Reactive oxygen species on transition metal-based catalysts for sustainable environmental applications

Abstract

Tailoring reactive oxygen species (ROS) in advanced oxidation processes (AOPs) in a controlled manner is essential in chemical synthesis and environmental applications. In this paper, we begin with an overview of four ROS, and summarize methods and various precursors (e.g., O₂, H₂O₂, and persulfate) to generate particular ROS. We then examine the use of the ROS in the degradation of pollutants and in the synthesis of value-added chemicals. We highlight the use and mechanism response of transition metal catalysts, generally developed with defect engineering, in advancing AOPs. We conclude with an outlook of current challenges and future perspectives of applying ROS in AOPs. We anticipate this review to inspire researchers to develop green, safe, and efficient ROS systems for sustainable environmental applications.

---

Yonglin Wen

Yonglin Wen received his BSc in materials science and engineering from Fuzhou University. He is currently pursuing his PhD degree at Fuzhou University under the supervision of Prof. Yu and Prof. Zhuang. His research focuses on the design of Mn-based catalysts for heterogeneous oxidation catalysis and biomass conversion.

Jiawei Yan

Jiawei Yan received his BSc in materials science and engineering from Fujian Agriculture and Forestry University. He is currently pursuing his PhD degree at Fuzhou University under the supervision of Prof. Zhuang and Prof. Yu. His research focuses on the design of transition metal-based materials deployed for advanced oxidation processes.

Zanyong Zhuang

Zanyong Zhuang received his BS degree (2006) in chemistry from Xiamen University, and PhD degree (2011) from the Fujian Institute of Research on the Structure of Matter (FJIRSM), Chinese Academy of Sciences (CAS). Currently, he is a full professor at Fuzhou University. His research interests mainly focus on the rational design of transition metal-based catalysts for energy and environmental applications, including advanced oxidation reactions and CO2 reduction reactions.

Yan Yu

Yan Yu received her BS, MS, and PhD degrees from Fuzhou University. She was a postdoctoral fellow from 2010–2013 at the Fujian Institue of Research on the Structure of Matter (FJIRSM), Chinese Academy of Sciences (CAS). Currently, she is a full professor at Fuzhou University. Her research interests include environmental remediation, water purification, ecological materials, and photocatalytic CO2 reduction, and H2 production.

---

1 Introduction

The growing consumption of petrochemical products is quickly depleting petroleum and fossil resources worldwide and discharging organic pollutants into the environment. Efficient green techniques are needed to reduce pollution and to produce value-added chemicals from renewable feedstock.^1–4 Advanced oxidation processes (AOPs) mediated by reactive oxygen species (ROS) have been demonstrated in the past few decades as the underpinning of many essential applications in environmental protection and sustainable development.^5,6

A key issue in developing suitable AOPs is finding the desired ROS with appropriate reactivity and selectivity.⁷ Fenton chemistry has been used extensively since 1894 to create the hydroxyl radical (HO˙), a powerful radical that can degrade toxic organic substrates thoroughly to CO₂ and H₂O. However, HO˙ has a short half-life and cannot be sustained in multi-step processes, and continuous addition of its precursor (e.g. H₂O₂) is required to produce a sufficient amount of HO˙ to ensure effective degradation of pollutants.⁸ Moreover, although it has long been used for the thorough degradation of recalcitrant organic contaminants to CO₂ and H₂O, it has limited applications in chemical synthesis or biomass conversion, as its high reactivity may cause undesired over-oxidation and degradation.⁹ Alternatively, researchers found that the superoxide radical (O₂˙⁻) and singlet oxygen (¹O₂) can also engage in AOPs, both of which have a longer half-life than HO˙ as well as a more moderate oxidation potential suitable for controlled oxidation. In addition to these ROS generated over the catalysts, the lattice oxygen (O_L) of transition metal oxides (TMO) can serve as a moderate ROS in heterogeneous AOPs to directly oxidize organic matter, and it has allowed successful oxidation reactions based on thermocatalysis, photocatalysis, electrocatalysis, etc.^10,11 Using O_L in chemical synthesis and biomass conversion is advantageous in that it can selectively oxidize the organic substrate and avoid over-oxidation.

Heterogeneous transition metal-based catalysts (e.g., oxides,¹² hydroxides,^13,14 oxyhydroxides,¹⁵ and sulfides¹⁶) are readily used in AOPs. Their advantages include earth abundance, high stability, low cost, etc.¹⁷ AOPs using heterogeneous catalysts manipulate the geometrical and/or electronic states of the precursors (e.g., H₂O₂, O₂, PMS, PDS, and TMO) to create ROS on the catalyst surfaces. Hence, surface and defect engineering has emerged in recent years as a popular and appealing technique to customize the catalyst and more effectively activate the percursors.^18–21 There are substantial developments in the rational design of deficient transition metal-based materials deployed for advancing AOP strategies.^22–26 Anionic vacancies (e.g., oxygen, sulfur, and nitrogen) introduced to transition metal-based catalysts as point defects always have a pronounced effect on the physical and chemical properties of the catalyst.^27,28 Engineering point defects creates new opportunities to afford advanced catalysts that can efficiently and selectively create ROS (HO˙, O₂˙⁻, ¹O₂, O_L) from the precursors to thus carry out the AOP as intended.²⁹

Understanding the reactivity and the selective generation of distinct ROS and identifying the mechanism response of deficient catalysts in AOPs are essential for the rational design of potent catalysts with fine-tuned oxidation behavior.³⁰ In this review, we firstly summarize the types and redox properties of ROS. We then systematically discuss the generation of ROS based on the precursor used and the methods to detect and identify the ROS. We also discuss defect engineering techniques that modify the catalyst surface and enable thorough or selective oxidation of organic matter. We next examine oxidation reactions that can be accomplished with the ROS. We finish with a brief comment on the current challenges and future developments of ROS in the burgeoning research area of AOPs.

2 Types of reactive oxygen species

2.1 Hydroxyl radical (HO˙)

The hydroxy radical is highly reactive and has demonstrated widespread use as a nonselective reagent in AOPs for wastewater treatment, as it can initiate complex chain oxidation reactions to convert organic matter to oxidized products or ultimately to CO₂ and H₂O. In acidic media, HO˙ has an extraordinarily high redox potential (2.8 V versus the normal hydrogen electrode, NHE) which is only second to that of fluorine.³¹ It can attack most organic pollutants with a high rate constant of 10⁶ to 10⁹ M⁻¹ s⁻¹, which is about 10³ to 10⁹ times faster than that of molecular O₂ (1–10³ M⁻¹ s⁻¹).³² However, the reactivity of HO˙ is pH-dependent according to the Nernst equation,³³ and its standard electrode potential drops to 1.55 V vs. NHE in basic media.³⁴ The HO˙ in aqueous solution has a half-life ranging from picoseconds to nanoseconds, and its self-diffusion coefficient in water is estimated to be 2.8 × 10⁻⁵ cm² s⁻¹. Hence, the diffusion distance of HO˙ is a few molecular diameters away from where the HO˙ is produced.^35,36

Mechanistically, the oxidation of organic matter by HO˙ may involve radical adduct formation, hydrogen abstraction, and electron transfer,³⁷ and is highly dependent on the types of functional groups in the organic substrate.³⁸ Because of its electrophilic nature, HO˙ normally prefers to react with electron-rich sites such as C=C and C=N bonds in the organic substrate rather than the C=O bonds whose carbon atom is electron-deficient.³⁹ The C–H bonds of saturated aliphatic compounds that have relatively small dissociation energy can experience hydrogen abstraction.⁴⁰

2.2 Superoxide radical (O₂˙⁻)

The superoxide radical, first proposed in 1934 as the radical anion in the potassium salt,⁴¹ is a relatively small univalent anion with an O–O bond distance between that of O₂ and peroxide.⁴² The ground-state triplet molecular oxygen (O₂) has two electrons occupying the separated π* orbitals of parallel spins (Fig. 1). When the ground-state O₂ participates in the redox reaction, the reductants need to offer a pair of electrons of the same spin states, so that the electrons from reductants can fit into the vacant spaces in the π* orbitals of oxygen.³³ Due to this limitation, the ground-state O₂ is normally of weak oxidizing capacity via the 2-electron transfer process. O₂˙⁻ can be formed by one-electron reduction of O₂, or by filling the two π* orbitals of O₂ with an electron followed by protonation (Fig. 1). The homonuclear diatomic radical O₂˙⁻ has a slightly lower redox potential (2.4 V vs. RHE) than HO˙,⁴³ and has demonstrated critical roles in chemical, biological, and environmental applications. Note that (somewhat perplexingly) O₂˙⁻ also has a reducing potential of E₀ (O₂/O₂˙⁻) = −0.33 V and can be oxidized by a ferricenium ion to give singlet oxygen (¹O₂),⁴⁴ because it has an extra electron in one of its π* antibonding orbitals compared to ground state dioxygen (³O₂). The redox potential of E₀ (O₂/O₂˙⁻) does no change with pH when the pH value is >4.8. At pH value <4.8, O₂˙⁻ will be protonated at pH < 4.8 which creates the perhydroxyl radical (˙O₂H) of reduction potential E₀ (O₂, H⁺/˙O₂H) = −0.046 V vs. NHE.^31,33 The half-life of O₂˙⁻ can reach up to 51–422 s at pH = 2–10,⁴⁵ much longer than those of HO˙ and ¹O₂.⁴⁶ It means O₂˙⁻ can be prepared and work under alkaline conditions. The diffusion distance of O₂˙⁻ has been estimated to be about 35 μm, assuming a general diffusion coefficient (about 105 cm² s⁻¹) of small molecules.⁴⁷

Fig. 1 Molecular π* orbitals for O₂, ¹O₂, and O₂˙⁻.³³

However HO˙ is electrophilic, and O₂˙⁻ can be both nucleophilic and electrophilic. The half-life of O₂˙⁻ is up to 51–422 s at pH = 2–10,⁴⁵ which is longer than that of HO˙ and ¹O₂.⁴⁶ Hence, O₂˙⁻ can be prepared and stabilized in strongly basic aqueous solutions. It is readily used to scavenge organic pollutants, such as treating pollutants in underground water under anoxic conditions,^33,48,49 and can serve as a major ROS for the degradation of organic compounds such as bisphenol A (BPA) and dyes.^50,51 However, HO˙ is highly active, and nonselective and tends to react with organic solvents; O₂˙⁻ can be used for the partial or selective oxidation of substrates in organic solvents.^33,52–54

The reaction of organic matter with O₂˙⁻ may involve radical adduct formation, hydrogen abstraction, and electron transfer. Transferable hydrogen atoms on the substrates can be readily removed by O₂˙⁻ (e.g., dihydrophenazine to phenazine radical, N-methylhydrophenazine to N-methylphenazine radical).⁵⁵ When O₂˙⁻ reacts with organic compounds via hydrogen abstraction, it generates carbon-based radicals that can further combine with oxygen to form peroxy intermediates that will decompose further. The nucleophilic substitution reaction of O₂˙⁻ was first reported in 1970,^56,57 and O₂˙⁻ can react with alkyl halides via nucleophilic substitution to break the C–Cl bond to give dechlorinated products.⁵⁸ Besides, O₂˙⁻ is found to also react with α-keto, α-hydroxy, and α-halo carbonyl compounds via nucleophilic addition at the carbonyl carbon, and the ensuing oxidative cleavage gives carboxylic acids.⁵⁹ As a nucleophile, O₂˙⁻ can attack the carbonyl carbon of esters to yield carboxylic acid anions and alcohols, and the carbonyl carbon of acyl halides to yield diacyl peroxides.⁴² Interestingly, O₂˙⁻ may also act as a reducing agent in one-electron transfer reductions, and reactions involving electron transfer from O₂˙⁻ to organic substrates (e.g., nitroblue tetrazolium) have been exploited to detect or quantify O₂˙⁻.³³

2.3 Singlet oxygen (¹O₂)

There have been reports on using ¹O₂ in the synthesis of chemicals and natural products since the 1980s.⁶⁰ Energy transfer to O₂ could create singlet oxygen (¹O₂). ¹O₂ has paired electrons of opposite spins (Fig. 1), manifesting higher reactivity compared with the ground-state O₂. As the unstable, excited state of molecular oxygen, ¹O₂ has a lower redox potential (0.81 V vs. NHE) than O₂˙⁻ and HO˙. The paired electrons in the external orbital of ¹O₂ have antiparallel spins.⁶¹ The reaction rate of ¹O₂ with organic compounds generally ranges in 10⁴ to 10⁷ M⁻¹ s⁻¹. The half-life of ¹O₂ (¹Δ_g state) is several tens of milliseconds in air but only as short as 3 μs in water.³³ A growing body of work has suggested that the redox potential of ¹O₂ does not change significantly with the pH value, so that the ¹O₂ can work in a wide pH range and exhibit considerable tolerance to environmental interference. Note that the half-life of ¹O₂ (¹Δ_g state) is several tens of milliseconds in air but only as short as 3 μs in water.³³ This very short half-life and the short diffusion distance (∼220 nm) restrict the reactivity of ¹O₂ to the proximity of where it is formed.⁶² As a result, it is less effective to rapidly remove organic pollutants in an aqueous environment by ¹O₂.^58,63

Nevertheless, ¹O₂ can be exploited as a moderate oxidant for selective or partial oxidations in chemical synthesis,⁴³ such as the selective oxidation of various organic substrates including amines, alcohols, olefins, and sulfides.^31,64–66 Studies have shown that ¹O₂ can oxidize unsaturated organic compounds by electrophilic attack and electron abstraction. For instance, the electrophilic attack of ¹O₂ to dioxin and furan gives endoperoxide intermediates,⁶⁷ and the photosensitized oxygenation of phenol proceeds via hydrogen abstraction by ¹O₂.⁶⁸ Because ¹O₂ is an excited state of O₂, it may be deactivated to the original stable ³O₂ without undergoing chemical reactions or electron transfer.

2.4 Surface lattice oxygen (O^2−x)

The lattice oxygen (O_L) of transition metal oxides (TMO) can directly oxidize organic matter through the Mars–van Krevelen (MvK) redox cycle.^69,70 In the MvK mechanism, the substrate is firstly oxidized by the O_L in the oxides, leaving behind an oxygen vacancy (OV) on the TMO surface that will be subsequently oxidized by O₂, i.e., the stoichiometric oxidant, to regenerate O_L. The reactivity of O_L depends strongly on the formation of OVs from the oxides, the local coordination environment of the O_L in oxides, and the strength of the M–O bond, and is sensitive to defects (or impurities), the composition, the surface, the interface, etc.^71,72 Normally TMO with transition metals of high valence state exhibit superior O_L activity. However reactions based on HO˙, O₂˙⁻, and ¹O₂ normally proceed at room temperature and the reactivity weakens at elevated temperatures;^73,74 O_L can serve as an appealing oxidation reagent at elevated temperatures.

Because O_L is nucleophilic and has a lower redox potential, the oxidation of hydrocarbons with O_L can give products in which all C–C bonds are retained (partial oxidation), which is in clear contrast to the electrophilic oxidation (e.g., by HO˙) that cleaves all C–C bonds.⁷⁵ Thus, O_L is readily used for the selective oxidation of biomass for sustainable development, as it can avoid undesired oxidative degradation. The use of O_L is also key in soot combustion, the oxidation of CO and NO_x, and the thermal oxidation of pollutants such as volatile organic compounds (VOC).^76–78

3 Generation of reactive oxygen species

Manipulating the geometrical and electron states of various precursors (H₂O₂, O₂, persulfates, and TMO) is key to generating ROS. In recent years, engineering point defects (e.g., anionic vacancies) on crystals (e.g., TMO) has emerged as an effective strategy to create catalytically active surfaces that help generate ROS. The deficient structure can strongly adsorb and activate the precursor oxidants (e.g., H₂O₂, O₂, PMS, and PDS) by elongating the chemical bond (e.g., the O–O bond in H₂O₂).⁷⁹ The electrons conducted from the bulk to the surface of the crystal may be trapped in the vacancies to create electron-rich regions⁸⁰ as additional carriers for surface reactions to generate ROS.⁸¹ For catalytic reactions using O_L in oxides as the ROS, the defect structure of oxides can help create O_L of high reactivity by increasing the O_L mobility and/or by promoting O_L regeneration.

(1) Activation of H₂O₂: hydrogen peroxide (H₂O₂) is a unique, versatile, green oxidizing agent for AOPs, as it has a high active oxygen content and gives H₂O as the only reduction product. However, due to its low redox potential (1.77 eV), it does not readily oxidize organic compounds on its own,⁸² and Fenton chemistry is exploited to activate the H₂O₂ to generate ROS. The classic, well-known homogeneous Fenton reaction system based on Fe²⁺/H₂O₂ suffers from several drawbacks that impede large-scale applications, such as a narrow working pH range (mostly acidic), generation of iron-containing sludge, and limited activity.⁸³ These hurdles prompted efforts to investigate the heterogeneous activation of H₂O₂ with transition metal-based catalysts, with which the type and amount of ROS can be regulated through distinct surface reactions. Although the decomposition of H₂O₂ on the catalyst surface is always spontaneous in 25–286 °C,⁸⁴ the activation energy (20.93–96.30 kJ mol⁻¹) and the rate of ROS generation depend heavily on the surface structure of the catalysts.^85,86

(2) Activation of persulfates: peroxymonosulfate (PMS and HSO₅⁻) and peroxydisulfate (PDS and S₂O₈²⁻) are relatively strong oxidizers with redox potentials of 1.82 and 2.01 V, respectively. AOPs with persulfates were introduced for groundwater remediation in the late 1990s, because methods to activate persulfates are more abundant and the availability of persulfate salts reduces storage and transportation costs.⁸⁷ Additionally, PMS and PDS attain a higher radical yield, and the treatment efficiency is less dependent on the operational parameters (e.g., pH, initial peroxide loading, and background constituents). Moreover, the activation of PMS/PDS on transition metal-based catalysts is usually carried out under alkali conditions.⁸⁸ Hence, persulfate AOPs are good substitutes of Fenton chemistry with H₂O₂. The energy of the O–O bond in PMS is estimated to be in 140–213.3 kJ mol⁻¹, and the activation requires adequate energy input. Appropriate activation of PMS and PDS to give ROS is imperative to ensure the desired oxidation performance, since PMS and PDS react directly with the organic contaminants only at a low rate.⁸⁹ They can be activated with metal catalysts,⁹⁰ carbon nanotubes,⁹¹ biochars,⁹² TMO,⁹³etc.

