Unravelling the redox behaviour of a carbon nitride polymer-based ternary heterostructure for CO₂ photoreduction promoted by amine coupling

Abstract

Imine compounds are essential commodity chemicals used in synthetic organic chemistry and pharmaceuticals. To date, the oxidative coupling of amines produces imines in a pure oxygen atmosphere. Here, we have presented a sustainable and more economical route for producing imines by replacing O₂ with inexpensive and likely inert CO₂. A noble metal-free ZnV₂O₆@WO₃–g–C₃N₄ heterojunction has dual photocatalysis ability, converting CO₂ into CO and CH₄ in 1674 and 683 μmol g⁻¹ yields, respectively, within six hours with the simultaneous oxidative coupling of amines into imines with a maximum of 55% yield for N-benzylidene benzylamine. Also, CO and CH₄ yields were 5.3 and 2.9 times higher than those of pristine g-C₃N₄. The reusability demonstrated the reliability of the present heterojunction, which showed four consecutive uses of the recovered photocatalyst with relatively similar yields. Besides these, extensive corroboration of mechanistic understanding was made through thorough characterization of the intermediates formed during this simultaneous photocatalytic transformation. Thus, replacing molecular oxygen with CO₂ to oxidize amines has vast opportunities, as CO₂ can be used as an oxidant for sustainable organic transformations into fine chemicals and fuels.

---

1 Introduction

The cost-effective production of fine chemicals and fuels is essential from an energy perspective.¹ The production of energy chemicals utilizing waste or unwanted chemicals as precursors is highly desirable. Meanwhile, imines are functional raw materials in natural product synthesis, medicinal chemistry, synthetic organic chemistry, catalysis, analytical chemistry, photochromism, etc.² Imines are produced from the oxidative coupling of amines under different catalytic conditions using oxidizing agents, such as oxygen and air. Traditionally, imines were produced from oxidative coupling of amines using carbonyl compounds and 2-iodoxybenzoic acid (or N-t-butylphenylsulfinimidoyl chloride).^3,4 These materials are expensive and produce harmful by-products under harsh conditions, increasing the complexity and energy consumption.

Later, the oxidative coupling of amines was improved by replacing expensive and hazardous oxidizing agents with more convenient and environmentally benign dioxygen and air as oxidizing agents. Furthermore, earth-abundant solar light replaced the harsh conditions, which decreased the cost of the oxidative coupling of amines to imines.⁵ Many pioneering studies on oxidative coupling of amines have been performed using various photocatalytic materials, including metal and non-metal semiconductors. Yang et al. reported a solvent-free ordered mesoporous Au/TiO₂ catalyzed oxidative coupling of amines under visible light.⁶ Khampuanbut et al. discussed the energy band alignment and mechanistic insight into visible light-assisted good conversion of amines into imines using a WO₃/BiOBr heterojunction photocatalyst.⁷ Xu et al. evidenced the charge separation in a porphyrin-based metal–organic framework for oxidative coupling of amines into imines under visible light irradiation.⁸ Battula et al. fabricated a truxene-based conjugated microporous polymer for homocoupling amines using O₂ under natural sunlight.⁹ Although the oxidative coupling of amines into imines is a meaningful organic transformation, improvements are still needed to make the process more convenient, economical, and energy-efficient.

Recently, the emission of greenhouse gases, mainly carbon dioxide (CO₂), emerged as a severe environmental issue. CO₂ is the most abundant component of greenhouse gas emitted by different chemical processes in various industries. The excessive emission of CO₂ from wetlands, crude oil exploration, rapid population growth, rapid industrialization, fossil fuel consumption, etc., has impacted the environment and raised the Earth's temperature.¹⁰ However, the non-toxicity, cheapness, and easy availability of CO₂ make it generous and a suitable C₁ feedstock for producing fine chemicals and fuels. The photoreduction of CO₂ produces different C₁ products, such as CO, CH₄, CH₃OH, HCOOH, and HCHO, using various hybrid and nanocomposite photocatalysts.¹¹ Water is used as the proton source, while amines are used as the sacrificial donor or hole scavenger for the photoreduction of CO₂. Several novel strategies have been developed for CO₂ as a C₁ feedstock with different organic transformations, such as cycloaddition with epoxides,¹² carboxylation of various olefinic and halogenated compounds,¹³ oxidants for alcohols,¹⁴ C–N couplings,¹⁵etc., under visible light irradiation.

Replacing O₂ with the major greenhouse gas CO₂ has attracted great interest because of the cost-effectiveness and ease of availability of CO₂ over expensive O₂. Very few reports are available in the literature on the oxidative coupling of amines with simultaneous reduction of CO₂ under UV-Vis light. Luo et al. achieved 82.1% conversion of benzylamine with CO₂ reduction to CH₃OH at 0.5 MPa using a solid solution of CuMo_xW_(1−x)O₄ under visible light.¹⁶ Markushyna et al. investigated benzylamine oxidation with CO₂ using a C₃N₄ semiconductor, but the maximum yield remained at only 34% for N-benzylidene benzylamine.¹⁷ Recently, Wang et al. fabricated CdS quantum dot decorated ordered mesoporous carbonaceous frameworks to efficiently reduce CO₂ with the oxidative coupling of amines into imines under visible light irradiation.¹⁸ Although the above-mentioned photocatalytic systems provided moderate conversion and yields of the corresponding products for the simultaneous reduction of CO₂ with oxidation of amines, significant development is desirable to achieve high efficiency for simultaneous photoreduction of CO₂ with the oxidative conversion of amines into imines.

