Selective heterogeneous photocatalytic activation for toluene oxidation: recent advances, challenges and perspective

Oxidation of toluene with photogenerated holes

During a typical photocatalytic process, the activation of inert C–H bonds can proceed through multiple mechanisms, with the direct involvement of photogenerated holes being particularly crucial^[56-58]. When the photocatalyst is exposed to an appropriate light source, it could absorb photons with energy equal to or greater than its band-gap energy, resulting in the generation of photogenerated electrons and holes, where the electrons in the valence band of the semiconductor photocatalyst are excited into the conduction band (CB), leaving behind photogenerated holes in the valence band^[59]. The photogenerated holes, with strong oxidative capability, can directly activate the C(sp³)–H bond in toluene molecules, resulting in the formation of benzyl radicals. Generally, toluene is trapped by photogenerated holes and the C(sp³)–H bond undergoes homolytic cleavage, leading to the formation of a benzyl radical and the concurrent generation of a proton. During the cleavage process of C(sp³)–H bonds, the toluene can be oxidized by a photogenerated hole into a benzyl radical and a proton. The obtained carbon radicals can rapidly interact with O₂, ^•O₂^-, or other ROS species, ultimately affording a series of oxygenated products, including benzyl alcohol, benzaldehyde, and benzoic acid. For example, Xue et al. synthesized TiO₂ supported single-atom yttrium photocatalyst for the selective photooxidation of toluene^[60]. In comparison to pristine TiO₂, on which the generated benzaldehyde would be overoxidized by excess ^•O₂^- to form benzoic acid and CO₂, resulting in poor selectivity towards benzaldehyde, the as-prepared Y₁/TiO₂ demonstrates a lattice oxygen-mediated mechanism for the selective oxidation of toluene molecules. It was revealed that the surface lattice O^2- species could directly react with benzyl radicals to produce benzaldehyde. Moreover, benzaldehyde molecules would quickly desorb from the catalyst surface, effectively preventing the overoxidation of benzaldehyde to the benzoic acid and CO₂, thereby leading to superior photocatalytic performance in the selective toluene oxidation.

Oxidation of toluene with superoxide radicals

When O₂ is employed within the photocatalytic system, it undergoes various transformation processes to form reactive ROS, which serve as primary oxidants for toluene oxidation^[61]. One of the typical ROS is the superoxide ion radical, which is a radical anion^[62]. In typical photocatalytic systems, photocatalysts with a CB edge level more negative than the O₂/^•O₂^- potential could deliver one photoactivated electron to the π^* orbitals of oxygen molecules via the reduction pathway, thereby generating superoxide radicals [Equation (1)]. As a highly reactive oxidant, ^•O₂^- could directly initiate the cleavage of C–H bonds or combine with the generated hydrocarbon radicals during the oxidation process to complete the photocatalytic cycle^[35]. For instance, ^•O₂^- could be generated over the as-prepared photochromic Bi₂WO_6-x/amorphous BiOCl (p-BWO) photocatalysts through two distinct pathways: (1) Direct reduction of oxygen molecules by photogenerated electrons from p-BWO, resulting in the formation of ^•O₂^-; and (2) photogenerated electrons initially captured by [W_(VI)O_6-x], reducing the W_(VI) to yield W_(V), displaying photochromic phenomenon, which facilitates the separation of charge carriers. Subsequently, the obtained W_(V) can be oxidized by O₂ to regenerate W_(VI), accompanied by the generation of ^•O₂^-[47]. The obtained ^•O₂^- subsequently combines with carbon radicals to form C₆H₅CH₂OOH intermediates, ultimately yielding benzaldehyde as the main product.

Oxidation of toluene with hydroxyl radicals

Upon light irradiation, the adsorbed water molecules and the surface-bound hydroxyl groups on the photocatalysts could be oxidized by photogenerated holes, resulting in the generation of hydroxyl radical (^•OH) [Equation (2)]. The obtained radicals subsequently interact immediately with the substrate molecules, further driving chemical reactions^[63-66]. Although the mechanistic implications remain controversial, ^•OH radicals are widely regarded as the most reactive species among all types of ROSs due to the powerful oxidative potential of the ^•OH/H₂O pair, which allows for the facile dissociation of the inert chemical bond in the reactant molecules (e.g., methane, ethylbenzene, microplastics, etc.) and further converts them into highly value-added products^[66-71]. However, given that the intermediates formed during photocatalysis are usually highly susceptible to overoxidation by ^•OH radicals, which unavoidably leads to the nonselective oxidation of organic reactants into CO₂ and water, increasing attention has been paid to controlling the content of ^•OH in the catalytic system and developing mild pathways for their production to improve the selectivity of targeted products while maintaining the conversion of reactant molecules. For example, Teng et al. investigated the mechanism of selective toluene oxidation using Sb(CN)₃-modified sp²-c-COFs^[48]. They revealed that the adsorbed water could be oxidized by the photogenerated holes to produce ^•OH radicals, which would lead to the generation of benzyl alcohol. The generated benzyl alcohol can further react with benzaldehyde to form hemiacetal, catalyzed by Brönsted acid, which could be further oxidized to generate ester by the oxygen reduction reaction (ORR)-related ROS in the reaction system.

Oxidation of toluene with singlet oxygen species

Singlet oxygen (¹O₂), a reactive form of molecular oxygen with high energy and electrophilic properties, can be generated through the energy transfer from the triplet state of a photosensitizer to the ground state oxygen molecule (³O₂) under light illumination, resulting in the formation of ¹O₂ [Equations (3) and (4)]^[72-77]. Moreover, it has been proposed that the oxidation of ^•O₂^- by photoactivated holes on suitable photocatalysts [e.g., TiO₂, ZnO, metal-organic frameworks (MOFs), covalent organic frameworks (COFs), polymers, and carbon-based nanomaterials] could also yield ¹O₂ species [Equations (5) and (6)]^[78-82]. Effective heterogeneous photocatalysis using ¹O₂ relies on satisfying two main criteria: (i) achieving efficient photosensitization and (ii) ensuring adequate exposure of active sites within photocatalyst, which can be achieved through the construction of hierarchical pores within photocatalysts or by controlling the crystal dimensions and sizes of the photocatalysts, thereby facilitating rapid diffusion and transfer of reactants and products to and from catalytically active sites^[83-87]. As a moderate oxidant, ¹O₂ can effectively promote the oxidation of organic molecules while preventing overoxidation. For instance, Li et al. reported the interfacial synergy of Pd sites and defective oxygen sites within the as-prepared Pd/BiOBr photocatalysts could promote the selective photooxidation of toluene^[88]. To investigate the ROS involved in the oxidation process, various scavengers [superoxide dismutase (SOD) to quench ^•O₂^-, carotene to quench ¹O₂, catalase to quench H₂O₂, mannitol to quench ^•OH] were applied within the photocatalysis process. The results showed that when Pd/BiOBr was used as the photocatalyst, both ^•O₂^- and ¹O₂ were the major ROSs contributing to the selective production of benzaldehyde during the photocatalytic oxidation reaction. Furthermore, in situ electron spin resonance (ESR) detection using 2,2,6,6,-tetramethylpiperidine (TEMP) as a radical-trapping agent provided a more direct clue for the generation of ¹O₂ during the reaction.

The photocatalytic conversion of O₂ to ROS proceeds through complex pathways, such as energy transfer and electron transfer, and is influenced by the type of photocatalyst employed and the catalytic environment within the system^[89-92]. Consequently, future efforts are expected to focus on the development of methods to control the O₂ activation process, aiming to achieve highly efficient and selective ROS generation. The selective generation of specific ROS is essential for understanding the underlying mechanisms and optimizing the performance and selectivity of photocatalytic systems, ensuring their effectiveness in various photocatalytic applications.

Although the specific reaction mechanism remains unclear, several key steps can be identified in the photocatalytic oxidation of toluene, providing readers with an overview of the photocatalytic mechanism: (i) Excitation of photocatalyst: Upon light irradiation, the photocatalyst absorbs photons, generating electron-hole pairs (i.e., e^-/h⁺); (ii) Efficient charge separation: The photogenerated holes would trap the adsorbed toluene molecules, facilitating their activation; (iii) Homolytic cleavage of C(sp³)–H bond: The reaction between the holes and the toluene molecules leads to the homolytic cleavage of C(sp³)–H bond, resulting in the formation of benzyl radicals (C₆H₅CH₂^•) and the release of protons (H⁺); (iv) Reactivity of benzyl radicals: The generated benzyl radicals can react with various ROS generated in the photocatalytic system, such as hydroxyl (^•OH) and superoxide radicals (^•O₂^-), yielding a range of oxidation products, including benzyl alcohol, benzaldehyde, and benzoic acid; (v) Termination reactions: The overall reaction pathway may involve further reactions of benzyl radicals, leading to overoxidation products (e.g., CO and CO₂) depending on the specific reaction conditions and available ROS.

Heterojunction construction

To enhance the photocatalytic performance of a specific semiconductor material, two primary challenges must be noted and addressed: (i) the severe and rapid recombination of photogenerated charge carriers and (ii) the relatively long distance for charge carriers to migrate and transfer to reach the catalyst surface to activate the substrates^[93,94]. In view of this, coupling two semiconductors with suitable band gaps to construct a heterostructure architecture is widely recognized as a promising strategy to shorten the diffusion length of charge carriers and prevent electron-hole recombination, thereby extending the lifetime of photogenerated charge carriers and enabling more timely and high-efficiency photocatalytic reactions^[95-99]. Specifically, the quality of the interface within a heterostructure, such as contact area, chemical bonds, and interfacial electric fields (E-fields), determines the charge separation efficiency. In this section, we will discuss recent advances in the fabrication of various composite photocatalysts with well-designed heterostructures for the selective photooxidation of toluene.

In 2018, He et al. reported the synthesis of organic/inorganic Cd₃(C₃N₃S₃)/CdS [denoted Cd₃(TMT)₂/CdS] composite with typical porous structures and rich in heterostructures for the selective toluene photooxidation, synthesized via a facile hydrothermal procedure^[100]. The in situ generation of porous CdS under elevated temperature offers abundant mesopores and macropores, enhancing the adsorption of reactants as the fabricated Cd₃(TMT)₂/CdS composite photocatalyst is used in the reaction systems. Moreover, the abundant heterostructures constructed by the direct contact of CdS and Cd₃(TMT)₂ promote an efficient separation of charge carriers. Consequently, the optimized Cd₃(TMT)₂/CdS synthesized at 155 °C with proper ratios of CdS to Cd₃(TMT)₂ demonstrated an excellent benzaldehyde formation rate of 787 mmol·g^-1·h^-1 without the need of organic solvent under visible-light irradiation (λ ≥ 420 nm). Following this work, a CdIn₂S₄-CdS sheet-to-sheet heterojunction structure enriched with abundant defect-induced S-vacancies on the photocatalyst surface was achieved via a facile hydrothermal procedure, as reported by Tan et al.^[101]. In the two-dimensional (2D) CdIn₂S₄-CdS composite photocatalyst, the photogenerated electrons could migrate from the CB of CdIn₂S₄ towards CB of CdS, leading to efficient spatial separation of photogenerated charge carriers. Additionally, the abundant S vacancies, characterized by their strong affinity for O₂ molecules, enable photogenerated electrons trapped in the vacancies to readily combine with O₂ to yield abundant ^•O₂^-, which serve as ROS for the oxidation of toluene molecules. Benefiting from the facile generation of ^•O₂^- species, the obtained CdIn₂S₄-CdS showed excellent photocatalytic performance, achieving a toluene conversion as high as 80% and a selectivity towards benzaldehyde of 99% within a 6-h photocatalytic reaction under visible light irradiation (λ ≥ 420 nm).

