Photocatalytic Synthesis of Aromatic Nitriles Using TiO₂ Nanotubes Decorated With Bismuth Halide Perovskite Nanodots

ABSTRACT

The aromatic nitrile synthesis via ammoxidation of methylarenes typically requires harsh reaction conditions (≥ 350°C or ≥ 15 bar O₂). Here, TiO₂ nanotubes (NTs) homogeneously decorated with halide perovskite Cs₃Bi₂Br₉ (CBB) nanodots (< 2 nm) were established as efficient photocatalysts for the sustainable synthesis of aromatic nitriles. A yield rate up to 353 µmol g⁻¹ h⁻¹ was achieved under mild conditions (100°C–120°C, 1–4 bar, and blue light irradiation) with decent nitrile selectivity (85%–88%). Enhanced photo‐induced charge diffusion and structural stability of CBB nanodots afforded excellent quantum efficiencies (up to 26.7%) and sustained photocatalytic activity in long‐term ammoxidation (45 h). Additionally, 11 methylarenes were successfully converted to the corresponding nitriles over the CBB/TiO₂ NT photocatalyst. Mechanism investigations revealed a synergistic catalytic effect between perovskite nanodots and TiO₂ NTs. CBB nanodots activated benzylic C(sp³)–H bonds and promoted the oxidation of aldimine intermediates, while TiO₂ NTs with unique adsorption of alcohol intermediates accelerated their partial oxidation to aldehydes.

TiO₂ nanotubes (NTs) anchored with lead‐free bismuth halide perovskite Cs₃Bi₂Br₉ nanodots (CBB/TiO₂ NT) are developed to drive photocatalytic ammoxidation, yielding aromatic nitriles with selectivity up to 99% under visible light. The close interfacial contact between CBB nanodots and TiO₂ NTs affords enhanced charge separation/diffusion, synergetic catalysis, and structural stability.

1. Introduction

Aromatic nitriles are key intermediates for preparing functional arenes (e.g., benzoic acid, benzamide, benzylamine), which are commonly used in pharmaceuticals, agrochemicals, dyes, and polymer materials [1, 2, 3, 4]. In industry, benzonitrile, as the simplest nitrile arene, is mainly produced via the toluene ammoxidation process over vanadium‐based catalysts [5, 6], requiring high reaction temperatures (≥ 350°C). Impressively, the Xiao group and Han group have developed Mn‐based catalysts to drive toluene ammoxidation at reduced temperatures (160°C–190°C) [7, 8]. However, these systems demanded elevated oxygen pressures (15 bar), limiting their practical applications due to safety and scalability problems. Therefore, ammoxidation of methylarenes under mild conditions (low temperature/pressure) remains challenging.

On the other hand, photocatalytic organic transformation offers a sustainable, green, and mild alternative to conventional thermal catalytic processes [9, 10]. Cs₃Bi₂Br₉ (CBB), a lead‐free perovskite derivative belonging to the A₃B₂X₉ family, exhibits a structure derived from the ABX₃ perovskite framework [11]. It features a suitable bandgap, visible‐light absorption, versatile surface chemical environment, and low toxicity, making it attractive for photocatalytic organic synthesis [12]. For instance, CBB‐based photocatalysts enabled the oxidation of toluene to benzaldehyde at room temperature under visible light irradiation [13, 14, 15, 16]. Instead, thermal catalysis for toluene oxidation required 160°C–300°C to activate the inert C(sp³)–H with a bond energy of ∼90 kcal mol⁻¹ [17, 18, 19, 20]. Meanwhile, bismuth‐based halide perovskite exhibited reversible interactions with NH₃ [21], which usually acted as the feedstock for nitrogen‐involving transformations. Accordingly, we reason that CBB may enable benzylic C(sp³)–H activation in the presence of NH₃ to produce nitriles. Notably, CBB nanoparticles (NPs) or microcrystals (from 20 nm to even 40 µm) may experience degradation of the perovskite structure in long‐term reactions, which would lead to declined photocatalytic properties [13, 22, 23]. Fortunately, the electronic or bonding interaction between halide perovskite nanodots (1–10 nm) and suitable supports (e.g., TiO₂ or C₃N₄) could enhance their photocatalytic stability [24, 25, 26]. Moreover, compared with their bulk counterpart, CBB nanodots can present abundant catalytically active centers and improved charge separation, benefiting photocatalytic activity [14, 26]. Thus, the combination of halide perovskite nanodots with nanostructured supports could afford a feasible strategy for constructing efficient and stable photocatalytic systems for organic transformations.

