English

• 1. Introduction

  Advanced oxidation processes (AOPs) have become indispensable in environmental remediation, achieving pollutant removal through the generation of highly reactive oxygen species (ROS) [1,2]. These ROS, including superoxide (O₂^•−), hydroxyl radicals (^•OH), and singlet oxygen (¹O₂), are typically derived from chemical precursors like hydrogen peroxide (H₂O₂), ozone (O₃), or peroxymonosulfate (PMS) [3,4]. While effective, the reliance on these precursors poses challenges due to their costly synthesis, unstable storage, and energy-intensive transport. In contrast, molecular oxygen (O₂), a naturally abundant and eco-friendly oxidant constituting 21% of the air by volume, holds immense potential for sustainable environmental applications [5]. However, its practical utilization in AOPs remains constrained by inherent stability, where its triplet ground state and high-energy molecular orbitals prevent direct interaction with most organic pollutants [6]. Consequently, only a small fraction (< 5%) of molecular O₂ in reaction systems is catalytically converted into ROS, starkly contrasting with the high efficiency achieved by synthetic precursors like H₂O₂ or O₃ [7,8]. Consequently, overcoming these limitations requires innovative strategies to activate molecular O₂ under ambient conditions, thereby unlocking its full potential as a sustainable ROS source.

  Conventional methods for molecular O₂ activation, including physical, chemical, and biological approaches, have been widely used to generate ROS for environmental remediation [9,10]. Despite their utility, these approaches often suffer from critical limitations, such as energy-intensive operations, secondary pollutant formation, and unsatisfactory activation efficiency, which hinder their practical application [11,12]. In response, (photo)electrocatalytic activation of molecular O₂ have emerged as a breakthrough strategy [13,14]. By synergistically integrating photonic and electrical energy, this approach breakthrough the inherent spin limitations of triplet-state molecular O₂, enabling efficient ROS generation under mild conditions [15]. A key advantage of (photo)electrocatalytic systems is their compatibility with renewable energy inputs, such as solar or wind power, which ensures cost-effective scalability for large-scale environmental remediation [16,17]. Therefore, the main orientation of this paper is the analysis of molecular O₂ activation via (photo)electrocatalytic system, offering a sustainable pathway to mitigate pollution.

  Recent reviews for molecular O₂ activation have predominantly concentrated on catalyst design, with particular emphasis on optimizing electronic properties and electron transfer through advanced strategies, including defect engineering, heterojunction construction, and nanoconfinement [18,19]. For instance, N-doped carbon electrodes derived from N-rich chitosan demonstrated enhanced molecular O₂ activation, where pyridinic and graphitic N sites doubled ^•OH and O₂^•− concentrations, enabling efficient degradation of persistent pollutants like perfluorooctanoic acid in complex aqueous environments [20]. Similarly, MIL-101(Fe)-based Z-scheme heterojunctions exhibited robust photocatalytic activity, generating O₂^•−, ^•OH, and ¹O₂ to drive pollutant degradation [21]. While, seldom of the reviews have targeted at the state-of-the-art (photo)electrocatalysts together with the (photo)electrochemical device designs for molecular O₂ activation. By addressing these limitations, this review bridges the disconnect between material-centric studies and device engineering, offering a holistic perspective on (photo)electrocatalytic molecular O₂ activation.

  Herein, we provide a thorough summary and in-depth analysis of current research on molecular O₂ activation utilizing (photo)electrocatalysts and (photo)electrochemical devices, with an emphasis on enhancing mass and electron transfer for the on-site production of ROS. Specifically, the reaction mechanisms, advantages, and disadvantages of (photo)electrochemical device designs, particularly those incorporating catalyst engineering, are analyzed in detail. By connecting fundamental science with practical engineering considerations, this work aims to inspire the development of robust, scalable technologies for ROS generation toward environmental remediation.

  2. Analysis of the molecular O₂ activation pathway

  ROS are broadly classified into two types: radicals (e.g., O₂^•− and ^•OH) and nonradicals (e.g., H₂O₂ and ¹O₂). The ground-state molecular O₂ (³Σ_g^-) exists in its lowest energy configuration, characterized by two unpaired electrons in antibonding orbitals with parallel spins (Figs. 1A and B) [22,23]. Contrast, excited-state oxygen species, such as O₂ (¹Δ_g) and O₂ (¹Σ_g⁺), possess distinct electronic configurations. In the O₂ (¹Δ_g) state, electrons pair within the same orbital, whereas in the higher-energy O₂ (¹Σ_g⁺) state, occupy separate orbitals with opposite spins. The O₂ (¹Σ_g⁺) state is highly unstable and short-lived, rapidly decaying to the lower-energy O₂ (¹Δ_g) state, which corresponds to the widely studied ¹O₂ [23,24]. Additionally, O₂^•− and ^•OH are often less stable than nonradicals because they contain a single unpaired electron.

  Figure 1

  

  Figure 1.  (A) Molecular orbital diagrams for a ground-state triplet O₂ molecule, two types of ¹O₂, O₂^•− and H₂O₂. Copied with permission [23]. Copyright 2013, Wiley. (B) Possibly the relationship or differences in generating different ROS.

  DownLoad: Full-Size Img PowerPoint

  2.1 Superoxide radical anion (O₂^•−)

  O₂^•− is a highly reactive intermediate formed when molecular O₂ undergoes single-electron reduction. Structurally, the O—O bond in O₂^•− (1.28–1.33 Å) is slightly elongated than that of ground-state O₂ (1.21 Å) due to the addition of an electron [25]. This structural enhances its instability and reactivity, enabling O₂^•− act as both an oxidant and a reductant across a broad redox potential range (−0.35 V to +0.94 V) in photo(electro)catalytic systems. As shown in Fig. 2A, a notable example involves Fe atoms coordinated within N-rich carbon frameworks (Fe–N₄), which can stabilize molecular O₂ adsorption and facilitate electron transfer from the Fe sites to O₂, yielding O₂^•− for ethylbenzene oxidation [26]. Although single-electron oxygen reduction dominates its formation, O₂^•− also serves as a precursor to other ROS, including H₂O₂ and ¹O₂, through further reduction or coupling steps. By optimizing electron transfer mechanisms and stabilizing reactive intermediates, catalytic systems can be engineered to enhance O₂^•− generation and environmental remediation processes.

  Figure 2

  

  Figure 2.  (A) Calculated charge difference surfaces of the Fe–N_x coordination systems, calculated energy of molecular O₂ adsorption on different sites, and schematic illustration of O₂ to O₂^•−. Copied with permission [26]. Copyright 2013, Wiley. (B) Proposed electrocatalytic activation mechanism of molecular O₂ to ¹O₂ over the CoFeCNT filter. Copied with permission [30]. Copyright 2025, Elsevier. (C) Selective electrocatalytic reduction of O₂ to ^•OH via 3e⁻ pathway with FeCo alloy encapsulated carbon aerogel. Copied with permission [32]. Copyright 2021, Wiley. (D) Binary heteroatom dopants enable carbon-based heterostructures for efficient (photo)electrocatalytic O₂ to H₂O₂ synthesis. Copied with permission [36]. Copyright 2024, Elsevier.

  DownLoad: Full-Size Img PowerPoint

  2.2 Singlet oxygen (¹O₂)

  ¹O₂ is an energized form of the molecular O₂, which was produced by elevating ground-state oxygen (³Σ_g⁻) to higher-energy configurations [27]. Two distinct excited states are possible: the lower-energy ¹Δg state (95 kJ/mol) and the higher-energy ¹Σ_g⁺ state (158 kJ/mol). The ¹Δ_g state forms when two electrons with opposing spins occupy the same π orbital, whereas the ¹Σ_g⁺ state arises when these electrons occupy separate orbitals [28]. Due to spin conservation constraints, transitions from the ¹Δ_g to the ¹Σ_g⁺ state are prohibited. In contrast, the ¹Σ_g⁺ state can revert to the ground state through a permitted transition, leaving the ¹Δg state relatively stable as "¹O₂" [29]. For instance, Jiang et al. demonstrated that alloying Co with Fe promotes electron transfer from Co to Fe, creating electron-deficient Co sites that efficiently catalyze O₂-to-¹O₂ conversion in flow-through systems Fig. 2B) [30]. However, the efficiency of ¹O₂ production is constrained by the fleeting lifespan of excited-state species, which limits their ability to drive molecular O₂ activation. Addressing this challenge requires strategies to stabilize these transient states, thereby improving the overall yield and utility of ¹O₂ in environmental remediations.

