Persulfate activation represents physicochemical strategies aimed at increasing the oxidizing capability of persulfate (collectively referring to peroxymonosulfate (PMS) and peroxydisulfate (PDS); E⁰(PMS) = +1.82 V_NHE; E⁰(PDS) = +2.08 V_NHE) via electron and energy transfer pathways (Fig. 2) [9]. Reduced transition metals with a high electron-donating capacity transform persulfate into SO₄^•− by inducing the reductive dissociation of the peroxide bond in persulfate, similar to how Fe²⁺ catalyzes the one-electron reduction of H₂O₂ to ^•OH in the Fenton reaction [15]. Photolysis [60] and thermolysis [61] of persulfate can also generate SO₄^•− through homolytic cleavage of the peroxide bond. It is widely acknowledged that SO₄^•− is effective in oxidatively degrading a wide range of organic pollutants due to its strong oxidation power, which is comparable to ^•OH. However, in contrast to ^•OH considered a non-selective oxidant, the substrate specificity of SO₄^•− for organic compounds arises in part because these two radical oxidants primarily rely on distinct oxidation mechanisms for organic transformation [62]. Specifically, the H-atom abstraction by which ^•OH oxidizes saturated hydrocarbons at the diffusion-limited reaction rates is not the preferred reaction mode for SO₄^•− [63]. Alternatively, SO₄^•− typically oxidizes both organic and inorganic substrates through electron-transfer reactions. This is supported by the frequent detection of phenyl radical cations that recombine to yield dimeric products, inorganic anion radicals (e.g., Cl^•) that behave as secondary oxidants, and decarboxylation intermediates originating from aromatic and aliphatic carboxylates [9]. Although the reactions of SO₄^•− with organic macromolecules (e.g., natural organic matters) are kinetically retarded due to steric hindrance [64], the anionic character of SO₄^•− favors the oxidative degradation of positively-charged target substrates, such as alkylammonium cations [65].

Persulfate activation can also occur via mechanisms independent of homolytic or heterolytic cleavage of the peroxide bond and subsequent SO₄^•− generation, facilitating non-radical organic transformation through (i) high-valent metal-induced oxygen-atom transfer, (ii) singlet oxygenation, and (iii) mediated electron transfer [9]. Reduced transition metals, whether present in isolated or supported states, oxidatively transform into the high-valent oxo intermediates (e.g., Co(IV)=O and Fe(IV)=O) upon exposure to persulfate, as evidenced by the indication reaction using PMSO as a probe (i.e., sulfoxide-to-sulfone conversion) and ¹⁸O isotope-labeling technique [66]. Singlet oxygen (¹O₂) as a non-radical oxidant is generated through the disproportionation of superoxide radical (O₂^•−) during redox reactions of PDS or via the self-decomposition of PMS, a process accelerated under alkaline conditions or when carbonyl and hydroxyl moieties are present [9]. Certain metallic and carbonaceous materials promote electron transfer from organic electron donors to the persulfate acceptor, as substantiated by the chronoamperometric current response upon successive addition of organic substrate followed by persulfate [67]. The mediated electron transfer ultimately enables oxidative degradation of organic compounds without the involvement of radical species.

The 3d transition and noble metal ions, including Mn²⁺, Co²⁺, Fe²⁺, and Ru³⁺, demonstrate selective reactivity in the activation of PMS with an asymmetric structure [68]. The major degradative pathway in transition metal-induced PMS activation varies depending on the electron configuration. In particular, certain 3d transition metals like Co and Fe form high-spin PMS complexes containing unpaired electrons, which facilitate radical generation via single-electron transfer processes. Conversely, low-spin PMS complexes, which are strongly bonded to metals possessing partially filled d-subshells (e.g., Cr and Mn with d-orbitals holding up to 4 electrons), promote electron transfer from organic compounds to persulfate, thereby initiating a non-radical oxidation pathway [69]. Non-radical degradation mechanisms are supported by (i) the negligible inhibitory effects observed when using alcohol-based radical scavengers, (ii) treatment efficiencies and distributions of intermediates that depend on the substrate and diverge from expected SO₄^•− reactivity patterns, and (iii) the absence of electron paramagnetic resonance signals typically associated with radical formation [9].