In this work, we report a CBB nanodots‐decorated TiO₂ nanotube (CBB/TiO₂ NT) photocatalyst via an incipient wetness impregnation method. Ultra‐small CBB nanodots (< 2 nm) were uniformly anchored onto TiO₂ NTs, exhibiting promoted charge transfer and visible‐light response. A linear and selective production of benzonitrile (353 µmol g⁻¹ h⁻¹) through toluene ammoxidation was successfully achieved over the CBB/TiO₂ NT photocatalyst in 2.5 bar NH₃/O₂ at 120°C under long‐term blue light irradiation (45 h). Satisfactory quantum efficiencies (up to 26.7%) and a broad substrate scope (11 methylarenes) demonstrated advances of the proposed photocatalytic protocol. Furthermore, photoelectronic analysis, kinetic isotope effect (KIE) experiments, and in situ spectroscopic studies revealed that the enhanced charge transfer and synergetic catalysis between CBB nanodots and TiO₂ NTs led to the selective ammoxidation of benzylic C(sp³)–H bonds to nitriles.

2. Results and Discussion

2.1. Catalyst Characterization

The TiO₂ NTs were synthesized via a hydrothermal route, exhibiting an outer diameter of ca. 11 nm, an inner diameter of ca. 5 nm, and an average length of 102 nm (Figures S1 and S2). By use of an incipient wetness impregnation method, CBB nanodots with different loadings (2.5–20 wt%) were deposited onto the as‐prepared TiO₂ NTs, using a dimethyl sulfoxide (DMSO) solution of pre‐synthesized bulk CBB (Figure S3a). The pure phase of the bulk CBB sample was well confirmed by x‐ray diffraction (XRD) and x‐ray photoelectron spectroscopy (XPS) analysis (Figure S3b,e). As shown in Figure 1a, high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) characterization demonstrated that the nanotubular architecture of TiO₂ NTs was well preserved after CBB loading. Ultra‐small CBB nanodots were uniformly anchored onto the surface of TiO₂ NTs with gradually increasing coverage at higher loadings. For the loadings of 2.5–10 wt%, the dispersion of CBB nanodots with sizes of 1.3–1.5 nm was clearly observed (Figure 1b). As expected, too high CBB loadings (15–20 wt%) led to the slightly enlarged nanodots (1.6–1.7 nm) and the growth of massive aggregates (up to ∼40 nm). Meanwhile, elemental mapping of the representative supported 7.5 wt% CBB/TiO₂ NT presented a homogeneous distribution of Cs, Bi, and Br elements along the nanotubular framework of TiO₂ (Figure 1c), suggesting the uniform dispersion of CBB nanodots onto TiO₂ NTs without segregation of CsBr or BiBr₃ phase.

FIGURE 1.

STEM analysis of CBB/TiO₂ NT samples. (a) HAADF‐STEM images of TiO₂ NTs and CBB/TiO₂ NT with various CBB loadings (2.5‐20 wt%). (b) Particle‐size distributions of CBB at various loadings. (c) HAADF‐STEM image and corresponding elemental mapping of Ti, O, Cs, Bi, and Br for the representative 7.5 wt% CBB/TiO₂ NT sample.

Consistent with the increase in CBB loading, the CBB/TiO₂ NT photocatalysts exhibit a gradual color evolution from white to pale yellow (Figure S4). Besides, the XPS data confirmed the single‐component chemical state of Bi in the CBB/TiO₂ NT photocatalyst, implying the absence of co‐precipitated BiBr₃ or other bismuth species (Figure S5). Inductively coupled plasma mass spectrometry (ICP‐MS) analysis confirmed that the measured CBB contents aligned well with the calculated loadings (Table S1, S2). Due to the nanotubular architecture, bare TiO₂ NTs exhibit a high specific surface area of 271 m² g⁻¹ (Figure S6 and Table S3), far higher than that of TiO₂ NPs (12 m² g⁻¹). Low‐level CBB loadings (2.5 and 5 wt%) presented an ignorable influence on the surface area, whose gradual decrease (from 258 to 200 m² g⁻¹) was observed for higher CBB loadings (from 7.5 to 20 wt%). The high surface area of CBB/TiO₂ NT photocatalysts would expose rich active sites, thus facilitating photocatalytic reactions at the interface.