  2.3 Hydroxyl radical (^•OH)

  ^•OH consist of a H atom bonded to an O atom, enabling them to extract H atoms from surrounding molecules to form water, thereby destabilizing contaminants [31]. The activation of molecular O₂ to generate ^•OH typically begins with a one-electron reduction step, producing O₂^•−. These intermediates can follow two primary pathways: direct dissociation and hydrogenation to form ^•OH, or further reduction and hydrogenation to yield H₂O₂. The latter route is often kinetically favorable, as it avoids the energy-intensive cleavage of the O–O bond, with H₂O₂ subsequently reacting with Fe²⁺ or other transition metal ions in a Fenton-like process to produce ^•OH. Recent advances highlight that coupling transition metals (e.g., Fe, Co, Ni, Cu) with carbon-based materials can precisely tune the selectivity of ^•OH generation via 3e^- oxygen reduction reaction (ORR) [5]. For example, Xiao et al. developed a carbon aerogel embedded with a core-shell FeCo alloy, where carboxyl groups on the carbon surface promoted H₂O₂ production via the 2e^- ORR pathway, while electrons from the FeCo alloy facilitated the reduction of H₂O₂ to ^•OH through a 1e^- process by altering the local electronic environment of the carbon shell (Fig. 2C) [32]. This cascade mechanism, converting O₂ to ^•OH through 3e^- ORR process, overcomes the slow electron transfer bottleneck in traditional Fenton redox cycles, significantly improving efficiency.

  2.4 Hydrogen peroxide (H₂O₂)

  H₂O₂, in which the two O atoms are bonded via a shared electron pair, forms an inherently unstable and transient −O—O− structural unit, serving as both oxidizing and reducing agents [33]. Utilizing solar energy and O₂ as feedstocks, photo(electro)catalytic H₂O₂ generation surpasses traditional photocatalysis in solar energy conversion efficiency and requires less electrical input than standalone electrocatalysis [34]. Photocathodes absorb light and produce carriers (e⁻ and h⁺), which are transferred to the catalyst surface. Molecular O₂ initially combines with 1e⁻ to form O₂^•−, which subsequently react with 2H⁺ and another 1e⁻ to produce H₂O₂ via a two-step 1e⁻ transfer pathway [35]. Alternatively, O₂ directly interacts with 2e⁻ and 2H⁺ to generate H₂O₂ through 2e⁻ transfer mechanism. Recent work by Zhu et al. demonstrated a metal-free carbon-based heterostructure material with N, S co-doping, in which binary heteroatom-doped carbon spheres are loaded onto carbon nitride polymer (Fig. 2D) [36]. Theoretical calculations and experimental results reveal that S- and N-sites suppress O–O cleavage and facilitate *OOH formation through their respective subtle microenvironments, mediating superior 2e^- ORR selectivity for H₂O₂ production in acidic and alkaline conditions. A critical challenge in H₂O₂ (photo)electrosynthesis lies in suppressing H₂O₂ self-decomposition. Mitigation strategies include physically isolating H₂O₂ from the counter electrode to minimize unwanted reactions or converting H₂O₂ into stable peroxide salts to reduce decomposition while enabling controlled release, storage, and transport.

  3. The (photo)electrocatalysts for molecular O₂ activation

  Advances in synthesis techniques and computational modeling have enabled the development of diverse (photo)electrocatalysts, including metal alloys, single-atom materials, and carbon-based systems, to drive the activation of molecular O₂ for ROS generation.

  3.1 Metal and alloy catalysts

  Noble metals, such as Au, Ag, Pd, and Pt, are widely studied in molecular O₂ activation within (photo)electrocatalytic systems [37]. However, their high binding affinity to molecular O₂ often demands significant energy input, creating a trade-off between catalytic activity and selectivity. To address this, researchers have engineered noble metal alloys, which adjust the O-binding properties to enhance catalytic performance by incorporating secondary metals. For example, Pd-Ag, Pd-Cu, Pt-Mn, and Au-Pd/ZnO hybrid systems exhibit superior molecular O₂ activation compared to single-metal catalysts, as demonstrated by experimental and theoretical studies [38,39]. As illustrated in Fig. 3A, analyses of the stable O₂ adsorption configurations and dissociation pathways on PdCu alloy surfaces reveal a composition-dependent trend: O₂ adsorption strength increases with higher Cu content but weakens when Cu exceeds Pd stoichiometry [39]. This behavior arises from the higher O₂ adsorption energy of Pd sites relative to Cu, confirming Pd as the preferential adsorption center in the alloy. Further innovation involves tailoring redox-active interfaces. Yang et al. designed a CuBi₂O₄ photocathode that exploits alternating Cu⁺/Cu²⁺ states to drive sequential molecular O₂ activation: Photogenerated electrons reduce O₂ at Cu²⁺ sites to form O₂^•−, which then react with protons and electrons to produce H₂O₂, an intermediate that subsequently interacts with Cu⁺ to yield ^•OH [40]. Another promising strategy involves nanoscale engineering of catalysts. Reducing metal nanoparticles to ultrasmall sizes, such as Pt clusters anchored on porous supports, significantly increases surface area, and improves selectivity, thereby enhancing the activation efficiency of molecular O₂ [41,42]. However, subsequent research should focus on the challenges of precious metal scarcity/cost, susceptibility to surface poisoning by intermediates (e.g., *OH, OOH), and controlling alloy homogeneity. Collectively, advancements in alloy design, nanostructure, and redox mediation highlight a multifaceted pathway toward developing scalable, energy-efficient systems for optimized molecular O₂ activation.

  Figure 3

  

  Figure 3.  (A) Transition states in the O₂ dissociation on four PdCu alloy surfaces. Copied with permission [39]. Copyright 2024, ACS. (B) Mechanism for (photo)electrocatalytic O₂ to ^•OH over Pd-SA/F-TiO₂ and Pd NPs/F-TiO₂. Copied with permission [49]. Copyright 2024, ACS. (C) N-doped graphitized carbon nanohorns in highly selective O₂ to H₂O₂. Copied with permission [54]. Copyright 2018, Elsevier. (D) The scheme for O₂ to H₂O₂ at NiO_x-TiO₂-PCN photocathode. Copied with permission [56]. Copyright 2023, Elsevier. (E) The mechanism of O₂ to H₂O₂ on Py-TD-COF and Py-TD-COF-NH. Copied with permission [59]. Copyright 2024, Elsevier.

  DownLoad: Full-Size Img PowerPoint

  3.2 Single atoms catalysts (SACs)

  SACs, characterized by near-complete metal atom utilization, have emerged as a promising platform for efficient molecular O₂ activation [43,44]. Their exceptional catalytic performance stems from the ability to precisely control electronic properties by tailoring interactions between isolated metal atoms and their support materials, which enhances the adsorption and activation of molecular O₂ and regulates its dissociation pathway to selectively form ROS [45,46]. Until now, various SACs based on different metals, including Fe, Co, Mn, Pt, and Pd, has been investigated for activation of molecular O₂ toward ROS generation [47]. Furthermore, modifying the coordination environment of SACs with heteroatom (e.g., N, O, S) optimizes electron density at active sites, further improving molecular O₂ activation efficiency [48]. For instance, Zhang et al. developed a F-modified TiO₂ catalyst with single Pd atoms bonded to F and O atoms [49]. This configuration promoted the 2e⁻ ORR to H₂O₂ generation with up to 99% selectivity, while the additional channel bond HO−O···Pd−F−TiO₂ facilitates the photogenerated electron transfer from the conduction band to Pd sites, reducing Pd···O−OH to ^•OH (Fig. 3B). Similarly, Qin et al. designed carbon-supported Ni SACs (Ni SA@C) featuring dual catalytic sites, where the Ni–N₄ (−2.70 eV) sites exhibited stronger adsorption energy for molecular O₂ than that of Ni–N₂O₂ sites (−1.59 eV), suggesting superior molecular O₂ activation at Ni–N₄ sites [50]. This was complemented by planar-averaged charge density difference analyses along the Z-direction, revealing facilitated directional electron transfer from Ni–N₄ sites to O₂ for ROS generation. Subsequent studies should focus on the limitations of SACs, such as stability issues (metal leaching/aggregation) and sensitivity to the coordination environment, where suboptimal ligand bonding can affect performance. These examples underscore how strategic engineering of SACs at the atomic scale can optimize interfacial adsorption and charge transfer dynamics, significantly advancing (photo)electrochemical systems for molecular O₂ activation.