Energy transfer processes such as photolysis and thermolysis are particularly effective for activating PDS, given its low peroxide bond dissociation energy [70]. PDS demonstrates a higher quantum yield for radical generation than PMS during UV photolysis [60], and increasing the temperature above 40 °C results in substantial SO₄^•− production from PDS, while only marginally promoting the homolytic cleavage of the peroxide bond in PMS [61]. Persulfate is considered a promising alternative to H₂O₂ for photochemical activation, due to (i) its activation under longer-wavelength UV irradiation, (ii) possessing a radical quenching activity two orders of magnitude lower than other agents, and superior resistance to the inhibitory effects of natural organic matter, which serve as natural radical scavengers [9]. Homogeneous persulfate activation can be performed with excess persulfate dosage, owing to its significantly reduced radical scavenging characteristics; this allows for the oxidative degradation of organic contaminants that are otherwise poorly reactive with oxidizing radicals (with bimolecular reaction rate constants in the range of 10⁶ to 10⁷ M⁻¹s⁻¹). Under realistic treatment conditions, heat-assisted activation leads to a preference shift in the persulfate precursor from PDS to PMS. This is because common background anions such as Cl⁻, HCO₃⁻, and HPO₄²⁻/H₂PO₄⁻ act as radical scavengers for PDS but function as nucleophiles towards the positively charged peroxide oxygen atom in PMS, resulting in the generation of secondary non-radical oxidants [9]. Furthermore, as the nucleophilic addition mechanism is facilitated by increased temperature, raising the temperature—even when below the threshold for direct peroxide bond cleavage—kinetically enhances the formation of anion-derived oxidants (via oxygen-atom transfer pathways), such as HOCl [71]. The complexation between PMS and inorganic nucleophiles (e.g., HPO₄²⁻/H₂PO₄⁻) weakens the strength of the peroxide bond, making it more susceptible to thermally induced dissociation and thereby increasing SO₄^•− production [72].

Heterogeneous metal-based materials, existing either as standalone entities or as supported structures in diverse chemical states such as zero-valent metal, oxide, sulfide, and phosphide, are capable of degrading organics through both radical and non-radical activation mechanisms of persulfate [15,73–75]. Co- and Fe-based activators (e.g., CoO, Fe⁰, and Fe_xP) facilitate redox reactions with PMS that involve heterolytic cleavage of the peroxide bond, thereby initiating radical-induced organic oxidation. Transition and noble metals significantly contribute to organic degradation irrespective of the persulfate precursor used, functioning as electron-transfer mediators that enable non-radical organic oxidation by enhancing electron exchanges between the target organic substrates and persulfate [76]. High-valent metal species function as non-radical oxidants whose efficacy in organic degradation varies significantly depending on the substrate during heterogeneous persulfate activation by metallic catalysts. Metal-oxo intermediates, frequently recognized as principal oxidants during Co- and Fe-mediated persulfate activation, facilitate the oxidative removal of organics via the oxygen-atom transfer pathway [66]. Alternatively, catalysts derived from Cu (e.g., CuO) produce non-oxo Cu(III) species exhibiting relatively high oxidizing power (E⁰ = +2.20, +2.3–1.57, and +1.68 V_NHE for Fe(VI) [77], Cu(III) [78], and Mn(VII) [77]) upon persulfate introduction, which then initiate electron-transfer oxidation of non-phenolic organics [79, 80]. The primary oxidant and the dominant degradative pathway in catalytic persulfate activation can shift based on the structural features of the metal-based activators and the type of persulfate precursor employed. Metal catalysts at molecular or atomic scale (i.e., single-atom catalysts) can generate SO₄^•− from persulfate at the nanoscale and subsequently undergo oxidative transformation into high-valent species, enabling organic treatment without the involvement of radicals [82]. The transformation of Ni(II) to Ni(IV) as a high-valent form of Ni, whose reactivity is largely independent of substrate type, occurs during PMS activation by NiO and Ni(OH)₂, while Ni(III), acting as a selective non-radical oxidant, forms exclusively when PDS is introduced instead [79].