Metal halide perovskites (MHP), featuring excellent photoelectronic properties, sufficient redox potentials, and abundant surface active sites, have garnered special attention as efficient photocatalysts for various photocatalytic applications^[102-106]. Further coupling MHP with appropriate semiconductor materials to construct heterojunction structures can enhance photocatalytic performance by promoting the efficient separation of the photogenerated charge carriers^[107-109]. In this context, Cui et al. prepared TiO₂/Cs₃Bi₂Br₉ (CBB) composite photocatalyst via a simple solvent precipitation method^[110]. The large specific surface area of TiO₂ provides substantial nucleation sites, effectively downsizing the CBB components from 1 μm in its pure microcrystal to 100 nm within the composite photocatalyst. Meanwhile, DFT calculations revealed a negative adsorption energy of TiO₂/CBB for toluene, indicating a strong adsorption capability of toluene molecules on the catalyst surface. The enhanced adsorption is conducive to the activation of adsorbed toluene and facilitates subsequent interaction with photogenerated charge carriers to form reaction intermediates, thereby boosting the overall reaction kinetics. The obtained TiO₂/CBB was found to demonstrate a type-II heterojunction, which promotes the separation of charge carriers while inhibiting the undesired severe recombination during photocatalysis, affording an excellent production rate (9,692.5 μmol·g^-1·h^-1) with high selectivity (91%) for benzaldehyde. Following that, Zhou et al. reported the synthesis of Cs₃Bi₂Br_9-x@AgBr core-shell photocatalyst with abundant adsorption-redox sites via an in situ light-assisted Ag⁺ insertion method [Figure 1A]^[111]. Experimental results and DFT calculations revealed that a type-II heterojunction structure was achieved within the resulting Cs₃Bi₂Br_9-x@AgBr photocatalyst [Figure 1B]. Benefiting from the enhanced spatial charge separation and the construction of unified adsorption-redox sites, which favors the adsorption and the activation of the substrates (i.e., toluene and oxygen molecules), the as-prepared Cs₃Bi₂Br_9-x@AgBr demonstrated an optimal production rate (5.61 mmol·g^-1·h^-1) of benzaldehyde, six times higher than the pure CBB, with high selectivity (91%) [Figure 1C].

Figure 1. (A) Diagram depicting the synthetic process of the Cs₃Bi₂Br_9-x@AgBr core–shell heterojunction; (B) Mechanism of electron transfer in the Cs₃Bi₂Br_9-x@AgBr photocatalyst under light illumination; (C) Photocatalytic performance comparison of CBB, AgBr, and various Cs₃Bi₂Br_9-x@AgBr-X composites (where X = 3, 6, 8). Reprinted with permission. Copyright 2024, American Chemical Society^[111]; (D) Diagram illustrating the preparation of the W₁₈O₄₉/CsPbBr₃ heterojunction; (E) Representation of the band structures between CPB and W₁₈O₄₉ under light irradiation. The E_f bending at the interface was considered; (F) Photocatalytic toluene oxidation and CO₂ reduction over pure W₁₈O₄₉ ultrathin NWs, pure CPB, and WOCPB heterojunction. Reproduced with permission, Copyright 2024, Elsevier^[117]. CBB: Cs₃Bi₂Br₉; CPB: CsPbBr₃; E_f: electric field; NWs: nanowires; WOCPB: W₁₈O₄₉/CsPbBr₃.

In a typical type-II heterojunction, the improved separation of photogenerated charge carriers generally comes at the cost of jeopardizing the redox potentials when two coupled semiconductors with staggered band structures come into contact^[112]. Regarding this, the step-scheme (S-scheme) heterojunction, consisting of an oxidation photocatalyst and a reduction photocatalyst, has emerged as a solution to efficiently separate photogenerated charge carriers while maximizing the oxidation and reduction potentials of the fabricated S-scheme photocatalyst^[37,113-116]. More recently, Jiang et al. reported the development of W₁₈O₄₉/CsPbBr₃ heterostructure via the in situ hot-injection growth method [Figure 1D]^[117]. Experimental results and DFT simulations revealed that an internal E-field (IEF) was established upon the W₁₈O₄₉/CsPbBr₃ heterostructure due to the staggered Fermi level and band alignment [Figure 1E]. The IEF induced by the construction of S-scheme heterojunction would cause the accumulation of photogenerated electrons in the CB of the reducing CsPbBr₃ component, while the holes would accumulate in the VB of oxidizing W₁₈O₄₉ component, affording the fabricated W₁₈O₄₉/CsPbBr₃ with strong redox potentials, enabling the simultaneous photocatalytic activation of both toluene and carbon dioxide molecules. Consequently, the as-prepared W₁₈O₄₉/CsPbBr₃ exhibited a high production rate of benzaldehyde (1,546 μmol·g^-1·h^-1) with high selectivity (80%), and a high CO production rate (143 μmol·g^-1·h^-1) [Figure 1F].

Control of crystal facets

Considering that photocatalytic reactions always occur on the surface of certain photocatalysts, where reaction molecules receive the photogenerated electrons/holes and undergo a series of intermediate steps to produce target product molecules, it is of significant importance to investigate the structures of the exposed surface (i.e., crystal facets) for the construction of high-efficiency photocatalytic materials^[118-121]. The facet effect can influence the performance of the photocatalyst in the following ways: (i) the arrangements of the surface atoms on different facets determine the distinct adsorption and activation pattern of reaction molecules, thereby modulating catalytic activity and selectivity; (ii) different facets can exhibit varied electronic band structures, providing photogenerated electrons/holes with tunable redox abilities, thus affecting the conversion of the substrates the selectivity towards target molecules for a specific photocatalytic reaction; (iii) the charge transfer and separation differ from one crystal facet to another, resulting in diverse charge densities over different facets. To conclude, investigating the facet effect is vital for the development of efficient photocatalysts with high activity, selectivity and durability. Since the cleavage of C–H bond within toluene is dominated by ROS and the oxidative photogenerated holes, it is essential to precisely engineer the exposed crystal facets and analyze their role in the activation of C–H bond.

By adjusting the solution pH, Yang et al. prepared Bi₂MoO₆ samples with exposed {010} and {001} facets to investigate the facet effect of Bi₂MoO₆ photocatalyst in the selective photooxidation of toluene^[122]. ESR measurements revealed that the self-induced IEFs are parallel to the {010} facet but perpendicular to the {001} facet of Bi₂MoO₆. Additionally, the diffusion distance was shorter in the {010} facet, ensuring efficient separation and transfer of photogenerated carriers in the Bi₂MoO₆ sample featuring the {010} facet. As a result, the as-prepared Bi₂MoO₆ sample dominated by the {010} facet exhibited a superior conversion rate of 375 μmol·g^-1·h^-1, which was 2.3 times higher than that dominated by the {001} facet (160 μmol·g^-1·h^-1) for C–H activation in toluene. Another work reported by Li et al. demonstrated that single-crystal oxygen-rich Bi₄O₅Br₂ nanosheets (Bi₄O₅Br₂ SCN) dominated by {10-1} crystal facets were superior to polycrystalline Bi₄O₅Br₂ nanosheets (Bi₄O₅Br₂ PCN) for the selective photooxidation of toluene^[123]. DFT calculations indicated that the exposed{10-1} crystal facets of Bi₄O₅Br₂ favor the physical and chemical adsorption of O₂ molecules. In addition, the single-crystal structure with abundant surface oxygen defect sites was found to significantly facilitate the separation and transfer of photogenerated charge carriers. Benefiting from the advantages in charge carrier dynamics and the unique adsorption properties, the obtained Bi₄O₅Br₂ SCN featured an ultrahigh toluene conversion rate of 1,876.66 μmol g^-1 h^-1, 21-fold higher than the Bi₄O₅Br₂ PCN counterparts (88.22 μmol·g^-1·h^-1) under the blue light irradiation.

More recently, Zhou et al. reported that EC-BiOBr, featuring well-exposed {110} facets, exhibited excellent activity in the selective photocatalytic toluene oxidation^[124]. DFT calculations and in situ Fourier transform infrared (FTIR) spectroscopy revealed that more Lewis acidic Bi sites were exposed on the {110} facets compared to {001} facets, which was found to be conducive to the oriented adsorption of toluene molecules [Figure 2A-F]. This directional adsorption of toluene on the exposed Lewis acid sites might facilitate the precise transfer of electrons from the C–H bond within toluene to the photogenerated holes (h⁺), thereby achieving efficient photo-activation of toluene molecules. As a result, an optimal toluene conversion rate as high as 2,460 μmol·g^-1·h^-1 was achieved on the obtained EC-BiOBr, which is 11-fold higher as compared to the (001) facet-exposed BiOBr (233 μmol·g^-1·h^-1).

Figure 2. Charge difference distribution between different BiOBr crystal facets [(A) 110 facets and (B) 001 facets] and toluene (insets are the side views and 2D displays), with charge accumulation in cerulean blue (Prussian blue in 2D displays) and depletion in yellow (red in 2D displays). HRTEM images of BiOBr crystals with (001) planes (C,E), and a mixture of (001) and (110) planes (D,F). Reproduced with permission, Copyright 2024, American Chemical Society^[124]. 2D: Two-dimensional; HRTEM: high-resolution transmission electron microscopy.

Control of crystal size

Controlling the size of semiconductor materials is an effective strategy for engineering efficient photocatalytic systems. Compared to three-dimensional (3D) bulk microcrystals, low-dimensional [i.e., two-, one- (1D), and zero-dimensional (0D)] nanosized photocatalytic materials exhibit significantly larger surface areas. This increase in surface area exposes more active sites and reduces the transfer distance of photogenerated carriers, thereby enhancing the charge separation efficiency. For instance, semiconductors with 1D nanostructures [e.g., nanotubes, nanorods and nanowires (NWs)] promote charge transfer along their axis direction, leading to linearly polarized light emission, longer carrier lifetimes, and efficient charge separation. These attributes result in outstanding performance in a wide range of applications, such as flexible electronics, chemical sensing, optoelectronic, and photocatalytic devices^[125-128]. Moreover, these nanostructures often exhibit unique morphologies due to crystal anisotropy, excitonic effects arising from reduced dielectric screening, and the preferential exposure of specific facets^[129,130]. These characteristics can be fine-tuned to enhance the visible-light photocatalytic activity for the selective oxidation of toluene.