(3) Activation of O₂: among all oxidants, oxygen is the most environmentally friendly, inexpensive, and accessible.⁹⁴ Due to the spin-flip restriction, molecular O₂ is normally at the ground triplet state (³O₂) and cannot directly engage in aerobic oxidation.⁹⁵ The activation of molecular O₂ into ROS (HO˙, O₂˙⁻, and ¹O₂) is normally a fundamental step in aerobic catalytic oxidation reactions.^96,97 However, the O=O bond of O₂ has a high bond energy (498 kJ mol⁻¹) that must be overcome to initiate O₂-mediated oxidation reactions. Transition metal-based catalysts (e.g., transition metal complexes,⁹⁸ TMO,⁹⁹ and transition metal sulfides¹⁰⁰) critically improve the reactivity of O₂ by effectively adsorbing the O₂ molecules and weakening the O=O bond of O₂ through efficient charge transfer, and photocatalysis is frequently used to promote electron transfer in the activation of O₂.¹⁰¹

(4) Lattice oxygen: metal oxides are the carrier of O_L and they maintain the lattice structure throughout the reaction. The MvK mechanism of O_L-mediating oxidation is well supported by isotope labelling, as using ¹⁸O₂ as the stoichiometric oxidant will install ¹⁸O in the lattice of the metal oxides.¹⁰² Heating is generally required to activate O_L and regenerate O_L from O₂, which can increase cost and make the reaction procedure tedious. The reactivity and regeneration of O_L can be tuned by regulating the strength of the M–O bond (M: metal) and the chemical states of the metal, to enable stronger oxidation performance at a lower temperature.¹⁰³ Anionic vacancies on the metal oxides change the electronic state and the surface structure, and they can improve the adsorption of O₂, reduce the energy of the M–O bond, modify the electron density distribution, change the ion transport in the crystal, and expedite the migration of O_L.¹⁰⁴ Doping metal oxides with heteroatoms is an appealing method to create such vacancies.

3.1 Hydroxyl radical

3.1.1 From H₂O₂. HO˙ can be deemed a reduction product of H₂O₂. The main ROS in the Fenton reaction involving H₂O₂ is the HO˙ radical generated in the Haber–Weiss cycle. A traditional homogeneous Fenton system requires a large input of Fe²⁺ (18–410 mmol L⁻¹) and excess H₂O₂ (30–6000 mmol L⁻¹) in the redox process to produce enough HO˙ for practical applications.¹⁰⁵ Moreover, it generates iron sludge, has limited activity, and requires a specific working pH range. In the search for heterogeneous Fenton systems as efficient alternatives, researchers developed transition metal-based catalysts,^106,107 clay-based catalysts,¹³ and zeolite-based catalysts^108–111 to accomplish Fenton chemistry, and external energy sources such as UV/visible irradiation, electricity, and ultrasound have also been used to boost the generation of HO˙.^112,113 The decomposition of H₂O₂ over transition metal-based catalysts to create ROS deserves special comments because these catalysts have high stability, low cost, earth abundance, and potentially high catalytic activity. The surface OVs of deficient TMO such as TiO_2−x can activate H₂O₂ to give HO˙.¹¹⁴ Zhang et al. reported that while the oxygen deficient TiO_2−x readily activates H₂O₂ to generate HO˙ to degrade methyl orange and p-nitrophenol, no HO˙ is detected in the pristine TiO₂/H₂O₂ system.¹¹⁵ Theoretical studies have been run to unveil how defect structures boost the H₂O₂ activation. For example, Density Functional Theory (DFT) calculations show that the adsorption of H₂O₂ is stronger on the deficient α-Fe₂O₃ (001) surface than on the pristine surface.¹¹⁶ The energy barrier of activation and the O–O bond length at the transition state are 3.65 kJ mol⁻¹ and 1.60 Å for H₂O₂ adsorbed on the α-Fe₂O₃ surface with OVs, but 44.89 kJ mol⁻¹ and 1.75 Å on the pristine α-Fe₂O₃ surface. Analogously, introducing S vacancies (SVs) in Fe₃S₄ (FS-100) creates strong peroxidase-like activity not available from SV-poor Fe₃S₄ (FS-0), which can be attributed to the efficient adsorption of H₂O₂ and its subsequent decomposition to HO˙ at the SVs of Fe₃S₄ (Fig. 2a).¹¹⁷ The adsorption energy of H₂O₂ is −7.15 eV on the (011) surface of SVs-free Fe₃S₄ but −6.01 eV for SVs-rich Fe₃S₄ (Fig. 2b). The length of the O–O bond is 1.515–1.525 Å for H₂O₂ on SVs-free Fe₃S₄ but 3.151–4.491 Å for H₂O₂ on SVs-Fe₃S₄ (Fig. 2c). In the case of deficient CeO₂, the key factor to the efficient generation of HO˙ from H₂O₂ is the electron density of Ce adjacent to the OVs, and the strong adsorption of H₂O₂ at the metastable OVs on the (110) surface of CeO₂ alone does not guarantee the subsequent activation of H₂O₂ (Fig. 3).¹¹⁸ Metastable OVs can be available for the catalytic cycle only when the electron density of the surface Ce is high enough to break the O–O bond of the adsorbed H₂O₂. In other words, a synergy exists between the geometry and the electronic state of the OVs to help adsorb and activate H₂O₂. The incorporation of Co into ZnO creates OV-rich Co–ZnO as a Fenton-like catalyst that features both electron-rich OVs with unpaired electrons and electron-deficient Co³⁺ sites.¹¹⁹ The electron-rich OVs help capture and reduce H₂O₂ to generate HO˙ that can quickly degrade pollutants. The electron-deficient Co³⁺ sites degrade the adsorbed organic pollutants through the electron transfer from the organics to Co³⁺, which helps offset the imbalanced electron distribution in the doped oxide.

Fig. 2 (a) Models of Fe₃S₄ and SVs-rich Fe₃S₄ with exposure of the (011) surface. (b) Energy diagrams of H₂O₂ adsorption on SVs-rich Fe₃S₄via end-on and side-on attachment models. (c) The O–O bond length of adsorbed H₂O₂ in different adsorption configurations. Adapted with permission from ref. 110 Copyright (2021) Elsevier.

Fig. 3 Schematic illustration of the synergy effect of Ce and OVs on the CeO₂ surface for the efficient activation of H₂O₂ into HO˙. Adapted with permission from ref. 118 Copyright (2020) American Chemical Society.

3.1.2 From O₂. Generating HO˙ directly from O₂ has also been demonstrated. For example, low-valence Mn ions can act as an efficient Fenton-like reagent to reduce atmospheric O₂ to H₂O₂, which can subsequently generate HO˙ (Fig. 4).¹²⁰ Catalyst surfaces with anionic vacancies can create HO˙ from O₂ because of enhanced adsorption and electron migration. According to DFT calculations, less energy is required to break the O=O bond to generate HO˙ when more electrons are transferred to O₂.¹²¹ Setvin et al. used scanning tunnelling microscopy to pull the vacancies of an anatase crystal to its surface and proved that the surface OVs of TiO₂ can serve as electron donation sites to activate the adsorbed O₂.¹²² In an example that is beyond the use of transition metals, Zhao et al. reported that the O₂ adsorbed on the surface of Pt SA/MgO with OVs can dissociate more easily to generate HO˙.¹²³ The presence of OVs notably decreases the adsorption energy of O₂ (E_ads = −3.25 eV), elongates the O=O bond (from 1.363 to 1.552 Å), and decreases the energy barrier of activating O₂ (down to 2.48 eV). The O₂ molecule adsorbed on the surface of Pt SA/MgO gives two adsorbed oxygen atoms [denoted as *O] after the O=O bond is broken, and a further reaction between *O and adsorbed H₂O gives HO˙ (i.e., O₂ → 2*O, *O + H₂O → 2HO˙).

Fig. 4 (a) Detection of H₂O₂ by activation of O₂ over Mn₃O₄ and Mn₂O₃, respectively. (b) Schematic illustration of the recycling strategy for Mn₃O₄ and Mn₂O₃ on O₂ activation to generate H₂O₂ that decomposes into HO˙. Adapted from ref. 120 with permission from The Royal Society of Chemistry.

3.1.3 From PMS and PDS. The cleavage of peroxide bonds in persulfates by one-electron reduction could create SO₄˙⁻ based on eqn (1) and (2).⁸⁸ The SO₄˙⁻ could react with H₂O to generate HO˙ viaeqn (3),^124–126 and the reaction rate constant of SO₄˙⁻ to give HO˙ can be significantly accelerated under alkaline conditions (e.g., at pH = 13.0) as described by eqn (4).¹²⁷

Mⁿ⁺ + HSO₅˙⁻ → M⁽ⁿ⁺¹⁾⁺ + SO₄˙⁻ + OH⁻ (1)

Mⁿ⁺ + S₂O₈²⁻ → M⁽ⁿ⁺¹⁾⁺ + SO₄˙⁻ + SO₄²⁻ (2)

SO₄˙⁻ + H₂O → HO˙ + SO₄²⁻ + H⁺ (3)

SO₄˙⁻ + OH⁻ → HO˙ + SO₄²⁻ (4)

3.2 Superoxide radical

3.2.1 From H₂O₂. The Fenton reaction that activates H₂O₂ may also create O₂˙⁻.⁶¹ Haber and Weiss proposed that in the Fenton reaction under neutral pH, the oxidation of Fe²⁺ by O₂ and the decomposition of H₂O₂ can give O₂˙⁻.¹²⁸ The generation of O₂˙⁻ can also be detected when deficient catalysts are used.¹²⁹ Introducing OVs to TMO can favor interfacial electron transfer to the adsorbed H₂O₂,¹³⁰ and the O–O bond of H₂O₂ elongated by OVs is more easily cleaved under neutral and even alkaline conditions.¹³¹ Wei et al. demonstrated enhanced oxidative behavior of a TiO₂–H₂O₂ system that can be attributed to the activation of H₂O₂ by TiO₂,¹³² and noted that O₂˙⁻ can be stably adsorbed on the TiO₂ surface but HO˙ cannot (Fig. 5). Wu et al. generated O₂˙⁻ and HO˙ from H₂O₂ using the single electron-trapped oxygen vacancies of bulk TiO₂ without any light irradiation,¹³⁰ although the mechanism of H₂O₂ activation and the role of ROS in the enhanced oxidation ability are still debatable.

Fig. 5 Side view of (a) H₂O₂ (b) initial configuration of two hydroxyl groups, and (c) superoxide radicals adsorbed on the TiO₂ predicted by the GGA-TS method. (d) The decomposition mechanism of H₂O₂ into superoxide radicals on TiO₂ NSs. Adapted with permission from ref. 132 Copyright (2017) American Chemical Society.

3.2.2 From O₂. Using O₂ or air as the precursor of O₂˙⁻ is preferred over using H₂O₂ because it is more environmentally friendly and more economically beneficial.^133–135 Photocatalysis is generally exploited to create O₂˙⁻ by injecting a photogenerated electron into O₂. Upon light irradiation, the photogenerated electrons and holes have the probability to recombine into excitons due to the strong Coulomb interaction. The defect structure prevalent in advanced catalysts can create defect levels in the band structure of catalysts, serving as the electron capturing center, to promote charge carrier separation and transfer to the catalytically active sites.³¹ Therefore, photocatalysts with engineered defects with stronger electron migration can enhance O₂ activation in the catalytic reaction.¹³⁶ For example, the OVs of blue TiO₂ with localized electrons can facilitate the activation of O₂ to generate O₂˙⁻ under visible-light irradiation.¹³⁷ OVs with localized electrons are the sites for the adsorption and activation of O₂, and the adsorption of O₂ on the perfect (101) surface of TiO₂ is much weaker. The adsorbed O₂ accumulates 1.1 e from the localized electrons of two unsaturated Ti atoms near the OVs of the TiO₂ surface and is activated into O₂˙⁻via a single-electron transfer pathway. In a similar manner, O₂˙⁻ is generated from the reduction of dissolved O₂ in water with the electrons trapped in the vacancy sites on the surface of the highly reactive CuFe₂O₄ catalyst, and O₂˙⁻ has stronger reactivity when it is on the solid surface than in the aqueous solution.¹³⁸ In another example, Co₃O₄ with OVs promotes the adsorption and reduction of O₂, and O₂˙⁻ can be detected over the surface of Co₃O₄.¹³⁹ In the case of Ni–Fe layered double hydroxide (NiFe-LDH), OVs facilitate the charge carrier transfer of NiFe-LDH, and activate O₂ into O₂˙⁻ for the selective oxidation of NO into nitrate (Fig. 6).¹⁴⁰ For the plasmonic catalyst of Au supported on OVs-BiOCl,¹⁴¹ the OVs on BiOCl facilitate the trapping of plasmonic hot electrons, which are transferred to the adsorbed O₂ to give O₂˙⁻ for the aerobic oxidation of benzyl alcohol. Although the presence of point defects could promote charge separation for enhanced ROS generation,²⁸ it must be noted that excessive OVs can be detrimental by trapping the electrons and restraining the electron mobility.¹⁴² Besides the modulation of charge transfer, the introduction of point defects could create additional defect levels rendering either a downshift of the conduction band (CB) minimum of catalysts, or an upshift of the valence band (VB) maximum.^143,144 It has been reported that the OVs can lead to a down shift of the CB band of BiO_2−x, which benefits the electron transfer to O₂ molecules for the generation of O₂˙⁻ efficiently.¹⁴⁵

Fig. 6 Schematic illustration of activation of O₂ into O₂˙⁻ on NiFe-LDH with an OV for the selective photocatalytic oxidation of NO. Adapted with permission from ref. 140 Copyright (2022) American Chemical Society.

In addition to OVs, S or N vacancies on the catalyst surface have also been exploited to tune the band structure and charge carrier transfer efficiency for O₂ activation on photocatalysts. For example, N vacancies can create electron-accumulated centers to activate O₂ over the Pt/C₃N₄ catalyst (Fig. 7b).¹⁴⁶ The length of the O=O bond is 1.496 Å for the activated O₂ over the N-doped catalyst but 1.260–1.365 Å in the control (Fig. 7a). The SVs on In₂S₃ nanosheets can activate O₂ into O₂˙⁻via enhanced electron transfer under visible light irradiation.¹⁴⁷ Theoretical calculations reveal that in the presence of the SVs, the density of states (DOS) of In₂S₃ is significantly enhanced at the valence band maximum (Fig. 7c and d). Consequently, electrons can be photoexcited to the conduction band and transferred to the defective surface more easily, and the defective In₂S₃ shows stronger interactions with O₂ than pristine In₂S₃ (Fig. 7e and f).

Fig. 7 (a) The structures of O₂ adsorbed on pure C₃N₄ (001), C₃N₄ (001) + Pt, Pt (111), and VN2c + Pt (gray, purple, white, and red represent C, N, Pt, and O). (b) Partial density of states (PDOS) of the adsorbed O₂ (black and red lines represent the O 2s and 2p orbitals, respectively). Adapted with permission from ref. 146 Copyright (2021) Springer Nature Group. Calculated DOS of (c) S-vacancy and (d) pristine In₂S₃ slabs (E_f = Fermi level; orange shading indicates the increased DOS around the VBM). (e) The adsorption of O₂ at the S vacancy of the deficient In₂S₃ slab. (f) O₂ floating on the pristine In₂S₃ slab. Adapted with permission from ref. 147 Copyright (2022) American Chemical Society.

While the vacancies can directly inject localized electrons into O₂ to elongate the O=O bond and activate O₂, the geometry of the vacancies may tune the pathway of ROS generation by changing the band structure of catalysts. There are two different crystallographic positions of the oxygen atoms in the [BiO]₂²⁺ layer of (BiO)₂CO₃ (referred to below as BOC), and Rao et al. constructed two kinds of OVs (i.e., OVs1 and OVs2) in BOC by breaking Bi1-O1 or Bi2-O2 from two distinct lattice oxygens (O1 and O2) binding to the Bi atoms (Fig. 8a–c).¹⁴⁸ Both OVs1 and OVs2 extend the lifetime of photogenerated charge carriers in their respective photocatalytic systems and promote the efficient separation of carriers (Fig. 8d and e). While both OVs1 and OVs2 enhance light absorption in the visible region by introducing an intermediate level in the band gap, only OVs1 can enhance light absorption in the infrared region (Fig. 8f). The photogenerated charge carriers in OVs1 react with the adsorbed O₂ and H₂O to form O₂˙⁻ and HO˙, respectively. In contrast, the observed redshift indicates that OVs2 decreases the optical bandgap energy of BOC, and HO˙ cannot be generated on the catalyst surface with OVs2.

Fig. 8 The structural models of (a) BOC, (b) OVs1-BOC, and (c) OVs2-BOC, respectively. Contradistinction of DOS/PDOS (Fermi levels are set to 0 eV) between (d) BOC and OVs1-BOC and (e) BOC and OVs2-BOC. (f) Calculated optical absorption of BOC, OVs1-BOC, and OVs2-BOC. Adapted with permission from ref. 148 Copyright (2021) American Chemical Society.

3.2.3 From PMS and PDS. PMS and PDS can be activated to generate O₂˙⁻ as described in eqn (5)–(9). However, it has been suggested that O₂˙⁻ frequently serves as an intermediate for the generation of other ROS (e.g., ¹O₂), rather than directly oxidizing the pollutant.¹⁴⁹

HSO₅⁻ + H₂O → HSO₄⁻ + H₂O₂ (5)

SO₅²⁻ + H₂O → SO₄²⁻ + H₂O₂ (6)

H₂O₂ + HO˙ → H₂O + H⁺ + O₂˙⁻ (7)

S₂O₈²⁻ + H₂O → SO₅²⁻ + 2H⁺ + SO₄²⁻ (8)

S₂O₈²⁻ + HO₂⁻ → SO₄˙⁻ + H⁺ + O₂˙⁻ + SO₄²⁻ (9)

3.3 Singlet oxygen

3.3.1 From H₂O₂. The decomposition pathway of H₂O₂ may also be regulated to give ¹O₂.^150,151 While the superoxide radical (O₂˙⁻) is produced from the catalytic activation of H₂O₂, in many cases it can undergo further disproportionation to produce ¹O₂.¹⁵² This pathway however cannot ensure the efficient and selective generation of ¹O₂, as the Haber–Weiss reaction (O₂˙⁻ + H₂O₂ → HO˙ + OH⁻ + O₂) significantly reduces the yield of ¹O₂.^43,153 Sels et al. prepared a catalyst containing MoO₄²⁻ exchanged on LDH to efficiently produce ¹O₂ from H₂O₂.¹⁵⁴ Yi et al. developed a molybdenum (Mo) co-catalytic Fenton system and revealed that the exposed Mo⁶⁺ can oxidize O₂˙⁻ to ¹O₂ in a Fenton-like reaction (Fig. 9a).⁴³ Intriguingly, Yang et al. built a catalyst with ∼2 nm Fe₂O₃ nanoparticles distributed inside multiwalled carbon nanotubes (CNTs) to selectively activate H₂O₂ to give ¹O₂ (Fig. 9b).¹⁵¹ The Fe(III) species on the surface of the Fe₂O₃ inside the CNT is reduced by H₂O₂ to produce HO₂˙/O₂˙⁻, which is then oxidized to give ¹O₂ in two pathways: (i) oxidation by Fe(III) in the favored spin state, and (ii) radical–radical reactions including the recombination of HO₂˙/O₂˙⁻ (the Gibbs free energy for the HO₂˙/O₂˙⁻ recombination reaction is −6.4 kcal mol⁻¹ at pH = 7). Defect structures may boost the generation of ¹O₂ from H₂O₂. Li et al. disclosed the production of ¹O₂ through rapid radical reactions in the vacancies of Fe–Co Prussian blue analogues.¹⁵⁵ Chong et al. found in the FeCeO_x–H₂O₂ Fenton-like system that the OVs on the metal oxide surface can participate in the H₂O₂ → ¹O₂ conversion to give ¹O₂ as the primary ROS.¹⁵⁶

Fig. 9 (a) Schematic illustration of Mo catalytic oxidation of O₂˙⁻ to ¹O₂. Adapted with permission from ref. 43 Copyright (2021) American Chemical Society. (b) Schematic illustration of the generation of ¹O₂ over Fe₂O₃@FCNT-H/H₂O₂. Adapted with permission from ref. 151 Copyright (2019) National Academy of Sciences.