  3.3 Carbon-based catalysts

  The high cost and limited durability of noble metal-based catalysts under harsh operational conditions remain significant barriers to their widespread adoption [51,52]. In response, carbon-based materials have gained prominence as cost-effective alternatives for molecular O₂ activation, offering robust stability and competitive catalytic performance for ROS generation [53]. For instance, defect-engineered polymeric carbon nitrides, where structural vacancies serve as active sites, efficiently adsorbs and directly converts molecular O₂ into H₂O₂ (Fig. 3C) [54]. Chemical doping of carbon materials further enhances their electronic properties, with catalytic activity dependent on the type and concentration of dopants. Chen et al. showed that carbonyl group-functionalized carbon nanotubes lower the energy barrier for molecular O₂ activation, thereby enhancing the production of ¹O₂, O₂^•−, and ^•OH, with ¹O₂ generation remaining stable across a broad pH range (7–11) [55]. Similarly, Braun et al. achieved unprecedented H₂O₂ yields using thermally treated carbon catalysts derived from polydopamine, attributing this success to optimized microporous structures and balanced N_pyridinic/N_pyrrolic ratios during synthesis (Fig. 3D) [56]. Besides N doping, incorporating heteroatoms like B, F, and S, have also been incorporated to improve the efficiency and selectivity of molecular O₂ activation.

  Among them, covalent organic frameworks (COFs), a class of porous carbon polymers with high surface areas and tunable pore structures, exhibit exceptional thermal and structural stability during molecular O₂ activation [57,58]. For instance, imine-linked COFs, which feature electron donor-acceptor characteristics, demonstrated 80%−92% selectivity for H₂O₂ production at low operating voltages (0.2–0.7 V vs. RHE) (Fig. 3E) [59]. Experimental and computational studies revealed that amine linkages in these frameworks promote H-bonding, elongating, and weakening O—O bonds to facilitate hydrogenation and enhance H₂O₂ selectivity in (photo)electrocatalytic systems. Current optimization strategies focus on tailoring functional groups, incorporating heteroatoms, and engineering linkage chemistry, all of which directly modulate catalytic pathways. Nevertheless, scalability remains constrained by complex synthetic requirements and limited electrical conductivity.

  3.4 (In)organic semiconductor photocathodes

  Organic variants, such as H-bonded frameworks and polyterthiophene polymers, contrast with inorganic counterparts like such as CuBi₂O₄, Cu₃BiS₃, and CuS/MnS. For example, Feng et al. designed a Cu-coordinated carbon-nitride (CuCN) photocathode, where Cu-N_2–C active sites reduced the energy barrier for critical reaction intermediates, significantly enhancing molecular O₂ activation for the generation of ^•OH and ¹O₂ via Cu-N_2–C active sites [60]. This approach improved efficiency by balancing charge transfer and catalytic activity. Sun et al. demonstrated this by integrating a NaCo₂O₄ layer beneath CuBi₂O₄, synergizing photothermal and photoelectric effects [61]. Under light exposure, this configuration created an internal electric field that boosted charge separation, achieving a notable H₂O₂ yield of 1192.9 µmol/L within 50 min. Such designs underscore the potential of hybrid systems to overcome limitations in charge-carrier dynamics. Beyond these strategies, Schottky junctions have emerged as promising candidates due to their ability to enhance charge separation [11]. These junctions form when semiconductors with differing work functions interact, driving electron transfer until equilibrium, which minimizes electron recombination, enhances charge separation, and further improves molecular O₂ activation. In addition to these approaches, researchers have explored oxides, chalcogenides, anthraquinone-based compounds, and molecular catalysts (Table 1) [62–70], all demonstrating strong molecular O₂ activation capabilities. These materials leverage diverse mechanisms, from bandgap engineering to surface modification, offering versatile pathways to improve molecular O₂ activation.

  Table 1

  

  Table 1.  Summary of ROS yields and their detection by photocatalysts applying different detection method.

  DownLoad: CSV

  Catalyst	Main ROS	Activity (mA/cm2)	Durability	Ref.
  B0.05—C3N4	O2•−, 1O2	/	3 h	[62]
  GBN	1O2	3 V vs. RHE	2 h	[63]
  MoS2/GDY-3	H2O2	/	4 h	[64]
  Co3O4-TiO2	H2O2	1.23 V vs. RHE	4 h	[65]
  CuxO@C	O2•−, •OH	−0.1 V vs. RHE	3 h	[66]
  Cu/TiO2	O2•−	−0.1 V vs. RHE	14 h	[67]
  Gd-doped BiVO4	H2O2	3.0 V vs. RHE	3 h	[68]
  Ni(OH)2/Ni foam	•OH	1.60 V vs. RHE	40 h	[69]
  Cu2O/CuO	•OH	2.2 V vs. RHE	8 h	[70]

  4. Strategies to improve the stability of (photo)electrocatalysts

  The stability of (photo)electrocatalysts is influenced by three primary factors: Catalyst composition, reaction conditions, and reactor design. Optimizing the catalyst itself presents a cost-effective pathway to enhance operational longevity without requiring costly infrastructure upgrades.

  4.1 Coating on oxide supports

  A promising strategy involves coating catalytic sites onto supports, such as TiO₂, ZnO, or SiO₂. These supports offer high surface area and structural stability, which help to anchor active sites and prevent leakage. While non-oxide supports (e.g., carbon nanotubes) may offer higher conductivity, their vulnerability to corrosion often results in activity loss in acidic O₂-rich environments. Thus, oxides remain preferred for long-term molecular O₂ activation. For example, Lyu et al. demonstrated that a carbon-coated TiO_2−x support stabilized interfacial oxygen vacancies, enabling activate molecular O₂ into ¹O₂ or O₂^•− during a 12 h of light exposure (Fig. 4A) [71]. Similarly, Zhang et al. addressed the challenge of Au nanoparticle sintering by applying a thin SiO₂ layer to Au/TiO₂ [72]. This modification not only prevented agglomeration but also enhanced molecular O₂ activation efficiency under prolonged operation. Protective oxide coatings mitigate structural hydrolysis, collapse, and metal leaching, thereby preserving catalytic activity. However, the coating method must be carefully tailored to ensure compatibility between the active sites, support material, and operational environment. For example, excessive coating thickness or improper thermal treatment during synthesis can block active sites or induce stress fractures. Thereby balancing material compatibility with functional design, these strategies pave the way for durable, high-performance (photo)electrocatalysts in molecular O₂-driven systems.

  Figure 4

  

  Figure 4.  (A) Self-driven ROS generation from O₂/H₂O molecules via interfacial oxygen vacancies on carbon-coated TiO_2–x. Copied with permission [71]. Copyright 2020, ACS. (B) The scheme for O₂ to ¹O₂ and the ratio of O₂ to individual ROS in the SA-Cu_1.0CA electrocatalytic system. Copied with permission [73]. Copyright 2023, ACS. (C) Illustration of ¹O₂ triggered by O₂^•– in the Mo cocatalytic Fenton reaction with enhanced REDOX activity. Copied with permission [74]. Copyright 2019, ACS. (D) Illustration of O₂ to O₂^•– by synergy of dual defects in dual defects in cubic BiO_2−x. Copied with permission [76]. Copyright 2019, Wiley. (E) Possible activation mechanism of molecular O₂ in the Fe₂O₃-in-CNT and Fe₂O₃-out-CNT system. Copied with permission [78]. Copyright 2021, ACS.