Among the various photocatalytic semiconductors, g-C₃N₄ has tremendous opportunities for photocatalytic applications, including CO₂ photoreduction and oxidation of organic compounds.¹⁹ The band positions and significant basic sites at the g-C₃N₄ surface provide excellent photocatalytic activity for CO₂ photoreduction for solar fuel production. Recently, g-C₃N₄-based metal oxide composites have been developed and exhibited excellent photocatalytic activity for CO, CH₄, CH₃OH, and HCOOH production from CO₂ photoreduction.²⁰ The heterojunction of g-C₃N₄ with metal oxides improved charge separation at the interface, resulting in the heterojunction photocatalyst's excellent oxidation and reduction ability.^21–23 The heterojunction of g-C₃N₄ with WO₃ was reported for a variety of CO₂ photoreduction due to the highly positive oxidation potential of the valence band (VB) of WO₃, which efficiently oxidized water or amine compounds provided necessary protons and electrons to reduce CO₂.^24,25 In addition, boosting the efficiency of CO₂ photoreduction, the composite of g-C₃N₄ and metal nanoparticles with efficient CO₂ adsorption and a suitable band position for CO₂ reduction is necessary.

A metal vanadate with strong visible light harvesting, a narrow band gap, and excellent photocurrent density could be a suitable semiconductor for various photocatalytic applications.²⁶ Among them, ZnV_xO_y-based semiconductors have vast opportunities for CO₂ photoreduction, which have not been explored yet.²⁷ Recently, ZnV₂O₆ proved to be a suitable component of heterojunction photocatalysts with g-C₃N₄ and graphene oxide (GO) for the photoreduction of CO₂ to solar fuels.^28,29 However, there is not much literature available for ZnV₂O₆ and g-C₃N₄-based heterojunctions. The favorable band gap and band positions of the individual components provide an impetus to develop a new heterojunction that augurs well for the simultaneous transformation, which is the need of the hour. Keeping this in mind, the construction of a heterojunction from ZnV₂O₆, g-C₃N₄, and WO₃ would be an excellent approach to improving charge transfer and subsequent efficient oxidation and reduction processes for such transformations.

Continuing our approach for the one-pot production of chemicals and fuels from CO₂, we constructed an efficient, strong oxidizing and reducing tendency and highly stable ZnV₂O₆@WO₃–g–C₃N₄ heterojunction via the sol–gel hydrothermal method. Different concentrations of ZnV₂O₆ on the WO₃–g–C₃N₄ composite were synthesized and studied for the simultaneous oxidative coupling of amines with the photoreduction of CO₂ to fine chemicals and solar fuels. We found that 20 wt% loading of ZnV₂O₆ on the WO₃–g–C₃N₄ nanocomposite, i.e., ZWC-20 nanocomposite, showed the highest photocatalytic activity under optimized conditions (Scheme 1). The heterojunction afforded good yields of the corresponding imines and C₁ products (CO and CH₄) within six hours of visible light irradiation. The detailed characterization of the heterojunction photocatalyst and the products was performed with various spectroscopic, microscopic, and chromatographic techniques.

Scheme 1 Photoreduction of CO₂ with an oxidative coupling of amines under visible light irradiation.

2 Experimental section

2.1 Catalysts' synthesis

2.1.1 g-C₃N₄ synthesis. g-C₃N₄ was synthesized following the literature reported.³⁰ Initially, a 1 : 4 ratio of dicyandiamide and thiourea was ground using a mortar and pestle for 30 min. This mixture was placed in a crucible and capped for the condensation process. The mixture in the crucible was annealed at 550 °C for 4 h under static air. After cooling, the yellow solid was ground and used further.

2.1.2 WO₃ synthesis. A 20 mL aqueous solution of 2 g sodium tungstate dihydrate was stirred for one hour in the typical synthesis. The reaction mixture temperature was maintained at 5 °C, and 5 mL of 6 N HCl solution was added dropwise for precipitation. The as-prepared precipitates were then washed with water by maintaining a pH of 6 and placed in an oven at 60 °C for 12 h. The dried pale-yellow powder was ground to obtain the desired heterojunctions.

2.1.3 WO₃–g–C₃N₄ heterojunction synthesis. The WO₃–g–C₃N₄ heterojunction was constructed by using a 1 : 5 ratio of WO₃ and g-C₃N₄. In the typical synthesis, 1 g of g-C₃N₄ was sonicated for 2 h to exfoliate the graphitic nanosheets in ethanol. Then, 200 mg of WO₃ nanosheets were added to this mixture and stirred overnight or till the evaporation of the solvent. Furthermore, the mixture was placed in an oven at 60 °C for 24 h to complete dryness and used for further modification. This heterojunction photocatalyst was coded as WCN.