3.3.2 From O₂. Energy transfer from the catalyst to the ground state oxygen (³O₂) may give ¹O₂.¹⁵⁷ Anionic vacancies on catalysts can serve as active sites to chemically adsorb O₂ and generate ¹O₂.¹³⁶ Wang et al. found that OVs in OVs-Bi₂O₃ favor the activation of O₂ to form ¹O₂ in the dark.¹⁵⁸ Calculations show that O₂ is chemically adsorbed on the surface of OVs-Bi₂O₃ but only physically adsorbed on pristine Bi₂O₃. The O=O bond length and the O₂ adsorption energy are 1.51 Å and 0.5 meV for O₂ on pristine Bi₂O₃, but 1.23 Å and 1.66 meV for O₂ on OVs-Bi₂O₃ (Fig. 10a and b). The O₂ adsorbed on OVs-Bi₂O₃ can be activated to the excited state (i.e., ¹O₂) because both the π*↑ and π*↓ states of O₂ are occupied (Fig. 10e and f).¹⁵⁸ Upon light irradiation, a faster generation of ¹O₂ occurs over OVs-Bi₂O₃ compared with pristine Bi₂O₃, and OVs-Bi₂O₃ displays stronger capability on O₂ activation and thus a higher ¹O₂ generation rate (Fig. 10c and d). However ¹O₂ is the only ROS over OVs-Bi₂O₃; the OVs-free Bi₂O₃ produces other ROS (e.g., H₂O₂ and HO˙) along with ¹O₂.¹⁵⁸ Ji et al. ultrasonically treated commercial CoS₂, and the SVs in the resulting deficient CoS_2−x can tune the redox reaction on the CoS_2−x surface to release ¹O₂ in a sustained manner.¹⁵⁹ As abundant electrons are confined in the vacant electron orbitals of the SVs, O₂ can capture electrons from the SVs to form O₂˙⁻ through a one-electron process (eqn (10)) or form H₂O₂ through a two-electron process (eqn (11)). O₂˙⁻ can be quickly oxidized to ¹O₂ by the highly oxidative Co⁴⁺ or Co³⁺ exposed on the S deficient surface (eqn (12)).

O₂ + e⁻ (defect) → O₂˙⁻ (10)

O₂ + 2e⁻ (defect) → H₂O₂ (11)

O₂˙⁻ + Co⁴⁺ (or Co³⁺) → ¹O₂ + Co²⁺ (12)

Fig. 10 Comparison of O₂ adsorption on the (120) surface of (a) pristine Bi₂O₃ and (b) OVs-Bi₂O₃ (O₂, blue; O atom of Bi₂O₃, red; Bi, purple); the total DOS of (c) Bi₂O₃ and (d) OVs-Bi₂O₃ with O₂; and partial DOS plot for O₂ absorbed on (e) Bi₂O₃ and (f) OVs-Bi₂O₃ surface; the up-spin states are represented by red curves and the down-spin states by blue curves. Adapted with permission from ref. 158 Copyright (2020) Wiley-VCH Verlag GmbH & Co. KGaA. Adapted with permission from ref. 159 Copyright (2021) Wiley-VCH Verlag GmbH & Co. KGaA.

3.3.3 From PMS and PDS. With traditional catalysts, the generation of SO₄˙⁻ and HO⁻ is the dominant process in the activation of PMS, and the production of O₂˙⁻ and ¹O₂ is just a competing process. However, with a suitable catalyst, the activation of PMS may also selectively generate ¹O₂ and O₂˙⁻ (Fig. 11b).¹⁶⁰ The activation energy for the self-decomposition of PMS in water to give ¹O₂ (HSO₅⁻ + SO₅²⁻ → SO₄²⁻ + HSO₄⁻ + ¹O₂) is low,^161,162 but the rate constant of this reaction is also low (0.2 M⁻¹ s⁻¹) and the selective generation of ¹O₂ is thus challenging. In activating PMS, OVs with localized electrons serve as the active sites for the heterolysis of PMS to generate ¹O₂.¹⁶³ The OVs in the oxides can decrease the adsorption energy of PMS, elongate and weaken its O–O bond, and accelerate the decomposition process.¹⁶⁴ Zeng et al. activated PMS with OV-rich CoAl hydroxide@hydrosulfide to give ¹O₂ as the main ROS to degrade sulfamethoxazole.¹⁶⁵ Zhao et al. found that the OV-rich AgFe_1−xNi_xO₂ selectively activates PMS into O₂˙⁻ and ¹O₂ rather than SO₄˙⁻ or HO˙.¹⁶⁶ Compared with AgFeO₂ having a relatively perfect lattice, the OV-rich AgFe_1−xNi_xO₂ has a more suitable redox potential for interacting with PMS and thus a lower energy barrier for the decomposition of PMS. Gao et al. reported that the evolution of ¹O₂ over perovskite is well correlated with the OV concentration.¹⁶⁷ The activation energies of PMS self-decomposition to give ¹O₂ are 48.4, 22.6, and 15 kJ mol⁻¹ for LaFeO₃, LaFe_0.6Zn_0.4O₃, and LaMnO₃, respectively.

Fig. 11 (a) Proposed mechanism of OVs-mediated activation of peroxydisulfate to create ¹O₂. Adapted with permission from ref. 168 Copyright (2021) American Chemical Society. (b) Schematic illustration of how the Fe single-atom catalyst (Fe¹/CN) activates PMS to generate ¹O₂. Adapted with permission from ref. 160 Copyright (2021) Wiley-VCH Verlag GmbH & Co. KGaA. (c) Activation of PDS over manganese oxides. Adapted with permission from ref. 169 Copyright (2018) American Chemical Society.

Bu et al. activated PDS to generate ¹O₂ on the deficient surface of BiOBr.¹⁶⁸ The OVs of BiOBr interact with hydroxyl ions (HO⁻) to form Bi^III–OH species, which are the major active sites for the adsorption and activation of PDS. The PDS adsorbed at the Bi^III–OH sites produces the metastable Bi^IV–O⁽⁻¹⁾–O⁽⁻¹⁾–SO₃⁻ intermediate, which reacts with another PDS to release O₂˙⁻ when the Bi^IV–O bond breaks. In this process, the O atom bonded to Bi^IV acts as the electron donor, and Bi^IV acts as the electron acceptor. Finally, Bi^(IV−x)–O⁽⁻¹⁾–O⁽⁻¹⁾–SO₃⁻ reacts with O₂˙⁻ to produce ¹O₂ under alkaline conditions (Fig. 11a).

Many studies have suggested the direct oxidation of O₂˙⁻ as a common route to generate ¹O₂,^43,169,170 and the oxidation of O₂˙⁻ to ¹O₂ requires an oxidant with a redox potential of at least 0.34 V vs. NHE. Hu et al. revealed that the OV concentration of Co₃O₄ impacts its efficiency in activating PMS to give ¹O₂.¹⁷¹ The PMS decomposition mainly depends on the adsorption energy of PMS on the Co₃O₄ surface, the cleavage of the O–O bond in the PMS molecule, and the electron transfer between PMS and Co₃O₄. The direct oxidation of O₂˙⁻ to give ¹O₂ may account for the generation of ¹O₂ in view of the thermodynamics, i.e., E₀ (Co³⁺/Co²⁺) = 1.93 V and E₀ (O₂˙⁻/¹O₂) = −0.34 V. Prior studies have also shown that the oxidization of O₂˙⁻ to ¹O₂ is thermodynamically feasible with Cu^III and Mn^IV (Fig. 11c),^169,170 as E₀ (Cu^III/Cu^II) = 2.3 V vs. NHE and E₀ (Mn^IV/Mn^III) = 0.95 V vs. NHE. The reduction of Mo⁶⁺ by O₂˙⁻ during the Fenton reaction can also give ¹O₂, i.e., Mo⁶⁺ + O₂˙⁻ → ¹O₂ + Mo⁴⁺.⁴³

3.4 Lattice oxygen in TMO: reactivity and regeneration

Just like HO˙, O₂˙⁻ and ¹O₂, the O_L of TMO itself can also serve as the ROS to participate in AOPs. Since O_L already exists stably in TMO, it is not newly generated from a precursor, but it needs to be constantly replenished in the oxidation reaction as described in the MvK mechanism. TMO with highly reactive O_L generally have low OV formation energy that arises from a weak M–O bond and strong oxygen migration ability in the TMO lattice.¹⁰⁴ Hayashi et al. found that β- and λ-MnO₂ are good catalysts for the oxidation of hydroxymethyl furfural (HMF).¹⁷² Compared with α- and γ-MnO₂, β- and λ-MnO₂ have lower OV formation energies due to their distinct crystal structures, and for both α- and γ-MnO₂ the OV formation energies are higher at the planar oxygen sites than at the bent oxygen sites (Fig. 12a).

Fig. 12 (a) Calculated OV formation energies of MnO₂ of distinct phase structures. Adapted with permission from ref. 172 Copyright (2019) American Chemical Society. (b) Relationship between the Mn–O bond force constant and the relative OV content and the initial reaction rates of various Mn₃Co₂O_x–zVC catalysts. (c) Top views of the (111) plane of the Mn–Co spinel oxide with one OV and two neighboring OVs. Adapted with permission from ref. 173 Copyright (2021) American Chemical Society.

The OVs of deficient TMO can regulate the coordination structure of the TMO and alter the reactivity of O_L.¹⁰³ Liu et al. presented a facile and green vitamin C-assisted solid-state grinding method to synthesize mesoporous Mn–Co spinel oxides with a well-defined OV concentration.¹⁷³ The Mn–Co oxides with a higher OV concentration have a weaker Mn–O bond and higher O_L reactivity (Fig. 12b). According to DFT calculations and in situ FT-IR studies, the removal energy of O_L decreases significantly from 3.53 to 2.67 eV when there is an adjacent OV (Fig. 12c). That is, the O_L becomes more active when OVs exist in its surrounding. The analyses of Raman spectroscopy, H₂-TPR, and O₂-TPD show that the Mn₃Co₂O_x–0.3VC catalyst with a higher OV concentration has stronger O_L reactivity than Mn₃Co₂O_x.

Doping is a common strategy to enhance the reactivity of O_L, as doping TMO with cations (e.g., Li⁺, Mg²⁺, Ca²⁺, Sr²⁺, Al³⁺, Zn²⁺, and Ln³⁺) can decrease the OV formation energy. Low-valence dopants may create OVs and enhance the reactivity of O_L.²⁷ Xu et al. prepared Li⁺-doped NiO which has remarkably enhanced O_L reactivity than pristine NiO.¹⁷⁴ The OV formation energy is 3.32 eV for pure NiO but only 2.12 eV for Li⁺-doped NiO. Schlexer et al. introduced Au to TiO₂ to capture the excess electrons associated with the vacancies formed at the Au/TiO₂ perimeter to stabilize the vacancies.¹⁷⁵ The OV formation energy at the Au/TiO₂ perimeter is reduced by 1.32 eV with respect to the Au-free surface. Dostagir et al. prepared a catalyst with isolated Co atoms in ZrO₂ and demonstrated with DFT calculations that the OV formation energy of ZrO₂ is decreased by Co doping to between −1.45 and −1.17 eV from 2.70–3.20 eV.¹⁷⁶ That is, OV formation is endothermic on pure ZrO₂ but becomes exothermic after Co doping. OV sites are created near Co atoms due to mismatch in the oxidation states of Co²⁺ and Zr⁴⁺ and thus facilitate the need for charge balance.

Doping may also enhance the O_L reactivity by inducing structural distortion in the oxides. Zhao et al. found that the Raman peaks of Ni-doped Co₃O₄ shift to lower frequencies and become broader because lattice distortion weakens the Co–O bond.¹⁷⁷ Consequently, more OVs are created in the oxide, the O_L has stronger mobility, and the active surface O_L become more abundant. Zhang et al. showed that the [FeO₆] octahedral distortion in La_1−xCe_xFeO₃ promotes the mobility and reactivity of O_L,¹⁷⁸ suggesting that it is feasible to tailor the reactivity of O_L in perovskites by modulating the distortion of BO₆ in ABO₃ (Fig. 13a).

Fig. 13 (a) The OV concentration and possible reaction mechanism of POM-CO₂ splitting over La_1−xCe_xFeO₃ (x = 0, 0.5, 1). Adapted with permission from ref. 178 Copyright (2020) American Chemical Society. (b) OV formation energy (E_Ov) and adsorption energy of O₂ (E_O2_-ads) over distinct Zr sites. Adapted with permission from ref. 179 Copyright (2018) American Chemical Society. (c) Isotopic ¹⁸O₂ exchange analysis of Bi₂Zr₂O₇ at 300 °C and 400 °C. Adapted with permission from ref. 102 Copyright (2021) American Chemical Society.

The efficient regeneration of O_L can also benefit from OVs. OVs can enhance the adsorption of O₂ and the storage of O₂²⁻ on the TMO surface in the MvK process.¹⁸⁰ Lu et al. incorporated Fe into CuO–CeO₂ to create the Fe–O–Ce structure with rich OVs and measured in situ the oxygen storage capacity (OSC) of the oxides by the CO pulse technique.¹⁸¹ The OVs readily trap the O₂ in the gas phase, which ensures a durable release of subsurface lattice oxygen to supply reactive O_L. The initial CuO–CeO₂ catalyst has a low OSC value of 46.4 μmol CO₂ per g, whereas the Fe-doped CuO–CeO₂ has a considerably higher OSC value of 94.8 μmol CO₂ per g thanks to the greater number of OVs introduced by Fe doping. Similarly, Wang et al. assembled ultrathin and vacancy-rich CoAl-LDH nanosheets prepared from exfoliating bulk CoAl-LDH with graphite oxide (GO) to obtain the CoAl-ELDH/GO hybrid catalyst that possesses abundant newly generated OVs.¹⁸² The OVs promote oxygen adsorption and storage over the deficient CoAl-ELDH/GO, which has larger O₂-TPD peaks than the corresponding bulk CoAl-LDH. Yang et al. showed with DFT calculations that O₂ can be strongly adsorbed and activated on the defective Mn-doped c-ZrO₂ (111) surface with OVs.¹⁷⁹ The OVs elongate the O=O bond from 1.24 Å on the perfect Mn-doped c-ZrO₂ (111) surface to 1.37 Å on the defective surface. The OV formation energy and the adsorption energy of O₂ are 5.84 eV and 0.37 eV on the perfect m-ZrO₂ (−111) surface, 0.18 eV and −0.05 eV on the perfect Mn-doped c-ZrO₂ (111) surface, and down to 0.18 eV and −0.74 eV on the defective Mn-doped c-ZrO₂ (111) surface, respectively (Fig. 13b).

Generally, there are two possible pathways (i.e., R¹ and R² mechanism) for the gas phase O₂ to oxidize the OVs and replenish O_L. In the R¹ mechanism, the exchange of oxygen species occurs at one surface vacant site. In the R² mechanism, two adjacent surface vacant sites are needed to adsorb and decompose the gaseous O₂. Feng et al. ran isotopic ¹⁸O₂ tracing experiments and demonstrated that the adsorption and activation of gas phase O₂ on Bi₂B₂O₇ mainly follow an R² mechanism that requires two adjacent surface vacancies (Fig. 13c).¹⁰²

3.5 Tailoring the reactive oxygen species generation

PMS, PDS and H₂O₂ are strong oxidants but normally have low reaction rates with most organic pollutants. Catalysts bearing transition metal ions (e.g., Fe, Mn, Co, Ni, V, Ce, etc.) are promising activators for these oxidants to give powerful ROS.¹⁸³ Transition metal catalysts can provide a diversity of metals with tunable valence states to afford high redox properties, and an appropriate d-band center to balance the catalyst-adsorbate binding strength, as well as controllable surface and electronic structures to promote light absorption and/or adsorbate adsorption, all crucial for tailoring and boosting the generation of ROS. Note that the catalysts containing different types of transition metals are frequently found to give the same type of ROS.^16,184–186 Hence, it should not just be the type of transition metal ions controlling the selectivity and efficiency of ROS generation.¹⁸⁷ Recent studies have shown that modulation of the geometrical and electronic structures of catalysts is a key to probe how to customize the ROS production.

Tuning the geometrical structure of catalysts can change the binding effect of catalysts and oxidants and monitor the ROS generation. For instance, transition metal based catalysts commonly drive the dissociation of PMS and PDS into SO₄˙⁻ and HO˙.¹⁶⁹ It has been reported that the catalysts of CoN₄ structure prefer to adsorb the O atom connected to S in the PMS, and the system produces a mixture of ROS including SO₄˙⁻, HO˙, O₂˙⁻ and ¹O₂.^188,189 In comparison, when Co is coordinated with N to form a CoN_[2 + 2] structure, CoN_[2 + 2] can bind the O from –OH in the PMS, while the loss of the H atom of –OH creates a SO₅˙⁻ intermediate.^188,189 Since SO₅˙⁻ has a high reaction rate (∼2 × 10⁸ M⁻¹ s⁻¹) and low activation energy (7.4 ± 2.4 kcal mol⁻¹), it can rapidly transfer into ¹O₂viaeqn (13)–(15).^188,189 Intriguingly, on the deficient surface of BiOBr the PDS can be activated into ¹O₂.¹⁶⁸ The OVs of BiOBr interact with HO⁻ to form Bi^III–OH species and produce the Bi^IV–O⁽⁻¹⁾–O⁽⁻¹⁾–SO₃⁻ intermediate, which reacts with another PDS to release O₂˙⁻, while Bi^(IV−x)–O⁽⁻¹⁾–O⁽⁻¹⁾–SO₃⁻ reacts with O₂˙⁻ to produce ¹O₂ under alkaline conditions (Fig. 11a).

HSO₅⁻ → SO₅˙⁻ + H⁺ + e⁻ (13)

SO₅˙⁻ + SO₅˙⁻ → S₂O₈²⁻ + ¹O₂ (14)

SO₅˙⁻ + SO₅˙⁻ → 2SO₄²⁻ + ¹O₂ (15)

Tuning the electronic structure of photocatalysts can boost the generation and transfer of electrons for catalytic reactions. Wu et al. employed OVs to mediate ¹O₂ generation in the Fe–Co LDH/PMS system, and found that PMS tends to donate one electron to the OV and is thus activated to form ¹O₂.¹⁹⁰ When oxidation reactions involve photocatalytic reactions, a key to tailor ROS generation is by tuning the band and electronic structure of photocatalysts. Introducing point defects (e.g. OVs) could create additional defect levels of photocatalysts, rendering a downshift of the conduction band (CB) minimum, and/or an upshift of the valence band (VB) maximum.^143,144 With an appropriate band structure, the electrons in the defect levels could transfer to the CB band through inter-band excitation and be captured by O₂ that generates O₂˙⁻,¹⁹¹ while upshift of the VB band may help suppress the oxidation of H₂O to HO˙. For instance, CdS and ZnIn₂S₄ catalysts have appropriate conduction and valence bands for O₂˙⁻ generation while avoiding HO˙ generation.¹⁹² The presence of OVs could introduce intermediate energy states, leading to a narrowed band gap and enhanced charge carrier separation efficiency,^193,194 and helps extend the light response to visible or even near infrared light,^195,196 Wang et al. showed that OV-rich sulfur-doped BiOBr has a narrowed band gap responsible for the visible light and can generate O₂˙⁻ and HO˙ efficiently for 4-chlorophenol degradation.¹⁹⁷

To date, it remains an open question to establish robust principles to develop effective transition-metal-based catalysts for fine control over ROS generation. The roles and mechanism responses of different metal groups in distinct reaction pathways await to be disclosed, by performing more systematical investigations under the assistance of operando spectroscopic studies and theoretical calculations.