  DownLoad: Full-Size Img PowerPoint

  4.2 Strengthening metal-substrate interaction

  The prolonged catalytic performance of materials used in molecular O₂ activation can be substantially enhanced by optimizing the interaction between catalytic sites and their supporting substrates. Key approaches, such as heteroatom doping, oxygen vacancy engineering, and interfacial modifications, which mitigate structural collapse including metal leaching or agglomeration.

  Heteroatoms doping: Incorporating heteroatoms such as N, B, and S into catalyst frameworks has proven effective in stabilizing catalytic sites. For instance, N-doped Cu-carbon aerogel electrodes create a N_4–Cu-OOH interface that selectively generates ¹O₂ at concentrations up to 2583 µmol/L during molecular O₂ activation, while maintaining 98% efficiency over 50 operational cycles (Fig. 4B) [73]. Additionally, doping can also introduce redox-active metal pairs, such as Fe³⁺/Fe²⁺ and Mo⁴⁺/Mo⁶⁺, which directly participate in molecular O₂ activation. The Fe³⁺/Fe²⁺ pair facilitates the one-electron reduction of O₂ to O₂^•−, while Mo⁴⁺/Mo⁶⁺ further converts O₂^•− to ¹O₂ (Fig. 4C) [74]. Research should focus how to precisely control the dopant species at the molecular level during the preparation process and adjust the distribution of different species according to the specific requirement and goals.

  Oxygen vacancies (OVs): OVs, defects created by missing O atoms in metal oxide lattices, enhance molecular O₂ activation by altering surface electronic states and improving charge transfer for surface reactivity [75]. For example, Mao et al. reported that C-doped BiO_2-x, engineered with abundant surface OVs and bulk oxygen substitutions, synergistically strengthened molecular O₂ adsorption and charge carrier separation for boosting photocatalytic O₂^•− generation (Fig. 4D) [76]. In another study, Yang et al. synthesized SnO₂ catalysts with abundant OVs, which demonstrated higher oxygen evolution potentials and faster electron transfer rates, resulting in greater ^•OH yields compared to unmodified SnO₂ [77]. Collectively, these strategies highlight OVs engineering is an effective approach to tune the electronic structure of (photo)electrocatalysts and the surface absorption/desorption of reactants toward molecular O₂ activation.

  Confining metal particles into the support: Advanced strategies, such as embedding or encapsulating metal particles within porous matrices, have emerged to address challenges like structural collapse and metal leaching during catalytic reactions. For example, Guo et al. embedded Fe₂O₃ nanoparticles directly into the framework of a carbon nanotube (CNT) filter, termed Fe₂O₃-in-CNT, which reduced iron leaching by a factor of 18.8 compared to a surface-coated counterpart (Fe₂O₃-out-CNT), releasing only 1.6 × 10⁻² mg/L of iron vs. 0.3 mg/L [78]. Beyond enhancing catalyst durability, spatial confinement within support materials can reshape reaction mechanisms. The embedded configuration improved ROS generation by enabling in situ H₂O₂ production and its conversion to ¹O₂ via iron redox cycling. In contrast, the surface-coated filter primarily generated ^•OH, highlighting how nanoparticle stability and spatial arrangement influence ROS pathways (Fig. 4E). Such findings emphasize the critical role of nanostructure engineering in directing molecular O₂ activation.

  Collectively, the strategies outlined above target the primary mechanisms of (photo)electrocatalyst deactivation: metal leaching, particle agglomeration/sintering, and structural collapse of the support. By mitigating these failure modes, these approaches significantly enhance operational longevity, which is paramount for the practical implementation of (photo)electrocatalytic processes in environmental remediation.

  5. (Photo)electrocatalytic device for molecular O₂ activation

  The development of stable and efficient photo(electro)catalytic system is one of the important components to industrial applications. The ideal device for molecular O₂ activation must balance key factors: sufficient active sites and efficient mass transfer to sustain high reaction rates.

  5.1 Flow-by (photo)electrochemical systems

  Conventional (photo)electrocatalytic approaches for ROS generation typically rely on stirred reactors with suspended catalyst powders, which require continuous molecular O₂ supply to maintain reaction efficiency. While effective in laboratory settings, these systems face practical limitations, including inefficient mass transfer, catalyst aggregation and recovery during prolonged operation [79,80]. To address these drawbacks, recent advancements have focused on immobilizing catalyst, such as fixed-bed reactors or electrode-integrated platforms. These configurations not only prevent catalyst loss and simplify reuse but also enhance scalability for real-world applications. For example, Fuku et al. developed a dual-compartment solar-driven reactor using a WO₃/BiVO₄ photoanode and an Au cathode [81]. This solar-driven dual-compartment system enables simultaneous H₂O₂ generation at both electrodes without requiring external voltage, with the photoanode and cathode achieving Faradaic efficiencies of 50% and 90% for H₂O₂ production, respectively. In a separate study, Zhou et al. demonstrated that immobilized BiVO₄-based composites functionalized with silver or copper nanoparticles generated ^•OH and O₂^•- at rates of 371 and 292 µmol g^-1 h^-1, respectively in an H-cell reactor separated by a Nafion 115 ion exchange membrane (Fig. 5A) [82]. A critical advantage of immobilized systems lies in their inherent charge-directed pollutant migration. Positively charged contaminants are electrostatically attracted to the negatively charged cathode, while anionic species migrate toward the photoanode. This directional transport ensures continuous adsorption of pollutants onto the electrode surfaces, where they undergo oxidative degradation via in situ-generated ROS. Such mechanisms not only enhance degradation efficiency but also minimize energy losses associated with random diffusion in conventional slurry reactors. By replacing suspended catalysts with immobilized architectures, researchers have resolved longstanding challenges in catalyst stability, recyclability, and energy efficiency, paving the way for scalable, energy-efficient ROS generation systems for environmental remediation.

  Figure 5

  

  Figure 5.  (A) Illustration of charge transfer processes for ROS generation on a flow-by BiVO4:I/BTO-metal electrode. Copied with permission [82]. Copyright 2021, Wiley. (B) Schematic illustrating the cross-flow filtration system. Copied with permission [85]. Copyright 2020, Springer Nature. (C) Schematic of the Sb-SA/MXene photocatalytic surface with H₂O₂ generation and the Fe-NC/MXene electrocatalytic surface with ^•OH production concurrently. Copied with permission [86]. Copyright 2023, Elsevier. (D) The (photo)electrocatalysis assisted filtration process occurring in the interaction of pollutant solute and reactive oxygen species generated in cross flow filtration mode. Copied with permission [87]. Copyright 2022, Elsevier.

  DownLoad: Full-Size Img PowerPoint

  5.2 Flow-through (photo)electrochemical systems

  Flow-through (photo)electrochemical systems have emerged as a paradigm shift in molecular O₂ activation for ROS generation, addressing critical limitations of traditional flow-by configurations, including inefficient mass transport, sluggish reaction kinetics, and unstable electrode performance [83,84]. By integrating layered membrane architectures, these systems create confined reaction zones at the nanoscale, where optimized electrical and hydrodynamic interactions minimize diffusion barriers, key features for practical environmental applications (Fig. 5B) [85]. A notable innovation in this field is the Sb-Fe/MXene Janus filter, which features a Ni-foam with spatially separated functional layers: a photocatalytic side (Sb single atoms on MXene) and an electrocatalytic side (Fe nanoclusters on MXene) [86]. The photocatalytic side drives water oxidation via a 4e^- pathway to generate O₂, while concurrently enabling non-sacrificial H₂O₂ production via a 2e^- oxygen reduction process. On the opposing side, the electrocatalytic side mediates Fe(Ⅲ)/Fe(Ⅱ) cycling, sustaining ^•OH generation through H₂O₂ dissociation (Fig. 5C). Further progress has been achieved through microporous (photo)electrocatalytic membranes, such as those developed by Kumari et al., where TiO₂ and ZnO coatings are deposited onto porous stainless-steel substrates using atomic layer deposition (Fig. 5D) [87]. These membranes suppress electron-hole recombination during (photo)electrocatalysis, significantly boosting the production of ^•OH and O₂^•−. Operation in the flow-through configuration outperformed the conventional batch reactor model (k = 0.17 min⁻¹) due to convection-enhanced mass transport. As compared to a conventional batch reactor, the flow-through (photo)electrochemical system exhibited > 4 times higher oxidation flux due to the convection-enhanced mass transfer and improved catalyst stability [84]. Such engineered decoupling of redox reactions not only enhances molecular O₂ activation efficiency but also ensures long-term stability under continuous operation, underscoring their viability for scalable environmental remediation technologies.