Two-dimensional nanosized semiconductor materials have garnered widespread application in the field of photocatalysis, owing to their large surface area, effective charge separation efficiency, and suitable energy band structure^[131,132]. Halide perovskites nanoplates Cs₄ZnSb₂Cl₁₂ (CZS), averaging 9.7 nm in size, were found to demonstrate superior photocatalytic activity compared to their bulk counterparts, as reported by Mai et al.^[133]. The size and morphology of the CZS could be precisely controlled by adjusting the starting materials, solvent composition, reaction temperature, and duration, enabling the formation of diverse nanostructures, including 0D CZS nanodots, 1D nanorods and NWs, 2D nanoplates, and 3D nanopolyhedra [Figure 3A]. By evaluating the photocatalytic activity of these CZS nanostructures, they discovered that reducing particle size could increase the surface area of the photocatalysts and induce a positive shift in the energy level of the valence band maximum (VBM), thereby enhancing the oxidation capability of these CZS nanocrystals [Figure 3B]. Compared to the 8.3 nm 0D nanodots, 9.7 nm 2D nanoplates exhibited superior catalytic performance for toluene oxidation to benzaldehyde (conversion rate = 1,998 μmol·g^-1·h^-1, selectivity towards benzaldehyde = 95%) [Figure 3C]. This enhancement was attributed to the anisotropic charge transport within the 2D CZS nanoplates, which inhibited charge recombination and prolonged the lifetime of the charge carriers. Moreover, these nanoplates exposed specific facets with reduced effective mass, further contributing to their enhanced catalytic activity.

Figure 3. (A) Diagram illustrating the control of size and shape in CZS nanocrystals; (B) Energy level schematic of photogenerated carriers in CZS nanocrystals of varying sizes, along with the mechanism for photocatalytic toluene oxidation; (C) Conversion rate of toluene over CZS nanocrystals with different sizes and morphologies. Reproduced with permission, Copyright 2023, American Chemical Society^[133]; (D) Cs₁₂Bi₁₄Br₅₄ cluster derived from crystal structure of CBB; (E) Optimized Cs₁₂Bi₁₄Br₅₄ cluster at the PBE-D3/def2-svp level; (F) Optimized geometry of the Cs₁₂Bi₁₄Br₅₄ cluster incorporating 17 toluene molecules; (G) Focus on a representative Br−H geometry; (H) Photocatalytic toluene oxidation over various photocatalysts; (I) Time-dependent toluene photooxidation over 10 wt % CBB/SBA-15 composite photocatalyst. Copyright 2020, Wiley-VCH^[134]. CZS: Cs₄ZnSb₂Cl₁₂; CBB: Cs₃Bi₂Br₉; PBE: Perdew-Burke-Ernzerhof.

Beyond the conventional colloidal synthesis of uniformly sized and shaped photoactive nanosized guest nanomaterials, employing porous materials such as mesoporous SBA-15 and microporous materials (e.g., silicates, carbons, polymers, COFs, and MOFs) to restrict the growth of these guest materials into low-dimensional nanostructures has proven to be a potent method for fabricating the robust and efficient photocatalysts^[134-137]. This approach leverages the combined properties of both guest photoactive materials and porous host materials, benefiting the overall photocatalytic performance. In this context, Dai et al. reported the synthesis of CBB halide perovskite nanoparticles (2-5 nm) confined within the pore channels of SBA-15, resulting in a high surface to bulk ratio and increased exposure of catalytically active centers on the crystal surface^[134]. DFT calculations further revealed that the formation of perovskite nanoclusters such as Cs₁₂Bi₁₄Br₅₄ could enhance charge separation and facilitate close interactions with hydrocarbon starting materials, thereby promoting the activation of stubborn C–H bonds [Figure 3D-G]. Consequently, the as-prepared CBB/SBA-15 demonstrated an optimal toluene conversion rate up to ca. 12,600 μmol·g_cat^-1·h^-1, significantly higher than that of pure CBB microcrystals (140 μmol·g_cat^-1·h^-1) [Figure 3H and I]. Taking advantage of the unique properties of microporous materials and the exceptional photoactive redox-catalytic properties of polyoxometalate (POM) nanoclusters, Yu et al. synthesized novel helical microporous nanorods (HMNRs) via a facile hydrothermal procedure, employing POM nanoclusters as the building blocks^[138]. The obtained HMNRs demonstrated the unique helical microporous structures, acting as highly optimized nanoreactors for facilitating the interactions between substrates and active sites within the POM units. Additionally, the extended alkyl chains of the decyltrimethylammonium (TTA) surfactants on the surface of HMNRs further enhanced interaction between the toluene molecules and the catalytic active sites. These characteristics collectively led to an optimized conversion rate of the reaction, achieving an impressive 2,254 μmol·g^-1·h^-1, representing a 50-fold increase as compared to that of K₆[α-P₂W₁₈O₆₂]·14H₂O.

Construction of SACs

Recently, SACs, characterized by uniform and isolated metal atoms/ions or mononuclear metal complexes anchored on suitable supporting material, have emerged as a hot topic in the field of photocatalysis^[139-143]. Theoretically, SACs maximize metal utilization by achieving 100% dispersion of metal active centers on the supporting material surface, offering a distinct advantage for developing efficient and cost-effective heterogeneous catalysts. The flexibility in combining single-atom metal centers with various host materials, coupled with precise control over the local coordination environments, opens up extensive possibilities for designing highly efficient SACs. More importantly, the study of SACs provides a unique platform for in-depth understanding of photocatalytic reaction mechanism at the atomic and molecular levels^[144-146]. This detailed insight would rationalize the design of SACs with optimized activity, selectivity, and stability, thereby pushing the boundaries of photocatalysis to new heights.

MOFs, comprising light-harvesting linkers and isolated metal ions/cluster nodes, represent a novel class of crystalline porous materials holding promise for various photocatalytic applications^[147-149]. To enhance the light absorbance and ensure the active sites of MOFs possess appropriate redox potentials for breaking chemically inert C–H bonds under ambient conditions, Xu et al. incorporated Fe³⁺ within the UiO-66 MOF materials by coordinatively binding them to Zr-oxo clusters, resulting in Fe-doped UiO-66 (Fe-UiO-66)^[150]. The anchoring of single Fe³⁺ ions onto UiO-66 induced an effective metal-to-cluster charge transfer (MCCT) process in the as-prepared Fe-UiO-66 photocatalyst. Unlike the traditional light-induced excitation mechanism of semiconductors, which relies on the energy band theory of semiconductors, the unique MCCT transition provides more stable charge separation states. This process not only extends the light absorbance of the Fe-UiO-66 photocatalyst into the visible light range but also facilitates the oriented transfer of photogenerated electrons from Fe³⁺ to Zr-oxo clusters. In the photooxidation system, photogenerated holes oxidize water to produce ^•OH radicals, initiating and driving the C–H activation of toluene, while the electrons reduce O₂ to ^•O₂^- radicals, further promoting the conversion of toluene. This work represents the first report of using MOF-supported SACs for highly efficient photocatalytic C–H bond activation via an MCCT process.

More recently, Xue et al. reported the development of rare-earth single-atom Y anchored on TiO₂ (Y₁/TiO₂) via a facile impregnation strategy for the selective photocatalytic oxidation of toluene [Figure 4A]^[60]. The obtained Y₁/TiO₂ featured substantial atomically dispersed Y₁^δ+-O-Ti³⁺ interfacial sites, facilitating the generation of highly spin-polarized electrons, effectively inhibiting the recombination of the photogenerated electron-hole pairs and thus optimizing the separation of charge carriers [Figure 4B-D]. Mechanism investigations showed that the presence of uniformly dispersed Y₁^δ+-O-Ti³⁺ sites can promote the dual activation of both molecular oxygen and surface lattice oxygen to form highly ROS, thereby leading to an optimal photocatalytic activity (conversion rate = 850 μmol·g^-1·h^-1, selectivity towards benzaldehyde = 94.1%) for Y₁/TiO₂, as compared to anatase TiO₂ (conversion rate = 630 μmol·g^-1·h^-1, selectivity towards benzaldehyde = 64.1%) [Figure 4E and F]. Similarly, da Silva et al. prepared poly(heptazine imides) (PHI)-stabilized iron/manganese SACs through a simple cation exchange procedure^[151]. Operando X-ray absorption spectroscopy (XAS) measurements revealed that the isolated metal sites could undergo oxidation to form high-valent oxo species (i.e., Fe=O and Mn=O) in the presence of photogenerated holes. Experimental results demonstrated that the obtained high-valent oxo species, rather than the conventional reactive free radicals (e.g., ^•O₂^-), were responsible for the efficient activation of C–H bonds followed by the formation of C–O bonds, leading to a high toluene conversion rate and high selectivity towards benzaldehyde.

Figure 4. (A) Diagram showing the synthesis process of the Y₁/TiO₂ photocatalyst; (B) AC-HAADF-STEM image of the as-prepared Y₁/TiO₂; (C) Fourier-transformed EXAFS spectra at the Y K-edge; (D) Schematic model of Y₁/TiO₂ structure; (E) Photocatalytic toluene oxidation performance for various photocatalyst samples; (F) Photocatalytic toluene conversion and the selectivity for benzoic acid, benzyl alcohol, and benzaldehyde after 48-h of photocatalytic reaction. Copyright 2024, American Chemical Society^[60]. AC-HAADF-STEM: Aberration-corrected high-angle-annular-dark-field scanning transmission electron microscopy; EXAFS: extended X-ray absorption fine structure.

In addition to facilitating the efficient transfer and separation of charge carriers, optimizing the surface kinetics of the photocatalyst through the creation of hydrophilic and oleophilic sites is crucial for C–H activation^[152]. Regarding this, Teng et al. undertook the engineering of isolated hydrophilic C≡N-Sb(CN)₃ sites within hydrophobic sp² carbon-conjugated COFs (sp²-c-COFs) specifically for toluene photooxidation^[48]. The resulting Sb5-sp²-c-COFs photocatalyst showcased a remarkable conversion rate of 2.248 mmol·g_catalyst^-1·h^-1 for toluene, representing a 54-fold increase compared to pristine sp²-c-COFs. Moreover, it demonstrated an optimal selectivity of 80% for benzyl benzoate, along with impressive durability over multiple reaction cycles (five cycles, 5 h for one cycle). Mechanism investigation revealed that the superior photocatalytic activity of Sb5-sp²-c-COFs photocatalyst could be attributed to the introduction of hydrophilic water-capturing C≡N-Sb(CN)₃ sites, which were found to enhance charge concentration and promote efficient charge separation in the presence of water.