4 Detection of reactive oxygen species

4.1 Hydroxyl radical

4.1.1 EPR measurement. The spin trapping method is a common method to detect ROS. Electron Paramagnetic Resonance (EPR) is a useful and facile method for the direct detection of free ROS at concentrations as low as 1 μM.¹⁹⁸ DMPO (5,5-dimethyl-1-pyrroline noxide) often severs as a trapping reagent to convert HO˙ into more stable DMPO-HO˙,¹⁹⁹ which displays quartet EPR signals with a relative intensity of 1 : 2:2 : 1.²⁰⁰

4.1.2 Spectroscopic measurement. Spectroscopic detection strategies, including absorbance (UV/Vis), fluorescence (FL) and chemiluminescence (CL), offer an alternative approach to identify ROS.²⁰¹ Simple molecules such as terephthalic acid,²⁰² coumarin²⁰³ and carboxyl derivatives of coumarin²⁰⁴ are normally employed to detect HO˙ as they can generate the corresponding fluorescent derivate in solution. For instance, HO˙ can react with terephthalic acid to produce 2-hydroxyterephthalic acid (TAOH),²⁰⁵ and the amount of HO˙ is quantified by analyzing the fluorescence intensity of TAOH (λ_exc = 320 nm; λ_em = 420 nm). Likewise, the coumarin (COU) and coumarin-3-carboxylic acid (3-CCA) reacting with HO˙ produce fluorescent derivates, i.e., 7-hydroxy-coumarin (λ_exc = ∼320–370 nm; λ_em = ∼450 nm) and 7-hydroxy-3-carboxycoumarinic acid (λ_exc = ∼380–400 nm; λ_em = ∼450 nm), respectively.²⁰⁶

4.2 Superoxide radical

4.2.1 EPR measurement. DMPO (5,5-dimethyl-1 pyrroline N-oxide) can also serve as the trapping reagent to detect O₂˙⁻ in aqueous solution during ESR measurements. DMPO-O₂˙⁻ normally displays an EPR signal with four major peaks of a relative intensity of 1 : 1:1 : 1.²⁰⁷ As the reaction rate of DMPO with O₂˙⁻ is much slower than that with HO˙ in aqueous solution, and DMPO-O₂˙⁻ is more stable than DMPO-HO˙ in methanol,²⁰⁸ it is preferred to perform the EPR measurements of O₂˙⁻ in methanol solvent.

4.2.2 Spectroscopic measurement. Additives such as nitro blue tetrazolium (NBT, 2,2′-di-p-nitrophenyl-5,5′-diphenyl-(3,3′-dimethy)-4,4′-bisphenyleneditetrazolium chloride) could be applied to detect O₂˙⁻.³³ O₂˙⁻ can react with yellow NBT to create blue formazan, leading to the decrease of the absorption peak of NBT at 260 nm in UV-visible spectra, as an indicator of the amount of O₂˙⁻.^209,210 Luminol and MCLA (6-(4-methoxyphenyl)-2-methyl-3,7-dihydroimidazo-[1,2-a]pyrazin-3-one hydrochloride) can also be used to trace O₂˙⁻via chemiluminescence (CL) analysis,^211,212 and the CL strategy could detect the trace amount of O₂˙⁻.²¹²

4.3 Singlet oxygen

4.3.1 EPR measurement. Sterically hindered cyclic amines can react with ¹O₂ to generate the 1-oxyl radical for identifying ¹O₂ in EPR analysis. For instance, 2,2,6,6-tetramethylpiperidine (TEMP) reacting with ¹O₂ gives a stable nitroxide radical (2,2,6,6-tetramethylpiperidine 1-oxyl, TEMPOL),²¹³ which displays EPR spectra with a 1 : 1 : 1 triplet signal.²¹⁴ For the detection of ¹O₂, the use of 4-oxo-tetramethylpiperidine,²¹⁵ and 4-hydroxy-tetramethylpiperidine²¹⁶ has also been reported.

4.3.2 Spectroscopic measurement. Both fluorescence and colorimetric methods can be used to detect ¹O₂. A singlet oxygen sensor green (SOSG) having a weak blue fluorescence peak with λ_em = 395 nm at λ_exc = 372 nm, and λ_em = 416 nm at λ_exc = 393 nm can be used for ¹O₂ detection. The reaction of ¹O₂ with SOSG produces endoperoxide that gives green fluorescence with λ_em = 525nm at λ_exc = 504 nm.²¹⁷ Besides, researchers have confirmed that 1,3-diphenylisobenzofuran (DPBF) can react with ¹O₂ giving endoperoxide (DPBF-EP),²¹⁸ and the amount of ¹O₂ is proportional to the decrease of DPBF which has absorption intensity at about 410 nm in UV-vis spectroscopy analysis.^219,220

4.4 Quenching technique

When two or more ROS are generated over the catalyst, a quenching experiment can be employed to identity the dominant ROS in the AOP system. Since many alcohols (e.g., isopropanol (IPA),²²¹ mannitol,²²²t-butyl alcohol (TBA)²²³ and methanol²²⁴) can efficiently react with HO˙, these substances could be good candidate reagents to quench HO˙. It is worth noting that methanol can also work as the scavenger for quenching SO₄˙⁻,²²⁵ while TBA is only effective for quenching OH˙ but not for SO₄˙⁻.²²⁶

Benzoquinone (BQ),^227,228 ascorbic acid,²²⁸ superoxide dismutase (SOD)²²⁹ and trichloromethane²³⁰ can be used to remove O₂˙⁻ in oxidation reactions. For instance, in many cases, O₂˙⁻ may serve as an intermediate for the generation of other ROS (e.g., ¹O₂). To find out whether O₂˙⁻ is the intermediate of ¹O₂ or the dominant ROS for pollutant degradation, Yi⁴³ and Zhang¹⁸⁹ introduced BQ as a quenching agent of O₂˙⁻, and showed that O₂˙⁻ served as an intermediate to form ¹O_2, the active species in their studies.

Useful quenching agents for ¹O₂ include NaN₃,²³¹ furfur alcohol (FFA),²³²L-tryptophan²³³etc. However, since FFA or NaN₃ can also react with free radicals,²³⁴ it is infeasible to use FFA/NaN₃ as a scavenger when the oxidation system contains both the radical and non-radical ROS. In addition, NaN₃ can react with PMS rapidly, and hence, it cannot apply to the PMS-involved system. Besides the quenching technique, sometimes, the solvent exchange method that replaces H₂O by D₂O can help identity the generation of ¹O₂, as the lifetime of ¹O₂ in D₂O is about ∼55 μs, more than 10 times that (∼4.2 μs) in H₂O.^235,236

5 Reactions of reactive oxygen species

5.1 Hydroxyl radical

5.1.1 Scavenging pollutants. Because of its high oxidation potential, HO˙ can react with the strong bonds of persistent organic matter (e.g., C–H bond in benzene) and oxidize tropospheric trace species or pollutants to CO₂ and H₂O.^237,238 It can react at a relatively fast rate with easily oxidizable organic compounds (e.g., methylene blue, promazine, and promethazine) through one-electron abstraction,²³⁹ or at a relatively slower reaction rate with saturated hydrocarbons (e.g., aliphatic compounds without double C=C bonds) by abstracting hydrogen atoms.²⁴⁰ Because of its electrophilic nature, HO˙ prefers to react with the C=C and C=N bonds in the organic substrate rather than the C=O bond whose carbon atom is electron-deficient.³⁹ Nevertheless, it can react with aliphatic carboxylic acids to create highly reactive carbon-centered radicals via abstraction of the hydrogen atom, and react with aromatic carboxyl acids via hydroxylation.

The surface defects of TMO can effectively enhance the oxidation performance of HO˙. For instance, the Co–Cu LDH with abundant OVs can boost the dissociation of H₂O₂ to generate a sufficient amount of HO˙ for the degradation of anthraquinones.²⁴¹ The highest COD and TOC removal values (89.9% and 71.3% respectively) are achieved with an initial H₂O₂/TOC ratio of 75 and an initial pH of 6.8. Zhang et al. examined a ceria-based heterogeneous Fenton-like catalyst.²⁴² Compared with pure iron oxide and ceria, Fe-doped CeO₂ with abundant OVs (FeCeO_x) degrades rhodamine B (RhB) at a higher efficiency (98%) over a wider working pH range (3.0–9.0). The surface HO˙ is the predominant ROS to oxidize RhB. Annealing the FeCeO_x in an oxygen atmosphere eliminates the OVs, and the annealed FeCeO_x with fewer OVs has poorer performance in activating H₂O₂ and degrading the target pollutant. Zhan et al. prepared OV-rich Co–ZnO microparticles (OV–CoZnO MPs) as a Fenton-like catalyst.¹¹⁹ The electron-rich OV sites on the catalyst are the key active sites responsible for the capture and reduction of H₂O₂ to generate HO˙ to degrade organic pollutants (e.g., BPA, PHT, 2,4-D, phenol, and methylene blue) over a wide pH range (4.5–9.5). In addition, the electron-deficient Co³⁺ sites of the catalyst appear to actively adsorb the pollutant, and the organic pollutant adsorbed on Co³⁺ acts as an electron donor to further accelerate the degradation process. The OV-CoZnO MPs/H₂O₂ system with dual reaction centers can degrade pollutants about ∼17 times faster than the ZnO/H₂O₂ system that has only one reaction center.

Li et al. performed quantitative analysis and reported an interesting case that shows the selectivity of HO˙ in the degradation of pollutants (Fig. 14e and f).²⁴³ The OVs of the BiOCl catalyst demonstrate an electron donor nature and serve as the “Fenton-catalytic” center to dissociate H₂O₂ to generate HO˙, but it occurs in a surface confinement pathway. The generated HO˙ quickly diffuses away from the (001) surface of BiOCl, because the steric hindrance due to the high density of O atoms on the (001) surface prevents the binding of newly generated HO˙ (Fig. 14a–d). However, free HO˙ prefers to oxidize dissolved pollutants (e.g., benzoic acid, benzene, and phenol) in the solution (Fig. 14g); the nascent HO˙ on the (010) surface of BiOCl prefers to bond and stay on the catalyst surface to react with strongly adsorbed pollutants (e.g., formic acid, rhodamine B, and methyl orange) (Fig. 14g).

Fig. 14 Schematic illustration of H₂O₂ dissociation on (a) BOC-001, (b) BOC-010, (c) BOC-001-OV and (d) BOC-010-OV, respectively. (e) Schematic illustration of H₂O₂ dissociation on different surfaces of BiOCl with OVs for scavenging benzoic acid and formic acid, respectively. (f) EPR spectra of spin-reactive HO˙ radicals generated over BOC-001-OV and BOC-010-OV in the presence of H₂O₂, respectively. (g) Reaction kinetic constants of BOC-001-OV and BOC-010-OV for the removal of group I and group II pollutants. Adapted with permission from ref. 243 Copyright (2017) American Chemical Society.

5.1.2 Generating value-added chemicals. The selective oxidation of organic substrates to value-added products is of prime importance in organic synthesis but can require the use of expensive and toxic dichromate or permanganate reagents. Although HO˙ typically mineralizes organic pollutants completely, it may also engage in selective oxidations to give value-added chemicals. For instance selective oxidation of alcohols to aldehydes remains one of the most important functional transformations in the synthesis of fine chemicals. Aldehydes such as benzaldehyde have been widely used in the synthesis of perfumes, dyes, and pharmaceuticals.²⁴⁴ However, the selective oxidation of alcohols to aldehydes without overoxidation to carboxylic acids remains challenging. Liu et al. found that photogenerated holes from the valence band of reconstructed carbon nitride nanosheets react with OH⁻ to form HO˙ as the ROS, which in turn selectively oxidizes phenylcarbinol to give phenyl methanal via radical intermediates.²⁴⁵ Luo et al. used ultrathin Co-based LDH and graphene to modify BiVO₄ photoanodes to accomplish photoelectrochemical (PEC) water oxidation, where HO˙ is generated to selectively oxidize benzyl alcohol to benzaldehyde (Fig. 15).²⁴⁶ Isotope-labeling experiments reveal that HO˙ is generated from the water oxidation reaction and binds to the LDH surface via hydrogen bonding. In situ electron spin resonance confirms that HO˙ is the main ROS responsible for the selective oxidation of benzyl alcohol. In addition, FT-IR spectroscopy shows that the catalyst surface preferentially adsorbs benzyl alcohol and readily desorbs the generated benzaldehyde, which is helpful for selective oxidation.

Fig. 15 (a) Mechanism of selective benzyl alcohol (BA) activation coupled with PEC water oxidation. (b) ESR spectra of HO˙ generated over the graphene@ultrathin-CoAl-LDH@BiVO₄ under different conditions (H₂O₂, 30%, 0.1 mM; DMSO, 0.5 mM). (c) Benzaldehyde yield (solid squares and lines) and selectivity (empty squares and dashed lines) at different reaction times in the PEC oxidation of BA under illumination (AM 1.5G, 100 mW cm⁻²) at 1.2 V vs. RHE. Solution: 2 mmol BA in PBS (pH = 7, 0.1 M, 10 mL) at 10 mV s⁻¹ scan rate. Adapted with permission from ref. 246 Copyright (2020) American Chemical Society.

5.2 Superoxide radical

5.2.1 Scavenging pollutants. Although O₂˙⁻ is less reactive than HO˙,³³ in recent years it has been frequently exploited as a green “aqua cleaner” for scavenging anthropogenic pollutants. Watts and coworkers pioneered in using O₂˙⁻ to degrade halogenated pollutants.^248–250 Zhang et al. found that the direct degradation of p-nitrophenol involves O₂˙⁻ rather than HO˙ as the main ROS, and the degree of O₂˙⁻ participation during the direct degradation mainly depends on the molecular structure of the pollutant.²⁵¹ The photocatalytic degradation of tetracycline over ZnO involves O₂˙⁻ as the primary ROS, as it is generated at much higher concentrations than HO˙ on the catalyst surface.²⁵²

As a free radical, O₂˙⁻ can work as both an electron transfer agent and an electron acceptor.²⁵³ Xiao et al. studied how the unpaired electrons fit into the resonance π* molecular orbitals in O₂˙⁻ and found that the HOMO and LUMO of O₂˙⁻ are degenerated.²⁵⁴ Thus, O₂˙⁻ is both electrophilic and nucleophilic, and can work as either a reducing agent or an oxidizing agent.^255,256 Because the carboxyl group is weakly electron-withdrawing and the hydroxyl group is electron-donating, the efficiency of degradation by O₂˙⁻ ranks as hydroquinone < p-hydroxybenzoic acid < p-nitrophenol, which is the exact opposite order of their electron cloud density.²⁵¹

Defect structures on the catalyst can help generate O₂˙⁻ for AOP. In the degradation of ciprofloxacin over W-doped CeO₂, W is beneficial for the formation of surface OVs, and the deficient surface effectively adsorbs and activates O₂ to O₂˙⁻ through a one electron transfer process.²⁵⁷ The 2% W/CeO₂ can degrade ciprofloxacin in a wide pH range with excellent efficiency. In some cases, O₂˙⁻ and other species in the reaction medium react with the substrate concurrently. For instance, in the oxidation of BPA over AgFe_1−xNi_xO₂, the dominant ROS are O₂˙⁻ and ¹O₂, rather than the traditional SO₄˙⁻ or HO˙.¹⁶⁶ Analogously, the MnO₂-mediated degradation of anthracene is mediated by both the non-radical oxidation by Mn³⁺ and the radical-based oxidation by O₂˙⁻.²⁵⁸

It is worth remarking that the geometry of O₂˙⁻ on the catalyst surface may affect the outcome of the oxidation reaction. Li et al. engineered deficient BiOCl that can oxidize NO completely into nitrate under visible light irradiation with >99% selectivity (Fig. 16a).²⁴⁷ When the reaction is carried out at ambient temperature, the carefully crafted OVs on the prototypical (001) surface of BiOCl preferentially give O₂˙⁻ in a side-on bridging mode and suppress the thermodynamically stable terminal end-on geometry that will be predominant at an elevated temperature (Fig. 16b–e). This newly developed O₂˙⁻-mediated technique thus avoids the partial oxidation of NO to NO₂.

Fig. 16 (a) Schematic illustration of photocatalytic NO removal by BiOCl-OV. (b) Free energy change against the reaction coordinate for the oxidation of NO by O₂˙⁻ on the BiOCl (001) surface in different geometries. (c) Geometric transition from peroxynitrite to nitrate. TS represents the transition state. Charge density difference and O₂ partial DOS of the BiOCl (001) surface adsorbed with (d) O₂ and (e) nitrate. The yellow and blue isosurfaces with an isovalue of 0.005 au represent charge accumulation and depletion in space. The vertical dashed line in the DOS shows the VBM. Adapted with permission from ref. 247 Copyright (2018) American Chemical Society.

5.2.2 Generating value-added chemicals. Thanks to its moderate oxidation potential, O₂˙⁻ allows the selective oxidation of organic substrates (e.g., olefins, alcohols, and amines) to give valuable chemical products.^259–261 Ito et al. thermally oxidized styrenes to benzaldehydes with H₂O₂ over TiO₂ without additional light irradiation and found that the hydroperoxy radical (˙OOH) and O₂˙⁻ are responsible for the selective oxidation.²⁶² Xie et al. oxidized aliphatic alcohols to aldehydes with O₂ at room temperature and found that BiOCl attains >90% conversion and nearly 100% selectivity.²⁶³ The BiOCl photocatalyst is excited to produce h⁺/e⁻ pairs. The adsorbed alcohol reacts with the photo-generated holes and subsequently deprotonates to form a carbon radical, which combines with O₂˙⁻ to form aldehydes. The adsorbed protons decompose the C–O and O–O bond to form aldehyde and H₂O₂. Chen et al. employed OV-rich Bi₂MoO₆ to selectively oxidize benzyl alcohol with O₂ to produce both benzaldehyde and H₂O₂ simultaneously.²⁶⁴ According to in situ ESR measurements, both carbon-centered radicals and O₂˙⁻ are key intermediates. The concentration of OVs is readily tunable, and the rate of benzaldehyde formation is 1310 μmol g⁻¹ h⁻¹ with OV-rich Bi₂MoO₆, 2.3 times that of OVs-poor Bi₂MoO₆. The OVs on Bi₂MoO₆ accelerate charge separation and charge transfer, and they also enhance the adsorption and activation of both benzyl alcohol and O₂.

Enhanced generation of O₂˙⁻ for selective photooxidation can also be accomplished on the defective surface with SVs. Sun et al. carried out a photocatalytic selective oxidation to convert alcohols to aldehydes with high efficiency and selectivity using defective In₂S₃ nanosheets (Fig. 17a and b).¹⁴⁷ The SVs-rich-In₂S₃ activates O₂ into O₂˙⁻via electron transfer, and the aldehyde selectivity is ≥98% for all alcohol substrates when SVs-rich In₂S₃ is used. For the oxidation of benzyl alcohol, the conversion rate of the substrate under ambient conditions is only 30% with SVs-poor In₂S₃ but 71% with SVs-rich In₂S₃ (Fig. 17c).