Carbonaceous materials, such as carbon nanotubes, graphene, and their analogues, which are composed of sp²-hybridized carbon atoms, have been shown to facilitate non-radical persulfate activation [9,83]. In particular, the redox reactions of persulfate over nanocarbons produce O₂^•− as the precursor of ¹O₂, while surface functionalities (e.g., carboxyl and hydroxyl groups) either inherently present or intentionally introduced onto the carbon surface accelerate the self-decay of PMS, thereby increasing ¹O₂ yield [9]. Alternatively, carbocatalytic persulfate activation enables selective degradation of phenolic organics without radical generation, similar to specific metal oxides; electrically conductive carbon-based activators enhance the electron transfer from organic electron donors to persulfate [84]. Nanodiamond, characterized as an sp³-hybridized carbon material with high mechanical and chemical stability, shows promise as a carbon-based persulfate activator by undergoing thermal annealing to induce carbon graphitization [85]. The integration of metal nanoparticles into carbon matrices significantly enhances persulfate activation through several mechanisms: inner metal cores boost the electron-donating capacity of outer carbon layers, both metal and carbon phases offer dual-active sites that contribute to persulfate activation, and the carbon phase encapsulates the metal components, thereby preventing metal dissolution [86].

Heterogeneous catalysis, designed to facilitate easy catalyst recovery and simultaneously reduce secondary contamination, often enables non-radical organic oxidation via persulfate activation, except when certain transition metals such as Co and Fe act as the catalytic centers. Due to the selective nature of non-radical degradation pathways, catalytic persulfate activation can achieve more efficient targeting of emerging contaminants at trace concentrations within complex water matrices, suppress the formation of undesirable toxic byproducts, and reduce persulfate consumption. Moreover, recent research has highlighted the potential of non-radical persulfate activation processes to generate polymeric products through the recombination of phenyl radical cations, which are then removed by adsorption [87]. Nonetheless, considering that non-radical persulfate activation deviates from the core technical advantage of AOPs—specifically, substrate-insensitive treatment efficiency—it is necessary to systematically evaluate the limitations of heterogeneous persulfate activation for non-radical oxidation. This evaluation should focus on the treatability of persistent organics, the adverse biological impacts of partially oxidized intermediates, and the feasibility of hybrid materials or processes capable of simultaneously triggering both radical and non-radical persulfate activation pathways to compensate for each approach’s technical limitations.

Transition metals, widely recognized as effective technical choices for catalytic water treatment, exhibit selective reactivity towards PMS activation. Although carbonaceous materials demonstrate treatment efficiency that is generally insensitive to persulfate precursors, they are particularly effective for the oxidative degradation of electron-rich organic contaminants when persulfate is added. Given that metal-mediated redox reactions can accomplish the heterolytic cleavage of peroxide bonds in a range of peroxides, such as H₂O₂, PMS, PDS, and peracetate, followed by the generation of oxidizing radicals, the judicious selection of redox-active metals and subsequent development of bimetallic or carbon-based composites has the potential to yield adaptable peroxide activators that can facilitate radical-induced oxidation upon introduction of any peroxide.

It is well established that persulfate activation processes facilitate the oxidative treatment of organic pollutants through multiple pathways that include SO₄^•−, ¹O₂, high-valent metals, and electron exchange between electron donor-acceptor pairs. Despite this recognition, significant uncertainties remain regarding the relative reactivity of radical and non-radical oxidants involved in persulfate activation. In particular, high-valent metal-oxo species, which have often been assumed to function as the key oxidants in metal-activated persulfate systems, attack specific classes of organics through oxygen-atom transfer—a mechanism that differs from the electron-transfer oxidation characteristic of non-oxo species’ interactions with organics. The variation in oxidizing strength among high-valent metals, determined by their distinct redox potentials, indicates that their effectiveness toward organic substrates depends on the particular metal employed as the persulfate activator. Both metal- and carbon-based materials can interact with persulfate to generate surface persulfate complexes, which play a central role as intermediates in the non-radical oxidation route, specifically the mediated electron transfer pathway. The reactivity of these surface persulfate complexes towards organic electron donors appears to be highly dependent on the choice of activator, as substantiated by comparative studies using metal oxides and CNTs to evaluate substrate-specific oxidizing capabilities. To bridge the knowledge gap linked to inconsistencies among persulfate studies regarding oxidant reactivity, it is essential to conduct comparative assessments of various persulfate activation processes, emphasizing differences in organic transformation rates and distributions of intermediates/products.

