, 2153–2164.
141. Y.-H. Yu, I.-H. Su and J. C. S. Wu, Environ. Technol., 2010, 31, 1449–1458.
142. J. Lasek, Y.-H. Yu and J. C. S. Wu, Environ. Technol., 2012, 33, 2133–2141.
143. Y.-H. Yu, Y.-T. Pan, Y.-T. Wu, J. Lasek and J. C. S. Wu, Catal. Today, 2011, 174, 141–147.
144. I. H. Su and J. C. S. Wu, Catal. Commun., 2009, 10, 1534–1537.
145. J. C.-C. Yu, V.-H. Nguyen, J. Lasek, S.-W. Chiang, D. X. Li and J. C. S. Wu, Appl. Catal., A, 2016, 523, 294–303.
146. J. C.-C. Yu, V.-H. Nguyen, J. Lasek and J. C. S. Wu, Appl. Catal., B, 2017, 219, 391–400.
147. J. C.-C. Yu, V.-H. Nguyen, J. Lasek, D. X. Li and J. C. S. Wu, Catal. Commun., 2016, 84, 40–43.
148. J. Miyawaki, T. Shimohara, N. Shirahama, A. Yasutake, M. Yoshikawa, I. Mochida and S.-H. Yoon, Appl. Catal., B, 2011, 110, 273–278.
149. G. Halasi, T. Bánsági and F. Solymosi, J. Catal., 2015, 325, 60–67.
150. L. Liao, S. Heylen, S. P. Sree, B. Vallaey, M. Keulemans, S. Lenaerts, M. B. J. Roeffaers and J. A. Martens, Appl. Catal., B, 2017, 202, 381–387.
151. G. Blyholder and K. Tanaka, J. Phys. Chem., 1971, 75, 1037–1043.
152. J. Cunningham, J. J. Kelly and A. L. Penny, J. Phys. Chem., 1971, 75, 617–625.
153. N. B. Wong, Y. B. Taarit and J. H. Lunsford, J. Chem. Phys., 1974, 60, 2148–2151.
154. M. Anpo, N. Aikawa and Y. Kubokawa, J. Chem. Soc., Chem. Commun., 1984, 644–645.
155. M. Anpo, N. Aikawa, Y. Kubokawa, M. Che, C. Louis and E. Giamello, J. Phys. Chem., 1985, 89, 5689–5694.
156. A. Kudo and T. Sakata, Chem. Lett., 1992, 21, 2381–2384.
157. A. Kudo and H. Nagayoshi, Catal. Lett., 1998, 52, 109–111.
158. T. Sano, N. Negishi, D. Mas and K. Takeuchi, J. Catal., 2000, 194, 71–79.
159. M. Matsuoka, W.-S. Ju, K. Takahashi, H. Yamashita and M. Anpo, J. Phys. Chem. B, 2000, 104, 4911–4915.
160. K. Ebitani, M. Morokuma, J.-H. Kim and A. Morikawa, J. Catal., 1993, 141, 725–728.
161. K. Ebitani, M. Morokuma, J.-H. Kim and A. Morikawa, J. Chem. Soc., Faraday Trans., 1994, 90, 377–381.
162. K. Ebitani, M. Morokuma and A. Morikawa, in Stud. Surf. Sci. Catal., ed. J. Weitkamp, H. G. Karge, H. Pfeifer and W. Hölderich, Elsevier, 1994, vol. 84, pp. 1501–1506.
163. W.-S. Ju, M. Matsuoka, K. Iino, H. Yamashita and M. Anpo, J. Phys. Chem. B, 2004, 108, 2128–2133.
164. W.-S. Ju, M. Matsuoka, H. Yamashita and M. Anpo, J. Synchrotron Radiat., 2001, 8, 608–609.
165. K. Ebitani, Y. Hirano and A. Morikawa, J. Catal., 1995, 157, 262–265.
166. L. Obalová, M. Šihor, P. Praus, M. Reli and K. Kočí, Catal. Today, 2014, 230, 61–66.
167. M. Valášková, K. Kočí and J. Kupková, Microporous Mesoporous Mater., 2015, 207, 120–125.
168. K. Kočí, M. Reli, I. Troppová, M. Šihor, J. Kupková, P. Kustrowski and P. Praus, Appl. Surf. Sci., 2017, 396, 1685–1695.