Based on the XRD analysis (Figure 2a), the CBB/TiO₂ NT samples with CBB loadings of 2.5–10 wt% only exhibited the diffraction peaks of TiO₂ support. No appearance of the CBB crystal phase in XRD indicated homogeneous dispersion of the ultra‐small CBB crystals on TiO₂ NTs. In contrast, weak and broadened characteristic diffraction peaks of CBB gradually emerged at high‐level CBB loadings (15–20 wt%), corresponding to the massive CBB NPs observed in TEM images. The decoration of CBB nanodots on TiO₂ NTs (bandgap E _g of 3.2 eV, Figure S7) introduced a visible light response extended up to 460 nm, where higher CBB loadings led to more pronounced light absorption (Figure 2b). More specifically, for the lowest loading (2.5 wt%), the light absorption from the CBB nanodots was extremely weak, presenting a slight change in the UV‐vis diffuse reflectance spectroscopy (DRS) profile compared with bare TiO₂ NTs. Moreover, the visible light absorption from CBB nanodots at moderate loadings (5–10 wt%) was gradually enhanced with loadings (Figure 2b). The negligible shift of absorption tails also aligned well with the similar size distribution (1.3–1.5 nm) at loadings of 5–10 wt% (Figures 1 and S7a). In addition, the high loadings (15 and 20 wt%) with larger CBB NPs (2–40 nm, Figure 1) demonstrated strong visible light absorption. Meanwhile, the bulk CBB sample (40–200 nm aggregated NPs, Figure S3c,d) afforded almost saturated light response covering 300 to 450 nm. As shown in Figure S7a, the E _g widening from the bulk CBB (2.6 eV) to CBB nanodots (2.8 eV; loadings of 5–10 wt%) was observed, reflecting size‐dependent optical changes. On the other hand, photoluminescence spectra (Figure 2c) showed a progressive red shift of the emission peak (455–465 nm) from CBB with the gradual loading increase, agreeing with the decrease of E _g due to the larger CBB particle size. Besides, almost total disappearance of this emission peak was observed in the cases of CBB loadings with 2.5–7.5 wt% (Figures 2c and S8a). On the other hand, the emission peak located at 380–400 nm from TiO₂ NTs was gradually quenched by introducing more CBB nanodots (Figures 2d and S8b,c). The bleached photoluminescence in the CBB/TiO₂ NT samples demonstrated suppressed radiative recombination of electron‐hole pairs, which can be ascribed to promoted charge transfer between perovskite and TiO₂. In contrast, the physical mixture retained much more intense photoluminescence from CBB (Figure S8a,b), indicating the poor charge separation due to limited interfacial contact.

FIGURE 2.

Crystal structure and optoelectronic properties of CBB/TiO₂ NT samples. (a) XRD patterns, (b) UV‐vis DRS spectra, and (c, d) photoluminescence profiles of CBB/TiO₂ NT samples with different CBB loadings, including the single component. (e) Transient photocurrent responses of 7.5 wt% CBB/TiO₂ NT with bare TiO₂ NTs and physical mixture of (CBB + TiO₂ NTs) for comparison. (f) Schematic band diagram of bulk CBB, CBB nanodots (7.5 wt% loading), and bare TiO₂ NTs.

Photoelectrochemical measurements (Figure 2e) demonstrated a dramatically enhanced photocurrent from the CBB/TiO₂ NT sample, which was 6‐fold higher than that of bare TiO₂ NTs. The promoted charge separation and diffusion may originate from charge transfer channels with short lengths between sub‐2 nm CBB nanodots and TiO₂ NT with abundant interfacial contact. Instead, the physical mixture of CBB with TiO₂ NTs presented the same photocurrent as bare TiO₂ NTs due to the absence of strong interfacial contact. Based on the E _g and valence‐band XPS (Figure S9), the valence band maxima (VBM) values of bulk CBB, CBB nanodots (7.5 wt%), and TiO₂ NTs were estimated to be 2.22, 2.35, and 3.03 eV (vs. NHE), respectively. Simultaneously, their corresponding conduction band minima (CBM) positions were located at −0.40, −0.44, and −0.20 eV (vs. NHE) (Figure 2f). Under visible light excitation, such staggered band structures favored hole accumulation at CBB and electron transfer to TiO₂, thus promoting spatial charge separation via a type‐II heterojunction at the interface [22, 27].