2.1.4 ZnV₂O₆@WO₃–g–C₃N₄ heterojunction synthesis. The desired ZnV₂O₆@WO₃–g–C₃N₄ heterostructure was constructed with a modification of a previous literature report.²⁹ A certain amount of ammonium metavanadate was dissolved in 30 mL of N,N-dimethylformamide (DMF) under vigorous stirring. Then, with continuous stirring, zinc acetate (5, 10, 20, 30, and 40 wt% ZnV₂O₆) was dissolved in this mixture. After one hour, oxalic acid dihydrate was added to this mixture in a 3 : 1 ratio of ammonium metavanadate. At the same time, 1 g of WO₃–g–C₃N₄ dispersed in 20 mL of DMF was added with vigorous stirring for 30 min. This solution was added to the zinc acetate and ammonium metavanadate solution and stirred for one hour. Finally, this mixture was transferred to a 100 mL Teflon-lined autoclave and heated at 200 °C for 24 h. After cooling, the precipitates were separated using centrifugation, followed by washing with absolute ethanol, and then the wet precipitates were placed in an oven at 80 °C for 12 h. The dried content was annealed at 550 °C for 4 h and collected after cooling. The composites of ZnV₂O₆ on WO₃–g–C₃N₄ were coded as ZWC-20, ZWC-10, ZWC-20, ZWC-30, and ZWC-40 based on the wt% of ZnV₂O₆.

2.2 Photocatalytic experiment

The photocatalytic experiments for one-pot production of imines from amines and C₁ products from CO₂ photoreduction were performed in a 10 mL round-bottom flask. The system was equipped with a 350 W xenon lamp solar simulator. Initially, 0.1 mmol of benzylamine as a model substrate, 10 mg of the photocatalyst, and 3 mL of solvent were charged to a 10 mL round-bottom flask. The round-bottom flask was closed with a rubber septum and purged with CO₂ gas continuously for 30 min. The reaction vessel was appropriately closed with parafilm wrapping on the mouth of the round-bottom flask to avoid any leakage. Then the light source was irradiated on the reaction mixture, and the reaction progress was monitored by collecting small aliquots from the reaction mixture and injecting them into a GC-MS and GC-FID to detect the corresponding products of the reaction. GC-MS and ¹H and ¹³C NMR spectroscopy confirmed the products of benzylamine oxidation.

2.3 Photoelectrode preparation

The synthesized semiconductors and nanocomposite materials were coated on fluorine-doped tin oxide (FTO) plates by a previously reported printing method.³¹ The materials' paste was prepared by blending 25 mg of photocatalyst with 50 mg of ethyl cellulose, 0.1 mL terpineol, 0.02 mL of acetic acid, and ∼1 mL of ethanol for homogeneity. This mixture was then manually printed over a FTO plate. The as-prepared plates were dried at 150 °C under a vacuum for 2 h to eliminate the binding agents. Furthermore, the modified FTO plates were utilized for photo-electrochemical measurements.

3 Results and discussion

3.1 Catalysts characterization

The desired heterostructure was fabricated via three-step synthesis (Scheme 2). Initially, melamine was heated at 550 °C to synthesize g-C₃N₄. WO₃ was synthesized using sodium tungstate as the precursor for precipitation under acidic conditions at low temperatures. Then, the composite of ZnV₂O₆, WO₃, and g-C₃N₄ was fabricated via the sol–gel-hydrothermal method and subsequent post-annealing process. The synthesized nanomaterials were characterized using various spectroscopic and microscopic techniques.

Scheme 2 Synthetic pathways of the ternary ZnV₂O₆@WO₃–g–C₃N₄ nanocomposite photocatalyst.

Powder X-ray diffraction was employed to analyze the crystal parameters of the synthesized nanomaterials (Fig. 1). XRD of g-C₃N₄ showed a (002) plane of conjugated aromatic units with the stacked interlayer structure because of its graphitic nanosheet structure, identified by the JCPDS card no. 87-1526.³² The high crystallinity of WO₃ confirmed its intense crystal planes of a cubic crystal lattice. The diffraction peaks at 2θ values 13.9°, 22.8°, 24.4°, 26.7°, 28.1°, 33.7°, 36.4°, 50.1°, and 55.3° were attributed to the (100), (001), (110), (101), (200), (111), (201), (220), and (202) planes for the hexagonal crystal lattice of WO₃ matching with the JCPDS card no. 33-1387.³³ The nanocomposite construction of WO₃ and g-C₃N₄ in a 1 : 5 ratio, i.e., WCN, showed the diffraction plane of both g-C₃N₄ and WO₃. However, due to the low concentration, the diffraction planes of WO₃ were observed with less intensity. In addition, the diffraction signals at 2θ values 20.3°, 27.3°, 28.9°, 29.4°, 32.8°, 37.3°, 38.8°, 44.4°, 47.4°, and 51.9° were attributed to the (−201), (110), (−202), (201), (111), (−112), (−311), (003), (311), and (020) planes that correspond to the orthorhombic crystal lattice of ZnV₂O₆, confirmed by the JCPDS card no. 01-074-1262.²⁹ As the construction of the heterojunction was reflected in its photocatalytic activity, the heterojunctions of ZnV₂O₆ and WCN, i.e., ZWC derivatives, were also analyzed by PXRD. ZWC derivatives showed the additional diffraction planes of ZnV₂O₆ along with WCN (Fig. S1†). Although ZnV₂O₆ was loaded on the WCN surface, the diffraction planes of ZnV₂O₆ were not observed in ZWC-05 and ZWC-10 nanocomposites due to the low loading content. The higher loading of ZnV₂O₆, i.e., ZWC-20 sample, exhibited (−202), (201), and (111) diffraction planes, indicating the incorporation of ZnV₂O₆ with WCN.