Fig. 17 (a) Proposed mechanism of the photooxidation of benzyl alcohol to benzaldehyde. (b) ESR spectra of SV-rich and SV-poor In₂S₃ in the presence of DMPO. (c) The dramatically different oxidation rates of the indicator molecule 3,3′,5,5′-tetramethylbenzidine in various gas environments (oxygen, air, and nitrogen) indicate that the ROS evolved from O₂. Adapted with permission from ref. 147 Copyright (2019) American Chemical Society.

The selective oxidation of primary amines to imines is essential to synthesize dyes, fragrances, fungicides, pharmaceuticals, and agrochemicals.^265–268 Imines are normally synthesized in industry by the direct oxidation of amines or by the condensation of amines with aldehydes, which can require using stoichiometric amounts of harmful or hazardous reagents. Khampuanbut et al. utilized a WO₃/BiOBr photocatalyst for oxidizing amines to imine, for which O₂˙⁻ is the main ROS and the OVs on WO₃/BiOBr promote the interfacial charge transfer.²⁶⁹ Carrying out the reaction at room temperature under an O₂ atmosphere with visible light irradiation converts nearly 74.1% amines into imines within 4 h. Shi et al. used the phenol–TiO₂ complex as an efficient photocatalyst for the selective oxidation of amines to imines with the aid of O₂ under visible light irradiation.²⁷⁰ Both primary and secondary amines can be oxidized to imines at >80% conversion rate in 30 min, and O₂˙⁻ again proves to be the main ROS.

5.3 Singlet oxygen

5.3.1 Scavenging pollutants. Singlet oxygen is about 94 kJ mol⁻¹ more energetic than the ground state O₂ and has a mild redox potential (E₀ = 2.2 V vs. NHE). Herzberg first defined ¹O₂ as the singlet oxygen in the excited state, but the importance of ¹O₂ remained overlooked until 1964 when scientists showcased its use in chemical oxidation.^271,272 Although ¹O₂ is not as strong as HO˙ for the depletion of wastewater COD or TOC, the AOP based on ¹O₂ generally use smaller amounts of oxidants and are less susceptible to radical scavengers.^158,273,274 Luo et al. used scavenger experiments and EPR analyses to prove that ¹O₂ is the primary ROS for the degradation of organic matter such as BPA (Fig. 18).²⁷⁵ Zhu et al. activated persulfate on β-MnO₂ nanorods to degrade phenol and verified that the ROS responsible for the oxidation is ¹O₂ instead of O₂˙⁻.¹⁶⁹ Liu et al. found that ¹O₂ is the main ROS for the degradation of phenol over a wide pH range (pH = 3.5–9) in the CPANI-9/PMS system.²⁷⁶ Sun et al. activated PDS with FeCO₃ and found that among the generated ROS (¹O₂, SO₄˙⁻, and HO˙), ¹O₂ is key to the initial degradation rate of sulfadiazine and it contributes more to the oxidation reaction at higher pH.²⁷⁷

Fig. 18 (a) Schematic illustration of BPA degradation. (b) EPR spectrum of PMS activation. (c) Effect of D₂O on BPA removal efficiency. Adapted with permission from ref. 275 Copyright (2019) Elsevier.

Zhao et al. showed that ¹O₂ plays a dominant role in the removal of BPA catalyzed by deficient ZnCoO_x.²⁷⁸ Bu et al. reported the OVs-mediated activation of PDS with BiOBr to degrade BPA, for which ¹O₂ is the main ROS under alkaline conditions.¹⁶⁸ The removal ratio of TOC reaches 57.3% after 15 min in the reaction system (pH = 11.5), suggesting that BPA can be effectively mineralized into CO₂ and H₂O by ¹O₂. Li et al. developed La_1.15FeO₃ with rich surface OVs.²⁷⁹ The catalyst has a high Fenton activity (0.0402 min⁻¹) for activating H₂O₂ to oxidize methyl orange, and the ¹O₂ formed at the surface OVs is the main ROS. Gao et al. found that the piperazinyl, oxazinyl, and carboxylic substituents of ofloxacin are readily attacked by the ¹O₂ generated from PMS with perovskites.¹⁶⁷ Both the amount of OVs of the perovskites and the degradation efficiency of ofloxacin fall in the order of LaFeO₃ < LaZnO₃ < LaMnO₃ < LaNiO₃.

5.3.2 Generating value-added chemicals. Many studies use ¹O₂ as an environmentally benign, versatile, and moderate ROS for the selective oxidation of organic substances,^169,280 as ¹O₂ shows high affinity for organic substrates with electron-rich moieties (e.g., olefins, sulfides, and amines).^281,282

Chen et al. developed a metal–organic framework based on Pt nanocrystals and porphyrin, denoted as Pt/PCN-224(M), to selectively oxidize aromatic alcohols to the corresponding aldehydes with O₂ (Fig. 19a).²⁸⁰ Benzyl alcohol is oxidized with 1 atm O₂ at ambient temperature (16 °C) in water under visible light irradiation and the reaction is complete within 50 min. The 100% aldehyde selectivity can be attributed to ¹O₂ as the main ROS, and extending the reaction time does not produce benzoic acid (Fig. 19b and c). Li et al. demonstrated the enhanced photocatalytic generation of ¹O₂ with Mn–NCs@SiO₂–Pt and the selective oxidation of primary benzylic alcohols to aldehydes.²⁸³ The energy transfer between the Mn-doped nanocrystals and molecular oxygen is made efficient thanks to the long half-life of the Mn excited state (in the order of milliseconds). Selective oxidization of glycerol to dihydroxyacetone (DHA) is a high-revenue chemical that has wide range of applications, e.g., as a sun tanning agent in the cosmetics industry and as a precursor of pharmaceuticals or a building block for fine chemical synthesis, and can combine with lactic acid for producing biodegradable polymers.²⁸⁴ Zhao et al. developed the Bi/Bi_3.64Mo_0.36O_6.55 heterostructure for the photocatalytic selective oxidization of glycerol.²⁸⁵ The presence of OVs in Bi/Bi_3.64Mo_0.36O_6.55 promotes the mass production of ¹O₂ as the ROS for the selective oxidation, and glycerol is oxidized in water to 1,3-dihydroxyacetone at ambient temperature (25 °C) with high selectivity (97–99%). Selective oxidation of organic sulfides has also sparked increasing attention as the resultant sulfoxides are key intermediates for the synthesis of agrochemicals, and pharmaceuticals as well as other valuable fine chemicals.²⁸⁶ Wang et al. demonstrated the highly efficient and selective photo-oxidation of phenyl methyl sulfide to phenyl methyl sulfoxide catalyzed by OVs-Bi₂O₃.¹⁵⁸ The OVs in OVs-Bi₂O₃ can activate O₂ into ¹O₂ in the dark, and the generation of ¹O₂ is accelerated over both OVs-Bi₂O₃ and Bi₂O₃ when there is light irradiation. OVs-Bi₂O₃ has a stronger capability for the selective oxidation of phenyl methyl sulfide to phenyl methyl sulfoxide owing to its higher ¹O₂ generation rate.

Fig. 19 (a) Schematic illustration showing the singlet oxygen-engaged selective oxidation of alcohols over Pt/PCN-224(M) using molecular oxygen under visible-light irradiation. (b) ESR spectra of the samples after mixing 4-oxo-TMP with PCN-224(Zn), Pt NCs, and Pt/PCN-224(Zn) in the absence or presence of carotene under visible-light irradiation or in the dark. (c) Oxidation yield of benzyl alcohol vs. the reaction time over Pt/PCN-224(Zn) in the presence or absence of light irradiation or external heating with other reaction parameters remaining identical. Adapted with permission from ref. 280 Copyright (2017) American Chemical Society.

In addition, the ¹O₂-mediated oxidation of HMF, a key biomass-derived intermediate, can give value-added products such as 2,5-furandicarboxylic acid (FDCA), which is an important monomer substitute for petrochemically-derived terephthalic acid to make polyethylene furanoate and other polymers (Fig. 20a).^287,288 Current technology for the oxidation of HMF to FDCA is costly as it normally needs to be run with noble metal catalysts at elevated temperatures (80–150 °C), thus needing high energy consumption and high oxygen pressure.²⁸⁹ Xu et al. created a photocatalyst of cobalt thioporphyrazine (CoPz) dispersed on g-C₃N₄ for the oxidation of HMF to FDCA under simulated sunlight (Fig. 20b and c).²⁸⁷ The reaction is performed at ambient temperature and air pressure, and the FDCA yield is 96.1%.

Fig. 20 (a) Possible mechanism of the photocatalytic oxidation of HMF with the CoPz/g-C₃N₄ catalyst. (b) EPR signals of the DMPO-HO˙ adduct in water in the presence of bulk g-C₃N₄ and the TMP-¹O₂ adduct in water in the presence of CoPz/g-C₃N₄ in the photocatalysis process. (c) Photocatalytic performance. Adapted with permission from ref. 287 Copyright (2017) American Chemical Society.

5.4 Surface lattice oxygen

5.4.1 Scavenging pollutants. Earth-abundant TMO have been used as good catalysts for environmental remediation applications. The oxidation capacity of OL is weak at a relatively low temperature (<200 °C), and O_L-mediated oxidations are generally carried out at elevated temperatures and a high partial pressure of oxygen.

Catalytic oxidation is the most appealing pathway to process effluent streams with dilute VOC (<0.5 vol%).^290,291 Conventional catalysts required for the oxidation of HCHO²⁹² normally contain noble metals (e.g., Pt, Au, Pd, and Ag) that are costly and poorly available, and cheap and thermally stable TMO catalysts are excellent alternatives although their efficiency may be somewhat lower. Zha et al. synthesized Co₃O₄ nanowires with abundant OVs grown in situ on the Ni foam (denoted as r-Co₃O₄ NW@Ni foam) as a stable catalyst for the highly efficient oxidation of HCHO to formate (Fig. 21a), and used the OVs-free Co₃O₄ NW@Ni foam and Co₃O₄ nanoparticles (Co₃O₄ NPs) to compare and characterize the catalytic performance.²⁹³ The T_10% (i.e., temperature at which HCHO conversion reaches 10%) of the catalysts ranks as r-Co₃O₄ NW@Ni foam (75 °C) < Co₃O₄ NW@Ni foam (100 °C) < Co₃O₄ NP (132 °C). According to in situ DRIFTs studies, HCHO and formate are slightly more easily adsorbed and formed on the OVs-free Co₃O₄ NW@Ni foam without O₂, but the formate species are more active on the surface of the r-Co₃O₄ NW@Ni foam (Fig. 21b). Further theoretical studies reveal that the abundant surface OVs reduce the adsorption energy of O₂, and r-Co₃O₄ NW@Ni can thus store more ROS and have a higher catalytic activity in HCHO oxidation (Fig. 21c). In another example, Ma et al. used MOF-derived Co₃O₄ with different shapes for the catalytic oxidation of o-xylene and studied how the surface twofold-coordinate lattice oxygen (O2f) creates higher catalyst reactivity (Fig. 21d).²⁹⁴ The rod-like Co₃O₄-R has a lower T_90% (i.e., temperature at which o-xylene conversion reaches at 90%) of 270 °C than the spherical Co₃O₄–S (about 290 °C). According to the OV formation energy and the adsorption energy of O₂, OVs are more easily formed and O₂ is more easily replenished on the Co₃O₄-R surface (Fig. 21e and f).

Fig. 21 (a) Schematic illustration of the catalytic oxidation of HCHO over r-Co₃O₄ NW@Ni with OVs; (b) In situ DRIFTs after HCHO adsorption over Co₃O₄ NW@Ni and r-Co₃O₄ NW@Ni catalysts as a function of temperature. (c) O₂ molecule adsorbed on the {111} surface of Co₃O₄ NW@Ni foam and r-Co₃O₄ NW@Ni catalysts. Adapted with permission from ref. 293 Copyright (2020) American Chemical Society. (d) Catalytic oxidation activities of o-xylene over MOF-derived Co₃O₄ with different shapes. (e) OV formation over the (220) and (311) surfaces of Co₃O₄, respectively; (f) side view of the adsorption configuration of O₂ over the (220) and (311) surfaces of Co₃O₄ with OVs. The Co and O atoms are in blue and red, respectively. Adapted with permission from ref. 294 Copyright (2021) American Chemical Society. (g) Formaldehyde conversion over reduced TiO₂ of different OV concentrations. Adapted with permission from ref. 296 Copyright (2019) Elsevier. (h) Activation of surface lattice oxygen induced by La vacancies and the amount of released surface O_latt (Dα-O₂  RO_latt) as a function of V_La ratio. Adapted with permission from ref. 297 Copyright (2021) Wiley-VCH Verlag GmbH & Co. KGaA.

Wang et al. performed catalytic HCHO oxidation with three kinds of Co₃O₄ with different OV concentrations and demonstrated the dependence of the O_L reactivity on the OV concentration.²⁹⁵ The reaction rate constant (k) is 0.14 min⁻¹ for the Co₃O₄ nanobelts with a high OV concentration, which is about 20 times that of Co₃O₄ nanoplates (0.0071 min⁻¹). However, excessive OVs may also impair reactivity. For example, He et al. found that the catalytic activity of reduced TiO₂ for HCHO oxidation ranks as R-TiO₂-30 > R-TiO₂-60 > R-TiO₂-15 > R-TiO₂-5 (Fig. 21g).²⁹⁶ In the catalytic reaction, O₂ is adsorbed on the surface of the reduced TiO₂, accepts electrons, and is transformed into active O_L to degrade HCHO. However, excessive OVs can reduce the adsorbed O₂ and trap electrons, and the formation of ROS from O₂ then becomes less effective.

Dong et al. developed a CoMn₂O₄ catalyst that has a lower activation energy (35.5 kJ mol⁻¹) for the oxidation of toluene than other metal oxides (Co₃O₄, MnO_x, and Co₃O₄/MnO_x).²⁹³ The obtained CoMn₂O₄ has a large surface area, rich cationic vacancy, and high mobility of oxygen species, all of which contribute to the high catalytic activity. According to in situ temperature-programmed experiments, the surface O_L induces the catalytic reaction, then gaseous O₂ is moved to the bulk phase lattice, and finally the bulk oxygen species is migrated to the surface to form more surface O_L species at 200–250 °C. Toluene oxidation occurs via the benzyl alcohol-benzoate-anhydride-acetate reaction pathway over spinel CoMn₂O₄, and the conversion of the surface anhydride is the rate-determining step, especially at 200–210 °C.

For perovskites of the ABO₃ general formula, modulation of A-site defects can create effective oxidation catalysts that can be used in clean air applications. Liu et al. regulated the La vacancies (V_La) in LaMnO_3.15 by simply introducing urea in the traditional citrate-based synthesis to examine the relationship between the creation of V_La and the activation of surface O_L (Fig. 21h).²⁹⁷ The LaMnO_3.15 catalyst with optimized V_La achieves toluene oxidation at above 220 °C which is 10–25 times faster than the LaMnO_3.15 prepared from the traditional urea-free citric acid sol–gel method. The easier charge transfer from Mn to O in the MnO₆ octahedron activates the surface O_L and enhances the redox properties of the LaMnO_3.15 perovskite, thus enabling superior performance.

5.4.2 Generating value-added chemicals. Chemists have long desired eliminating the use of oxidizing reagents that are expensive, toxic, or harmful in synthesis.²⁹⁸ In this regard, the O_L of oxides can serve as a mild oxidant to access specific products. For example, O_L is readily used in the oxidation of alcohols to aldehydes (RCH₂OH → RCHO). The selective oxidation of alcohols to aldehydes, one of the most fundamental oxidation reactions in organic synthesis for making drugs, fine chemicals, perfume, etc., can be complicated by the formation of the corresponding acids or esters. Sabaté et al. found that Ru-doped KMn₈O₁₆ can efficiently catalyze the selective oxidation of alcohols to aldehydes with nearly 100% selectivity.²⁹⁹ Dong et al. designed a CeO₂ decorated Au/CNT to catalyze the aerobic oxidization of benzyl alcohol.³⁰⁰ The reaction gives benzaldehyde in 77.9% yield in 3 h under atmospheric pressure at 40 °C with 40 mL min⁻¹ O₂, and the O_L of CeO₂ is the ROS for the reaction. Dai et al. reported that the Bi₂₄O₃₁Br₁₀(OH)_δ photocatalyst with active O_L can selectively oxidize primary alcohols to aldehydes under visible light irradiation at room temperature using air.³⁰¹ The XPS analysis of the O1s spectra reveals that O_L is converted to the adsorbed water on the catalyst surface after the reaction. With the developed catalyst, (1) aromatic alcohols are oxidized considerably faster than aliphatic alcohols, (2) benzyl alcohols with different functional groups (e.g., p-NO₂, p-F, p-Cl, p-CH₃, p-OCH₃) are fully air-oxidized to the corresponding aldehydes with ≥99% selectivity within 4 h of illumination, and (3) there is no over-oxidation to carboxylic acids, CO, or CO₂ in all cases. Note that aldehydes are normally afforded from activated alcohols in which the carbon bears a phenyl group, such as benzyl alcohol,^302,303 probably because the electron-rich conjugated aromatic rings may lead to stronger adsorption on the catalyst surface.³⁰⁴

Conversely, O_L is also used in various cases for the thorough oxidation of alcohols to carboxylic acids (RCH₂OH → RCOOH). The aerobic oxidation of biomass to carboxylic acids under mild conditions is environmentally and economically advantageous to generate value-added chemicals from renewable feedstocks. Jin et al. found that Co-doped Mg₃Al(OH)_y(CO₃)_z can promote the aerobic oxidation of glycerol into tartronic acid,³⁰⁵ which is a high value-added chemical that has been widely used as a pharmaceutical and anti-corrosive protective agent, as well as for the monomer of biopolymers.³⁰⁶ The complete conversion of glycerol is achieved with 64% selectivity of tartronic acid under mild conditions (70 °C, 0.1 MPa O₂). Hayashi et al. used MnO₂ for the aerobic oxidation of HMF and obtained FDCA in 95% yield under mild conditions (3 eq. NaHCO₃, 10 bar O₂, 120 °C, 24 h).³⁰⁷ Han et al. prepared a mixed oxide MnO_x-CeO₂ (Mn/Ce = 6) catalyst via co-precipitation for the aerobic oxidation of HMF.³⁰⁸ After 15 h at 110 °C in the presence of KHCO₃, the catalyst gives 91% FDCA yield and 98% HMF conversion. Zhang et al. prepared a series of Mn–Co binary oxides for the catalytic oxidation of HMF to FDCA and found that the MnCo₂O₄ catalyst is significantly better than Mn₃O₄, Co₃O₄, and other Mn–Co binary oxides with different Mn/Co molar ratios.³⁰⁹ The MnCo₂O₄ catalyst has a Mn/Co molar ratio of 0.5, and gives 99.5% HMF conversion and 70.9% FDCA yield when the reaction is carried out at 100 °C under 10 bar O₂ with KHCO₃ for 24 h.