As outlined previously, NZVI offers considerable potential for the reductive transformation and remediation of contaminants. NZVI (Fe⁰) exhibits a standard reduction potential of E⁰ = −0.44 V_NHE (Fe²⁺ + 2e⁻ → Fe⁰), enabling the electron transfer to aqueous contaminants. The predominant reductive decontamination mechanisms of NZVI involve complex surface reactions, typically beginning with the migration of contaminants toward the outer Fe^II-Fe^III oxide or hydroxide shell, and proceeding with electron transfer from the inner Fe⁰ core to this shell. A principal direct reduction pathway for NZVI is the dechlorination of chlorinated organic compounds (e.g., trichloroethylene (TCE) [94], perchloroethylene (PCE) [95], and carbon tetrachloride (CT) [96]).

These compounds are converted into non-toxic end products, including acetylene, ethene, and ethane, via β-elimination and hydrogenolysis pathways. To improve dechlorination efficiency, researchers have developed bimetallic NZVI particles with Pd or Ni, which catalyze H₂ transformation into reactive hydrogen species [97]. This is facilitated by NZVI’s interaction with water in anoxic and anaerobic conditions through the “anaerobic corrosion process” (Fe⁰ + 2H₂O → Fe²⁺ + H₂ + 2OH⁻), producing hydrogen available to drive additional reductive decontamination steps. The generated H₂ can participate in further reductive decontamination processes when H-active sites such as Pd and Ni are present. Venkateshaiah et al. demonstrated that Pd@NZVI exhibited the fastest dechlorination kinetics for PCE removal, followed sequentially by Ni@NZVI, Ag@NZVI, Cu@NZVI, and NZVI alone, attributable to additional catalytic surface reactions involving adsorbed atomic hydrogen [97].

For the removal of heavy metal, cations such as Cu²⁺, possessing more positive standard reduction potentials than Fe, are effectively and rapidly reduced (Fe⁰ + 2Cu²⁺ + H₂O → Fe²⁺ + Cu₂O + 2H⁺). Oxyanions like arsenate and chromate are also reduced by NZVI, as chemical reduction is thermodynamically favored over precipitation or adsorption in these cases [98]. In contrast, metal cations with standard potentials that are negative or similar to Fe, such as Zn(II) and Ni(II), are predominantly removed via electrostatic attraction, chemical adsorption, and precipitation [99].

Nitrogen and phosphorus are fundamental nutrients for living organisms, yet their excessive discharge can lead to severe water eutrophication. Recently, research has increasingly emphasized the removal of nitrate and phosphate from wastewater with the intention of resource recovery. For nitrate removal in aqueous systems, NZVI reduces nitrate primarily to ammonia gas as the final product (4Fe⁰ + NO₃⁻ + 10H⁺ → 4Fe²⁺ + NH₄⁺ + 3H₂O ΔG^o = −997.71 kJ) due to the ammonia stripping at high pH caused by H⁺ consumption and OH⁻ formation during anaerobic corrosion [100]. For phosphate removal from aqueous solutions, the estimated maximum adsorption capacity of NZVI approaches 245 mg/g, indicating robust phosphate uptake, which occurs primarily via adsorption and co-precipitation rather than by chemical redox reaction [101].

Although there has been a global increase in the need and demand for renewable energy sources, nuclear energy continues to be considered an economical and efficient energy option in recent years. However, the recycling and recovery of spent fuel can generate a variety of radioactive nuclides (such as Pu, U, Np, and other fission products like ⁹⁹Tc, ¹³⁷Cs, and ⁹⁰Sr), and there remains a significant risk of catastrophic nuclear accidents, exemplified by the Fukushima Daiichi accident in 2011. Notably, NZVI has demonstrated a fast adsorption rate for various radioactive nuclides, along with high removal efficiency and stable immobilization of radionuclides [102]. Hua et al. reported that aqueous uranium ions (U(VI)) are initially attracted to the NZVI surface through negatively charged surface (OH⁻ groups), leading to the formation of surface complexes with the Fe shell and U(VI) ions. Subsequently, U(VI) diffuses across the surface layer, readily obtaining electrons from the inner Fe⁰, which facilitates its rapid reduction to U(IV) forms [103]. For both Pu and U, NZVI reduces Pu aqueous phases (Pu(V) and Pu(VI)) to PuO₂ (Pu(IV)) and converts U(VI) to U(IV), such as UO₂, confirming the chemical reduction of sorbed Pu and U by NZVI [101].

Overall, the removal mechanisms of contaminants by NZVI are highly complex due to the core-shell structure, which provides effective Fe^II-Fe^III adsorption sites and serves as a potential Fe⁰ electron reservoir. This distinctive property of NZVI alters the removal pathways for different contaminants, emphasizing the necessity to thoroughly examine NZVI behavior during and after decontamination processes for achieving efficient and cost-effective environmental applications.