Fig. 1 Powder XRD of g-C₃N₄, WO₃, WCN, ZnV₂O₆, and ZWC-20 nanocomposites. HR-TEM images and SAED pattern of the ZWC-20 nanocomposite.

The morphological aspects of ZnV₂O₆, WCN, and ZWC-20 nanocomposites were analyzed through field-emission scanning electron microscopy (FE-SEM) (Fig. S2†). FE-SEM of ZnV₂O₆ showed that nanorods are embedded in ZnV₂O₆ granules. The WCN composite showed rectangular sheets of WO₃ settled on g-C₃N₄ sheets. Some sheets of g-C₃N₄ were found cracked during WCN nanocomposite formation. These cracked sheets of g-C₃N₄ with rectangular WO₃ nanosheets doped with ZnV₂O₆ in the ZWC-20 nanocomposite were observed. The nanorods of ZnV₂O₆ were observed at WCN sheets, thus confirming the formation of a ZWC-20 heterojunction. High-resolution transmittance electron microscopy (HR-TEM) scanned the rough imperfection and morphology of WCN and ZWC-20 nanocomposites. The nanosheets of WO₃ were observed at the g-C₃N₄ surface, and the composite WCN showed some crystalline features, confirmed in the SAED pattern with bright dots of the corresponding planes (Fig. S3†). Also, the elemental mapping indicates the amount of W, C, N, and O in WCN. In the ZWC-20 nanocomposite, the nanorods of ZnV₂O₆ were observed on g-C₃N₄ nanosheets. The TEM images of the ZWC-20 nanocomposite confirmed the nanorod morphology of ZnV₂O₆. The high-resolution TEM image exhibited the interplanar spacing of 0.32 and 0.24 nm matching with the (110) and (−112) planes of ZnV₂O₆, respectively. In addition, the SAED pattern of ZWC-20 showed bright dots indicating the highly crystalline nature of the composite. Also, the elemental mapping and EDX pattern showed Zn, V, W, C, N, and O in the ZWC-20 nanocomposite (Fig. S4†).

X-ray photoelectron spectroscopy (XPS) determined the surface properties of the synthesized nanomaterials (Fig. 2). The deconvoluted C-1s XPS spectra for g-C₃N₄ showed three peaks. The three signals at 288.0, 287.6, and 284.4 eV correspond to N–CN, CN, and CC in the graphitic sheets of g-C₃N₄, respectively. In the heterojunction of g-C₃N₄ with WO₃ and ZnV₂O₆, there was a shift in the binding energies of C-1s XPS peaks. Especially when compared to g-C₃N₄, C-1s XPS signals shifted to relatively higher binding energies in the WCN composite. In contrast, as an interesting observation, C-1s XPS peaks were shifted to lower binding energies in the ZWC-20 nanocomposite than in WCN, indicating that ZnV₂O₆ and WO₃ constructed an interface that shifted the C-1s peaks to lower binding energies. Still, they remained slightly higher (288.2 eV for N–CN and 287.8 eV for CN) than g-C₃N₄. In addition, an additional peak observed at 285.5 eV in C-1s XPS spectra might be due to the interaction between ZnV₂O₆ and g-C₃N₄. The deconvoluted N-1s XPS spectra indicated three characteristic signals of –NH₂ groups, graphitic N–(C)₃, and C–NC at 400.4, 399.5, and 398.2 eV, respectively. In WCN, –NH₂ and C–NC peaks shifted to relatively higher binding energies, i.e., 400.6 and 398.3 eV, respectively, while the graphitic N–(C)₃ XPS peak moved to 399.3 eV, which might be due to the construction of an interface between g-C₃N₄ and WO₃.³⁴ The graphitic N–(C)₃ signal was more pronounced and shifted to lower binding energy in ZWC-20 than in g-C₃N₄ and WCN. In addition, the C–NC peak shifted by −0.5 eV in the ZWC-20 nanocomposite, indicating a change in the environment of graphitic C₃N₄ sheets. The characteristic two XPS signals, W-4f_5/2 and W-4f_7/2, were found at 37.5 and 35.4 eV in the XPS spectra of WO₃. These two signals correspond to W⁶⁺ without any other oxidation state in WO₃.³⁵ Similar binding energies were obtained for the elements in WCN and ZWC-20 nanocomposites, indicating that the heterojunction remained intact with the chemical properties of WO₃. However, a shift to lower binding energies was observed in WCN and ZWC-20 nanocomposites, which corresponds to the facile excitation of the electrons at the surface of the photocatalyst. ZnV₂O₆ and ZWC-20 nanocomposites showed two characteristic XPS peaks for Zn-2p, 2p_1/2, and 2p_3/2 signals at 1044.9 and 1021.7 eV corresponding to Zn²⁺ with no significant change in the binding energies of 2p_1/2 and 2p_3/2 peaks.³⁶ XPS of V-2p for ZnV₂O₆ and ZWC-20 nanocomposites showed two characteristic signals for 2p_1/2 and 2p_3/2. The peaks 2p_1/2 and 2p_3/2 were found at 524.4 and 517.0 eV, respectively, in the XPS of ZnV₂O₆. There was no marginal change in the binding energies of V-2p_1/2 and 2p_3/2, indicating that the environment of V-2p was relatively similar to the ZWC-20 nanocomposite. It confirms that there was no change in the crystal lattice of ZnV₂O₆. The ZnV₂O₆ nanorods were decorated at the WCN composite in the ZWC-20 nanocomposite. All the components, ZnV₂O₆, WO₃, WCN, and ZWC-20 nanocomposites, showed two deconvoluted XPS signals for surface oxygen and lattice oxygen. There was no marginal change in the binding energies, and these two XPS peaks were found at 531.8 and 530.2 eV, respectively, for the ZWC-20 nanocomposite. This indicates no change in the crystal lattice of WO₃ and ZnV₂O₆ in WCN and ZWC-20. ZnV₂O₆, WO₃, g-C₃N₄, WCN, and ZWC-20 nanocomposite survey spectra confirmed all the elements (Fig. S5†).