  6. Application in environmental protection

  The activation of molecular O₂ to generate ROS has gained significant attention in environmental remediation due to its cost-effectiveness, eco-friendly nature, and potent oxidative capacity, offering a sustainable alternative to conventional remediation methods.

  6.1 Pollutants degradation

  Persistent organic pollutants, including antibiotics and pharmaceuticals, threaten ecosystems and human health due to their resistance to natural degradation [88]. Central to this process is the identification of molecular sites to oxidative attack. Computational tools like the Fukui index provide critical insights by quantifying atomic reactivity within a molecule [89]. Three indices are pivotal:

  ${\rm{Nucleophilic}}\ {\rm{attack}}\ {\rm{index}}\ f^+ =q({\rm{N}} + 1) – q({\rm{N}})$

  (1)

  ${\text{Electrophilicity attack index}}\ f^− = q({\rm{N}}) – q({\rm{N}} – 1)$

  (2)

  ${\text{Free radical attack index}}\ f^0 = 1/2[q({\rm{N}} +1) – q({\rm{N}} – 1)]$

  (3)

  where q represents the electron density of the atom, and the electron density values q(N), q(N-1) and q(N + 1) are derived from Natural Bond Orbital charge calculations [3]. Regions with higher f ⁺ or f ⁰ values are prone to nucleophilic radicals like ^•OH, while sites with elevated f ^– value are targeted by electrophilic species such as ¹O₂. For example, computational analyses of the antibiotic tetracycline (TC) reveal that ^•OH radicals preferentially attack C, O, and N atoms (e.g., C8, O23, N29), forming hydroxylated intermediates. Simultaneously, O₂^•− cleave critical bonds at high- f ⁰ sites (e.g., C7, O23, N29), initiating ring-opening reactions at positions such as C16 and C18. These fragmented intermediates are further oxidized to harmless end products like CO₂ and H₂O [11]. Recent advances highlight the synergy between experimental systems and computational modeling. For instance, a TiO₂ photoanode with engineered surface facets demonstrated enhanced cleavage of resilient carbon-fluorine bonds in fluorinated pollutants under (photo)electrocatalytic conditions, leveraging ^•OH radicals for efficient degradation [90]. Demonstrating practical applicability, a modified biochar electrocatalytic system achieved effective degradation of diverse emerging contaminants in real wastewater [91]. Driven primarily by ^•OH, O₂^•–, and ¹O₂ generation from molecular O₂ activation, removal efficiencies reached 59.2% (antibiotics), 89.2% (quinoline), 88.3% (atrazine), and 100% (bisphenol A) within a 5 h treatment period. Such integrated approaches enable precise targeting of pollutants, underscoring the potential of ROS-mediated pollutants remediation.

  6.2 Sterilization and disinfection

  Traditional sterilization and disinfection methods, such as chlorine, ozone, or persulfate treatments, often produce harmful chemical byproducts, posing risks to ecosystems and human health [92]. In response, ROS generated through (photo)electrocatalytic activation of molecular O₂ have emerged as a safer, residue-free alternative, aligning with the growing demand for sustainable solutions in environmental remediation [93,94]. Materials such as metals (e.g., Ag, Cu), metal oxides (e.g., TiO₂, CuO, Fe₂O₃, ZnO), and carbon-based structures (e.g., graphene oxide, carbon nanotubes) can efficiently activate molecular O₂ for ROS generation to disrupt microbial cells while reducing ecological risks. For example, Shi et al. designed a (photo)electrochemical system that combines light-driven oxidation with on-site H₂O₂ production, powered by renewable energy, to disinfect water contaminated with Escherichia coli [95]. This flow-through method achieved bacterial inactivation rates 2.19 times higher than conventional electrochemical H₂O₂ production and 2.11 times greater than standalone flow-by processes, highlighting the enhanced efficiency of ROS-mediated mechanisms such as h⁺, ^•OH, O₂^•−. Further demonstrating the resilience of ROS-based systems, Valenzuela et al. demonstrated the capacity of electrospray TiO₂/Ce-TiO₂/graphene oxide anodes in a (photo)electrocatalytic arrangement to inhibit bacterial growth and boost ROS generation even under the very unfavorable conditions [96]. By leveraging renewable energy to eliminate harmful byproducts and, this approach offers a promising pathway for advancing environmental remediation technologies.

  6.3 Volatile organic compounds (VOC) oxidation

  Unlike pollutants dissolved in water, the effective elimination of gas-phase VOCs necessitates managing complex multiphase interactions (gas, liquid, solid) [97]. A critical prerequisite for degradation via ROS is the efficient transfer of VOCs from exhaust streams to the gas-liquid interface, followed by their dissolution into the aqueous phase. Crucially, because VOC oxidation operate continuously, ROS generation must occur rapidly and with sufficient persistence to sustain the degradation process, a stark contrast to batch-based treatments used for water pollutants. For example, Qiao et al. developed a (photo)electrocatalytic system that integrates a gas-permeable electrode with a light-responsive anode, effectively converting toxic H₂S into benign sulfur and H₂O₂ [98]. This design achieved rapid molecular O₂ diffusion and selective 2e^- ORR, enabling a high H₂O₂ production rate of 0.8 mmol L^-1 h^-1 at the cathode. Meanwhile, the anode fully oxidized H₂S to sulfur using an iodide-based mediator, recovering sulfur at 0.60 mmol/h. Thus, the gas diffusion electrode was pivotal in this process, establishing optimized gas/water interfaces within its porous structure. This configuration significantly enhances the transport of electrons, ions, and gaseous reactants, thereby accelerating reaction kinetics. Furthermore, the formation of gas cavities adherent to the electrode surface can promote the more effective oxidation of dissolved species. Collectively, such engineered interfaces enable continuous VOC degradation via ROS-driven mechanisms, providing a promising foundation for scalable and sustainable strategies to mitigate gas-phase environmental pollutants.

  6.4 Soil preservation

  Great concern has been paid to polycyclic aromatic hydrocarbons (PAHs) polluted soil because they can enter the food chains and post health risks on ecological receptors [99]. However, conventional methods require substantial exogenous supplies of H₂O₂, incurring high costs and logistical challenges related to its procurement, transport, and storage. To address these limitations, (photo)electrocatalytic in-situ H₂O₂ generation has emerged as a critical advancement. Particularly, gas diffusion electrodes consisting of a catalyst layer and a diffusion layer have been just developed in last decade, which can continuously produce H₂O₂ via 2e^- ORR, offering significant potential for environmental remediation. Chu et al. engineered a modified ordered mesoporous carbon-based gas diffusion electrode, which can in-situ produce H₂O₂ and further to generate ^•OH radicals locally [100]. The practicability of the system was further validated using a contaminated soil spiked with anthracene, pyrene, and benzanthracene; remediation efficiency was quantified, and PAH degradation pathways were elucidated. Additionally, ROS exert a dual influence on soil carbon cycling: They enhance organic matter mineralization while simultaneously promoting long-term carbon stabilization through increased recalcitrance of residual organic matter and organo-mineral complex formation [101]. Consequently, understanding ROS generation from molecular O₂ activation and their subsequent effects on organic matter transformation represents an emerging research frontier, providing fundamental chemical insights into oxidative enzymatic processes central to soil biogeochemistry.

  7. Conclusions and outlook

  This review comprehensively examines the molecular O₂ activation pathways in (photo)electrocatalytic systems and elucidates the mechanisms driving ROS generation. By highlighting efficient ROS-producing catalysts, we clarify correlations between activation routes and dominant ROS species while establishing a framework for optimizing catalyst design and reaction device. To bridge lab-scale innovations and industrial deployment, future work must address four key challenges:

  (1) Scalable catalyst synthesis. While oxygen vacancy engineering, heterojunctions, and doping enhance molecular O₂ activation, these strategies face intrinsic constraints: doping may block active sites, and facet engineering often involves complex syntheses. Prioritizing scalable, low-cost production of robust catalysts is imperative. Refining synthesis protocols to balance performance with industrial manufacturability will be essential for overcoming efficiency-stability trade-offs, particularly in large-scale environmental or energy applications.