In the process of contaminant adsorption and reduction, Fe⁰ undergoes the surface oxidation (Fe⁰ → Fe^II or Fe^III), precipitation (in the forms of Fe^II, Fe^III, and Fe^II/Fe^III oxides and hydroxides) and coprecipitation with anionic species such as Cl⁻, SO₄²⁻ and CO₃²⁻ [11]. Consequently, various Fe-containing minerals can be fully or partially formed during the decontamination process, including vivianite (Fe^II₃(PO₄)₂), Fe(OH)₂, green rust (Fe^II–Fe^III hydroxides containing interlayer anions (e.g., Cl⁻, SO₄²⁻ and CO₃²⁻)), ferrihydrite, magnetite (Fe^II₁^III₂O₄), lepidocrocite (γ-Fe^IIIOOH), goethite (α-Fe^IIIOOH), mackinawite (Fe^IIS), and siderite (Fe^IICO₃) (Table 2) [11]. Since secondary Fe-containing minerals are capable of further removing various aqueous pollutants through reduction reactions (for Fe^II-containing minerals) as well as adsorption [105,106], both the direct reduction of pollutants and their secondary removal must be considered in NZVI applications. For example, Fe^II-containing minerals such as magnetite, vivianite, and green rust have demonstrated effective removal of organic (e.g., chlorinated compounds, halogenated compounds, and phenolic compounds) as well as inorganic (e.g., heavy metals, and radioactive nuclides) contaminants [107–110]. Fe^III-containing minerals (e.g., maghemite (γ-Fe^III₂O₃), goethite and hematite) have been employed for the removal of heavy metals and phosphate primarily via adsorption processes [111,112].

The oxidation of NZVI during reductive decontamination can also reveal a noteworthy natural reaction mediated by dissimilatory iron reducing bacteria (DIRB). DIRB are subsurface microorganisms capable of transferring electrons to extracellular objects (e.g., Fe^III-containing minerals), utilizing electrons acquired through intracellular organic metabolism [113]. This biological reaction is critically important in the geochemical Fe cycle and influences the environmental fate of contaminants. For example, Fe^III-containing minerals such as magnetite, lepidocrocite, and goethite were biologically reduced by Shewanella putrefaciens CN32 in the presence of phosphate ions, resulting in the formation of biogenic Fe-P containing mineral (i.e., vivianite) [114]. When compared experiments with individual minerals, mixed cultures containing CN32 and each mineral exhibited a substantial enhancement in CT dechlorination, with magnetite in combination with CN32 yielding the highest formation of biogenic vivianite [114]. Additionally, biogenic Fe^II mineral phases (i.e., magnetite, green rust, siderite, and vivianite) showed effective U(VI) removal, with biogenic green rust achieving the most significant U(VI) reduction, followed by magnetite and siderite [115]. These experimental findings provide valuable perspective on the significance of NZVI passivation byproducts for the behavior and movement of environmental contaminants in wastewater and groundwater systems.

Declining activity and concerns about long-term stability remain major barriers to the practical deployment of NZVI. Weak vander Waals force and intrinsic interactions promote aggregation, reducing accessible surface area and overall performance. Surface passivation further diminishes reducing capacity. Although certain passivation products can contribute to secondary contaminant removal, heavily passivated NZVI provides insufficient electron transfer capacity for efficient reduction. The nature and extent of passivation depend strongly on the composition of the water matrix, complicating mechanistic interpretation and process design.

Future work should prioritize improving reductivity and durability to enable treatment of refractory contaminants and expand application scope. Promising strategies include immobilizing NZVI on solid supports or encapsulating it within porous hosts to suppress aggregation and improve stability, as well as applying surface-protection chemistries that preserve active sites during operation. In parallel, deeper mechanistic studies of passivation are needed to enable rational depassivation approaches and to better leverage Fe-containing secondary phases.

With the broadening of the AOPs concept, investigations are increasingly focused on radicals beyond ^•OH, like SO₄^•−, which offer unique kinetics for targeting specific contaminants. To this end, anodes capable of generating ^•OH, such as BDD, can also yield peroxy-species including peroxyacetate, peroxodiphosphate (P₂O₈⁴⁻), peroxodicarbonate (C₂O₆²⁻), and peroxodisulfate (S₂O₈²⁻) [212–214], which act both as redox mediators and as sources of various radicals. For instance, anodically generated S₂O₈²⁻ (produced by oxidizing SO₄²⁻) can be further activated at the cathode to form SO₄^•−. Moreover, the concurrent generation of ROS at both the anode and cathode has the potential to significantly enhance degradation processes at a fixed current density (Fig. 4e).