Fig. 2 HR-XPS spectra of the C-1s, N-1s, W-4f, Zn-2p, V-2p, and O-1s of ZnV₂O₆, WO₃, g-C₃N₄, WCN, and ZWC-20 nanocomposites.

Ultra-violet-visible (UV-Vis) spectroscopy determined the optical behaviour of the synthesized nanomaterials (Fig. 3a). g-C₃N₄ exhibited a steep and strong absorption in the UV region due to π → π* and n → π* transitions in the heptazine ring. A broad absorption above 400 nm arose from the electronic transition from the N-2p orbitals of the valence band (VB) to the C-2p orbitals of the conduction band (CB) in g-C₃N₄ with a band gap of 2.68 eV responsible for its photocatalytic activity (Fig. S6†).³⁷ WO₃ absorbs in the UV region as well as in the visible region, showing a band gap of 2.53 eV. ZnV₂O₆ had multiple absorption bands; two absorption bands at 260 and 350 nm in the UV-region and another at 490 nm were observed in the visible region showing a band gap of 2.09 eV. The broad absorption tail, i.e., the Urbach tail, was observed due to the excitation of defect states with energies in the forbidden energy region.^38,39 Constructing a nanocomposite of g-C₃N₄ and WO₃, i.e., WCN, exhibited relatively similar absorption in the UV and visible regions. Nevertheless, the ternary heterojunction ZWC-20 showed good absorption in visible light with a band gap of 2.3 eV. This indicates that incorporating ZnV₂O₆ and WO₃ with g-C₃N₄ would improve the visible light absorption and be a suitable candidate for photocatalytic application.

Fig. 3 (a) UV-Vis spectra, (b) PL spectra, (c) time-resolved PL spectra, and (d) EIS spectra of the samples used in this study.

As mentioned above, the band gaps of g-C₃N₄, WO₃, and ZnV₂O₆ were 2.68, 2.53, and 2.09 eV, respectively. Mott–Schottky (M–S) and VB-XPS were employed to calculate the potentials of CB and VB to predict the electron transfer mechanism. M–S plots could determine the semiconductors' flat-band potentials (E_fb), with the help of which the band edge potentials of the CB and VB can be calculated. VB-XPS is another crucial technique to determine the band edge potentials of the CB and VB, as the binding energy determined by VB-XPS refers to the energy difference between the Fermi and VB edge levels of the particular semiconductor.⁴⁰ M–S plots determined that the individual components g-C₃N₄, WO₃, and ZnV₂O₆ were n-type semiconductors. The E_fb of g-C₃N₄, WO₃, and ZnV₂O₆ was calculated to be −0.71, −0.06, and −0.46 V (vs. Ag/AgCl), respectively, from M–S plots (Fig. S7a–c†) using the equation:

E_RHE = E^θ_(Ag/AgCl) + 0.0591 × pH + E_fb(Ag/AgCl)

E _fb was −0.10, 0.54, and 0.16 V (vs. NHE) for g-C₃N₄, WO₃, and ZnV₂O₆, respectively, where E^θ_(Ag/AgCl) was 0.1989 and pH was 7.⁴¹ The energy differences between the VB edge level and Fermi level determined by VB-XPS for g-C₃N₄, WO₃, and ZnV₂O₆ were 1.71, 2.56, 1.28 eV, respectively (Fig. S7d–f†). The calculated VB for g-C₃N₄, WO₃, and ZnV₂O₆ was 1.60, 3.10, and 1.44 V (vs. NHE), respectively.⁴² Therefore, the CB calculated was −1.08, 0.57, and −0.65 V (vs. NHE), respectively, from the equation E_g = E_VB − E_CB.