  (2) Enhancing ROS yields. The low solubility of molecular O₂ in electrolytes often limits ROS yields in (photo)electrocatalytic processes. Integrating gas diffusion (photo)electrodes to create a stable gas-electrode-electrolyte interface could enhance molecular O₂ transport. Combining computational modeling with catalyst optimization will improve molecular O₂ delivery to catalytic surfaces and promote intermediates desorption, thereby increasing ROS yields.

  (3) Advanced reactor engineering. Flow-through configuration offers distinct advantages, significantly enhancing mass transfer and consequently accelerating reaction kinetics while improving electron utilization efficiency. Strategically embedding nanoconfined catalytic architectures within structured (photo)electrodes presents a synergistic pathway to enhance activity, ROS selectivity, and long-term stability.

  CRediT authorship contribution statement

  Limin Jin: Data curation, Conceptualization. Jie Xu: Conceptualization. Huayue Zhu: Supervision, Conceptualization. Chunjuan Li: Writing – review & editing, Conceptualization. Zhengfeng Hu: Supervision, Conceptualization. Qi Wang: Writing – review & editing, Formal analysis.

  Declaration of competing interest

  The authors declare that they have no known competing financial interest or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (No. 22276168) and "Pioneer" and "Leading Goose" R&D Program of Zhejiang (No. 2025C02240).

  
  
• 1. [1]

     X. Zhou, X. Li, Y. Xiang, et al., Chin. Chem. Lett. 36 (2025) 110664. doi: 10.1016/j.cclet.2024.110664

  2. [2]

     J. Guo, B. Gao, Q. Li, et al., Adv. Mater. 36 (2024) 2403965. doi: 10.1002/adma.202403965

  3. [3]

     L.M. Jin, S.J. You, X.G. Duan, et al., J. Hazard. Mater. 423 (2022) 127111. doi: 10.1016/j.jhazmat.2021.127111

  4. [4]

     K. Lu, T. Ren, N. Yan, et al., Environ. Sci. Technol. 57 (2023) 18888–18897. doi: 10.1021/acs.est.3c00595

  5. [5]

     Z. Wang, N. Hu, L. Wang, et al., Angew. Chem. Int. Ed. 63 (2024) e202407628. doi: 10.1002/anie.202407628

  6. [6]

     S. Nazia, N. Sahu, V. Jegatheesan, et al., Chem. Eng. J. 412 (2021) 128598. doi: 10.1016/j.cej.2021.128598

  7. [7]

     J.M. Anglada, M. Martins Costa, J.S. Francisco, M.F. Ruiz López, Acc. Chem. Res. 48 (2015) 575–583. doi: 10.1021/ar500412p

  8. [8]

     L.L. Zhou, Y.T. Liu, H. Shi, et al., Water Res. 250 (2024) 121075. doi: 10.1016/j.watres.2023.121075

  9. [9]

     S. Yang, S. Sun, Z. Xie, et al., Environ. Sci. Technol. 58 (2024) 19571–19583. doi: 10.1021/acs.est.4c06676

  10. [10]

      H. Yi, E. Almatrafi, D. Ma, et al., Water Res. 233 (2023) 119719. doi: 10.1016/j.watres.2023.119719

  11. [11]

      Q. Li, F.T. Li, Chem. Eng. J. 421 (2021) 129915. doi: 10.1016/j.cej.2021.129915

  12. [12]

      Q. Li, Q. Zhou, Y. Wu, et al., J. Environ. Chem. Eng. 155 (2025) 111–126.

  13. [13]

      J. Chang, B. Yu, X. Peng, et al., Water Res. 272 (2025) 122960. doi: 10.1016/j.watres.2024.122960

  14. [14]

      Q. Li, M. Deng, J. Gao, et al., Sep. Purif. Technol. 362 (2025) 131722. doi: 10.1016/j.seppur.2025.131722

  15. [15]

      Q. Wang, Y. Hui, X. Ma, et al., Chem. Eng. J. 507 (2025) 160487. doi: 10.1016/j.cej.2025.160487

  16. [16]

      H. Zhang, J. He, C. Zhai, M. Zhu, Chin. Chem. Lett. 30 (2019) 2338–2342. doi: 10.1016/j.cclet.2019.07.021

  17. [17]

      L. Jin, Y. Wang, Q. Li, et al., Appl. Catal. B: Environ. 373 (2025) 125328. doi: 10.1016/j.apcatb.2025.125328

  18. [18]

      K. Wang, Y. Li, G. Zhang, et al., Appl. Catal. B: Environ. 240 (2019) 39–49. doi: 10.1016/j.apcatb.2018.08.063

  19. [19]

      H.Y. Gu, X. Liu, X.F. Liu, et al., Nat. Commun. 12 (2021) 5422. doi: 10.1038/s41467-021-25726-w

  20. [20]

      Y. Wang, H. Li, Y. Chen, et al., Appl. Catal. B: Environ. 374 (2025) 125388. doi: 10.1016/j.apcatb.2025.125388

  21. [21]

      X.Y. Li, H.Y. Chu, P. Wang, et al., Sep. Purif. Technol. 372 (2025) 133525. doi: 10.1016/j.seppur.2025.133525

  22. [22]

      Q. Wang, L. Ren, J. Zhang, et al., Adv. Energy Mater. 13 (2023) 2301543. doi: 10.1002/aenm.202301543

  23. [23]

      R. Long, H. Huang, Y. Li, et al., Adv. Mater. 27 (2015) 7025–7042. doi: 10.1002/adma.201502068

  24. [24]

      R. Long, K. Mao, X. Ye, et al., J. Am. Chem. Soc. 135 (2013) 3200–3207. doi: 10.1021/ja311739v

  25. [25]

      X. Lu, X. Zhou, W. Qiu, et al., J. Hazard. Mater. 438 (2022) 129420. doi: 10.1016/j.jhazmat.2022.129420

  26. [26]

      X. Xiao, Z. Ruan, Q. Li, et al., Adv. Mater. 34 (2022) 2200612. doi: 10.1002/adma.202200612

  27. [27]

      Y. Wang, Y. Lin, S. He, et al., J. Hazard. Mater. 461 (2024) 132538. doi: 10.1016/j.jhazmat.2023.132538

  28. [28]

      M.C. DeRosa, R.J. Crutchley, Coord. Chem. Rev. 233-234 (2002) 351–371. doi: 10.1016/S0010-8545(02)00034-6

  29. [29]

      J. Al-Nu’airat, I. Oluwoye, N. Zeinali, et al., Chem. Rec. 21 (2021) 315–342. doi: 10.1002/tcr.202000143

  30. [30]

      S. Jiang, M. Xie, L. Jin, et al., Chin. Chem. Lett. 36 (2025) 110293. doi: 10.1016/j.cclet.2024.110293

  31. [31]

      J. Xie, C. Zhang, T.D. Waite, Water Res. 217 (2022) 118425. doi: 10.1016/j.watres.2022.118425

  32. [32]

      F. Xiao, Z.N. Wang, J.Q. Fan, et al., Angew. Chem. Int. Ed. 60 (2021) 10375–10383. doi: 10.1002/anie.202101804

  33. [33]

      Y. Kim, H. Lee, H. Oh, et al., Environ. Sci. Technol. 56 (2022) 5763–5774. doi: 10.1021/acs.est.1c08502

  34. [34]

      C. Chu, Z. Chen, D. Yao, et al., Angew. Chem. Int. Ed. 63 (2024) e202317214. doi: 10.1002/anie.202317214

  35. [35]

      Q. Wang, A. Liu, B. Shen, et al., Chem. Eng. J. 502 (2024) 157806. doi: 10.1016/j.cej.2024.157806

  36. [36]

      Q. Zhu, Y. Shi, Y. Tao, et al., Chem. Eng. J. 482 (2024) 149066. doi: 10.1016/j.cej.2024.149066

  37. [37]

      J.H. Lee, D.Y. Jo, J.W. Choung, et al., J. Hazard. Mater. 403 (2021) 124085. doi: 10.1016/j.jhazmat.2020.124085