The advancement of SACs offers valuable insights into the atomic-scale mechanisms of photocatalytic C–H bond activation. Nonetheless, achieving precise construction of uniform active sites on the photocatalyst surface poses challenges, as the exposure and reactivity of individual atom sites can vary significantly. Further research is necessary to explore the intricate engineering of truly active single-atom sites and comprehensively elucidate the structure-activity relationship using multiple state-of-the-art in situ and ex situ techniques.

Defect engineering

The engineering of defects within a semiconductor crystal is widely recognized as an effective strategy for modulating the catalytic performance for specific photocatalytic reactions. The critical role of defects within semiconductors has been systematically investigated in many excellent review works^[153-157]. In summary, the introduction of defect sites can influence the photocatalytic process in the following aspects: (i) defect sites can act as catalytically active sites, facilitating the adsorption and subsequent dissociation of substrate molecules; (ii) the creation of defects can facilitate the efficient transfer of photogenerated electrons between the semiconductor and the reactant molecules; (iii) the presence of defect sites can alter the band structure, extending the light absorption range and thereby promoting light utilization. Herein, with a focus on the photocatalytic activation of C–H bond in toluene, we provided a summary of recent advancements in several representative defective photocatalytic materials, aiming to illustrate how defect sites impact each step of the selective photocatalytic conversion of toluene towards benzaldehyde.

Dating back to 2018, Cao et al. reported the fabrication of Bi₂WO_6-x/amorphous BiOCl (p-BWO) nanosheets via a facile hydrothermal strategy^[47]. The obtained p-BWO demonstrated a unique reversible photochromic phenomenon, which could be attributed to the rapid capture and consumption of photogenerated electrons by abundant [W_(VI)O_6-x] units exposed at the crystalline-amorphous boundaries [Figure 5A]. Comprehensive experimental results confirmed that the introduction of amorphous BiOCl resulted in the disruption of ordered W-O-W bridge bonds within WO₆ octahedral layers, replacing them with W=O terminal bonds, thereby generating numerous [W_(VI)O_6-x] active sites. These defect sites on the p-BWO surface trap and subsequently consume photogenerated electrons for W_(VI) reduction towards W_(V), while W_(V) can be further regenerated to form W_(VI) in the presence of oxygen molecules, completing the overall photochromism cycle [Figure 5B and C]. The unique photochromism phenomenon, induced by the construction of distorted [W_(VI)O_6-x] sites on the catalyst’s outer boundaries, not only helps enhance the efficient separation of photogenerated charge carriers, but also shortens the electron transfer pathlength between photocatalysts and the substrates, thus endowing the p-BWO nanosheets with an impressive toluene conversion rate of 4,388 μmol·g^-1·h^-1 and high selectivity for benzaldehyde (> 80%) [Figure 5D].

Figure 5. (A) Digital images showing p-BWO in its initial state versus its colored state; (B) Diagram illustrating the separation of photogenerated carriers in p-BWO, and the mechanism behind photocatalytic toluene oxidation and photochromism; (C) Schematic depiction of the microscopic structure and operational mechanism of p-BWO; (D) Toluene conversion rate for BWO, and p-BWO at varying substrate loadings. Reprinted with permission, Copyright 2018, Springer Nature^[47]; (E) Illustration of the band structure for 0.01BOC/TiO₂; (F) Toluene conversion rate and the benzaldehyde formation rate for TiO₂, BOC, and 0.01BOC/TiO₂. Reproduced with permission, Copyright 2023, American Chemical Society^[159]. p-BWO: Photochromic Bi₂WO_6-x/amorphous BiOCl; BOC: BiOCl; NHE: normal hydrogen electrode.

While defect sites are crucial for the robust adsorption of reactant molecules, they might also result in strong interactions with product molecules, introducing unnecessary kinetic barriers that hinder the overall reaction process. Given this, Li et al. prepared ultrathin BWO nanosheets modified with moderate Bi defects (denoted as BT-48) via an optimized defect-doping strategy for the photoactivation of toluene^[158]. DFT calculations and physiochemical measurements confirmed that the introduction of Bi defects can not only enhance the adsorption and activation of toluene but also facilitate the desorption of the benzaldehyde product. Moreover, the construction of Bi defects intricately modulates the local electronic structure of the BWO surface, which is conducive to promoting the separation and transfer of photogenerated charge carriers. These characteristics collectively contribute to an optimized production rate (6,781 μmol·g^-1·h^-1) with high selectivity (96%) for benzaldehyde achieved on the as-prepared BT-48 photocatalyst. More recently, Wang et al. reported the synthesis of amorphous BiOCl/TiO₂ (0.01BOC/TiO₂), in which only a small amount of BiOCl (0.01 mol) was strategically assembled on the surface of TiO₂^[159]. In addition to constructing a type-II heterojunction at the interface of the composite photocatalyst [Figure 5E], which facilitates charge carrier separation and transfer, the amorphous BiOCl structure introduces abundant oxygen defect sites on its surface, further promoting the separation efficiency of charge carriers, thereby contributing to the overall improved photocatalytic performance. Moreover, the abundant oxygen defects in the amorphous BiOCl component exhibited strong adsorption and activation of O₂ molecules; in the meanwhile, the product benzaldehyde can be easily desorbed from the amorphous BiOCl surface, further enhancing the selectivity for benzaldehyde. Consequently, the as-prepared 0.01BOC/TiO₂ demonstrated an excellent production rate of 1.7 mmol·g^-1·h^-1 with high selectivity (80%) for benzaldehyde, outperforming the individual BiOCl and TiO₂ [Figure 5F].

It is widely acknowledged that coupling noble metal species with certain semiconductor materials results in the formation of an efficient Schottky barrier at the interface, effectively preventing the severe recombination of photogenerated charge carriers^[160,161]. Furthermore, the incorporation of metal species would generate point defect sites (e.g., O- and S-vacancies) onto the semiconductor surface, which serve as highly active sites for diverse photocatalytic reactions. Regarding this, Li et al. fabricated a Pd/BiOBr composite photocatalyst by dispersing Pd nanoparticles onto the BiOBr flower-like sphere via a facile photo-deposition method^[88]. Comprehensive experimental results revealed that the integration of Pd within the Pd/BiOBr leads to the formation of abundant O-vacancies on the BiOBr surface, enabling the efficient adsorption of both the oxygen and toluene substrates. As a result, the as-prepared Pd/BiOBr demonstrated an optimized toluene conversion rate (6,022.5 μmol·g_cat^-1·h^-1) and high selectivity (> 99%).

Heteroatom doping

Doping heteroatoms, including F, B, N, P, and S atoms, into semiconductor materials is widely recognized as a potent strategy for modulating the local structures and electronic microenvironment of the active sites. Additionally, these dopants play a crucial role in determining the path of charge transfer, as some dopants have the ability to generate additional electronic states within the energy levels of the semiconductor hosts, thereby customizing the dynamics of charge transfer. In this section, we will explore how heteroatom dopants within different semiconductor materials influence the photocatalytic behavior of the fabricated photocatalyst.

Doping Nitrogen atoms into Nb₂O₅ nanomeshes semiconductors (Nb₂O₅-N) was found to significantly enhance the photocatalytic activity, resulting in a 37-fold increase in the benzaldehyde production rate (25 μmol·g^-1·h^-1) as compared to the commercial unmodified Nb₂O₅ (0.66 μmol·g^-1·h^-1), in the photocatalytic oxidation of toluene into benzaldehyde under visible light irradiation, as reported by Su et al.[Figure 6A]^[162]. Based on comprehensive experimental results, the superior photocatalytic performance can be attributed to the synergistic effect of the following reasons [Figure 6B]: (i) N-modification extends visible-light absorption range into the visible region; (ii) the nanomesh nanostructure promotes the transfer and separation of charge carriers, as demonstrated by transient photocurrent response spectra [Figure 6C]; (iii) the presence of Nb⁴⁺ and oxygen vacancies created by N-doping enhances O₂ adsorption and subsequent activation. Recently, Ding et al. investigated several F-doping models within polymeric carbon nitride (PCN) using first-principles calculations^[163]. Simulation results revealed that among various F-doping types in PCN, two stable and cooperative F-doping types were identified and analyzed. One involves the formation of a C(sp³)–F (F_corner type) bond, and the other involves the replacement of an amino group -NH₂ by an F atom (F_N3 type), resulting in the formation of a covalent C–F bond. In the F_corner and F_N3 co-doped PCN model, the F_corner doping reduces the energy level of VBM, leading to a polarized distribution of excited electron-hole pairs, and increases the photogenerated hole distribution on F atoms, thereby strengthening the photocatalytic oxidation capacity and promoting the electron-hole separation efficiency. Meanwhile, the F_N3 doping contributes to a slight reduction in the band gap and improves the light-harvesting capability [Figure 6D and E]. Moreover, the adsorption behavior of toluene is greatly enhanced on the fabricated F-codoped PCN model (E_ads = -0.445 eV) compared to pristine PCN model (E_ads = -0.225 eV), further facilitating the subsequent activation of toluene on the photocatalytic active sites [Figure 6F and G]. However, due to limited computational resources, the study failed to account for the crucial factor of the concentration of different F-doping sites within PCN. Additionally, future research endeavors are expected to leverage more advanced theoretical methodologies and substantial computational resources to provide a more comprehensive picture for understanding the influence of various dopants on modulating the photocatalytic behavior of typical semiconductor materials.

Figure 6. (A) Diagram illustrating and comparing various Nb-based photocatalysts; (B) UV-vis diffuse reflectance spectra for the as-synthesized Nb-based photocatalysts; (C) Transient photocurrent response spectra for the as-prepared Nb-based photocatalysts. Reprinted with permission, Copyright 2020, American Chemical Society^[162]; (D) Diagram showing the photocatalytic oxidation of toluene over F-doped PCN photocatalysts; (E) Band-edge positions of both F-doped and pristine PCN relative to SHE; (F) Structure of toluene adsorbed onto pristine PCN, with E_ads representing the toluene adsorption energy; (G) Structure of toluene adsorbed onto F-codoped PCN, with E_ads representing toluene adsorption energy. Reproduced with permission, Copyright 2023, Elsevier^[163]. UV-vis: Ultraviolet-visible spectroscopy; PCN: polymeric carbon nitride; SHE: standard hydrogen electrode.

Development of novel photocatalysts

The quest for novel types of photocatalysts has long been a focal point in the field of photocatalysis^[40,164,165]. Not only does it expand the family of available photocatalysts, but it also deepens our understanding of the intricate photocatalytic reaction mechanism, involving multiple steps throughout the overall reaction process (e.g., construction of photocatalytic active sites, generation of photogenerated charge carriers, separation and transfer of the charge carriers, adsorption and activation of substrates, stabilization and observation of reaction intermediates, and desorption of products). In this section, we will discuss several representative works on the development of novel photocatalysts, such as MOFs with novel topology, supramolecular Pd₆L₄¹²⁺ porous materials, ultrathin rare earth (RE) oxide NWs, and lead halide perovskites with different ratios of A-site cations^[49,166-170].