Moreover, photoluminescence (PL) spectroscopy provided information regarding the charge transfer and charge carrier's recombination of the semiconductors and composite photocatalysts correlated with the photocatalytic activity (Fig. 3b). The three bands at 420, 439, and 459 nm in the PL spectra of WO₃ correspond to the near band–edge transition and oxygen vacancies (Fig. S8†).⁴³ ZnV₂O₆ exhibited three emission bands at 411, 434, and 457 nm due to band gap excitation, interstitial Zn atoms, and excitations between Zn and singly charged surface oxygen, respectively.⁴⁴ An intense PL emission peak at 447 nm for g-C₃N₄ was found due to the n → π* electronic transition, showing the poor separation of charge carriers and rapid recombination of electron–hole pairs, which may refer to decreased catalytic activity in visible light.⁴⁵ The constructed WCN composite exhibited two emission peaks at 441 and 461 nm with relatively less intensity, indicating an improvement in the separation of charge carriers.⁴⁶ As expected, the ternary heterojunction ZWC-20 showed a less intense and broad emission peak corresponding to the lowest recombination of charge carriers. This refers to the highest separation of electron–hole pairs, and the highest activity is expected under visible light conditions. Thus, the heterojunction ZWC-20 showed the best optical properties with band edge potentials of the CB and VB suitable for oxidation and reduction reactions.

Furthermore, the time-resolved photoluminescence (TRPL) decay curves were recorded at an emission wavelength of 460 nm (Fig. 3c). The decay curves of g-C₃N₄, ZnV₂O₆, WO₃, and WCN comprised two components, short-lived (τ₁) and long-lived lifetime (τ₂). The fast decay with a time constant of τ₁ was 10.45–26.67 ns, followed by a slow decay (τ₂ = 26.48–40.00 ns). The average lifetimes (τ_av) of g-C₃N₄, ZnV₂O₆, WO₃, and WCN were 5.94, 4.37, 7.00, and 6.09 ns, respectively. This indicates a slightly extended excitation diffusion lifetime. The π*–n transition of sp² bonded carbon in g-C₃N₄ resulted in a shorter carrier lifetime (5.94 ns).⁴⁷ In the case of ZnV₂O₆, we observed the shortest lifetime of 4.37 ns. In comparison, WO₃ showed a slightly extended lifetime of 7.00 ns. The shorter lifetime of pristine g-C₃N₄, ZnV₂O₆, and WO₃ indicates the less probability of the charge carrier's participation in the photocatalytic reaction. The prolonged charge carrier lifetime of 6.09 ns for WCN compared to g-C₃N₄ comes from the interface between g-C₃N₄ and WO₃. In ZWC-20, a fast decay with a time constant of τ₁ ranged between 10.41–14.27 ns, followed by a slow decay with a time constant (τ₂) of 20.01–40.00 ns, was found. The τ_av for ZWC-20 is unprecedented and was about 42.47 ns, a long extended excitation diffusion lifetime. The decoration of ZnV₂O₆ on the WCN surface inhibited the charge transfer interaction, resulting in an extended lifetime in ZWC-20 heterojunctions.⁴⁸ This means that the more extensive decay time of ZWC-20 corresponds to the prolonged lifetime of charge carriers, which refers to its higher photocatalytic activity than g-C₃N₄ and WCN nanocomposites.

Electrochemical impedance spectroscopy (EIS) could further study the charge transfer kinetics and charge carrier's separation in the nanomaterials (Fig. 3d). The arc radius is the facial charge transfer resistance function across the catalyst's interface. The large arc radius of g-C₃N₄ indicates extensive facial charge transfer resistance, which corresponds to the minimum efficiency of the charge transfer. Consequently, the photocatalytic activity of g-C₃N₄ showed the lowest photocatalytic activity. The composite of WO₃ and g-C₃N₄, i.e., WCN, showed a smaller semicircle arc radius with decreased facial charge transfer resistance, which implies improved charge transfer efficiency. Furthermore, the decoration of ZnV₂O₆ on the WCN composite, i.e., ZWC-20 nanocomposite, showed the smallest semicircle arc radius with minimum facial charge transfer resistance, which corresponded to the high efficiency of the separation of charge carriers and inhibited the recombination of electron–hole pairs. This suggested that the ZWC-20 nanocomposite with minimum facial charge transfer resistance has the highest photocatalytic activity among the synthesized nanomaterials. This result was also supported by the PL spectra of the synthesized nanomaterials, which showed that the ZWC-20 nanocomposite photocatalyst would be superior in activity compared to the other photocatalysts studied in the present system.