169. M. Reli, L. Svoboda, M. Šihor, I. Troppová, J. Pavlovský, P. Praus and K. Kočí, Environ. Sci. Pollut. Res. Int., 2018, 25, 34839–34850.
170. K. Kočí, M. Reli, I. Troppová, M. Šihor, T. Bajcarová, M. Ritz, J. Pavlovský and P. Praus, Catalysts, 2019, 9, 735.
171. V. Matějka, M. Šihor, M. Reli, A. Martaus, K. Kočí, M. Kormunda and P. Praus, Mater. Sci. Semicond. Process., 2019, 100, 113–122.
172. P. Praus, J. Lang, A. Martaus, L. Svoboda, V. Matějka, M. Kormunda, M. Šihor, M. Reli and K. Kočí, J. Inorg. Organomet. Polym., 2019, 29, 1219–1234.
173. L. Wang, W. Song, J. Deng, H. Zheng, J. Liu, Z. Zhao, M. Gao and Y. Wei, Nanoscale, 2018, 10, 6024–6038.
174. W. Song, L. Wang, Y. Gao, J. Deng, M. Jing, H. Zheng, J. Liu, Z. Zhao, M. Gao and Y. Wei, J. Mater. Chem. A, 2018, 6, 19241–19255.
175. L. Wang, J. Liu, W. Song, H. Wang, Y. Li, J. Liu, Z. Zhao, J. Tan, Z. Duan and J. Deng, Chem. Eng. J., 2019, 366, 504–513.
176. J. Liu, L. Wang, W. Song, M. Zhao, J. Liu, H. Wang, Z. Zhao, C. Xu and Z. Duan, ACS Sustainable Chem. Eng., 2019, 7, 2811–2820.
177. D. F. Swearer, H. Robatjazi, J. M. P. Martirez, M. Zhang, L. Zhou, E. A. Carter, P. Nordlander and N. J. Halas, ACS Nano, 2019, 13, 8076–8086.
178. A. Klerke, C. H. Christensen, J. K. Nørskov and T. Vegge, J. Mater. Chem., 2008, 18, 2304–2310.
179. Q. Wang, J. Guo and P. Chen, J. Energy Chem., 2019, 36, 25–36.
180. F. Schüth, R. Palkovits, R. Schlögl and D. S. Su, Energy Environ. Sci., 2012, 5, 6278–6289.
181. X. Cui, C. Tang and Q. Zhang, Adv. Energy Mater., 2018, 8, 1800369.
182. Q. Han, H. Jiao, L. Xiong and J. Tang, Adv. Mater., 2021, 2, 564–581.
183. E. Stokstad, Science, 2014, 343, 238.
184. K. Vikrant, K.-H. Kim, F. Dong and D. A. Giannakoudakis, ACS Catal., 2020, 10, 8683–8716.
185. S. Zhang, Z. He, X. Li, J. Zhang, Q. Zang and S. Wang, Nanoscale Adv., 2020, 2, 3610–3623.
186. H. Yuzawa, T. Mori, H. Itoh and H. Yoshida, J. Phys. Chem. C, 2012, 116, 4126–4136.
187. A. Utsunomiya, A. Okemoto, Y. Nishino, K. Kitagawa, H. Kobayashi, K. Taniya, Y. Ichihashi and S. Nishiyama, Appl. Catal., B, 2017, 206, 378–383.
188. A. C. Sola, D. Garzón Sousa, J. Araña, O. González Díaz, J. M. Doña Rodríguez, P. Ramírez de la Piscina and N. Homs, Catal. Today, 2016, 266, 53–61.