Electrocatalyst development should aim to maximize the current efficiency (CE) for DET or the targeted generation of mediators. For metal oxide electrocatalysts, the metal-oxygen binding strength serves as a useful descriptor for selectivity. According to Sabatier’s principle, there is an optimal O adsorption energy described by a volcano relationship for each ROS. Implementing nonprecious electrocatalysts could expand the range of applications, but their long-term stability in the wastewater matrix must be verified. From a process engineering perspective, the CE of chemical oxygen demand (COD) removal is a key determinant of energy efficiency [215]. In mediated processes, CE values should be independently measured for both mediator generation and COD removal to identify possible side reactions. With increasing emphasis on the management of total organic carbon (TOC) and recalcitrant micropollutants like PFAS, it is essential to establish appropriate figures-of-merit to evaluate the effectiveness of electrochemical processes in degrading TOC and micropollutants.

Heterogeneous semiconductor photocatalysis provides a means of advanced oxidation by absorbing photons, which results in the generation of charge carriers. When photons with energies equal to or greater than the bandgap of the photocatalyst are absorbed, electrons are excited from the valence band (VB) to the conduction band (CB), leaving holes in the VB. Although the majority of electron-hole pairs recombine before reaching the surface, a limited portion successfully migrates and participates in pollutant degradation at the surface [19]. Electrons located at the catalyst surface react with dissolved oxygen (E⁰ = −0.33 V_NHE for O₂/O₂^•−), producing O₂^•−, while holes facilitate the oxidation of water to ^•OH (E₀ = 2.8 V_NHE for H₂O/^•OH) [216,217]. Owing to its high oxidative potential, ^•OH is widely recognized as the predominant oxidant in photocatalytic water treatment (Fig. 5a). These ROS exhibit a strong oxidation potential, promoting the decomposition of complex organic pollutants into smaller intermediates, which are subsequently mineralized to CO₂, H₂O, or other less harmful byproducts. The heterogeneous nature of this catalytic process eliminates the need for continuous addition of precursor chemicals, which is particularly beneficial for use in remote or resource-constrained environments. While optimizing operational conditions such as light intensity, wavelength, pH, temperature, and the effects of competing aqueous species can improve the performance of photocatalytic water treatment, the advancement of the technology is chiefly dependent on improving photon utilization through materials innovation.

Among various photocatalytic materials, TiO₂ is considered the benchmark commercial semiconductor material due to its structural stability, cost-effectiveness (~$1/kg), and relatively low toxicity [218]. TiO₂, with its 3.2 eV bandgap, requires UV irradiation for excitation, making it applicable for solar-driven and LED systems to generate ROS via both hole- and electron-mediated processes. Nonetheless, its practical use is constrained by two significant drawbacks: limited light utilization and a high propensity for charge carrier recombination. Various enhancement strategies have therefore been developed to expand light absorption properties and increase charge separation efficiency (Fig. 5b). Metal (e.g., Fe, Pt, Mn, Cr, V) and nonmetal (e.g., N, F, C, S) dopants can be introduced into the TiO₂ crystal lattice to extend its light absorption region [219]. Nonmetal doping leads to the formation of oxygen vacancies, which serve as additional active sites for dissolved oxygen activation [220]. Surface hydrogenation introduces disorder-induced mid-gap states that enhance light absorption, allowing the material to utilize the visible and infrared regions more efficiently [221]. Additionally, sensitization using organic chromophores further extends light absorption into the visible spectrum [222]. To promote efficient charge separation, noble metal cocatalysts are commonly deposited on the TiO₂ surface, functioning as electron-withdrawing centers through the creation of a rectifying Schottky barrier [223]. Carbon-based cocatalysts, such as graphene oxide, are efficient electron reservoirs due to their excellent electrical conductivity [224]. Current strategies are increasingly emphasizing sustainability by minimizing noble metal usage. The employment of single-atom catalysts as cocatalysts has arisen as a promising approach, as these materials both improve charge separation and drastically lower noble metal requirements [225,226]. Architectural modifications, including the development of nanotube and hierarchical porous structures, result in organized open channels that facilitate effective charge-carrier mobility and mass transfer for aqueous reactants and products [227].