The additional optical properties of the semiconductors and the nanocomposites were investigated using cyclic voltammetry (CV) and linear sweep voltammetry (LSV) in the potential range of −1.5 to 1.5 V and −1.5 to 0 V (Fig. S9†). In CV, g-C₃N₄ showed a strong anodic peak at positive and a cathodic peak at negative potentials (Fig. S9a†). The anodic peak at 1.37 V corresponds to the oxidation of the electrolyte, while the cathodic peak at −1.12 V corresponds to the reduction with a current density of −0.89 mA.⁴⁹ This indicates that the g-C₃N₄ surface would be favorable for oxidation and reduction reactions.⁵⁰ In ZnV₂O₆, two oxidation peaks in the forward scan were found at 0.31 and 1.09 V with relatively weak oxidation properties. Also, a reduction peak at −0.7 V was found with a current density of −0.16 mA. WO₃ showed a current density of 0.22 mA at −0.28 V in the forward scan, ascribed to its strong oxidation properties under reaction conditions.⁵¹ In addition, WO₃ had a −0.95 mA current density, higher than g-C₃N₄ and ZnV₂O₆. However, in WCN and ZWC-20 nanocomposites, an anodic oxidation peak and cathodic reduction peak were found. WCN showed a reduction peak at −1.09 V with a current density of −0.46 mA, while ZWC-20 had a reduction peak at −1.08 V with a current density of −0.52 mA. In addition, ZWC-20 had a higher current density of −1.45 mA than WCN, which had a current density of −1.15 mA.⁵² WCN and ZWC-20 nanocomposite photocatalysts have the highest current density of −1.15 mA and −1.45 mA, respectively, which correspond to a large amount of the charge carriers' transportation efficiency. This suggested that the charge carriers can transport easily through WCN and ZWC-20 nanocomposites and thus enhanced the mobility values compared to g-C₃N₄, WO₃, and ZnV₂O₆.⁵³ In addition, the ZWC-20 nanocomposite showed improved absorption with a reduced band gap with increase in the number of charge carriers and revealed better mobility than the other samples. Thus, the ZWC-20 nanocomposite photocatalyst having a large number of charge carriers with high current density exhibited the highest photocatalytic activity for the desired photocatalytic experiments. Similarly, LSV showed the highest current density of −1.51 mA with the highest photocatalytic activity under experimental conditions (Fig. S9b†).

Fourier transform infrared (FTIR) spectroscopy was utilized to determine the surface functionalities of the synthesized nanomaterials (Fig. 4a). g-C₃N₄ exhibited a vibrational peak at 807 cm⁻¹, characteristic of the breathing mode of tri-s-triazine units, confirming its heptazine basic units.⁵⁴ The stretching bands between 1250 and 1650 cm⁻¹ were assigned to the C–N and CN stretching vibrational modes with a broad IR peak above 3300 cm⁻¹ for some uncondensed amine groups in the framework of g-C₃N₄. The intense vibrations between 500–1000 cm⁻¹ were ascribed to the O–W–O stretching modes in WO₃. Another weak vibrational peak at 1635 cm⁻¹ corresponds to the hydroxyl bending modes, including hydroxyl stretching modes found above 3300 cm⁻¹. Thus, the composite WCN showed the vibrational peaks of WO₃ and g-C₃N₄, indicating heterostructure formation. ZnV₂O₆ showed the characteristic V–O–Zn stretching modes at 484 and 607 cm⁻¹.⁵⁵ Also, the V–O–V asymmetric vibrations of VO₄³⁻ tetrahedral units were found at 823 and 1018 cm⁻¹.⁵⁶ In addition, adsorbed H₂O stretching and bending modes were obtained at 3436 and 1631 cm⁻¹. Furthermore, the ternary heterojunctions of ZnV₂O₆ and WCN, i.e., ZWC derivatives, showed two stretching modes of Zn–O–V below 700 cm⁻¹. However, the intensity was low with lower loading of ZnV₂O₆ (5 and 10 wt%) (Fig. S10†). A significant increase in the vibrational frequencies was found in composites with higher ZnV₂O₆, i.e., ZWC-20 to ZWC-40 heterojunctions, indicating the deposition of ZnV₂O₆ on the WCN nanocomposite, thus confirming the formation of a ternary heterojunction.

Fig. 4 (a) FTIR and (b) Raman spectra of g-C₃N₄, WO₃, WCN, ZnV₂O₆, and ZWC-20 heterojunction photocatalysts.

The Raman spectra of the synthesized nanomaterials are shown in Fig. 4b. The Raman peak at 469 cm⁻¹ was a group of multiple peaks showing the in-plane bending mode of the heptazine linkage and the symmetric N-breathing mode of heptazine units.⁵⁷ The broad peak centred at 1621 cm⁻¹ is ascribed to the stretching vibrations of C–N and analogous to the typical D- and G-bands of graphitic carbon nitride-based materials.⁵⁸ The Raman spectra exhibited various characteristic peaks of WO₃. The peaks below 200 cm⁻¹, i.e., 143 and 194 cm⁻¹, are the characteristics of (W₂O₂)_n chains, while the Raman bands at 280 and 334 cm⁻¹ are ascribed the W–O–W bending modes of bridging oxide ions. In addition, the highly intense Raman bands at 723 and 818 cm⁻¹ demonstrated the W–O–W stretching mode of the WO₃ network.⁵⁹ The composite of WO₃ and g-C₃N₄, i.e., WCN, showed similar Raman bands to g-C₃N₄ along with WO₃ Raman bands. The intense Raman bands of WO₃ have shifted slightly towards lower frequencies (695 and 797 cm⁻¹) in WCN, indicating the construction of an interface between WO₃ and g-C₃N₄. The Raman spectra of ZnV₂O₆ showed two characteristic peaks, 702 and 900 cm⁻¹, with multiple peaks ranging between 100 and 1100 cm⁻¹.⁶⁰ In addition, the ternary nanocomposite ZWC-20 showed the Raman band at a relatively similar frequency of 900 cm⁻¹ for ZnV₂O₆ along with strong Raman bands at 695 and 794 cm⁻¹ for WO₃. This confirmed the formation of the ZWC-20 heterostructure of g-C₃N₄, WO₃, and ZnV₂O₆ semiconductors.