Recently, Khoo et al. designed and prepared a novel Fe-doped Zr-MOF featuring a (4,12)-connected ith topology [i.e., Nebraska porous framework (NPF)-520], incorporating rare Zr₉ nodes instead of the widely reported 12-connected Zr₆ nodes, for the selective photooxidation of toluene [Figure 7A and B]^[168]. The incorporation of Fe^III induced a robust MCCT effect, resulting in a red shift in the ultraviolet-visible spectroscopy (UV-vis) and enhanced visible light absorption of the composite photocatalyst as compared to the pristine NPF-520 [Figure 7C]. Moreover, it was revealed that the selectivity of the products could be modulated by adjusting the water content in the photocatalysis reactor. In the presence of water, multiple ROS species (i.e., ^•O₂^-, and ^•OH) are generated, which could oxidize toluene into the partially oxidized derivatives, leading to the formation of benzyl alcohol, benzaldehyde, and benzoic acid [Figure 7D]. In contrast, in the absence of water in the reaction system, only toluene radicals are formed, which then combine with ^•O₂^- to produce solely benzaldehyde, resulting in high selectivity for benzaldehyde production [Figure 7E]. Another work reported by Zhang et al. investigated the effect of amorphization on the photocatalytic performance of phosphonate-based MOF (FePPA) materials [Figure 7F]^[170]. In comparison to crystalline FePPA (c-FePPA), amorphous FePPA (a-FePPA) features more distorted metal nodes (i.e., Fe-oxo clusters), optimizing the electronic structure and thereby promoting the separation and transfer of charge carriers. Additionally, the presence of distorted Fe-oxo clusters provides the amorphous FePPA with enhanced Lewis acidity, thus promoting the adsorption and the subsequent activation of O₂ molecules. As a result, the obtained a-FePPA exhibited an excellent production rate (274 μmol·g^-1·h^-1) for benzaldehyde, much higher than the crystalline counterparts (112 μmol·g^-1·h^-1).

Figure 7. (A) Illustration showing the preparation of (4,8)-connected flu and (4,12)-connected ith topologies from a tetrahedral ligand, involving Zr₆ and Zr₉ nodes, respectively; (B) Stepwise process for the rigidification and steric modification of tetrahedral ligands to create NPF-520 with ith topology; (C) Diagram depicting the charge transfer mechanism from Fe^III to Zr-oxo cluster in the Fe^III-decorated Zr₉ node (with Zr in violet, O in red, Fe in orange, and Cl in green); (D) Development of reaction intermediates when water is present versus in an anhydrous environment; (E) Proposed pathway for the photocatalytic oxidation of toluene. Reprinted with permission, Copyright 2023, American Chemical Society^[168]; (F) Diagram illustrating the synthesis of both amorphous and crystalline FePPA photocatalysts. Reproduced with permission, Copyright 2024, Wiley-VCH^[170]. NPF: Nebraska porous framework.

To mimic the concept of C–H bond activation within a natural enzyme’s hydrophobic cavity, the construction of supramolecular cavities with diverse shapes, sizes, and electronic properties has been explored as a viable method for hosting photocatalytic reactions^[171-173]. In this context, Das et al. employed a cationic Pd₆L₄¹²⁺ (L = ligand) nanocage to encapsulate toluene molecules and convert them into benzaldehyde under visible light irradiation^[49]. Comprehensive measurements revealed that the C–H bond in toluene can be preorganized and polarized within the water-soluble cationic Pd₆L₄¹²⁺ nanocavities. This preorganization facilitates the photoactivation of toluene molecules via an ultrafast proton-coupled electron transfer (PCET) reaction under ambient conditions. Further investigation indicated that the preorganization was induced by the strong E-fields generated by the electrostatic field of the six Pd²⁺ ions in the Pd₆L₄¹²⁺ nanocage. Modulating the E-fields was proposed as a promising approach to enable selective bond polarization in complex polyatomic molecules encapsulated within the well-designed nano-hosts.

Metal-halide perovskites have emerged as highly effective photocatalysts due to their exceptional light-harvesting capabilities, suitable band gaps, and excellent charge transfer properties. They have been successfully employed within several photocatalytic processes, demonstrating significant potential for selective photoactivation of various organic/inorganic molecules^[174-177]. The substitution of B-site Pb²⁺ cations with In³⁺, Sb³⁺, Bi³⁺ has been widely investigated due to their low toxicity and their beneficial effects on optimizing the band gap structures, thereby improving light-harvesting ability and redox potentials of photogenerated charge carriers. However, A-site cation substitution is rarely considered, as these cations do not directly participate in the construction of the perovskite band gaps. Regarding this, Zhang et al. prepared lead-free A₂Sb₂Br₉ perovskite nanocrystals with varying ratios of Cs and CH₃NH₃ (MA) to explore how changes in the A-site composition influence the photocatalytic behavior of the fabricated perovskite photocatalysts^[166]. Experimental results revealed that the distortion of [SbBr₆] octahedron in the as-prepared Cs_xMA_3-xSb₂Br₉ increases with the incorporation of larger MA ions. The A-site-induced distortion facilitates the electron depletion of Sb-sites and accumulation on the Br-sites, resulting in decreased ability on the C–H bond activation and reduced stability for photocatalytic toluene oxidation. The work elucidated the relationship between A-sites induced octahedron distortion and photocatalytic performance, offering new insights for designing highly efficient halide perovskite-based photocatalysts.

Ultrathin inorganic NWs have garnered significant attention, showcasing unique chemical/physical properties as their diameters decrease to only a few crystal cells. However, their controlled and facile preparation remains a great challenge. In this context, Fu et al. developed a novel and general method, which enables the synthesis of a series of RE oxide ultrathin NWs at atmospheric pressure and low temperature (50 °C), assisted by POM^[167]. The low temperature was found to prevent crystal growth in the 2D/3D directions, leading to the formation of 1D NWs, while the introduction of POM facilitated the smooth 1D growth of 0D monomers. The as-prepared RE oxide NWs exhibited polymer-like behaviors, displaying high viscosity and rheological properties, and novel structure of nanosized red blood cells (RBCs) could be obtained through electro-spinning using only concentrated NWs without the addition of any polymers. Moreover, the Ce-Mo-O NWs were found to be a potential photocatalyst, affording an excellent toluene conversion rate and durability in photocatalytic toluene oxidation reaction.

In the above content, we have provided a systematic description and classification of different modification methods for the fabrication of efficient photocatalysts used in toluene photooxidation, along with an analysis of the characteristics associated with each method. To facilitate a more intuitive comparison between different modification methods. Table 1 presents the characteristics of different methods achievable in various photocatalytic systems. This comprehensive summary enables researchers to identify the most suitable approach based on their specific research needs.

In situ characterization techniques

In situ characterization techniques, capable of monitoring the dynamic evolution of photocatalysts during simulated/real catalytic reaction processes, are instrumental in designing highly efficient photocatalysts and elucidating related reaction mechanisms. Many excellent review works have highlighted the crucial role of various in situ techniques in the field of photocatalysis^[180-184]. In this section, we will present and summarize several key applications of in situ/operando characterization techniques aimed at selective photocatalytic toluene conversion into benzaldehyde, including probing of chemical structure of photocatalysts, tracking transfer of photogenerated charges, assessing spatial distribution of charge carriers, and most importantly, determining the surface reaction pathways within the reaction environment.

In situ electron paramagnetic resonance

Determining the actual reaction pathways and mechanisms is crucial for the rational design and optimization of photocatalytic conversion of toluene towards benzaldehyde. During the process, multiple reactive paramagnetic species with unpaired electrons, such as ^•OH, ^•O₂^-, carbon-centered radicals (^•R), are generated on the photocatalyst surfaces. To identify the formation and evolution of specific free radicals, in situ electron paramagnetic resonance (EPR) or ESR measurement has been developed and employed under simulated reaction conditions, aiding researchers in predicting the ongoing reaction pathways and elucidating reaction mechanisms^[185]. Besides, it can also help monitor the generation of unpaired electrons in solid photocatalysts under light irradiation and detect the presence of crystal defects.

Since most free radicals are unstable, highly reactive, and short-lived, their detection is often performed using spin trap agents [e.g., 5,5-dimethyl-1-pyrroline-N-oxide (DMPO), 5-tert-butoxycabony 5-methyl-1-pyrroline-N-oxide (BMPO), and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO)] that selectively react covalently with a highly transient radical to generate a more stable radical (i.e., spin adduct). For instance, Cui et al. employed in situ EPR spectroscopy to directly confirm the presence of ^•O₂ and h⁺ during the reaction process by using DMPO and TEMPO as spin traps^[110]. As depicted in Figure 8A, four characteristic peaks corresponding to DMPO-^•O₂ could be detected after 3 min light irradiation. The signals recorded at the same illumination time for TiO₂/CBB were much more intense than those for pure TiO₂ and CBB, which accounts for the higher photocatalytic activity of TiO₂/CBB composite photocatalyst [Figure 8B]. Moreover, three characteristic peaks corresponding to TEMPO-h⁺ with similar intensities could be observed among the TiO₂, CBB, and TiO₂/CBB photocatalysts under dark conditions. However, the signals for TiO₂/CBB were much lower than those for TiO₂ and CBB after 5 min of light irradiation, indicating that more holes were consumed in TiO₂/CBB in the same period [Figure 8C and D]. The in situ EPR results revealed that the synergistic effect between ^•O₂ and h⁺ promotes the selective conversion of toluene, demonstrating an efficient separation of photogenerated charge carriers over the TiO₂/CBB.

Figure 8. In situ ESR spectra of radical adducts captured by (A) DMPO (^•O₂^-) and (C) TEMPO (h⁺) in TiO₂/CBB dispersion under both dark and light irradiation conditions. A comparison of the ESR signal intensities for (B) DMPO (^•O₂^-) and (D) TEMPO (h⁺) in TiO₂, CBB and TiO₂/CBB photocatalysts after the same exposure to light. Reproduced with permission, Copyright 2023, Elsevier^[110]. ESR: Electron spin resonance; DMPO: 5,5-dimethyl-1-pyrroline-N-oxide; TEMPO: 2,2,6,6-tetramethyl-1-piperidinyloxy; CBB: Cs₃Bi₂Br₉.

Besides identifying the exact radical species, in situ EPR measurements allow for the direct observation and monitoring of the paramagnetic active species within solid photocatalysts during photocatalytic processes. For example, Xu et al. utilized in situ EPR techniques to investigate the electron transfer mechanism within the obtained Fe-UiO-66 under light irradiation^[150]. The EPR peak at g = 4.0, corresponding to Fe(III) with a high-spin state, decreased after light irradiation, accompanied by the presence of a new peak at g = 2.02. This new peak corresponds to the reduction product of ^•O₂ adsorbed onto Zr-oxo clusters. Upon adding methanol as a hole sacrificial agent, the signal at g = 2.00, representing oxygen-centered active sites in Zr-oxo clusters that accept electrons from Fe(III), became evident under light irradiation upon N₂ injection. This observation demonstrates the efficient electron transfer from Fe(III) sites to Zr-oxo clusters (i.e., MCCT process) in Fe-UiO-66.