  38. [38]

      Q. Zhang, G. Li, W. Liu, et al., Appl. Catal. B: Environ. 352 (2024) 124051. doi: 10.1016/j.apcatb.2024.124051

  39. [39]

      X. Quan, F. Guo, R. Li, et al., ACS Appl. Nano Mater. 7 (2024) 23228–23239. doi: 10.1021/acsanm.4c04695

  40. [40]

      Y. Yang, X. Yang, L. Tang, et al., Adv. Mater. 37 (2025) 2502321. doi: 10.1002/adma.202502321

  41. [41]

      C.H. Choi, M. Kim, H.C. Kwon, et al., Nat. Commun. 7 (2016) 10922. doi: 10.1038/ncomms10922

  42. [42]

      G. Zhang, R. Hao, J. Zhang, et al., J. Colloid Interface Sci. 658 (2024) 301–312. doi: 10.1016/j.jcis.2023.12.074

  43. [43]

      Y. Shang, X. Xu, B. Gao, et al., Chem. Soc. Rev. 50 (2021) 5281–5322. doi: 10.1039/d0cs01032d

  44. [44]

      Y. Yang, Y. Yang, Z. Pei, et al., Matter 3 (2020) 1442–1476. doi: 10.1016/j.matt.2020.07.032

  45. [45]

      H. Jeong, S. Shin, H. Lee, ACS Nano 14 (2020) 14355–14374. doi: 10.1021/acsnano.0c06610

  46. [46]

      H. Li, T. Li, R. Zhao, et al., Small 21 (2025) 2412234. doi: 10.1002/smll.202412234

  47. [47]

      H. Li, H. Ji, J. Liu, et al., Appl. Catal. B: Environ. 328 (2023) 122481. doi: 10.1016/j.apcatb.2023.122481

  48. [48]

      J. Zhang, H. Yang, B. Liu, Adv. Energy Mater. 11 (2021) 2002473. doi: 10.1002/aenm.202002473

  49. [49]

      J. Zhang, Z. Zhou, Z. Feng, et al., Environ. Sci. Technol. 56 (2022) 1331–1340. doi: 10.1021/acs.est.1c06368

  50. [50]

      Y. Qin, R. Tan, J. Wen, et al., Chem. Sci. 14 (2023) 7346–7354. doi: 10.1039/d3sc01523h

  51. [51]

      S. Liu, D. Lu, X. Wang, et al., J. Mater. Chem. B 5 (2017) 4918–4925. doi: 10.1039/C7TB00954B

  52. [52]

      J. Liu, Z. Yao, G. Qiu, et al., Chem. Eng. J. 455 (2023) 140602. doi: 10.1016/j.cej.2022.140602

  53. [53]

      A. Liu, X. Ma, B. Shen, et al., Chem. Eng. J. 504 (2025) 158929. doi: 10.1016/j.cej.2024.158929

  54. [54]

      D. Iglesias, A. Giuliani, M. Melchionna, et al., Chem 4 (2018) 106–123. doi: 10.1016/j.chempr.2017.10.013

  55. [55]

      C.Y. Chen, C.T. Jafvert, Environ. Sci. Technol. 44 (2010) 6674–6679. doi: 10.1021/es101073p

  56. [56]

      H. Braun, D. Mitoraj, J. Kuncewicz, et al., Appl. Catal. A: Gen. 660 (2023) 119173. doi: 10.1016/j.apcata.2023.119173

  57. [57]

      L. Wang, H. Du, X. Wang, et al., Environ. Res. 272 (2025) 121166. doi: 10.1016/j.envres.2025.121166

  58. [58]

      H. Wang, C. Yang, F. Chen, et al., Angew. Chem. Int. Ed. 61 (2022) e202202328. doi: 10.1002/anie.202202328

  59. [59]

      S. Huang, B. Zhang, D. Wu, et al., Appl. Catal. B: Environ. 340 (2024) 123216. doi: 10.1016/j.apcatb.2023.123216

  60. [60]

      Z. Feng, Q. Tian, Q. Yang, et al., Appl. Catal. B: Environ. 286 (2021) 119908. doi: 10.1016/j.apcatb.2021.119908

  61. [61]

      M. Sun, B. Liu, W. Han, et al., Appl. Catal. B: Environ. 304 (2022) 120980. doi: 10.1016/j.apcatb.2021.120980

  62. [62]

      H. Zhan, Q. Zhou, M. Li, et al., Appl. Catal. B: Environ. 310 (2022) 121329. doi: 10.1016/j.apcatb.2022.121329

  63. [63]

      Y. Peng, W. Xiao, H. Wang, Z. Bian, Appl. Catal. B: Environ. 363 (2025) 124820. doi: 10.1016/j.apcatb.2024.124820

  64. [64]

      B. Huo, F. Meng, J. Yang, et al., Chem. Eng. J. 436 (2022) 135173. doi: 10.1016/j.cej.2022.135173

  65. [65]

      J. Zhang, X. Chang, Z. Luo, et al., Chem. Commun. 54 (2018) 7026–7029. doi: 10.1039/c8cc03303j

  66. [66]

      J. Zhang, G. Zhang, H. Lan, et al., Environ. Sci. Technol. 57 (2023) 1134–1143. doi: 10.1021/acs.est.2c08172

  67. [67]

      Z. Lv, L. Kang, X. Pan, et al., ACS Catal. 14 (2024) 10172–10180. doi: 10.1021/acscatal.4c01749

  68. [68]

      J.H. Baek, T.M. Gill, H. Abroshan, et al., ACS Energy Lett. 4 (2019) 720–728. doi: 10.1021/acsenergylett.9b00277

  69. [69]

      K. Liu, Y. Wang, F. Liu, et al., Chem. Eng. J. 473 (2023) 145292. doi: 10.1016/j.cej.2023.145292

  70. [70]

      A. Li, H. Qiu, Z. Wang, et al., ACS Sustain. Chem. Eng. 12 (2024) 9558–9567. doi: 10.1021/acssuschemeng.4c03759

  71. [71]

      P. Lyu, J. Zhu, C. Han, et al., ACS Appl. Mater. Interfaces 13 (2021) 2033–2043. doi: 10.1021/acsami.0c19414

  72. [72]

      Y. Zhang, J. Zhang, B. Zhang, et al., Nat. Commun. 11 (2020) 558. doi: 10.1038/s41467-019-14241-8

  73. [73]

      Z. Feng, M. Chen, Q. Yang, et al., Environ. Sci. Technol. 57 (2023) 17123–17131. doi: 10.1021/acs.est.3c06336

  74. [74]

      Q.Y. Yi, J.H. Ji, B. Shen, et al., Environ. Sci. Technol. 53 (2019) 9725–9733. doi: 10.1021/acs.est.9b01676

  75. [75]

      Y. Rao, Y. Zhang, J. Fan, et al., Chem. Eng. J. 435 (2022) 134882. doi: 10.1016/j.cej.2022.134882

  76. [76]

      Y. Mao, P. Wang, L. Li, et al., Angew. Chem. Int. Ed. 59 (2020) 3685–3690. doi: 10.1002/anie.201914001

  77. [77]

      C. Yang, S.S. Shang, X.Y. Li, J. Hazard. Mater. 436 (2022) 129212. doi: 10.1016/j.jhazmat.2022.129212

  78. [78]

      D. Guo, Y. Liu, H. Ji, et al., Environ. Sci. Technol. 55 (2021) 4045–4053. doi: 10.1021/acs.est.1c00349

  79. [79]

      S. Dominguez, P. Ribao, M.J. Rivero, I. Ortiz, Appl. Catal. B: Environ. 178 (2015) 165–169. doi: 10.1016/j.apcatb.2014.09.072

  80. [80]

      S. Qing, X. Lu, Y. Jiang, et al., Chin. Chem. Lett. 37 (2026) 110576. doi: 10.1016/j.cclet.2024.110576

  81. [81]

      K. Fuku, Y. Miyase, Y. Miseki, et al., Chem. -Asian J 12 (2017) 1111–1119. doi: 10.1002/asia.201700292

  82. [82]

      X. Zhou, B. Shen, J. Zhai, N. Hedin, Adv. Funct. Mater. 31 (2021) 2009594. doi: 10.1002/adfm.202009594

  83. [83]

      M. Sun, X. Wang, L.R. Winter, et al., ACS ES & T Eng. 1 (2021) 725–752. doi: 10.1021/acsestengg.1c00015

  84. [84]

      Y.B. Liu, G.D. Gao, C.D. Vecitis, Acc. Chem. Res. 53 (2020) 2892–2902. doi: 10.1021/acs.accounts.0c00544

  85. [85]

      Y.M. Zhao, M. Sun, X.X. Wang, et al., Nat. Commun. 11 (2020) 6228.