Beyond modifications to TiO₂, alternative semiconductor materials have been explored as substitutes, with their respective bandgap positions and variations presented in Fig. 5a. Distinct bandgap ranges and band positions result in each photocatalyst possessing characteristic properties. As the bandgap narrows, visible-light response is improved; however, this generally exacerbates charge carrier recombination, highlighting the inherent trade-off between increased light harvesting and shorter charge carrier lifetimes. WO₃ exhibits excellent oxidative capability attributable to its VB position, but its performance depends on efficient utilization of residual electrons to suppress charge recombination [228]. BiVO₄ offers both strong oxidative potential and visible-light absorption, yet electron mobility is relatively low, which restricts efficient charge transport. C₃N₄, a metal-free photocatalyst with tunable band structures, predominantly drives reduction reactions but shows limited oxidizing ability. CdS provides high visible-light absorption but is susceptible to photocorrosion, necessitating robust stabilization approaches. Covalent organic frameworks (COFs), featured by their porous architectures, strong π-bridges, and 1D open channels, facilitate superior charge transport properties and demonstrate significant promise for photocatalytic degradation of emerging contaminants [229]. Similar modification strategies used for TiO₂ can also be applied to other semiconductor photocatalysts to enhance both light absorption and charge separation, contributing to improved ROS generation. Heterojunctions comprising multiple semiconductors and operating via Z-scheme or S-scheme charge transfer mechanisms preserve high redox potential (Fig. 5c) [230]. Such heterostructured systems support selective ROS production under visible-light irradiation, thereby enabling comprehensive degradation of persistent micropollutants in water matrices [231].

Emerging micropollutants, including PPCPs, PFAS, and microplastics (MPs) are increasingly recognized as significant environmental concerns because of their persistence in aquatic environments and their potential long-term health impacts. These contaminants often demonstrate resistance to conventional oxidation processes in contrast to traditional organic pollutants. Selective photocatalytic strategies are thus required for effective elimination. The fabrication of Bi₂MoO₆/CoWO₄ Z-scheme heterojunctions has demonstrated superior photocatalytic activity for norfloxacin degradation under visible light irradiation [232]. The Z-scheme heterojunction facilitates more efficient charge carrier separation, enabling the preservation of highly reductive electrons and oxidative holes. These strong oxidative holes preferentially cleave the C-N bond linking the piperazine and benzene rings, while ROS such as ^•OH, O₂^•−, and ¹O₂ subsequently drive further mineralization. Furthermore, in high-salinity wastewater, activated chloride ions can yield chloride radicals that selectively interact with electron-rich moieties in trimethoprim derivatives [233]. The exceptional recalcitrance of PFAS arises from the high bond strength of the C-F bond (~485 kJ/mol), rendering them largely unreactive to conventional oxidation methods. During photocatalysis, ROS tend to attack more labile bonds, such as C–H, C–O, and C–C, rather than the robust C–F bond, which limits the degradation efficiency for PFAS [234]. Alternatively, photocatalytic reductive processes have demonstrated promise for PFAS defluorination, wherein photogenerated electrons function as the principal reductive agents (Fig. 5d) [235,236]. To address the inherently slow kinetics of photocatalysis, a combined approach involving adsorption followed by solid-phase degradation, as exemplified by Bi/TNTs@AC for GenX treatment, has been proposed [237]. Nevertheless, carbon-based adsorbents are vulnerable to degradation from ROS, leading to the proposal of inorganic adsorbents, such as layered double hydroxides, for use in advanced adsorbent–photocatalyst hybrid systems (Fig. 5e) [238]. MPs, which are solid, non-dissolved micropollutants, are most commonly separated via membrane filtration. Photocatalytically generated ROS have been shown to attack the C-C and C-H bonds within polymeric structures, inducing chain scission and generating molecular fragments [239]. Despite this, achieving total mineralization of MPs is still a considerable challenge. Contemporary studies have examined SO₄^•−-mediated photocatalytic reactions as supplementary strategies. In this system, SO₄^•− is generated by activating PMS and PDS with photoexcited conduction band electrons. Compared with ^•OH, which are less selective and readily scavenged by background constituents, SO₄^•− shows greater selectivity for the degradation of target polymeric materials and maintains activity in complex water matrices [240].