Reducing the OV formation energy increases the O_L reactivity. Hayashi et al. found that compared with α- and γ-MnO₂, β- and λ-MnO₂ have lower OV formation energies due to their distinct crystal structures, and the OV formation energies of α- and γ-MnO₂ are higher both at the planar oxygen sites and at the bent oxygen sites.¹⁷² The initial reaction rate per surface area of the catalysts for the aerobic oxidation of HMF to FDCA decreases in the order of β-MnO₂ (16.4 μmol h⁻¹ m⁻²) > λ-MnO₂ (12.2 μmol h⁻¹ m⁻²) > α-MnO₂ (7.6 μmol h⁻¹ m⁻²) ≈ γ-MnO₂ (7.4 μmol h⁻¹ m⁻²) > δ-MnO₂ (5.3 μmol h⁻¹ m⁻²), which agrees well with what is expected from computational results. Wan et al. found that the high OV concentration of CuMn₂O₄ promotes the mobility and adsorption of oxygen species in the oxidation of HMF to FDCA.³¹⁰ Liu et al. demonstrated with DFT calculations that an increasing OV concentration boosts the O_L reactivity of manganese oxide by reducing the OV formation energy, and also strengthens the adsorption and activation of O₂ to regenerate O_L by significantly cutting down the O₂ adsorption energy, ultimately increasing the catalytic activity for the oxidation of HMF.¹⁸⁴

Regulating the coordination environment of the metal also affects the activity of O_L. The coordination structures affect the M–O bond strength, and thu results in different mobilities of the lattice oxygen. From the Rietveld refinement of the X-ray diffractogram, Sabaté et al. determined that the interplanar spacing of the (100) plane of pristine KMn₈O₁₆ increased from 7.1 to 7.7 Å after doping with Ru.^299,311 With the lattice expansion and the elongation of the Mn–O bond, the Ru-doped KMn₈O₁₆ can more easily lose its O_L on the surface, thus allowing selective oxidation of alcohols to aldehydes.

Ishikawa et al. oxidized acrolein to acrylic acid with up to 98% selectivity by using orthorhombic Mo–V–Cu oxide as the catalyst.³¹² The location of Cu under the oxygen defect site moderates the oxygen activation and prevents side reactions such as C–C scission to form CO_x and acetic acid, thus improving the selectivity for acrylic acid.³¹³ The dispersion of noble metals (Au, Pt, Pd, Ru and Ag) can also regulate the coordination environment of TMO and generate O_L with high reactivity thanks to the enhanced electronic metal-support interactions. Li et al. reported that the O_L activated at the interfacial sites in Au₁/CeO₂ has lower OV formation energy (0.36 eV) and is more selective in oxidizing the alcohol substrate compared with the O_L in Au NPs/CeO₂.³¹⁴

Various studies have demonstrated reactions in which O_L is the ROS directly responsible for the oxidation, and the OVs on the catalyst surface commonly expedite the dissociation of the O–H bond in the alcohol substrate.^315,316 Wang et al. used ultrathin CoAl-ELDH to catalyze the oxidation of benzyl alcohol to benzaldehyde.¹⁸² The adsorption of benzyl alcohol is spontaneous and exothermic on the catalyst surface. The main active sites are OVs and V_Co–Co–OH^δ−, where alcohol can be captured through linkage via its –OH group and the –OH group can then be activated and oxidized by the O_L (Fig. 22). Dai et al. found that the O_L-mediated selective photo-oxidation of alcohols over Bi₂₄O₃₁Br₁₀(OH)_δ starts with the photo-dehydrogenation of alcohols to form adsorbed hydrogen species on the surface of the photocatalyst.³⁰¹ The O_L of Bi₂₄O₃₁Br₁₀(OH)_δ then oxidizes the alcohol to form aldehydes and OVs, and the OVs are subsequently healed by mild oxidants (O₂ or benzoquinone) to complete the photocatalytic cycle.

Fig. 22 DFT-calculated conformation of adsorption states and adsorption energies of a benzyl alcohol molecule on (a) an OV, (b) a V_Co–Co–OH^δ− site, and (c) defect-free Co–OH. Top views of the optimized structures of (d) an OV and (e) a Co vacancy. (f) Schematic diagram of Co-containing LDH for the aerobic oxidation of alcohols. Reproduced with permission. Adapted with permission from ref. 182 Copyright (2018) American Chemical Society.

While both ¹O₂ and O_L allow selective oxidations, oxidations mediated by O_L are typically carried out under conditions such as high temperature and high humidity. Catalysts may collapse or agglomerate to a certain extent in these harsh environments, which will decrease the specific surface area and the number of exposed active sites. As a result, it may not be possible to maintain high catalytic efficiency and the catalyst may become deactivated. In contrast, ¹O₂-mediated AOP are more favorable for selective oxidations; its catalytic conditions are milder, thus avoiding the agglomeration, collapse, and thus the deactivation of catalysts.

6 Challenges and perspectives

Various knowledge gaps remain in developing novel AOP and exploring the controllable reactivity of ROS. For example, although defect engineering has proven fruitful in accessing ROS for the desired AOP, some routes are sluggish and give interfering reactive species that do not contribute to the intended chemical transformation. To accomplish the rational design of catalysts and reactions, more understanding is needed regarding how the ROS contribute to the AOP and affect the selectivity and reactivity. To this end, in situ characterization techniques (e.g., Raman, FTIR, and EPR) may be helpful for the on-line detection of ROS, and monitoring the transition states of the ROS, as well as verifying the interactions of the ROS with the organic substrate and oxidation pathways.

Controversy still exists regarding how the structures of catalysts, particularly the ubiquitous point defects, participate in the AOP. While in many cases OVs engage directly as active defect sites to accelerate the catalytic reaction,³⁰⁵ at times the intrinsic properties of the catalyst altered by OVs are no less important.¹¹⁸ The selective generation of ROS toward the specified oxidation reaction is still a growing research area, and various questions remain to be answered. For example, given a certain TMO catalyst, what kind of selectivity should be expected? What is the mechanistic aspect that can account for differences in the catalyst's behavior? How do researchers exercise control in designing catalysts? More research is needed on how the surface structure of catalysts regulate the generation and reactivity of ROS for intended reactions.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported primarily by the National Key Research and Development Program of China (2020YFA0710303) and the National Natural Science Foundation of China (No. U1905215, 51772053 and 52072076). We thank Jay Wang for suggestions in writing.