2.2. Photocatalytic Methylarene Ammoxidation

For the photocatalytic synthesis of aromatic nitriles, toluene ammoxidation was employed as the model by use of our home‐made photoreactors (Figure S10) to screen out the active photocatalyst. Specifically, the photocatalytic toluene ammoxidation was conducted under 450 nm LED irradiation at 120°C for 9 h in the presence of a 2.5 bar gas mixture of (40% NH₃ + 60% O₂) with products identified by GC‐MS (Figure S11). As shown in Figure 3a, pristine CBB exhibited decent activity for toluene ammoxidation with the production of 160 µmol benzonitrile. However, substantial byproducts of benzyl alcohol (116 µmol) and benzaldehyde (32.5 µmol) were observed, leading to the low selectivity of benzonitrile (50.4%). Control experiments demonstrated that both CsBr and BiBr₃ were catalytically inactive, while the extra introduction of BiBr₃ into CBB failed to improve the catalytic performance (Figure S12). It further identified the perovskite phase as the main active site. In contrast, bare TiO₂ support in the form of either NTs or NPs hardly activated toluene, which was mainly attributed to the wide E _g (3.2 eV) with negligible visible light response.

FIGURE 3.

Photocatalytic performance of CBB/TiO₂ NT for toluene ammoxidation. (a) Evaluation of various photocatalysts. (b) The action spectrum. (c) The influence of reaction pressure. (d) The long‐term test. Standard reaction conditions: 5 mL toluene, 75 mg photocatalyst, 2.5 bar (40% NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 9 h.

Delightedly, the decoration of CBB nanodots onto TiO₂ NTs significantly improved the selectivity of benzonitrile up to 88%. A linear growth of benzonitrile amount (from 125 to 380 µmol) appeared when CBB loading was increased from 2.5 to 10 wt% (Figure S13), indicating the uniform dispersion of CBB nanodots with a similar size. Although more benzonitrile (520–657 µmol) was yielded at higher loadings (15 and 20 wt%), its linear production gradually started to be deviated due to the presence of large CBB NPs with less accessible active sites. Among them, the 7.5 wt% CBB/TiO₂ NT photocatalyst displayed the optimal performance, in which 260 µmol benzonitrile was produced with a satisfactory selectivity of 88%, with 29 µmol benzaldehyde dominating the remaining 36 µmol byproducts (Figure S11). The presence of tiny benzenemethanimine (PhCH = NH) in Figure S11 implied the reaction pathway involving aldimine intermediates from aldehydes and NH₃ [28]. In addition, neither overoxidation products from toluene (e.g., CO₂ or benzoates) nor nitrogen‐containing oxidation products derived from NH₃ (e.g., N₂, NO, N₂O, and NO₂, Figure S14) were detected in the present reaction conditions.

To explore the role of CBB and TiO₂ in photocatalytic toluene ammoxidation, we designed a series of photocatalysts for comparison (Figure 3a). A mesoporous silica SBA‐15 decorated with 7.5 wt% CBB nanodots activated 265 µmol toluene with benzyl alcohol as the main product (selectivity of 58%) and benzonitrile accounting for 40%. It suggested the toluene activation by CBB and the conversion of alcohol intermediates by TiO₂. Accordingly, either bulk Cs₃Bi₂I₉ (CBI) or CBI/TiO₂ NT (Figure S15) failed to activate benzylic C(sp³)–H bonds of toluene (< 25 µmol) due to the weak oxidative capability of CBI [29]. Furthermore, 7.5 wt% CBB/TiO₂ NP was prepared as the reference photocatalyst, where the size of CBB nanodots was larger (∼3 nm), and the surface area of TiO₂ NPs was low (12 m² g⁻¹) in contrast to CBB/TiO₂ NT (Figures S6 and S16). Unexpectedly, much less toluene (157 µmol) was activated with a poor benzonitrile selectivity of 47%, where benzyl alcohol appeared as the main byproduct (selectivity of 42%). The inferior activity of CBB/TiO₂ NP was probably related to the limited accessible active sites of larger 3 nm CBB nanodots, which only approached 43% of the specific surface area of ultra‐small 1.3 nm nanodots in CBB/TiO₂ NT. Moreover, TiO₂ NPs may experience excessive coverage by CBB nanodots, blocking the active sites on TiO₂ for the conversion of alcohol intermediates. The comparison experiments demonstrated the necessity of CBB nanodots and TiO₂ NTs with well‐defined nanoarchitectures for selective light‐driven synthesis of aromatic nitriles.