Although EPR is highly sensitive to unpaired electrons and can detect changes of paramagnetic centers under light irradiation, its application often requires the use of a spin trap and/or low temperatures to ensure the detection of these unstable, low-yield, short-lived, reactive species with unpaired electrons. Due to these limitations, the application of EPR in photocatalysis under operando conditions remains scarce.

In situ FTIR spectroscopy

FTIR spectroscopy has been widely used for both quantitative and qualitative identification of the molecular structure in various organic and inorganic materials. Nevertheless, conventional FTIR is limited to offline analysis, making it challenging to monitor intermediates and dynamic changes of catalytically active sites over photocatalysts in real time. With the advent of high-sensitivity detectors, in situ FTIR has rapidly advanced, allowing for real-time exploration of the evolution of adsorbed molecules and reaction intermediates during photocatalytic reactions^[186-191]. Additionally, it facilitates real-time visualization of reactant adsorption and specific product desorption, two crucial steps in the photocatalytic process, thereby deepening the understanding towards the superior activity and selectivity of certain photocatalysts, offering insights that drive the development of more efficient and targeted photocatalytic systems.

For instance, Zhang et al. utilized in situ FTIR spectroscopy to investigate the adsorption of toluene and the formation of various oxygenated intermediates over the as-prepared amorphous FePPA photocatalyst [Figure 9A]^[170]. As depicted in Figure 9B, typical signals (3,120-3,040, 1,550-1,440, and 1,603 cm^-1) attributed to C–H bonds and aromatic rings, along with signals (2,970-2,830 and 1,420-1,360 cm^-1) assigned to the stretching and bending vibration of C–H bonds within the methyl group, both increased on prolonging the irradiation time, demonstrating effective adsorption of toluene on the catalyst surface. With extended light irradiation, peaks associated with the aldehyde group (1,698 cm^-1 for C=O bonds and 2,733 cm^-1 for C–H bonds) emerged and gradually intensified, indicating the generation of benzaldehyde. Additionally, the peak at 1,176 cm^-1 could be ascribed to the peroxide intermediate, which was formed through the interaction between the ^•O₂^- radical and the ^•R radical. Furthermore, two weak peaks at 1,652 and 1,243 cm^-1 were identified, corresponding to water generated during the dehydration of the peroxide intermediate and the by-product benzyl alcohol, respectively.

Figure 9. (A) Proposed mechanism for photocatalytic toluene oxidation using the FePPA photocatalyst; (B) In situ FTIR spectra showing the photocatalytic toluene oxidation using a-FePPA. Reprinted with permission, Copyright 2024, Wiley-VCH^[170]. Proposed reaction pathways for (C) pristine TiO₂ and (D) Y₁/TiO₂ photocatalysts. In situ DRIFTS spectra of toluene adsorbed on (E) pure TiO₂ and (F) Y₁/TiO₂ during a 30-minute Ar flushing procedure. In situ DRIFTS spectra of toluene oxidation over (G) pure TiO₂ and (H) Y₁/TiO₂. Reprinted with permission, Copyright 2024, American Chemical Society^[60]. FTIR: Fourier transform infrared; DRIFTS: diffuse reflectance FTIR spectroscopy.

Diffuse reflectance FTIR spectroscopy (DRIFTS), which enables the direct use of powdered photocatalysts and thus mitigates the inconvenience of external transport compared to transmission IR, has become an increasingly popular tool for investigating heterogeneous photocatalysts and screening the photocatalytic reaction processes in the in situ/operando mode^[192-194]. For example, Xue et al. employed in situ DRIFTS to elucidate the mechanism of the photooxidation of toluene over the two photocatalysts: pristine TiO₂ and TiO₂ supported yttrium single atom photocatalyst (Y₁/TiO₂) [Figure 9C and D]^[60]. For toluene adsorption over TiO₂ photocatalyst, strong adsorption bands corresponding to the stretching vibrations of C–H bonds (3,079 and 3,025 cm^-1) and the C=C bonds (1,600 and 1,494 cm^-1) of the aromatic ring were detected [Figure 9E]. In contrast, only weak adsorption bands were observed for the C–H stretching vibrations (2,926 and 2,876 cm^-1) and CH₃ bending vibrations (1,460 and 1,380 cm^-1) of the methyl group, indicating that pristine TiO₂ prefers to adsorb the benzene ring rather than the methyl group of toluene. For the Y₁/TiO₂ catalyst, distinctive C=O stretching vibrations (1,687 and 1,280 cm^-1) attributed to the aldehyde group of benzaldehyde were detected [Figure 9F]. Concurrently, the bending vibrations of O–H bonds (1,326 cm^-1) decreased with prolonged irradiation, indicating the oxidation of benzyl alcohol to benzaldehyde. The comparison between TiO₂ and Y₁/TiO₂ reveals that toluene undergoes partial oxidation to benzaldehyde via a lattice oxygen oxidation process over Y₁/TiO₂, as evidenced by in situ DRIFTS measurements performed under Ar conditions. To establish a plausible reaction mechanism for the selective photooxidation of toluene over the two photocatalysts, in situ time-resolved DRIFTS were further conducted under the simulated photocatalytic reaction conditions. In the case of pristine TiO₂, characteristic bands corresponding to CO₂ increased with the irradiation time, suggesting the overoxidation of toluene into CO₂ over TiO₂ [Figure 9G]. However, for the as-prepared Y₁/TiO₂, the absence of CO₂ bands suggests an effective inhibition of toluene overoxidation, affirming the high selectivity over Y₁/TiO₂. Moreover, the stretching band of C=O bond corresponding to benzaldehyde adsorption over Y₁/TiO₂ was blue-shifted as compared to pristine TiO₂, indicating a relatively weak adsorption of aldehyde group in the benzaldehyde product, further illustrating the inhibition of benzaldehyde overoxidation [Figure 9H].

In situ X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS), known for its high sensitivity to the chemical and electronic states of most elements, has been applied to the mechanistic elucidation of several catalytic processes^[195-198]. With in situ light illumination, the transfer and migration of photogenerated electrons and holes under simulated light irradiation could be visualized through the shifts of binding energies, as the dynamics of photogenerated charge carriers could alter the chemical and electron environment of the targeted elements during photocatalytic processes^[199-202]. The construction of heterojunctions by coupling two different semiconductors with suitable band structures has long been regarded as an effective strategy to improve the performance of the photocatalytic system. However, the band alignment could vary significantly among diverse types of heterojunction structures (e.g., type-II and Z-scheme heterostructures). To characterize the specific type of heterojunction structure within a fabricated photocatalyst, in situ XPS measurements have been employed, providing robust and direct evidence.

For instance, Bai et al. investigated the charge transfer behavior over the CBB/d-BiOBr photocatalyst through in situ XPS under visible light irradiation^[178]. As depicted in Figure 10A, two peaks in the O 1s XPS spectra underwent a slight positive shift of 0.07 eV upon light irradiation, indicating a decrease in electron density of d-BiOBr component. Meanwhile, two characteristic peaks in Cs 3d XPS spectra underwent a negative shift of 0.2 eV, indicating the accumulation of electrons in the CBB component [Figure 10B]. Since the Fermi level of CBB is higher than that of d-BiOBr, when the two components come into close contact, free electrons in CBB would spontaneously transfer toward d-BiOBr to reach a new equilibrium state. This electron migration results in the alignment of the Fermi levels of CBB and d-BiOBr, with CBB becoming positively charged and d-BiOBr negatively charged. Consequently, an internal built-in E-field directed from CBB to d-BiOBr is formed at the composite interface. The photogenerated electrons in the CB of d-BiOBr would migrate to the VB of CBB and recombine with the photogenerated holes. Therefore, the charge transfer follows a direct Z-scheme structure within CBB/d-BiOBr, rather than a type-II one, as evidenced by in situ XPS. Similarly, with the aid of in situ light irradiation XPS measurements, a distinct S-scheme heterostructure was identified in the W₁₈O₄₉/CsPbBr₃ composite photocatalyst, which was found to demonstrate improved photocatalytic activity for selective toluene oxidation^[117].

Figure 10. Operando XPS spectra (A) O 1s and (B) Cs 3d of CBB(1.7)/d-BiOBr in both dark and visible light irradiation conditions. Reproduced with permission, Copyright 2022, American Chemical Society^[178]. XPS: X-ray photoelectron spectroscopy; CBB: Cs₃Bi₂Br₉; d-BiOBr: defective BiOBr nanosheets.

In situ/operando XAS

XAS is an effective technique for providing detailed electronic and geometric information of active sites over photocatalysts, including inter-atomic distances, types of chemical bonds, coordination numbers, and structural disorder^[203-205]. Generally, XAS can be divided into two parts: X-ray absorption near edge structure (XANES) and EXAFS. XANES extends to approximately 50 eV above the absorption edge and is highly sensitive to the electronic and geometric configuration of the absorbing atom. It reflects the density of electronic states near the Fermi level, thereby revealing the oxidation states, local symmetry, and chemical bonding of the absorbing atom. The region of EXAFS follows XANES, beginning at roughly 50 eV above the absorption edge and extending to thousands of electron volts beyond it, reflecting the coordination number, coordination distance, and types of atoms surrounding the target atom. In situ/operando time-resolved XAS can monitor the charge transfer, evolution of geometric and electronic structures of selected atomic components, offering valuable insights into the relationship between structure and catalytic performance in photocatalytic systems under simulated or realistic reaction conditions^[206].

For example, the oxidation of Mn and Fe sites into active high-valent oxo species during photocatalysis was investigated by the use of operando XAS measurements, as reported by da Silva et al.^[151]. The XAS spectrum of the dehydrated Mn-PHI photocatalyst at 250 °C in flowing He is fully consistent with the presence of only the Mn(II) oxidation state [Figure 11A]. At the same temperature, when exposed to oxygen, water, and light illumination, a notable increase in the spectral weight can be observed in the high-energy regions, specifically between 640 and 645 eV [Figure 11B and C]. Furthermore, under these conditions, a noticeable shift towards higher energies could be detected for the main peak at ca. 640 eV. These changes were reversible upon repeated cycling of light on and off, also indicating the generation of Mn in higher oxidation states, possibly Mn(IV), induced by light irradiation [Figure 11D-F]. Similar results were observed at the Fe L_2,3-edges for the Fe-PHI catalyst, suggesting that both Fe and Mn atoms interact with photogenerated holes to produce high-valent species, which act as active centers for the selective photooxidation of toluene.