Notes and references

1. C. Xu, E. Paone, D. Rodriguez-Padron, R. Luque and F. Mauriello, Chem. Soc. Rev., 2020, 49, 4273–4306.
2. H. Luo, J. Barrio, N. Sunny, A. Li, L. Steier, N. Shah, I. E. L. Stephens and M. M. Titirici, Adv. Energy Mater., 2021, 11, 2101180–2101231.
3. Q. Chen, D. Yang, L. Yu, X. Jing and Y. Chen, Mater. Horiz., 2020, 7, 317–337.
4. Y. Shang, X. Xu, B. Gao, S. Wang and X. Duan, Chem. Soc. Rev., 2021, 50, 5281–5322.
5. F. Franco, C. Rettenmaier, H. S. Jeon and B. R. Cuenya, Chem. Soc. Rev., 2020, 49, 6884–6946.
6. H. Dong, G. Wei, D. Yin and X. Guan, J. Hazard. Mater., 2020, 384, 121497.
7. L. Luo, T. Zhang, M. Wang, R. Yun and X. Xiang, ChemSusChem, 2020, 13, 5173–5184.
8. S. Wu, P. Wang, J. Qin, Y. Pei and Y. Wang, Adv. Funct. Mater., 2021, 31, 2102160.
9. J. L. Wang, X. C. Dai, H. L. Wang, H. L. Liu, J. Rabeah, A. Brückner, F. Shi, M. Gong and X. J. Yang, Nat. Commun., 2021, 12, 6840.
10. N. Zhang, Y. Zou, L. Tao, W. Chen, L. Zhou, Z. Liu, B. Zhou, G. Huang, H. Lin and S. Wang, Angew. Chem., Int. Ed., 2019, 58, 15895–15903.
11. B. H. Ko, B. Hasa, H. Shin, Y. Zhao and F. Jiao, J. Am. Chem. Soc., 2022, 144, 1258–1266.
12. K. Wang, C. Han, Z. Shao, J. Qiu, S. Wang and S. Liu, Adv. Funct. Mater., 2021, 31, 2102089.
13. Z. Z. Yang, C. Zhang, G. M. Zeng, X. F. Tan, H. Wang, D. l. Huang, K. H. Yang, J. J. Wei, C. Ma and K. Nie, J. Mater. Chem. A, 2020, 8, 4141–4173.
14. Q. Zhang, D. He, X. Li, W. Feng, C. Lyu and Y. Zhang, J. Hazard. Mater., 2020, 384, 121350.
15. R. Pelalak, Z. Heidari, R. Alizadeh, E. Ghareshabani, N. Nasseh, A. Marjani, A. B. Albadarin and S. Shirazian, J. Hazard. Mater., 2021, 411, 125074.
16. Y. Sun, R. Li, C. Song, H. Zhang, Y. Cheng, A. Nie, H. Li, D. D. Dionysiou, J. Qian and B. Pan, Appl. Catal., B, 2021, 298, 120537.
17. D. Wang and D. Astruc, Chem. Soc. Rev., 2017, 46, 816–854.
18. J. Wang and S. Wang, Chem. Eng. J., 2021, 411, 128392.
19. S. O. Ganiyu, S. Sable and M. Gamal El Din, Chem. Eng. J., 2022, 429, 132492.
20. Y. Ding, L. Fu, X. Peng, M. Lei, C. Wang and J. Jiang, Chem. Eng. J., 2022, 427, 131776.
21. J. Ye, G. Zhuang, Y. Wen, J. Wei, J. Chen, Z. Zhuang and Y. Yu, Chem. Eng. J., 2019, 372, 260–268.
22. Y. Zhu, X. Liu, S. Jin, H. Chen, W. Lee, M. Liu and Y. Chen, J. Mater. Chem. A, 2019, 7, 5875–5897.
23. S. Wu, G. Zhuang, J. Wei, Z. Zhuang and Y. Yu, J. Mater. Chem. A, 2018, 6, 18234–18241.
24. J. Azadmanjiri, V. K. Srivastava, P. Kumar, J. Wang and A. Yu, J. Mater. Chem. A, 2018, 6, 13509–13537.
25. C. Xie, D. Yan, W. Chen, Y. Zou, R. Chen, S. Zang, Y. Wang, X. Yao and S. Wang, Mater. Today, 2019, 31, 47–68.
26. S. Guo, Y. Jiang, L. Li, X. Huang, Z. Zhuang and Y. Yu, J. Mater. Chem. A, 2018, 6, 4167–4178.
27. G. Zhuang, Y. Chen, Z. Zhuang, Y. Yu and J. Yu, Sci. China Mater., 2020, 63, 2089–2118.
28. M. Ai, J. W. Zhang, Y. W. Wu, L. Pan, C. Shi and J. J. Zou, Chem.–Asian J., 2020, 15, 3599.
29. J. Miao, X. Duan, J. Li, J. Dai, B. Liu, S. Wang, W. Zhou and Z. Shao, Chem. Eng. J., 2019, 355, 721–730.
30. Q. Lin and Y. Deng, Environ. Sci. Technol., 2021, 55, 15010–15012.
31. L. Xiong and J. Tang, Adv. Energy Mater., 2021, 11, 2003216.
32. J. Hoigné and H. Bader, Water Res., 1983, 17, 185–194.
33. Y. Nosaka and A. Y. Nosaka, Chem. Rev., 2017, 117, 11302–11336.
34. W. D. Oh, Z. Dong and T. T. Lim, Appl. Catal., B, 2016, 194, 169–201.
35. W. A. Pryor, Annu. Rev. Phytopathol., 1986, 48, 657–667.
36. E. Codorniu-Hernández and P. G. Kusalik, J. Am. Chem. Soc., 2012, 134, 532–538.
37. J. L. Wang and L. J. Xu, Crit. Rev. Environ. Sci. Technol., 2012, 42, 251–325.
38. S. Luo, Z. Wei, D. D. Dionysiou, R. Spinney, W. P. Hu, L. Chai, Z. Yang, T. Ye and R. Xiao, Chem. Eng. J., 2017, 327, 1056–1065.
39. L. Wojnarovits and E. Takacs, Radiat. Phys. Chem., 2014, 96, 120–134.
40. D. Minakata, K. Li, P. Westerhoff and J. Crittenden, Environ. Sci. Technol., 2009, 43, 6220–6227.
41. E. W. Neuman, J. Chem. Phys., 1934, 2, 31–33.
42. D. T. Sawyer and J. S. Valentine, Acc. Chem. Res., 1981, 14, 393–400.
43. Q. Yi, J. Ji, B. Shen, C. Dong, J. Liu, J. Zhang and M. Xing, Environ. Sci. Technol., 2019, 53, 9725–9733.
44. E. A. Mayeda and A. J. Bard, J. Am. Chem. Soc., 1973, 95, 6223–6226.
45. C. Marques, J. M. F. Ferreira, E. Andronescu, D. Ficai, M. Sonmez and A. Ficai, Int. J. Nanomed., 2014, 9, 2713–2725.
46. L. De Trizio and L. Manna, Chem. Rev., 2016, 116, 10852–10887.
47. S. Korshunov and A. Imlay James, J. Bacteriol., 2006, 188, 6326–6334.
48. J. Ma, Z. Wei, R. Spinney, D. D. Dionysiou and R. Xiao, Environ. Sci.: Water Res. Technol., 2021, 7, 1966–1970.
49. Y. Pan, H. Xu, M. Chen, K. Wu, Y. Zhang and D. Long, ACS Catal., 2021, 11, 5974–5983.
50. C. Pan, L. Fu, F. Lide, Y. Ding, C. Wang, J. Huang and S. Wang, Chem. Eng. J., 2022, 431, 133957.
51. B. Cai, J. F. Feng, Q. Y. Peng, H. F. Zhao, Y. C. Miao and H. Pan, J. Hazard. Mater., 2020, 392, 122279.
52. M. Q. Yang, N. Zhang and Y. J. Xu, ACS Appl. Mater. Interfaces, 2013, 5, 1156–1164.
53. C. Zhang, F. R. F. Fan and A. J. Bard, J. Am. Chem. Soc., 2009, 131, 177–181.
54. X. Ma, X. Li, J. Zhou, Y. Wang and X. Lang, Chem. Eng. J., 2021, 426, 131418.
55. E. J. Nanni and D. T. Sawyer, J. Am. Chem. Soc., 1980, 102, 7591–7593.
56. R. Dietz, A. E. J. Forno, B. E. Larcombe and M. E. Peover, J. Chem. Soc. B, 1970, 816–820.
57. M. V. Merritt and D. T. Sawyer, J. Org. Chem., 1970, 35, 2157–2159.
58. J. Wang and S. Wang, Chem. Eng. J., 2020, 401, 126158.
59. J. S. Filippo, C. I. Chern and J. S. Valentine, J. Org. Chem., 1976, 41, 1077–1078.
60. A. A. Ghogare and A. Greer, Chem. Rev., 2016, 116, 9994–10034.
61. M. Hayyan, M. A. Hashim and I. M. AlNashef, Chem. Rev., 2016, 116, 3029–3085.
62. R. W. Redmond and I. E. Kochevar, Photochem. Photobiol., 2006, 82, 1178–1186.
63. L. Chen, S. Wang, Z. Yang, J. Qian and B. Pan, Appl. Catal., B, 2021, 292, 120193.
64. J. Lin, Q. Dai, H. Zhao, H. Cao, T. Wang, G. Wang and C. Chen, Environ. Sci. Technol., 2021, 55, 8683–8690.
65. X. Liu, A. Mazel, S. Marschner, Z. Fu, M. Muth, F. Kirschhöfer, G. Brenner-Weiss, S. Bräse, S. Diring and F. Odobel, ACS Appl. Mater. Interfaces, 2021, 13, 57768–57773.
66. R. Ma, N. Liu, T. T. Lin, T. Zhao, S. L. Huang and G. Y. Yang, J. Mater. Chem. A, 2020, 8, 8548–8553.
67. N. Zeinali, I. Oluwoye, M. Altarawneh and B. Z. Dlugogorski, Ecotoxicol. Environ. Saf., 2019, 184, 109605.
68. T. Matsuura, N. Yoshimura, A. Nishinaga and I. Saito, Tetrahedron, 1972, 28, 4933–4938.
69. Y. Liu, W. Shu, K. Chen, Z. Peng and W. Chen, ACS Catal., 2012, 2, 2557–2565.
70. Y. Li, C. Wang, H. Zheng, F. Wan, F. Yu, X. Zhang and Y. Liu, Appl. Surf. Sci., 2017, 391, 654–661.
71. W. Li, J. Shi, K. H. L. Zhang and J. L. MacManus-Driscoll, Mater. Horiz., 2020, 7, 2832–2859.
72. S. Huang, T. Ouyang, B. F. Zheng, M. Dan and Z. Q. Liu, Angew. Chem., Int. Ed., 2021, 60, 9546–9552.
73. Y. Li, J. Huang, T. Peng, J. Xu and X. Zhao, ChemCatChem, 2010, 2, 1082–1087.
74. L. Ren, M. Mao, Y. Li, L. Lan, Z. Zhang and X. Zhao, Appl. Catal., B, 2016, 198, 303–310.
75. S. Albonetti, F. Cavani and F. TrifirÒ, Catal. Rev., 1996, 38, 413–438.
76. H. Xu, N. Yan, Z. Qu, W. Liu, J. Mei, W. Huang and S. Zhao, Environ. Sci. Technol., 2017, 51, 8879–8892.
77. C. H. Chung, F. Y. Tu, T. A. Chiu, T. T. Wu and W. Y. Yu, Chem. Lett., 2021, 50, 856–865.
78. M. Konsolakis, Appl. Catal., B, 2016, 198, 49–66.
79. H. Jin, X. Tian, Y. Nie, Z. Zhou, C. Yang, Y. Li and L. Lu, Environ. Sci. Technol., 2017, 51, 12699–12706.
80. W. Dai, J. Long, L. Yang, S. Zhang, Y. Xu, X. Luo, J. Zou and S. Luo, J. Energy Chem., 2021, 61, 281–289.
81. S. Guan, L. Wang, S. M. Xu, D. Yang, G. I. N. Waterhouse, X. Qu and S. Zhou, Chem. Sci., 2019, 10, 2336–2341.
82. Z. Jiang, F. Yang, G. Yang, L. Kong, M. O. Jones, T. Xiao and P. P. Edwards, J. Photochem. Photobiol., A, 2010, 212, 8–13.
83. Y. Zhu, R. Zhu, Y. Xi, J. Zhu, G. Zhu and H. He, Appl. Catal., B, 2019, 255, 117739.
84. C. M. Lousada, A. J. Johansson, T. Brinck and M. Jonsson, J. Phys. Chem. C, 2012, 116, 9533–9543.
85. C. C. Lin, F. R. Smith, N. Ichikawa, T. Baba and M. Itow, Int. J. Chem. Kinet., 1991, 23, 971–987.
86. A. Hiroki and J. A. LaVerne, J. Phys. Chem. B, 2005, 109, 3364–3370.
87. J. Lee, U. von Gunten and J. H. Kim, Environ. Sci. Technol., 2020, 54, 3064–3081.
88. R. L. Johnson, P. G. Tratnyek and R. O. B. Johnson, Environ. Sci. Technol., 2008, 42, 9350–9356.
89. J. Wang and S. Wang, Chem. Eng. J., 2018, 334, 1502–1517.
90. J. Zhu, S. Wang, H. Li, J. Qian, L. Lv and B. Pan, Water Res., 2021, 202, 117397.
91. W. Ren, L. Xiong, X. Yuan, Z. Yu, H. Zhang, X. Duan and S. Wang, Environ. Sci. Technol., 2019, 53, 14595–14603.
92. S. Cai, Q. Zhang, Z. Wang, S. Hua, D. Ding, T. Cai and R. Zhang, Appl. Catal., B, 2021, 291, 120093.
93. J. Hu, H. Dong, J. Qu and Z. Qiang, Water Res., 2017, 112, 1–8.
94. M. M. Montemore, M. A. van Spronsen, R. J. Madix and C. M. Friend, Chem. Rev., 2018, 118, 2816–2862.
95. C. Zhang, Z. Huang, J. Lu, N. Luo and F. Wang, J. Am. Chem. Soc., 2018, 140, 2032–2035.
96. Z. Xie, J. Zhang, Y. Xiao, Y. Xie, W. Zhu, S. Yu, T. Hou, S. Liang and L. Wang, Energy Environ. Sci., 2021, 4, 444–450.
97. L. Wenqian, H. Xu, L. Boda, Z. Bin, L. Ting, H. Ying and M. Jun, Appl. Catal., B, 2021, 305, 121019.
98. H. Zhang, G. P. Hatzis, C. E. Moore, D. A. Dickie, M. W. Bezpalko, B. M. Foxman and C. M. Thomas, J. Am. Chem. Soc., 2019, 141, 9516–9520.
99. Y. Yang, K. Hu, P. Zhang, P. Zhou, X. Duan, H. Sun and S. Wang, Small, 2021, 17, 2100927.
100. Y. Li, H. Dong, L. Li, L. Tang, R. Tian, R. Li, J. Chen, Q. Xie, Z. Jin, J. Xiao, S. Xiao and G. Zeng, Water Res., 2021, 192, 116850.
101. X. Chu, Y. Qu, A. Zada, L. Bai, Z. Li, F. Yang, L. Zhao, G. Zhang, X. Sun, Z. D. Yang and L. Jing, Adv. Sci., 2020, 7, 2001543.
102. X. Feng, J. Xu, X. Xu, S. Zhang, J. Ma, X. Fang and X. Wang, ACS Catal., 2021, 11, 12112–12122.
103. Z. Guo, B. Liu, Q. Zhang, W. Deng, Y. Wang and Y. Yang, Chem. Soc. Rev., 2014, 43, 3480–3524.
104. H. Arandiyan, S. S. Mofarah, C. C. Sorrell, E. Doustkhah, B. Sajjadi, D. Hao, Y. Wang, H. Sun, B. J. Ni, M. Rezaei, Z. Shao and T. Maschmeyer, Chem. Soc. Rev., 2021, 50, 10116–10211.
105. M. Xing, W. Xu, C. Dong, Y. Bai, J. Zeng, Y. Zhou, J. Zhang and Y. Yin, Chem, 2018, 4, 1359–1372.
106. R. Yang, Y. Fan, R. Ye, Y. Tang, X. Cao, Z. Yin and Z. Zeng, Adv. Mater., 2021, 33, 2004862.
107. N. Thomas, D. D. Dionysiou and S. C. Pillai, J. Hazard. Mater., 2021, 404, 124082.
108. X. Wang, W. Liu, J. Qin and L. Lei, Ind. Eng. Chem. Res., 2020, 59, 2192–2202.
109. Q. Wu, H. Yang, L. Kang, Z. Gao and F. Ren, Appl. Catal., B, 2020, 263, 118282.
110. W. Jiang, D. D. Dionysiou, M. Kong, Z. Liu, Q. Sui and S. Lyu, Chem. Eng. J., 2020, 380, 122537.
111. M. E. Lindsey and M. A. Tarr, Environ. Sci. Technol., 2000, 34, 444–449.
112. L. Lyu, L. Zhang, G. He, H. He and C. Hu, J. Mater. Chem. A, 2017, 5, 7153–7164.
113. L. Z. Huang, M. Zhu, Z. Liu, Z. Wang and H. C. B. Hansen, J. Hazard. Mater., 2019, 364, 39–47.
114. J. Y. Qiu, J. H. Chen, B. Y. Xiao, X. X. Li, T. Wan, F. H. Qin, Y. Mi and Z. Y. Huang, Catal. Lett., 2020, 150, 222–233.
115. A. Y. Zhang, T. Lin, Y. Y. He and Y. X. Mou, J. Hazard. Mater., 2016, 311, 81–90.
116. B. Gong, C. Ku, H. Q. Yu and P. H. L. Sit, J. Phys. Chem. C, 2021, 125, 8508–8517.
117. W. Liu, J. Tian, C. Mao, Z. Wang, J. Liu, R. A. Dahlgren, L. Zhang and X. Wang, Anal. Chim. Acta, 2020, 1127, 246–255.
118. Z. Tan, J. Zhang, Y. C. Chen, J. P. Chou and Y. K. Peng, J. Phys. Chem. Lett., 2020, 11, 5390–5396.
119. S. Zhan, H. Zhang, X. Mi, Y. Zhao, C. Hu and L. Lyu, Environ. Sci. Technol., 2020, 54, 8333–8343.
120. Z. Weng, J. Li, Y. Weng, M. Feng, Z. Zhuang and Y. Yu, J. Mater. Chem. A, 2017, 5, 15650–15660.
121. X. Zhang, T. Li, M. Guan, R. Li, J. Liu, Y. Wang, C. Zhang and C. Fan, Appl. Surf. Sci., 2021, 567, 150828.
122. M. Setvin, U. Aschauer, P. Scheiber, Y. F. Li, W. Hou, M. Schmid, A. Selloni and U. Diebold, Science, 2013, 341, 988–991.
123. S. Zhao, Y. Wen, X. Liu, X. Pen, F. Lu, F. Gao, X. Xie, C. Du, H. Yi, D. Kang and X. Tang, Nano Res., 2020, 13, 1544–1551.
124. H. Y. Gao, C. H. Huang, L. Mao, B. Shao, J. Shao, Z. Y. Yan, M. Tang and B.-Z. Zhu, Environ. Sci. Technol., 2020, 54, 14046–14056.
125. D. E. Pennington and A. Haim, J. Am. Chem. Soc., 1968, 90, 3700–3704.
126. R. O. C. Norman, P. M. Storey and P. R. West, J. Chem. Soc. B, 1970, 1087–1095,  10.1039/J29700001087.
127. N. Chen, D. Lee, H. Kang, D. Cha, J. Lee and C. Lee, J. Environ. Chem. Eng., 2022, 10, 107654.
128. J. D. Rush and B. H. J. Bielski, J. Phys. Chem., 1985, 89, 5062–5066.
129. K. Xu and U. von Gunten, ACS ES&T Engineering, 2021, 1, 1410–1419.
130. Z. Wu, K. Guo, S. Cao, W. Yao and L. Piao, Nano Res., 2020, 13, 551–556.
131. E. S. Kumar, S. Venkatesh and M. S. R. Rao, Appl. Phys. Lett., 2010, 96, 232504.
132. Z. Wei, D. Liu, W. Wei, X. Chen, Q. Han, W. Yao, X. Ma and Y. Zhu, ACS Appl. Mater. Interfaces, 2017, 9, 15533–15540.
133. T. Mallat and A. Baiker, Chem. Rev., 2004, 104, 3037–3058.
134. T. Matsumoto, M. Ueno, N. Wang and S. Kobayashi, Chem.–Asian J., 2008, 3, 196–214.
135. B. Z. Zhan and A. Thompson, Tetrahedron, 2004, 60, 2917–2935.
136. P. Lyu, J. Zhu, C. Han, L. Qiang, L. Zhang, B. Mei, J. He, X. Liu, Z. Bian and H. Li, ACS Appl. Mater. Interfaces, 2021, 13, 2033–2043.
137. H. Shang, M. Li, H. Li, S. Huang, C. Mao, Z. Ai and L. Zhang, Environ. Sci. Technol., 2019, 53, 6444–6453.
138. W. Qin, G. Fang, Y. Wang and D. Zhou, Chem. Eng. J., 2018, 348, 526–534.
139. W. Li, X. He, B. Li, B. Zhang, T. Liu, Y. Hu and J. Ma, Appl. Catal., B, 2022, 305, 121019.
140. H. Li, H. Zhu, Y. Shi, H. Shang, L. Zhang and J. Wang, Environ. Sci. Technol., 2022, 56, 1771–1779.
141. H. Li, F. Qin, Z. Yang, X. Cui, J. Wang and L. Zhang, J. Am. Chem. Soc., 2017, 139, 3513–3521.
142. J. Li, C. Shu, C. Liu, X. Chen, A. Hu and J. Long, Small, 2020, 16, 2001812.
143. J. Wu, X. Li, W. Shi, P. Ling, Y. Sun, X. Jiao, S. Gao, L. Liang, J. Xu, W. Yan, C. Wang and Y. Xie, Angew. Chem., Int. Ed., 2018, 57, 8719–8723.
144. M. Kim, B. Lee, H. Ju, J. Y. Kim, J. Kim and S. W. Lee, Adv. Mater., 2019, 31, 1903316.
145. Y. Mao, P. Wang, L. Li, Z. Chen, H. Wang, Y. Li and S. Zhan, Angew. Chem., Int. Ed., 2020, 59, 3685–3690.
146. T. Gan, J. Yang, D. Morris, X. Chu, P. Zhang, W. Zhang, Y. Zou, W. Yan, S. H. Wei and G. Liu, Nat. Commun., 2021, 12, 2741.
147. X. Sun, X. Luo, X. Zhang, J. Xie, S. Jin, H. Wang, X. Zheng, X. Wu and Y. Xie, J. Am. Chem. Soc., 2019, 141, 3797–3801.
148. F. Rao, G. Zhu, W. Zhang, Y. Xu, B. Cao, X. Shi, J. Gao, Y. Huang, Y. Huang and M. Hojamberdiev, ACS Catal., 2021, 11, 7735–7749.
149. Y. Zhao and H. Wang, Langmuir, 2021, 37, 13969–13975.
150. Y. Yin, R. Lv, X. Li, L. Lv and W. Zhang, Appl. Catal., B, 2021, 299, 120685.
151. Z. Yang, J. Qian, A. Yu and B. Pan, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 6659–6664.
152. Y. Yin, Y. Ren, J. Lu, W. Zhang, C. Shan, M. Hua, L. Lv and B. Pan, Appl. Catal., B, 2021, 286, 120282.
153. X. Li, J. Wang, X. Duan, Y. Li, X. Fan, G. Zhang, F. Zhang and W. Peng, ACS Catal., 2021, 11, 4848–4861.
154. B. F. Sels, D. E. De Vos and P. A. Jacobs, J. Am. Chem. Soc., 2007, 129, 6916–6926.
155. X. Li, J. Liu, A. I. Rykov, H. Han, C. Jin, X. Liu and J. Wang, Appl. Catal., B, 2015, 179, 196–205.
156. S. Chong, G. Zhang, N. Zhang, Y. Liu, T. Huang and H. Chang, J. Hazard. Mater., 2017, 334, 150–159.
157. M. C. DeRosa and R. J. Crutchley, Coord. Chem. Rev., 2002, 233–234, 351–371.
158. J. Wang, X. Xu, Y. Liu, Z. Wang, P. Wang, Z. Zheng, H. Cheng, Y. Dai and B. Huang, ChemSusChem, 2020, 13, 3488–3494.
159. J. Ji, Q. Yan, P. Yin, S. Mine, M. Matsuoka and M. Xing, Angew. Chem., Int. Ed., 2021, 60, 2903–2908.
160. W. J. Duan, J. L. He, Z. L. Wei, Z. R. Dai and C. H. Feng, Environ. Sci.: Nano, 2020, 7, 2982–2994.
161. T. Liu, D. Zhang, K. Yin, C. Yang, S. Luo and J. C. Crittenden, Chem. Eng. J., 2020, 388, 124264.
162. Q. Ye, J. Wu, P. Wu, S. Rehman, Z. Ahmed and N. Zhu, Chem. Eng. J., 2021, 417, 12911.
163. L. Kong, G. Fang, Z. Fang, Y. Zou, F. Zhu, D. Zhou and J. Zhan, Chem. Eng. J., 2021, 416, 128996.
164. M. I. Akhtar, F. Junejo, H. I. Naqvi, M. Hamid, Y. Wang, M. Liu, C. Hu, Y. Xin, D. Ma, M. Gao and H. Xie, Chem. Eng. J., 2022, 433, 134048.
165. H. Zeng, L. Deng, H. Zhang, C. Zhou and Z. Shi, J. Hazard. Mater., 2020, 400, 123297.
166. Y. Zhao, H. An, G. Dong, J. Feng, T. Wei, Y. Ren and J. Ma, Chem. Eng. J., 2020, 388, 124371.