The action spectrum was obtained by using monochromatic LED lamps (2 mW cm⁻²; light wavelengths varying from 365 to 700 nm, Figure S17). An apparent quantum efficiency (AQE) of 26.7% was achieved under UV (365 nm) light irradiation (Figure 3b). In alignment with the light absorption profile of CBB/TiO₂ NT, lower AQEs of 7.6% and 0.9% corresponded to purple (410 nm) and blue (450 nm) light excitation, respectively. It further implied the essential light absorption from CBB nanodots for toluene activation under blue light irradiation during ammoxidation. Meanwhile, the activity disappeared at longer wavelengths (520 and 700 nm) due to a lack of light absorption. Besides, no products in the dark and only trace amounts of oxygenated products (13 µmol) in the absence of photocatalyst also supported the photocatalytic nature of CBB/TiO₂ NT for toluene ammoxidation (Figures 3a and S18). Additionally, the hot filtration experiment excluded the catalytic contribution from the possible soluble species (Figure S19). Regarding the temperature, we observed the dramatic acceleration of toluene ammoxidation from 16 µmol consumed toluene at 80°C to 492 µmol at 130°C, accompanied by a lower selectivity of benzonitrile (69%) at 130°C (Figure S20). The requirement of external heating (100°C–120°C) could correspond to the reversible NH₃ adsorption/desorption of bismuth‐based halide perovskites, exposing the active perovskite phase at higher temperatures.

On the other hand, the amount of consumed toluene decreased markedly from 296 µmol at 2.5 bar to 163 µmol at 4 bar NH₃/O₂, while the benzonitrile selectivity stabilized at 88% (Figure 3c). It may be related to the excessive NH₃ adsorption on the CBB phase at a higher pressure, thus occupying some active sites of CBB nanodots [30, 31]. Consistently, higher NH₃ partial pressure at a constant total pressure (2.5 bar) presented decreased photocatalytic activity from CBB/TiO₂ NT (Figure S21). Besides, the drop of total pressure from 2.5 to 1.0 bar led to a poor selectivity of benzonitrile (48%) with numerous benzaldehydes produced (46%), reflecting that the conversion of aldehyde intermediates to nitriles required abundant NH₃ [32, 33]. The importance of gas feedstock availability was further evidenced in long‐term reactions. As shown in Figures 3d and S22a, the selectivity of benzonitrile gradually declined from 88% to 63% during the period of 9–18 h, accompanied by an increase of benzaldehyde selectivity (up to 27%). Based on the amounts of organic products and involved reactions, the conversions of O₂ and NH₃ were accurately calculated (Figure S22b,c). Specifically, the conversion of O₂/NH₃ reached 55%–60% after 12 h irradiation, indicating the total pressure was reduced to 1.0 bar (∼0.6 bar O₂ and 0.4 bar NH₃). Consequently, insufficient gaseous reactants led to the accumulation of benzaldehyde and a decline in nitrile selectivity under prolonged irradiation, coinciding with the photocatalytic performance at lower total pressures. Fortunately, by recharging NH₃/O₂ every 9 h, the photocatalytic properties were well recovered with a linear yield of benzonitrile (∼26.5 µmol h⁻¹, 353 µmol g⁻¹ h⁻¹) and a high selectivity of ≥ 85% over 45 h (Figures 3d and S22d). Post‐reaction XRD, UV–vis DRS, and HAADF‐STEM analyses confirmed that the used photocatalyst retained its crystal structure, optical properties, and nanostructure, demonstrating excellent stability (Figure S23).