  86. [86]

      L. Jin, M. Sun, J. Yang, et al., Appl. Catal. B: Environ. 339 (2023) 123150. doi: 10.1016/j.apcatb.2023.123150

  87. [87]

      P. Kumari, N. Bahadur, X.A. Conlan, et al., Water Res. 218 (2022) 118519. doi: 10.1016/j.watres.2022.118519

  88. [88]

      S. Naghdi, M.M. Shahrestani, M. Zendehbad, et al., J. Hazard. Mater. 442 (2023) 130127. doi: 10.1016/j.jhazmat.2022.130127

  89. [89]

      J. Guo, H. Sun, X. Yuan, et al., Water Res. 219 (2022) 118558. doi: 10.1016/j.watres.2022.118558

  90. [90]

      H. Chi, J. Ma, R. Duan, et al., Water Res. 262 (2024) 122101. doi: 10.1016/j.watres.2024.122101

  91. [91]

      S. Qu, H. Yang, R. Wang, et al., Sep. Purif. Technol. 371 (2025) 133288. doi: 10.1016/j.seppur.2025.133288

  92. [92]

      Z. Tang, R. Yin, W. Qu, et al., ACS ES & T Eng. 2 (2022) 629–641. doi: 10.1021/acsestengg.1c00314

  93. [93]

      N. Kaushik, S. Mitra, E.J. Baek, et al., J. Adv. Res. 43 (2023) 59–71. doi: 10.1016/j.jare.2022.03.002

  94. [94]

      Y. Fu, J. Chi, Y. Wu, et al., Appl. Surf. Sci. 692 (2025) 162711. doi: 10.1016/j.apsusc.2025.162711

  95. [95]

      Y. Shi, Y. Qian, J. Guo, et al., J. Clean Prod. 416 (2023) 138001. doi: 10.1016/j.jclepro.2023.138001

  96. [96]

      L. Valenzuela, M. Faraldos, A. Bahamonde, R. Rosal, Electrochim. Acta 395 (2021) 139203. doi: 10.1016/j.electacta.2021.139203

  97. [97]

      Z. Jiang, Y. Wang, C. Chen, et al., Chin. Chem. Lett. 35 (2024) 109400. doi: 10.1016/j.cclet.2023.109400

  98. [98]

      L. Qiao, J. Bai, T. Luo, et al., Appl. Catal. B: Environ. 238 (2018) 491–497. doi: 10.1016/j.apcatb.2018.07.051

  99. [99]

      S.P. Maletić, J.M. Beljin, S.D. Rončević, et al., J. Hazard. Mater. 365 (2019) 467–482. doi: 10.1016/j.jhazmat.2018.11.020

  100. [100]

       L. Chu, L. Cang, G. Fang, et al., J. Hazard. Mater. 428 (2022) 128273. doi: 10.1016/j.jhazmat.2022.128273

  101. [101]

       G.H. Yu, Y. Kuzyakov, Earth Sci. Rev. 214 (2021) 103525. doi: 10.1016/j.earscirev.2021.103525

• 

  Figure 1  (A) Molecular orbital diagrams for a ground-state triplet O₂ molecule, two types of ¹O₂, O₂^•− and H₂O₂. Copied with permission [23]. Copyright 2013, Wiley. (B) Possibly the relationship or differences in generating different ROS.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (A) Calculated charge difference surfaces of the Fe–N_x coordination systems, calculated energy of molecular O₂ adsorption on different sites, and schematic illustration of O₂ to O₂^•−. Copied with permission [26]. Copyright 2013, Wiley. (B) Proposed electrocatalytic activation mechanism of molecular O₂ to ¹O₂ over the CoFeCNT filter. Copied with permission [30]. Copyright 2025, Elsevier. (C) Selective electrocatalytic reduction of O₂ to ^•OH via 3e⁻ pathway with FeCo alloy encapsulated carbon aerogel. Copied with permission [32]. Copyright 2021, Wiley. (D) Binary heteroatom dopants enable carbon-based heterostructures for efficient (photo)electrocatalytic O₂ to H₂O₂ synthesis. Copied with permission [36]. Copyright 2024, Elsevier.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (A) Transition states in the O₂ dissociation on four PdCu alloy surfaces. Copied with permission [39]. Copyright 2024, ACS. (B) Mechanism for (photo)electrocatalytic O₂ to ^•OH over Pd-SA/F-TiO₂ and Pd NPs/F-TiO₂. Copied with permission [49]. Copyright 2024, ACS. (C) N-doped graphitized carbon nanohorns in highly selective O₂ to H₂O₂. Copied with permission [54]. Copyright 2018, Elsevier. (D) The scheme for O₂ to H₂O₂ at NiO_x-TiO₂-PCN photocathode. Copied with permission [56]. Copyright 2023, Elsevier. (E) The mechanism of O₂ to H₂O₂ on Py-TD-COF and Py-TD-COF-NH. Copied with permission [59]. Copyright 2024, Elsevier.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (A) Self-driven ROS generation from O₂/H₂O molecules via interfacial oxygen vacancies on carbon-coated TiO_2–x. Copied with permission [71]. Copyright 2020, ACS. (B) The scheme for O₂ to ¹O₂ and the ratio of O₂ to individual ROS in the SA-Cu_1.0CA electrocatalytic system. Copied with permission [73]. Copyright 2023, ACS. (C) Illustration of ¹O₂ triggered by O₂^•– in the Mo cocatalytic Fenton reaction with enhanced REDOX activity. Copied with permission [74]. Copyright 2019, ACS. (D) Illustration of O₂ to O₂^•– by synergy of dual defects in dual defects in cubic BiO_2−x. Copied with permission [76]. Copyright 2019, Wiley. (E) Possible activation mechanism of molecular O₂ in the Fe₂O₃-in-CNT and Fe₂O₃-out-CNT system. Copied with permission [78]. Copyright 2021, ACS.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (A) Illustration of charge transfer processes for ROS generation on a flow-by BiVO4:I/BTO-metal electrode. Copied with permission [82]. Copyright 2021, Wiley. (B) Schematic illustrating the cross-flow filtration system. Copied with permission [85]. Copyright 2020, Springer Nature. (C) Schematic of the Sb-SA/MXene photocatalytic surface with H₂O₂ generation and the Fe-NC/MXene electrocatalytic surface with ^•OH production concurrently. Copied with permission [86]. Copyright 2023, Elsevier. (D) The (photo)electrocatalysis assisted filtration process occurring in the interaction of pollutant solute and reactive oxygen species generated in cross flow filtration mode. Copied with permission [87]. Copyright 2022, Elsevier.

  下载: 全尺寸图片 幻灯片

  Table 1.  Summary of ROS yields and their detection by photocatalysts applying different detection method.

  Catalyst	Main ROS	Activity (mA/cm2)	Durability	Ref.
  B0.05—C3N4	O2•−, 1O2	/	3 h	[62]
  GBN	1O2	3 V vs. RHE	2 h	[63]
  MoS2/GDY-3	H2O2	/	4 h	[64]
  Co3O4-TiO2	H2O2	1.23 V vs. RHE	4 h	[65]
  CuxO@C	O2•−, •OH	−0.1 V vs. RHE	3 h	[66]
  Cu/TiO2	O2•−	−0.1 V vs. RHE	14 h	[67]
  Gd-doped BiVO4	H2O2	3.0 V vs. RHE	3 h	[68]
  Ni(OH)2/Ni foam	•OH	1.60 V vs. RHE	40 h	[69]
  Cu2O/CuO	•OH	2.2 V vs. RHE	8 h	[70]

  下载: 导出CSV
•