Figure 11. (A) XAS spectra for the L₃ and L₂ edges of Mn-PHI photocatalyst; (B) Operando XAS spectra at the Mn L_3,2-edges of Mn-PHI under various conditions; (C) Intensity of high-energy peaks at the Mn L₃-edges under various conditions, with yellow boxes marking the illumination periods during the photocatalytic reaction; (D) Proposed photocatalytic mechanism for the formation of hydrogen peroxide and high-valent metal species; (E) Mechanism for the generation of metal-oxo species via photogenerated holes and H₂O₂; (F) Photocatalytic mechanism for the oxidation of toluene into benzaldehyde through benzyl alcohol, catalyzed by iron- and manganese-oxo species. Reproduced with permission, Copyright 2023, Wiley-VCH^[151]. XAS: X-ray absorption spectroscopy; PHI: poly(heptazine imides).

DFT calculations

DFT calculations are extensively used to supplement the experimental findings in the field of photocatalysis^[207-209]. Specifically, leveraging DFT calculations provides useful information to probe the mechanism of toluene photooxidation, encompassing substrate adsorption, C–H bond activation, intermediate migration, and product desorption^{[134,210,211]}. This theoretical approach contributes to a deeper understanding of the reaction mechanism and also aids in the design of novel photocatalysts with improved performance. In this section, we will present several representative works that employ DFT calculations for toluene photooxidation over various photocatalysts, aiming to elucidate essential theoretical aspects that researchers should consider when investigating the selective photooxidation of toluene.

O₂ adsorption on photocatalyst surface

The generation of ROS plays a crucial role in enhancing the performance of the photocatalyst during the catalytic conversion of toluene into high-value-added benzaldehyde. However, the activation of gaseous oxygen into ROS is strongly dependent on the extent of oxygen adsorption on the photocatalyst surface^[212-214]. Therefore, it is of great significance to investigate the adsorption sites of O₂ within the as-prepared photocatalyst during photocatalysis and correlate the behavior with the enhanced catalytic performance. Experimentally, temperature-programmed desorption with oxygen (O₂-TPD) and the O₂ adsorption isotherms are widely applied to investigate the oxygen absorption capacity of the target catalyst^[215-219]. In addition to experimental techniques, DFT calculations can also provide insights into the adsorption of oxygen molecules over a simulated photocatalyst model.

For example, Li et al. used DFT calculations to reveal the active sites for the adsorption and activation of oxygen molecules over the as-synthesized Pd/BiOBr photocatalyst^[88]. First, the adsorption energies of O₂ on various optimized photocatalyst models were calculated to demonstrate the likelihood of a selected model adsorbing oxygen molecules. It was revealed that the Pd/BiOBr-Vo model exhibits stronger interactions with both O₂ and toluene molecules compared to the BiOBr-Vo model. In the simulated Pd/BiOBr-Vo surface, the interfacial Pd sites show significantly stronger chemical interactions with oxygen compared to the BiOBr-Vo surface. On the Pd/BiOBr-Vo model, oxygen molecules can be adsorbed onto either interfacial or top Pd clusters, enabling more favorable oxygen adsorption on the photocatalyst surface [Figure 12A-C]. Since both ^•O₂^- and ¹O₂ are the dominant reactive species reacting with the adsorbed toluene molecules during the photocatalytic conversion process, it is reasonable to consider that the two adsorption modes of O₂ on the interfacial Pd site and top Pd site exist simultaneously, which could facilitate the formation of both ^•O₂^- and ¹O₂ through the transfer of photogenerated electrons and energy, respectively. According to the calculated results, the introduction of Pd nanoparticles onto the BiOBr surfaces could enhance the chemical adsorption of O₂ molecules on the composite photocatalyst, ensuring efficient electron and energy transfer between the reactants and the Pd/BiOBr photocatalyst [Figure 12D].

Figure 12. DFT-calculated local configurations for O₂ molecule adsorbed onto the (A) V_O site in the BiOBr–V_O surface; (B) interfacial Pd site (site 1) and (C) top Pd site (site 2) in the Pd/BiOBr–V_O surface; (D) The adsorption energies (Ead) for O₂ and toluene absorbed onto the different sites. Reproduced with permission, Copyright 2020, Royal Society of Chemistry^[88]. The models for O₂ adsorption on (E) c-FePPA and (F) a-FePPA (Fe, orange; C, gray; P, yellow; H, white; O, red; adsorbed O, blue). Reproduced with permission, Copyright 2024, Wiley-VCH^[170]. DFT: Density functional theory.

Another study showed that the amorphization of phosphonate-based MOF photocatalysts would enhance oxygen adsorption and promote its activation for the subsequent oxidation process, as reported by Zhang et al.^[170]. The O₂ adsorption isotherms revealed that a-FePPA exhibits both a higher maximum O₂ adsorption amount (3.13 cm³·g^-1) and O₂ adsorption amount per unit of specific surface area (0.036 cm³·m^-2) compared to c-FePPA (maximum O₂ adsorption amounts = 1.66 cm³·g^-1 and O₂ adsorption amount per unit of specific surface area = 0.032 cm³·m^-2). DFT calculations were performed to further evidence the experimental results [Figure 12E and F]. The model of a-FePPA, featuring distorted Fe-oxo clusters, demonstrated a greater adsorption energy of -3.07 eV as compared to the c-FePPA model (-1.54 eV), confirming that O₂ molecules are more readily adsorbed on the simulated a-FePPA model. The combined experimental and computational results suggested that amorphization could efficiently facilitate the adsorption of oxygen and its further activation to initiate the photooxidation process.

C–H bond activation on the photocatalyst surface

It is widely acknowledged that the rate-limiting step in the oxidation of toluene to yield benzaldehyde is the dissociation of C(sp³)–H bond within toluene molecules. To estimate the ease of C(sp³)–H bond on various photocatalyst active sites, dissociation energy of C(sp³)–H bond (E_C-H) was calculated by DFT calculations. For instance, Li et al. used DFT calculations to investigate the dissociation of C(sp³)–H bond in toluene on the simulated (001) surfaces of Cs₂AgBiBr₆ (CABB), 3.13% Fe-CABB, and 1.56% Zn-CABB by measuring the E_C-H values in these three computational models [Figure 13A-C]^[220]. Without light irradiation (i.e., no photogenerated holes produced on the photocatalyst surfaces), the E_C-H values were determined to be 3.23, 3.17, and 3.02 eV for CABB, 1.56% Zn-CABB, and 3.13% Fe-CABB, respectively. Such high E_C-H values indicated that efficient C(sp³)–H bond activation and dissociation would not occur in the absence of photogenerated holes. In order to simulate the photocatalyst behavior under light irradiation, one electron was manually transferred from the highest occupied orbital of the catalyst system to the lowest unoccupied orbital for each k-point, thus creating a hole in the VB. The E_C-H values were calculated as 1.89, 1.69, and 2.21 eV for CABB, 1.56% Zn-CABB, and 3.13% Fe-CABB under light irradiation, which are significantly lower compared to those under dark conditions. Among the three models, 1.56% Zn-CABB exhibited the lowest E_C-H, indicating its superior capability for the activation of C(sp³)–H bond in toluene, further facilitating the generation of benzyl radical, thereby accelerating the overall photooxidation process. To understand the role of active Br^- sites in generating photogenerated holes and facilitating the dissociation of C(sp³)–H bond in toluene molecules, the electron densities of the surface Br^- were modulated over diverse photocatalyst systems, and the electron density differences (Δρ) between the excited state and ground state were subsequently analyzed [Figure 13D-F]. According to the computational results, the surface electron density at the selected Br1 site in 1.56% Zn-CABB was revealed to be significantly lower than at the Br1 sites in both the 3.13% Fe-CABB and the pristine CABB in their excited states. Benefiting from its lower surface electron density in the excited state, the Br1 in 1.56% Zn-CABB demonstrated a stronger oxidation capability in dissociating the C(sp³)–H bonds under light irradiation, thereby yielding the lowest E_C-H value and achieving the highest toluene conversion rate.

Figure 13. DFT calculations showing the energy changes associated with toluene adsorption and C(sp³)–H bond dissociation for the GS and ES of (A) CABB, (B) 1.56% Zn-CABB and (C) 3.13% Fe-CABB; The electron density colored along the (010) plane based on the density difference between the GS and ES, as depicted in (D) CABB, (E) 1.56% Zn-CABB and (F) Fe-CABB. Red circles indicate Br atoms adjacent to the substituent. Reproduced with permission, Copyright 2023, Wiley-VCH^[220]; (G) DFT calculations showing the influence of surface-localized holes on toluene adsorption (E_ads) and C(sp³)–H bond dissociation. Reproduced with permission, Copyright 2022, American Chemical Society^[178]; (H) Geometries and energies of transition state models (TS1, TS2, TS3, and TS4); (I) Calculated energy profiles of toluene adsorbed on different structure models. Reproduced with permission, Copyright 2024, Springer Nature^[36]. DFT: Density functional theory; GS: ground state; ES: excited state; CABB: Cs₂AgBiBr₆; TS: transition state.

To elucidate the superior oxidation capability of Z-scheme CBB/d-BiOBr compared to pristine d-BiOBr, Bai et al. employed DFT calculations to explore the impact of accumulated photogenerated holes on the photocatalytic performance^[178]. Two computational models were first simulated: one for d-BiOBr with one hole and another for CBB/d-BiOBr with two photogenerated holes. Then, the energy barriers for the rate-limiting step (i.e., ^*C₇H₈ + h⁺ → ^*C₇H₇ + H⁺) were calculated on the constructed photocatalyst models [Figure 13G]. In the case of d-BiOBr with no holes, the energy barrier was determined to be 1.96 eV, suggesting that the activation of C(sp³)–H bond is rather challenging under dark conditions. For the d-BiOBr model with one hole, the energy barrier is only slightly reduced to 1.79 eV, indicating that the step is still not significantly promoted. In contrast, for the CBB/d-BiOBr model with two holes, the energy barrier was dramatically lowered to 0.85 eV, significantly lower than that of the d-BiOBr under the conditions. This substantial reduction in the energy barrier demonstrates that the transfer of photogenerated holes to the catalyst surface effectively facilitates the dissociation of C(sp³)–H bond in toluene molecules.