The substrate scope of the CBB/TiO₂ NT photocatalyst successfully covered diverse methylarenes (Table 1). The toluene derivatives with electron‐donating groups (e.g., ‐CH₃, ‐OCH₃) or electron‐withdrawing groups (e.g., ‐F/Cl/Br, ‐COOC₂H₅, and ‐CF₃) were converted to the corresponding nitriles with selectivity up to 99%. As expected, the electron‐withdrawing substituents would destabilize benzyl radicals during ammoxidation, thus resulting in lower photocatalytic activity (< 170 µmol in 9 h [34]). Additionally, the position of ‐Cl and ‐Br groups presented a steric effect, since methylarenes bearing para substituents exhibited much higher reaction rates than ortho counterparts. The methylarene ammoxidation over CBB/TiO₂ NT under visible light irradiation afforded an effective, sustainable, and mild protocol for the selective synthesis of aromatic nitriles.

TABLE 1.

Ammoxidation of methylarenes over the CBB/TiO₂ NT photocatalyst under blue light irradiation. ^a

Entry	Substrate	Product	Selectivity (%)	Production (µmol)
1			88	260
2 b			78	182
3 b			80	65
4			90	153
5			99	81
6			93	122
7			99	61
8			94	111
9			98	84
10			89	136
11			96	50

^a Standard reaction conditions: 5 mL methylarenes, 75 mg 7.5 wt% CBB/TiO₂ NT, 2.5 bar (40% NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 9 h.

^b 5 mL methylarenes, 75 mg 2.5 wt% CBB/TiO₂ NT, 4 bar (40% NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 3 h.

2.3. Mechanistic Investigations

An apparent KIE was observed when deuterated toluene was employed as the substrate, affording a k_H/k_D value of 4.3. It confirmed the activation of benzylic C(sp³)–H bond as the rate‐determining step in photocatalytic toluene ammoxidation (Figures 4a and S24) [14]. In contrast, substitution of NH₃ with ¹⁵NH₃ presented a negligible impact on the reaction rate, indicating that NH₃‐involved steps proceeded rapidly. Moreover, the addition of triethanolamine (TEOA), as a sacrificial electron donor, drastically suppressed the activity (Figure 4b), suggesting the indispensable role of photogenerated holes in toluene activation [35]. Likewise, the reaction was quenched in the presence of butylated hydroxytoluene (BHT), indicating the generation of carbon‐centered radicals [36]. The benzyl radicals were experimentally captured by 2,2,6,6‐tetramethylpiperidine‐1‐oxyl (TEMPO), which afforded the corresponding radical adduct (Figure S25a). Besides, the benzyl radical intermediate was also supported by the formation of bibenzyl when the reaction was conducted in the absence of O₂, where the ammoxidation was prohibited as well (Figure S25b). On the contrary, the addition of CCl₄ and tert‐butanol (TBA) exhibited a slight effect on the activity, implying photo‐generated electrons and possible hydroxyl radicals would not be involved in the rate‐determining step [37, 38]. In combination with the photocatalytic ammoxidation/oxidation performance (Figures 3a and S26), photo‐generated holes at the VBM level of CBB, instead of TiO₂ NTs, were primarily responsible for activating the benzylic C(sp³)–H bond via a concerted proton‐electron transfer (CPET) process [14], yielding benzyl radicals. The initial toluene activation under visible light was consistent with previous reports as well [15, 39, 40]. These carbon‐centered radicals could be rapidly trapped by O₂ to form peroxy species (ROO·), which subsequently evolved into benzyl alcohol and benzaldehyde intermediates [13, 14, 41]. Therefore, almost no oxygenated products were obtained under anaerobic conditions (Figure 4b).

FIGURE 4.

Mechanism experiments. (a) The isotope labeling experiments. (b) Radical scavenging experiments with the addition of various sacrificial reagents, including CCl₄, TBA, BHT, and TEOA. (c) In situ DRIFTS analysis of photocatalytic toluene ammoxidation. Reaction conditions: (a) 5 mL toluene or deuterated toluene, 75 mg 7.5 wt% CBB/TiO₂ NT, 2.5 bar (40% ¹⁴NH₃ or ¹⁵NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 9 h. (b) 5 mL toluene, 75 mg 7.5 wt% CBB/TiO₂ NT, 10 mg scavengers, 2.5 bar (40% NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 9 h.