167. P. Gao, X. Tian, Y. Nie, C. Yang, Z. Zhou and Y. Wang, Chem. Eng. J., 2019, 359, 828–839.
168. Y. Bu, H. Li, W. Yu, Y. Pan, L. Li, Y. Wang, L. Pu, J. Ding, G. Gao and B. Pan, Environ. Sci. Technol., 2021, 55, 2110–2120.
169. S. Zhu, X. Li, J. Kang, X. Duan and S. Wang, Environ. Sci. Technol., 2019, 53, 307–315.
170. A. Jawad, K. Zhan, H. Wang, A. Shahzad, Z. Zeng, J. Wang, X. Zhou, H. Ullah, Z. Chen and Z. Chen, Environ. Sci. Technol., 2020, 54, 2476–2488.
171. J. Hu, X. Zeng, G. Wang, B. Qian, Y. Liu, X. Hu, B. He, L. Zhang and X. Zhang, Chem. Eng. J., 2020, 400, 125869.
172. E. Hayashi, Y. Yamaguchi, K. Kamata, N. Tsunoda, Y. Kumagai, F. Oba and M. Hara, J. Am. Chem. Soc., 2019, 141, 890–900.
173. H. Liu, W. Jia, X. Yu, X. Tang, X. Zeng, Y. Sun, T. Lei, H. Fang, T. Li and L. Lin, ACS Catal., 2021, 11, 7828–7844.
174. X. Xu, L. Li, J. Huang, H. Jin, X. Fang, W. Liu, N. Zhang, H. Wang and X. Wang, ACS Catal., 2018, 8, 8033–8045.
175. P. Schlexer, D. Widmann, R. J. Behm and G. Pacchioni, ACS Catal., 2018, 8, 6513–6525.
176. N. H. M. Dostagir, R. Rattanawan, M. Gao, J. Ota, J. Y. Hasegawa, K. Asakura, A. Fukouka and A. Shrotri, ACS Catal., 2021, 11, 9450–9461.
177. M. Zhao, J. Deng, J. Liu, Y. Li, J. Liu, Z. Duan, J. Xiong, Z. Zhao, Y. Wei, W. Song and Y. Sun, ACS Catal., 2019, 9, 7548–7567.
178. X. Zhang, C. Pei, X. Chang, S. Chen, R. Liu, Z. J. Zhao, R. Mu and J. Gong, J. Am. Chem. Soc., 2020, 142, 11540–11549.
179. X. Yang, X. Yu, M. Jing, W. Song, J. Liu and M. Ge, ACS Appl. Mater. Interfaces, 2019, 11, 730–739.
180. A. D. Quinonero, E. B. Garcia, S. L. Rodriguez, J. Juan, D. L. Castello, M. G. Melchor, F. C. Herrera, E. Pellegrin, C. Escudero and A. B. Lopez, ACS Catal., 2020, 10, 6532–6545.
181. J. Lu, J. Wang, Q. Zou, D. He, L. Zhang, Z. Xu, S. He and Y. Luo, ACS Catal., 2019, 9, 2177–2195.
182. Q. Wang, L. Chen, S. Guan, X. Zhang, B. Wang, X. Cao, Z. Yu, Y. He, D. G. Evans, J. Feng and D. Li, ACS Catal., 2018, 8, 3104–3115.
183. Q. Gao, G. Wang, Y. Chen, B. Han, K. Xia and C. Zhou, Environ. Sci.: Water Res. Technol., 2021, 7, 1197–1211.
184. W. Sang, Z. Li, M. Huang, X. Wu, D. Li, L. Mei and J. Cui, Chem. Eng. J., 2020, 383, 123057.
185. S. Hussain, E. Aneggi and D. Goi, Environ. Chem. Lett., 2021, 19, 2405–2424.
186. F. Chen, X. L. Wu, C. Shi, H. Lin, J. Chen, Y. Shi, S. Wang and X. Duan, Adv. Funct. Mater., 2021, 31, 2007877.
187. G. P. Anipsitakis and D. D. Dionysiou, Environ. Sci. Technol., 2004, 38, 3705–3712.
188. X. Zhao, X. Li, Z. Zhu, W. Hu, H. Zhang, J. Xu, X. Hu, Y. Zhou, M. Xu, H. Zhang and G. Hu, Appl. Catal., B, 2022, 300, 120759.
189. L. S. Zhang, X. H. Jiang, Z. A. Zhong, L. Tian, Q. Sun, Y. T. Cui, X. Lu, J. P. Zou and S. L. Luo, Angew. Chem., Int. Ed., 2021, 60, 21751–21755.
190. L. Wu, Z. Sun, Y. Zhen, S. Zhu, C. Yang, J. Lu, Y. Tian, D. Zhong and J. Ma, Environ. Sci. Technol., 2021, 55, 15400–15411.
191. H. Zhao, X. Liu, Y. Dong, Y. Xia and H. Wang, Appl. Catal., B, 2019, 256, 117872.
192. C. Y. Toe, C. Tsounis, J. Zhang, H. Masood, D. Gunawan, J. Scott and R. Amal, Environ. Sci. Technol., 2021, 14, 1140–1175.
193. Z. Geng, X. Kong, W. Chen, H. Su, Y. Liu, F. Cai, G. Wang and J. Zeng, Angew. Chem., Int. Ed., 2018, 57, 6054–6059.
194. J. Xiang, T. Zhang, R. Cao, M. Lin, B. Yang, Y. Wen, Z. Zhuang and Y. Yu, Sol. RRL, 2021, 5, 2100703.
195. M. Yang, T. Xu, X. Jin, Q. Shen and C. Sun, Appl. Surf. Sci., 2022, 581, 152439.
196. J. Sahu, S. Kumar, V. S. Vats, P. A. Alvi, B. Dalela, S. Kumar and S. Dalela, J. Lumin., 2022, 243, 118673.
197. Q. Wang, Z. Liu, D. Liu, W. Wang, Z. Zhao, F. Cui and G. Li, Chem. Eng. J., 2019, 360, 838–847.
198. M. J. Jackson, Proc. Nutr. Soc., 1999, 58, 1001–1006.
199. C. D. Jaeger and A. J. Bard, J. Phys. Chem., 1979, 83, 3146–3152.
200. K. Hiramoto, H. Johkoh, K. I. Sako and K. Kikugawa, Free Radical Res. Commun., 1993, 19, 323–332.
201. J. M. Burns, W. J. Cooper, J. L. Ferry, D. W. King, B. P. DiMento, K. McNeill, C. J. Miller, W. L. Miller, B. M. Peake, S. A. Rusak, A. L. Rose and T. D. Waite, Aquat. Sci., 2012, 74, 683–734.
202. Y. Jing and B. P. Chaplin, Environ. Sci. Technol., 2017, 51, 2355–2365.
203. C. C. Perry, V. J. Tang, K. M. Konigsfeld, J. A. Aguilera and J. R. Milligan, J. Phys. Chem. B, 2011, 115, 9889–9897.
204. C. Yu, Y. Wu, F. Zeng and S. Wu, J. Mater. Chem. B, 2013, 1, 4152–4159.
205. T. Charbouillot, M. Brigante, G. Mailhot, P. R. Maddigapu, C. Minero and D. Vione, J. Photochem. Photobiol., A, 2011, 222, 70–76.
206. M. Náfrádi, L. Farkas, T. Alapi, K. Hernádi, K. Kovács, L. Wojnárovits and E. Takács, Radiat. Phys. Chem., 2020, 170, 108610.
207. C. He, Y. Wang, Z. Li, Y. Huang, Y. Liao, D. Xia and S. Lee, Environ. Sci. Technol., 2020, 54, 12771–12783.
208. T. Wu, T. Lin, J. Zhao, H. Hidaka and N. Serpone, Environ. Sci. Technol., 1999, 33, 1379–1387.
209. L. Liang, Q. Chang, T. Cai, N. Li, C. Xue, J. Yang and S. Hu, Chem. Eng. J., 2022, 428, 131139.
210. M. I. Muniz-Junqueira and V. N. de Paula-Coelho, Int. Immunopharmacol., 2008, 8, 1633–1638.
211. Y. Guo and B. Li, Carbon, 2015, 82, 459–469.
212. Y. Kakuma, A. Y. Nosaka and Y. Nosaka, Phys. Chem. Chem. Phys., 2015, 17, 18691–18698.
213. K. K. Mothilal, J. Johnson Inbaraj, R. Gandhidasan and R. Murugesan, J. Photochem. Photobiol., A, 2004, 162, 9–16.
214. R. Konaka, E. Kasahara, W. C. Dunlap, Y. Yamamoto, K. C. Chien and M. Inoue, Free Radical Biol. Med., 1999, 27, 294–300.
215. L. R. Barclay, M. C. Basque and M. R. Vinqvist, Can. J. Chem., 2003, 81, 457–467.
216. J. A. Rengifo-Herrera, K. Pierzchała, A. Sienkiewicz, L. Forró, J. Kiwi and C. Pulgarin, Appl. Catal., B, 2009, 88, 398–406.
217. X. Ragàs, A. Jiménez-Banzo, D. Sánchez-García, X. Batllori and S. Nonell, Chem. Commun., 2009, 2920–2922,  10.1039/B822776D.
218. A. M. Asadirad, Z. Erno and N. R. Branda, Chem. Commun., 2013, 49, 5639–5641.
219. H. Wang, X. Yang, W. Shao, S. Chen, J. Xie, X. Zhang, J. Wang and Y. Xie, J. Am. Chem. Soc., 2015, 137, 11376–11382.
220. M. Wozniak, F. Tanfani, E. Bertoli, G. Zolese and J. Antosiewicz, Biochim. Biophys. Acta, Lipids Lipid Metab., 1991, 1082, 94–100.
221. Z. C. Redman, J. Wesolowski and P. L. Tomco, Environ. Sci. Technol., 2021, 55, 4974–4983.
222. C. Liang, H. P. Feng, H. Y. Niu, C. G. Niu, J. S. Li, D. W. Huang, L. Zhang, H. Guo, N. Tang and H. Y. Liu, Chem. Eng. J., 2020, 384, 123236.
223. Z. Wang, J. Wang, B. Xiong, F. Bai, S. Wang, Y. Wan, L. Zhang, P. Xie and M. R. Wiesner, Environ. Sci. Technol., 2020, 54, 464–475.
224. S. Kang, L. Zhang, C. Yin, Y. Li, L. Cui and Y. Wang, Appl. Catal., B, 2017, 211, 266–274.
225. P. Neta, V. Madhavan, H. Zemel and R. W. Fessenden, J. Am. Chem. Soc., 1977, 99, 163–164.
226. S. Gligorovski, R. Strekowski, S. Barbati and D. Vione, Chem. Rev., 2015, 115, 13051–13092.
227. K. Bao, S. Zhang, P. Ni, Z. Zhang, K. Zhang, L. Wang, L. X. Sun, W. Mao, Q. Zhou and Y. Qian, Catal. Today, 2020, 340, 311–317.
228. L. Hu, Y. Liao, D. Xia, F. Peng, L. Tan, S. Hu, C. Zheng, X. Lu, C. He and D. Shu, Chem. Eng. J., 2020, 385, 123824.
229. H. Guo, C. G. Niu, C. Y. Feng, C. Liang, L. Zhang, X. J. Wen, Y. Yang, H. Y. Liu, L. Li and L. S. Lin, Chem. Eng. J., 2020, 385, 123919.
230. H. Xu, Q. Ye, J. Zhang, Q. Li, M. Wang, P. Zhou, G. Zhou and Q. Wang, Sci. Total Environ., 2021, 778, 146280.
231. Y. Liu, X. Liu, Y. Zhao and D. D. Dionysiou, Appl. Catal., B, 2017, 213, 74–86.
232. A. A. Gorman and M. A. J. Rodgers, Chem. Soc. Rev., 1981, 10, 205–231.
233. X. Chen, W. D. Oh and T. T. Lim, Chem. Eng. J., 2018, 354, 941–976.
234. Z. Xiong, Y. Jiang, Z. Wu, G. Yao and B. Lai, Chem. Eng. J., 2021, 421, 127863.
235. M. Zhang, R. Luo, C. Wang, W. Zhang, X. Yan, X. Sun, L. Wang and J. Li, J. Mater. Chem. A, 2019, 7, 12547–12555.
236. W. R. Haag and J. Hoigne, Environ. Sci. Technol., 1986, 20, 341–348.
237. M. A. Oturan and J. J. Aaron, Crit. Rev. Environ. Sci. Technol., 2014, 44, 2577–2641.
238. X. Ye, Y. Cui, X. Qiu and X. Wang, Appl. Catal., B, 2014, 152, 383–389.
239. G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross, J. Phys. Chem. Ref. Data, 1988, 17, 513–886.
240. B. Ervens, S. Gligorovski and H. Herrmann, Phys. Chem. Chem. Phys., 2003, 5, 1811–1824.
241. X. X. Guo, T. T. Hu, B. Meng, Y. Sun and Y. F. Han, Appl. Catal., B, 2020, 260, 118157.
242. N. Zhang, E. P. Tsang, J. Chen, Z. Fang and D. Zhao, J. Colloid Interface Sci., 2020, 558, 163–172.
243. H. Li, J. Shang, Z. Yang, W. Shen, Z. Ai and L. Zhang, Environ. Sci. Technol., 2017, 51, 5685–5694.
244. M. T. Bender, Y. C. Lam, S. Hammes-Schiffer and K. S. Choi, J. Am. Chem. Soc., 2020, 142, 21538–21547.
245. Z. Liu, Z. Zhao, W. Jiang, Z. Zhu, Y. Wang, Z. Liu, W. Gu, Y. Yang, Y. Yang, L. Zhang, W. Yao and F. Teng, Appl. Catal., B, 2021, 298, 120517.
246. L. Luo, Z. J. Wang, X. Xiang, D. Yan and J. Ye, ACS Catal., 2020, 10, 4906–4913.
247. H. Li, H. Shang, X. Cao, Z. Yang, Z. Ai and L. Zhang, Environ. Sci. Technol., 2018, 52, 8659–8665.
248. A. L. Teel, S. A. Khotz and R. J. Watts, J. Adv. Oxid. Technol., 2007, 10, 282–286.
249. R. J. Watts and A. L. Teel, Water Res., 2019, 159, 46–54.
250. M. Ahmad, M. A. Simon, A. Sherrin, M. E. Tuccillo, J. L. Ullman, A. L. Teel and R. J. Watts, Chemosphere, 2011, 84, 855–862.
251. C. Zhang, T. Li, J. Zhang, S. Yan and C. Qin, Appl. Catal., B, 2019, 259, 118030.
252. Q. Wang, H. Zhou, X. Liu, T. Li, C. Jiang, W. Song and W. Chen, Environ. Sci.: Nano, 2018, 5, 2864–2875.
253. R. J. Watts, B. C. Bottenberg, T. F. Hess, M. D. Jensen and A. L. Teel, Environ. Sci. Technol., 1999, 33, 3432–3437.
254. J. Xiao, Y. Xie, Q. Han, H. Cao, Y. Wang, F. Nawaz and F. Duan, J. Hazard. Mater., 2016, 304, 126–133.
255. J. Jiang, Z. Lu, M. Zhang, J. Duan, W. Zhang, Y. Pan and J. Bai, J. Am. Chem. Soc., 2018, 140, 17825–17829.
256. A. Ishak, K. Dayana, M. H. Mamat, M. F. Malek and M. Rusop, Optik, 2015, 126, 1610–1612.
257. Z. Li, W. Yang, L. Xie, Y. Li, Y. Liu, Y. Sun, Y. Bu, X. Mi, S. Zhan and W. Hu, Appl. Surf. Sci., 2021, 549, 149262.
258. Z. Wang, H. Jia, T. Zheng, Y. Dai, C. Zhang, X. Guo, T. Wang and L. Zhu, Appl. Catal., B, 2020, 272, 119030.
259. Y. H. Kim and B. C. Chung, J. Org. Chem., 1983, 48, 1562–1564.
260. X. Xiao, J. Jiang and L. Zhang, Appl. Catal., B, 2013, 142, 487–493.
261. A. Yuan, H. Lei, Z. Wang and X. Dong, J. Colloid Interface Sci., 2020, 560, 40–49.
262. S. Ito, Y. Kon, T. Nakashima, D. Hong, H. Konno, D. Ino and K. Sato, Molecules, 2019, 24, 2520.
263. F. Xie, Y. Zhang, X. He, H. Li, X. Qiu, W. Zhou, S. Huo and Z. Tang, J. Mater. Chem. A, 2018, 6, 13236–13243.
264. C. Chen, G. Qiu, T. Wang, Z. Zheng, M. Huang and B. Li, J. Colloid Interface Sci., 2021, 592, 1–12.
265. V. R. Akhmetova, G. R. Khabibullina, E. B. Rakhimova, R. A. Vagapov, R. R. Khairullina, Z. T. Niatshina and N. N. Murzakova, Mol. Diversity, 2010, 14, 463–471.
266. S. Kobayashi, Y. Mori, J. S. Fossey and M. M. Salter, Chem. Rev., 2011, 111, 2626–2704.
267. J. H. Xie, S. F. Zhu and Q. L. Zhou, Chem. Rev., 2011, 111, 1713–1760.
268. Q. Li, X. Huang, G. Su, M. Zheng, C. Huang, M. Wang, C. Ma and D. Wei, Environ. Sci. Technol., 2018, 52, 13351–13360.
269. A. Khampuanbut, S. Santalelat, A. Pankiew, D. Channei, S. Pornsuwan, K. Faungnawakij, S. Phanichphant and B. Inceesungvorn, J. Colloid Interface Sci., 2020, 560, 213–224.
270. J. L. Shi, H. Hao and X. Lang, Sustainable Energy Fuels, 2019, 3, 488–498.
271. E. J. Corey and W. C. Taylor, J. Am. Chem. Soc., 1964, 86, 3881–3882.
272. C. S. Foote and S. Wexler, J. Am. Chem. Soc., 1964, 86, 3880–3881.
273. T. Li, L. Ge, X. Peng, W. Wang and W. Zhang, Water Res., 2021, 190, 116777.
274. L. He, Z. W. Jiang, W. Li, C. M. Li, C. Z. Huang and Y. F. Li, ACS Appl. Mater. Interfaces, 2018, 10, 28868–28876.
275. R. Luo, M. Li, C. Wang, M. Zhang, M. A. Nasir Khan, X. Sun, J. Shen, W. Han, L. Wang and J. Li, Water Res., 2019, 148, 416–424.
276. S. Liu, Z. Zhang, F. Huang, Y. Liu, L. Feng, J. Jiang, L. Zhang, F. Qi and C. Liu, Appl. Catal., B, 2021, 286, 119921.
277. F. Sun, T. Chen, H. Liu, X. Zou, P. Zhai, Z. Chu, D. Shu, H. Wang and D. Chen, Sci. Total Environ., 2021, 784, 147117.
278. J. Zhao, F. Li, H. Wei, H. Ai, L. Gu, J. Chen, L. Zhang, M. Chi and J. Zhai, Chem. Eng. J., 2021, 409, 128150.
279. J. Li, J. Miao, X. Duan, J. Dai, Q. Liu, S. Wang, W. Zhou and Z. Shao, Adv. Funct. Mater., 2018, 28, 1804654.
280. Y. X. Chen, J. Y. Gao, Z. W. Huang, M. J. Zhou, J. X. Chen, C. Li, Z. Ma, J. M. Chen and X. F. Tang, Environ. Sci. Technol., 2017, 51, 7084–7090.
281. J. Lee, S. Hong, Y. Mackeyev, C. Lee, E. Chung, L. J. Wilson, J. H. Kim and P. J. J. Alvarez, Environ. Sci. Technol., 2011, 45, 10598–10604.
282. A. Kashyap, E. Ramasamy, V. Ramalingam and M. Pattabiraman, Molecules, 2021, 26, 2673.
283. Z.-J. Li, S. Li, A. H. Davis, E. Hofman, G. Leem and W. Zheng, Nano Res., 2020, 13, 1668–1676.
284. P. K. Dikshit, S. K. Padhi and V. S. Moholkar, Bioresour. Technol., 2017, 244, 362–370.
285. S. Zhao, Z. Dai, W. Guo, F. Chen, Y. Liu and R. Chen, Appl. Catal., B, 2019, 244, 206–214.
286. H. Li, X. Li, J. Zhou, W. Sheng and X. Lang, Chin. Chem. Lett., 2021 DOI:10.1016/j.cclet.2021.10.068.
287. S. Xu, P. Zhou, Z. Zhang, C. Yang, B. Zhang, K. Deng, S. Bottle and H. Zhu, J. Am. Chem. Soc., 2017, 139, 14775–14782.
288. N. Zheng, X. He, R. Hu, R. Wang, Q. Zhou, Y. Lian and Z. Hu, Appl. Catal., B, 2022, 307, 121157.
289. M. Sajid, X. Zhao and D. Liu, Green Chem., 2018, 20, 5427–5453.
290. C. He, J. Cheng, X. Zhang, M. Douthwaite, S. Pattisson and Z. Hao, Chem. Rev., 2019, 119, 4471–4568.
291. Y. Li, C. Wang, C. Zhang and H. He, Top. Catal., 2020, 63, 810–816.
292. P. Pinto Joseph, G. R. Gladstone and L. Yung Yuk, Science, 1980, 210, 183–185.
293. C. Dong, Z. Qu, Y. Qin, Q. Fu, H. Sun and X. Duan, ACS Catal., 2019, 9, 6698–6710.
294. Y. Ma, L. Wang, J. Ma, H. Wang, C. Zhang, H. Deng and H. He, ACS Catal., 2021, 11, 6614–6625.
295. Z. Wang, W. Wang, L. Zhang and D. Jiang, Catal. Sci. Technol., 2016, 6, 3845–3853.
296. M. He, J. Ji, B. Liu and H. Huang, Appl. Surf. Sci., 2019, 473, 934–942.
297. X. Liu, J. Mi, L. Shi, H. Liu, J. Liu, Y. Ding, J. Shi, M. He, Z. Wang, S. Xiong, Q. Zhang, Y. Liu, Z. S. Wu, J. Chen and J. Li, Angew. Chem., Int. Ed., 2021, 60, 26747–26754.
298. T. Brink Gerd Jan, W. C. E. Arends Isabel and A. Sheldon Roger, Science, 2000, 287, 1636–1639.
299. F. Sabaté, J. L. Jordá, M. J. Sabater and A. Corma, J. Mater. Chem. A, 2020, 8, 3771–3784.
300. Y. Dong, J. Luo, S. Li and C. Liang, Catal. Commun., 2020, 133, 105843.
301. Y. Dai, P. Ren, Y. Li, D. Lv, Y. Shen, Y. Li, H. Niemantsverdriet, F. Besenbacher, H. Xiang and W. Hao, Angew. Chem., Int. Ed., 2019, 131, 6331–6336.
302. E. Markó István, R. Giles Paul, M. Tsukazaki, M. Brown Stephen and J. Urch Chistopher, Science, 1996, 274, 2044–2046.
303. I. Enache Dan, K. Edwards Jennifer, P. Landon, B. Solsona-Espriu, F. Carley Albert, A. Herzing Andrew, M. Watanabe, J. Kiely Christopher, W. Knight David and J. Hutchings Graham, Science, 2006, 311, 362–365.
304. M. J. Schultz, R. S. Adler, W. Zierkiewicz, T. Privalov and M. S. Sigman, J. Am. Chem. Soc., 2005, 127, 8499–8507.
305. H. Yan, Q. Shen, Y. Sun, S. Zhao, R. Lu, M. Gong, Y. Liu, X. Zhou, X. Jin, X. Feng, X. Chen, D. Chen and C. Yang, ACS Catal., 2021, 11, 6371–6383.
306. Y. Wan and J. M. Lee, ACS Catal., 2021, 11, 2524–2560.
307. E. Hayashi, T. Komanoya, K. Kamata and M. Hara, ChemSusChem, 2017, 10, 654–658.
308. X. Han, C. Li, X. Liu, Q. Xia and Y. Wang, Green Chem., 2017, 19, 996–1004.
309. S. Zhang, X. Sun, Z. Zheng and L. Zhang, Catal. Commun., 2018, 113, 19–22.
310. X. Wan, N. Tang, Q. Xie, S. Zhao, C. Zhou, Y. Dai and Y. Yang, Catal. Sci. Technol., 2021, 11, 1497–1509.
311. F. Sabaté, J. L. Jordà and M. J. Sabater, Catal. Today, 2021 DOI:10.1016/j.cattod.2021.06.033.
312. S. Ishikawa, Y. Yamada, C. Qiu, Y. Kawahara, N. Hiyoshi, A. Yoshida and W. Ueda, Chem. Mater., 2019, 31, 1408–1417.
313. L. Bui, R. Chakrabarti and A. Bhan, ACS Catal., 2016, 6, 6567–6580.
314. T. Li, F. Liu, Y. Tang, L. Li, S. Miao, Y. Su, J. Zhang, J. Huang, H. Sun, M. Haruta, A. Wang, B. Qiao, J. Li and T. Zhang, Angew. Chem., Int. Ed., 2018, 57, 7795–7799.
315. J. S. Chung, R. Miranda and C. O. Bennett, J. Catal., 1988, 114, 398–410.
316. S. P. Bates, M. J. Gillan and G. Kresse, J. Phys. Chem. B, 1998, 102, 2017–2026.

---

Footnote

† Y. Wen and J. Yan contributed equally to this work.

---