189. H. Mozzanega, J. M. Herrmann and P. Pichat, J. Phys. Chem., 1979, 83, 2251–2255.
190. M. A. Kebede, N. K. Scharko, L. E. Appelt and J. D. Raff, J. Phys. Chem. Lett., 2013, 4, 2618–2623.
191. B. Boulinguiez, A. Bouzaza, S. Merabet and D. Wolbert, J. Photochem. Photobiol., A, 2008, 200, 254–261.
192. M. Chen, J. Ma, B. Zhang, F. Wang, Y. Li, C. Zhang and H. He, Appl. Catal., B, 2018, 223, 209–215.
193. G. Zhang, J. Ruan and T. Du, ACS ES&T Eng., 2021, 1, 310–325.
194. J. Feng, X. Zhang, G. Zhang, J. Li, W. Song and Z. Xu, Chemosphere, 2021, 274, 129689.
195. X. Zhu, S. R. Castleberry, M. A. Nanny and E. C. Butler, Environ. Sci. Technol., 2005, 39, 3784–3791.
196. R. Zellner and I. W. M. Smith, Chem. Phys. Lett., 1974, 26, 72–74.
197. E.-M. Bonsen, S. Schroeter, H. Jacobs and J. C. Broekaert, Chemosphere, 1997, 35, 1431–1445.
198. S. Kim and W. Choi, Environ. Sci. Technol., 2002, 36, 2019–2025.
199. J. Lee, H. Park and W. Choi, Environ. Sci. Technol., 2002, 36, 5462–5468.
200. H.-H. Ou, M. R. Hoffmann, C.-H. Liao, J.-H. Hong and S.-L. Lo, Appl. Catal., B, 2010, 99, 74–80.
201. H. Kominami, H. Nishimune, Y. Ohta, Y. Arakawa and T. Inaba, Appl. Catal., B, 2012, 111–112, 297–302.
202. A. C. A. de Vooys, M. T. M. Koper, R. A. van Santen and J. A. R. van Veen, J. Electroanal. Chem., 2001, 506, 127–137.
203. J. Nemoto, N. Gokan, H. Ueno and M. Kaneko, J. Photochem. Photobiol., A, 2007, 185, 295–300.
204. Y. Shiraishi, S. Toi, S. Ichikawa and T. Hirai, ACS Appl. Nano Mater., 2020, 3, 1612–1620.
205. L. Qing-shui, D. Kazunari, N. Shuichi, O. Takaharu and T. Kenzi, Chem. Lett., 1983, 12, 321–324.
206. M. Reli, M. Edelmannová, M. Šihor, P. Praus, L. Svoboda, K. K. Mamulová, H. Otoupalíková, L. Čapek, A. Hospodková, L. Obalová and K. Kočí, Int. J. Hydrogen Energy, 2015, 40, 8530–8538.
207. A. Iwase, K. Ii and A. Kudo, Chem. Commun., 2018, 54, 6117–6119.
208. M. Kaneko, N. Katakura, C. Harada, Y. Takei and M. Hoshino, Chem. Commun., 2005, 3436–3438.
209. J. Zheng, T. Lu, T. M. Cotton and G. Chumanov, J. Phys. Chem. B, 1999, 103, 6567–6572.
210. Y. Kim, E. B. Creel, E. R. Corson, B. D. McCloskey, J. J. Urban and R. Kostecki, Adv. Energy Mater., 2018, 8, 1800363.
211. H. Wang and J. A. Turner, Energy Environ. Sci., 2013, 6, 1802–1805.
212. M. Kan, D. Yue, J. Jia and Y. Zhao, Electrochim. Acta, 2015, 177, 366–369.
213. E. Kecsenovity, S. T. Kochuveedu, J.-P. Chou, D. Lukács, Á. Gali and C. Janáky, Sol. RRL, 2021, 5, 2000418.
214. R. Saito, H. Ueno, J. Nemoto, Y. Fujii, A. Izuoka and M. Kaneko, Chem. Commun., 2009, 3231–3233.
215. H.-Y. Cheng, X.-D. Tian, C.-H. Li, S.-S. Wang, S.-G. Su, H.-C. Wang, B. Zhang, H. M. A. Sharif and A.-J. Wang, Environ. Sci. Technol., 2017, 51, 12948–12955.
216. S. Su, Y. Zhang, W. Hu, X. Zhang, D. Ju, C. Jia and J. Liu, Bioelectrochemistry, 2020, 132, 107439.
217. M. Kaneko, N. Gokan, N. Katakura, Y. Takei and M. Hoshino, Chem. Commun., 2005, 1625–1627.
218. X. Fan, Y. Zhou, G. Zhang, T. Liu and W. Dong, Appl. Catal., B, 2019, 244, 396–406.
219. C. Yan and L. Liu, J. Hazard. Mater., 2020, 388, 121793.
220. Y. Qu, X. Song, X. Chen, X. Fan and G. Zhang, Chem. Eng. J., 2020, 382, 123048.

---

Footnote

† These authors contributed equally.

---