The photocatalytic roles of CBB nanodots and TiO₂ NTs in toluene ammoxidation were further investigated by in situ DRIFTS experiments (setup shown in Figure S27). As shown in Figure 4c, pristine TiO₂ NTs exhibit no discernible spectral changes even after 25 min irradiation, indicating that TiO₂ could not activate toluene under visible light. Instead, the CBB/TiO₂ NT photocatalyst offered rapid emergence of a distinct band at 2230 cm⁻¹ after 10 min irradiation, which was attributed to the C≡N stretching vibration of benzonitrile (Figure S28) [42]. Notably, no obvious vibrational features associated with benzyl alcohol or benzaldehyde were detected in the case of CBB/TiO₂ NT, matching well with its high selectivity towards nitriles. Furthermore, for the physical mixture of (CBB + TiO₂ NTs), only characteristic bands from benzaldehyde and benzyl alcohol (∼1699 and 1206 cm⁻¹) were detected without benzonitrile signals [43, 44]. The lack of a strong interfacial contact between CBB and TiO₂ in the physical mixture cut off the vital intermediate transfer and charge diffusion (Figure 2e) for the domino‐type multiple steps from toluene to benzonitrile via alcohol/aldehydes.

To further explore the reaction pathway involving alcohol/aldehydes, we used the ammoxidation of benzyl alcohol as the model reaction. Notably, the conversion of benzyl alcohol from CBB/TiO₂ NT with different CBB loadings was almost equal to that from bare TiO₂ NTs (Figure 5a). Thus, the oxidation of alcohol was dominantly driven by the TiO₂ phase, which was also evidenced in the photocatalytic oxidation of benzyl alcohol with pure O₂ (Figure S29a). However, the poor selectivity of benzonitrile over bare TiO₂ was observed with benzoic acid and ammonium benzoate as main byproducts due to the overoxidation of benzaldehydes in the presence of an NH₃‐O₂ mixture. Delightedly, the benzonitrile selectivity was dramatically improved after loading CBB nanodots, which may facilitate the conversion of aldehydes to nitriles. Additionally, no conversion in the dark guaranteed the photocatalytic nature for the ammoxidation of benzyl alcohol. Here, the activity from bare TiO₂ under blue light was related to the formation of surface complexation between TiO₂ and alcohol [45, 46], affording visible absorption tails beyond 500 nm (Figure S29b,c). In contrast to TiO₂ NTs, TiO₂ NPs exhibited much lower activity in benzyl alcohol ammoxidation/oxidation, which was diminished after CBB loading (Figures 5a and S29a). It further suggested the occupation of active sites on TiO₂ NPs by CBB nanodots due to the excessive surface coverage, consequently suppressing the conversion of alcohols.

FIGURE 5.

Investigation of the reaction pathway. (a) Photocatalytic ammoxidation of benzyl alcohol over various photocatalysts. (b) DRIFTS analysis of benzyl alcohol adsorbed on CBB, TiO₂ NPs, and TiO₂ NTs. In situ DRIFTS analysis of the ammoxidation of benzyl alcohol over (c) bare TiO₂ NTs and (d) CBB/TiO₂ NT. (e) The proposed photocatalytic cycle for toluene ammoxidation over CBB/TiO₂ NT. Reaction conditions: 5 mL chlorobenzene containing 200 µmol benzyl alcohol, 75 mg photocatalysts, 2.5 bar (40% NH₃ + 60% O₂), LED light irradiation (450 nm, 60 mW cm⁻²), 120°C, 9 h.