Similarly, Shi et al. prepared Ni-doped BWO (Ni/BWO) photocatalyst oxidation via a facile solvothermal method for efficient photocatalytic oxidation of toluene^[36]. DFT calculations were employed to demonstrate the superiority of the as-synthesized Ni/BWO over the pristine BWO in terms of C–H bond activation in toluene molecules. As shown in Figure 13H and I, four computational models including Ni/BWO (001), Ni/BWO (010), BWO-O_v (001), and BWO-O_v (010) were established for calculating the activation path of C–H bond in toluene. The energy barriers for the dehydrogenation of toluene on the four models were calculated to be 0.23, 0.28, 1.02, and 1.16 eV for Ni/BWO (001), Ni/BWO (010), BWO-O_v (001), and BWO-O_v (010), respectively, indicating that the Ni/BWO photocatalysts exhibit higher activity due to the integration of Ni dopants on the BWO substrates. Notably, toluene dehydrogenation was more easily achieved on the Ni/BWO (001) model, which can be explained by the formation of Bi···C and O···H coordination between frustrated Lewis pairs (FLP) and the C–H bond. The transient charge density difference image of transition state (TS)-4 showed that the coordination led to the transfer of electrons from the H atoms in the C–H bond to the O atoms and from the Bi atoms to the C atoms, resulting in effective activation of C–H bonds.

Screening of intermediates during toluene photooxidation process

To elucidate the photooxidation pathways and intermediates involved in the photocatalytic conversion of toluene to benzaldehyde, DFT calculations were conducted to offer valuable atomistic insights into the mechanisms underlying the photocatalytic oxidation process over certain photocatalysts. For instance, Zhou et al. employed DFT calculations to simulate the potential reaction pathways following the dissociation of C–H bonds over EC-BiOBr, based on the detection of intermediate products^[124]. As illustrated in Figure 14A and B, the oxidation of absorbed benzyl radical (^*PhCH₂^•) to ^*PhCH₂OO^- by ^•O₂^- was exothermic (ΔG = -0.82 eV), while its oxidation to ^*PhCH₂OO^• was endothermic by O₂ (ΔG = 2.0 eV) under standard conditions (atmospheric pressure and room temperature). Subsequently, ^*PhCH₂OO^- could spontaneously form ^*PhCH₂OOH (ΔG = -1.76 eV), whereas the reaction between ^*PhCH₂OO^• and toluene to generate ^*PhCH₂OOH appeared to be challenging (ΔG = 0.89 eV). Following the formation of ^*PhCH₂OOH, it undergoes an autonomous conversion into benzaldehyde (ΔG = -2.90 eV). However, the production of benzaldehyde and benzyl alcohol from ^*PhCH₂OOH was calculated to be highly unfavorable (ΔG = 27.31 eV). The obtained results indicate that ^•O₂^- was the primary oxidizing species for the conversion of benzyl radical to benzaldehyde, rather than O₂ molecules.

Figure 14. (A) Proposed reaction mechanism of benzyl radical oxidation over EC-BiOBr according to DFT calculations. The asterisk (*) denotes the adsorption site on the substrate; (B) Schematic illustrating the toluene oxidation mechanism over EC-BiOBr photocatalyst. Reproduced with permission, Copyright 2024, American Chemical Society^[124]; (C) Gibbs free energy profiles of photocatalytic toluene oxidation over pure CBB and TiO₂/CBB. Reproduced with permission, Copyright 2023, Elsevier^[110]. EC: (110) facet-exposed; DFT: density functional theory; CBB: Cs₃Bi₂Br₉.

Cui et al. used DFT calculations to model the interactions of CBB and TiO₂/CBB with each intermediate during the photocatalytic conversion of toluene^[110]. The Gibbs free energy (ΔG) for each reaction step, as shown in Figure 14C, was calculated. The results indicated that the formation of benzyl radicals derived from toluene is the rate-determining step during the photocatalytic toluene conversion reaction, with TiO₂/CBB exhibiting a lower benzyl formation energy than that over pristine CBB. After dehydrogenation, the benzyl radical reacted with the generated ^•O₂^- to form ^*PhCH₂OH intermediates. Benzaldehyde was then produced from ^*PhCH₂OH intermediates via two successive dehydrogenation steps, involving the dissociation of the O–H bond within the -OH structure and the C–H bond within the methyl structure.

Machine learning in novel photocatalyst discovery

Machine learning (ML) is driving transformative advancements in material science, particularly in the realm of discovering and optimizing novel photocatalysts^[40,221,222]. This technology optimizes the entire process, from collecting and preparing data to selecting features, training models, making predictions, and testing performance. By efficiently navigating the expansive chemical space, ML stands to significantly reduce the time and costs traditionally associated with identifying new photocatalysts. However, several challenges remain unsolvable. Ensuring the quality and quantity of data across the research community is crucial for improving model accuracy and reproducibility. Additionally, developing interpretable ML models is essential for gaining deeper insights into the intricate mechanisms underlying photocatalytic activity, thereby facilitating knowledge transfer and innovation. Integrating ML seamlessly into experimental workflows to create automated systems that continuously refine predictions through feedback loops is another critical challenge. Moreover, enhancing the ability of ML models to generalize knowledge across different photocatalytic systems through techniques such as transfer learning promises to broaden their applicability and impact. Despite these challenges, ML holds immense promise in revolutionizing the discovery and optimization of photocatalysts, enabling researchers to accelerate the development of next-generation materials that offer superior performance and sustainability benefits.

Development of novel In situ photocatalytic characterization techniques

In situ characterization techniques represent pivotal tools in the quest to unravel the intricate dynamics occurring on photocatalyst surfaces during chemical reactions^{[180,223,224]}. These methods provide invaluable real-time insights into reaction mechanisms, the formation of intermediate species, and the behavior of catalysts under diverse operational conditions. By offering a direct view into the complexities of photocatalytic processes, these techniques are indispensable for guiding the development and optimization of photocatalysts aimed at achieving enhanced efficiency and stability.

The current landscape of in situ characterization, while highly impactful, faces notable challenges, particularly in its ability to monitor catalytic processes at the atomic and molecular scales. This limitation underscores a pressing need for more advanced and versatile methodologies. Advanced techniques such as in situ scanning/transmission electron microscopy (STEM/TEM), XAS, and vibrational spectroscopy (e.g., infrared and Raman spectroscopy) are increasingly recognized as essential for achieving finer spatial and temporal resolution. These tools enable researchers not only to observe catalysts in action but also to probe the intricate details of their surface chemistry and electronic structure with unprecedented precision.

By bridging the gap between static characterization techniques and real-time, atomic-scale observations, ongoing advancements in in situ characterization technologies promise to catalyze significant innovation in photocatalysis. They facilitate a deeper understanding of fundamental processes such as charge carrier dynamics, surface adsorption-desorption kinetics, and interface reactions, which are pivotal for optimizing photocatalytic performance. Moreover, these techniques play a crucial role in addressing critical challenges facing photocatalysis, including improving selectivity, durability, and overall efficiency in energy conversion and environmental remediation applications. Looking forward, the continued evolution and integration of advanced in situ characterization techniques are poised to transform the field of photocatalysis. By empowering researchers to tailor the properties and functionalities of photocatalysts with unprecedented precision, these technologies hold immense potential for advancing sustainable energy technologies and addressing pressing global challenges related to clean water, air quality, and renewable energy production. As such, investments in further research and development of these methodologies are crucial for unlocking the full promise of photocatalysis in the pursuit of a more sustainable future.

The reaction mechanism of toluene selective oxidation remains unclear

Although researchers have now demonstrated that ROS are important reactive intermediates and thus involved in the reaction, more details of the reaction remain obscure^[35,51,225]. The selective oxidation of toluene molecules to yield benzaldehyde or benzoic acid is a complex process governed by the interaction between toluene molecules and metal oxide catalysts. Catalysts such as titanium and molybdenum oxides play a crucial role in activating molecular oxygen, which initiates the oxidation sequence^[159]. Initially, toluene adsorbs onto the catalyst surface, where it undergoes activation through interactions with the metal centers. The oxidation process typically begins with the abstraction of a hydrogen atom from the methyl group of toluene, forming benzyl radicals or benzyl cations as key intermediates. These reactive species then react with oxygen species adsorbed on the catalyst surface, resulting in the formation of benzyl peroxides. The fate of these peroxides determines the final products: benzaldehyde can be formed directly through the decomposition of the benzyl peroxides, while further oxidation steps can yield benzoic acid.

The selectivity of these transformations depends on several factors, including the catalyst composition, structure, and surface properties, as well as the reaction conditions such as temperature, pressure, and the partial pressures of toluene and oxygen. Catalysts are often tailored to enhance specific reaction pathways, aiming to maximize yields of desired products while minimizing the formation of by-products such as benzyl alcohol and carbon dioxide. Understanding and controlling these mechanisms are critical for optimizing the efficiency and selectivity of toluene oxidation processes in industrial applications, where these chemicals are vital for pharmaceuticals, fragrances, and other fine chemical manufacturing. Advanced spectroscopic and computational techniques play a significant role in elucidating these complex mechanisms and guiding catalyst design and process optimization efforts.

Development of additional tandem/coupling reactions based on oxidation

The development of tandem/coupling reactions based on toluene oxidation represents a promising area of research in synthetic organic chemistry. Toluene can undergo various oxidative transformations to yield valuable intermediates for further chemical reactions^[226,227]. These reactions include oxidative cross-coupling, oxidative cyclization, and one-pot oxidation and coupling.

For example, benzyl alcohol obtained from the oxidation of toluene can be coupled with amines to form secondary amines or imines using metal or organocatalysts^[228]. Benzaldehyde, derived from toluene oxidation, can participate in aldol reactions with ketones or other aldehydes to form β-hydroxy ketones or aldehydes^[229]. Additionally, direct oxidation of toluene to benzyl alcohol, followed by coupling with aryl halides using Ni catalysts, can form diarylmethane structures in a single pot^[230]. Despite significant progress, several challenges remain in the development of tandem/coupling reactions based on toluene oxidation. Achieving high selectivity and precise control over oxidation states and product distribution is a major hurdle. Research needs to focus on sustainable catalysis, developing environmentally friendly systems using non-toxic oxidants and renewable energy sources such as visible light. A deeper mechanistic understanding of these reactions will aid in designing more efficient catalytic systems. Additionally, applying these reactions to the synthesis of complex natural products, pharmaceuticals, and materials will demonstrate their practical utility and drive further advancements. Addressing these challenges will enable continued progress and new opportunities for efficient and selective synthesis in organic chemistry.

Promoting the industrialization of photocatalytic reactions

Promoting the industrialization of photocatalytic reactions involves several critical steps to bridge the gap between laboratory research and large-scale industrial applications. The first principle is the improvement of photocatalyst efficiency, which includes material innovation, nanostructuring, and doping and the formation of composites. Second, reaction conditions should be optimized for industrial methods, such as reactor design, light source optimization, and operational parameters. Third, scalability and process integration should be firmly organized in industrial processes. Pilot plants should be established to test the scalability of photocatalytic processes and address any unforeseen issues that may arise during scaling up. Meanwhile, photocatalytic processes should be integrated with existing industrial systems to utilize waste streams and existing infrastructure. Fourth, economic viability is an important indicator of business investment. Reducing the cost of photocatalyst production and the overall process can be achieved through material innovations, efficient manufacturing processes, and economies of scale. Moreover, comprehensive techno-economic analyses should be performed to evaluate the feasibility and profitability of the industrialized processes.