On the other hand, TiO₂ NTs demonstrated unique adsorption of benzyl alcohol based on the IR data in comparison with CBB and TiO₂ NPs (Figure 5b). A new band at 1107 cm⁻¹ was observed in the case of TiO₂ NTs, which could be assigned to the C‐O stretching vibration of alcoholate species formed between benzyl alcohol and TiO₂ [47, 48]. In contrast, CBB interacts mainly with the phenyl ring via relatively weak π‐type interactions. The unique adsorption mode of benzyl alcohol on TiO₂ NTs may explain its superior photocatalytic activity in alcohol ammoxidation/oxidation. Correspondingly, the in situ DRIFTS data showed the fast generation of benzaldehyde within 5 min irradiation (at 1699 cm⁻¹) in the case of bare TiO₂ NTs with the adsorption of benzyl alcohol (Figures 5c and S30). Another 10 min irradiation led to the deep oxidation product of benzoic acid/benzoate (at 1640 cm⁻¹), while no benzonitrile was detected even after 30 min [49, 50]. In comparison, the CBB/TiO₂ NT photocatalyst exhibited the rapid appearance of benzonitrile (at 2230 cm⁻¹) within 5 min, accompanied by benzaldehyde signals at 1699 cm⁻¹ (Figures 5d and S30). Notably, a much smaller amount of aldehyde intermediates was accumulated in the case of CBB/TiO₂ NT in comparison with bare TiO₂ NTs (Figure S30). In agreement with the photocatalytic performance, in situ IR data indicated that TiO₂ NTs promoted partial oxidation of alcohols to aldehydes, which were subsequently selectively converted to nitriles with the assistance of CBB nanodots. The conversion of benzaldehyde to benzonitrile underwent aldimine (PhCH = NH) intermediates (Figures S11 and S31), formed by fast condensation between aldehydes and NH₃ [28, 51]. Different from other reported studies involving toluene activation via oxidation or amination [14, 39], our system demonstrated a photocatalytic ammoxidation process in the presence of an NH₃‐O₂ mixture. After the formation of alcohol/aldehyde intermediates from benzyl radicals and O₂, a cascade pathway appeared to realize the selective ammoxidation of toluene. Specifically, benzyl alcohols were selectively oxidized to benzaldehyde intermediates on TiO₂ NTs, followed by non‐catalytic reversible condensation of aldehydes with NH₃ to form aldimines. Meanwhile, CBB nanodots accelerated the photo‐oxidation of aldimine intermediates to generate nitrile products, thus suppressing the accumulation of active aldehydes for other possible side reactions (e.g., overoxidation).

Based on the above mechanistic investigations, a plausible photocatalytic cycle for toluene ammoxidation over CBB/TiO₂ NT was proposed (Figure 5e). First, the C(sp³)–H bonds of toluene were activated by photogenerated holes at CBB nanodots under visible light irradiation to form benzyl radicals, which was identified as the rate‐determining step. Upon reaction with O₂, the benzyl radicals are subsequently converted into benzyl alcohol and benzaldehyde intermediates via peroxide species. Meanwhile, alcohols were selectively oxidized to aldehydes by TiO₂ NTs. Later, the reversible condensation reaction between benzaldehyde and NH₃ afforded aldimine intermediates (PhCH = NH). Finally, CBB nanodots promoted the oxidation of aldimines to the desired nitriles, shifting the equilibrium of the condensation step and avoiding the accumulation of aldehyde intermediates as well. Overall, the reaction proceeded via a radical‐mediated pathway, in which CBB nanodots dominated the initial C(sp³)–H activation and the final aldimine oxidation steps, while TiO₂ NTs mainly facilitated the selective oxidation of alcohol to aldehyde intermediates.

3. Conclusion

In summary, we have developed TiO₂ NTs decorated with CBB nanodots to drive selective ammoxidation of diverse methylarenes under visible light irradiation. The close interfacial contact between ultra‐small CBB nanodots (< 2 nm) and TiO₂ NTs support provided efficient charge separation/diffusion and structural stability. Under blue light irradiation at 100°C–120°C in the presence of 1–4 bar NH₃/O₂, the CBB/TiO₂ NT photocatalyst afforded a linear yield of benzonitrile (353 µmol g⁻¹ h⁻¹) with satisfactory selectivity (85%–88%) in long‐term ammoxidation (45 h). Mechanistic studies revealed the synergetic catalytic roles of CBB and TiO₂ NTs, where the former activated benzylic C(sp³)–H bonds of toluene and the latter accelerated alcohol‐to‐aldehyde conversion. The homogeneous dispersion of CBB nanodots onto TiO₂ NTs enables a sustainable and efficient photocatalytic route for the synthesis of aromatic nitriles via methylarene ammoxidation.

Author Contributions

Y.‐T.D. supervised the project and conceived the research. T.X. and Y.‐T.D. designed the experiments. T.X., J.L., and Y.‐T.D. conducted the experiments/characterizations, analyzed the data, and produced figures/tables. T.X. and Y.‐T.D. drafted the manuscript, and Y.‐T.D. finalized it. All authors contributed to the revision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: The authors have cited additional references within the Supporting Information [52, 53].

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.