The surface area of the nanomaterials could be determined using the Brunauer–Emmett–Teller (BET) theory of multilayer adsorption (Fig. S11†). The N₂ adsorption–desorption isotherm of g-C₃N₄ was identified as type II with an H3 hysteresis loop. The BET surface area of g-C₃N₄ was 66 m² g⁻¹, with an average pore size of 23 nm and a total pore volume of 0.35 cm³ g⁻¹. This indicates that g-C₃N₄ had mesopores with a significant surface area. The decoration of WO₃ on g-C₃N₄ in WCN changed the isotherm from type-II to type-III with H3 hysteresis. The incorporation of WO₃ on g-C₃N₄ reduced the surface area to 15 m² g⁻¹. Also, the total pore volume and average pore diameter were reduced to 0.14 cm³ g⁻¹ and 3.8 nm, respectively. The reduced pore volume and diameter are ascribed to the deposition of WO₃ particles on the surface of g-C₃N₄. In the ZWC-20 nanocomposite, i.e., the decoration of ZnV₂O₆ on the WCN composite, the surface area was diminished to 4 m² g⁻¹ with a total pore volume of 0.03 cm³ g⁻¹ and an average pore diameter of 3.4 nm. This indicates that the decoration of WO₃ and ZnV₂O₆ on g-C₃N₄ nanosheets diminished the surface area, total pore volume, and average pore diameter.

3.2 Visible-light-driven production of imines and solar fuels

The photocatalytic experiments were performed in a 10 mL round-bottom flask equipped with a 350 W xenon lamp. Benzylamine was used as a model sacrificial donor for CO₂ reduction to CO. In addition, the in situ generated hydrogen from benzylamine oxidation reduces CO₂ to CH₄. Blank experiments with just benzylamine and irradiation with a Xe-lamp without a photocatalyst showed no reaction (Table 1, entry 1). While adding a catalytic amount of g-C₃N₄, we observed the formation of N-benzylidene benzylamine along with CO and CH₄. Using 10 mg of g-C₃N₄, we achieved 27% N-benzylidene benzylamine as the coupled product with 316 and 236 μmol g⁻¹ of CO and CH₄, respectively (entry 2) (Fig. 5a). This indicates that the catalyst was necessary for the reaction to take place. Other semiconductors, WO₃ and ZnV₂O₆, were screened, and WO₃ showed a low yield (5%) of imine and, as expected, did not generate CO (entry 3). However, a low yield of CH₄ (84 μmol g⁻¹) was observed, and it might be due to some basic nature of benzylamine, which absorbed more CO₂ in the solution. In contrast, ZnV₂O₆ has suitable CB and VB positions for CO₂ reduction and amine oxidation; thus, it resulted in a relatively higher N-benzylidene benzylamine yield (14%) but afforded CO (151 μmol g⁻¹) and CH₄ (120 μmol g⁻¹) (entry 4). Therefore, the semiconductors used in this reaction were suitable for visible light-assisted oxidation and reduction (redox) reactions.

Table 1 Investigation of the photocatalytic activity for a redox reaction by individual, binary and ternary composites^a

Entry	Photocatalyst	Solvent	Yield
			Imine^b (%)	CO^c (μmol g^−1)	CH4^c (μmol g^−1)
a Reaction conditions: benzylamine: 0.1 mmol; solvent: 3 mL; photocatalyst: 10 mg; simulated solar light; temperature: 25 °C; atmospheric pressure; reaction time: 6 h. b Yield determined by GC-MS. c Yield determined by GC-FID.
1	—	CH3CN	0	0	0
2	g-C3N4	CH3CN	27	316	236
3	WO3	CH3CN	5	0	84
4	ZnV2O6	CH3CN	14	151	120
5	WO3–g–C3N4	CH3CN	29	892	219
6	ZnV2O6@WO3	CH3CN	27	492	137
7	ZnV2O6@g-C3N4	CH3CN	9	728	194
8	ZWC-05	CH3CN	29	218	741
9	ZWC-10	CH3CN	35	236	455
10	ZWC-20	CH3CN	55	1674	683
11	ZWC-30	CH3CN	52	754	513
12	ZWC-40	CH3CN	52	580	433
13	ZWC-20	H2O	0	633	477
14	ZWC-20	THF	10	102	93
15	ZWC-20	1,4-Dioxane	38	500	433
16	ZWC-20	DMF	50	705	590

---

Fig. 5 Optimization using (a) individual semiconductors, (b) ZnV₂O₆ loadings, (c) solvents, and (d) timeline for the redox reaction.

The construction of the heterojunction always helps improve the photocatalytic conversion of the substrate molecules by inhibiting the recombination of charge carriers at the interface between the semiconductors. Thus, the constructed heterojunctions of g-C₃N₄, WO₃, and ZnV₂O₆ were screened for the current photocatalytic reaction (Fig. 5b). At first, the constructed WCN nanocomposite resulted in a 29% yield of N-benzylidene benzylamine with 892 and 219 μmol g⁻¹ yields of CO and CH₄, respectively (entry 5). This indicates that the WCN heterojunction did not improve the imine yield; however, the CO yield improved much compared to CH₄. In addition, other nanocomposites ZnV₂O₆@WO₃ and ZnV₂O₆@g-C₃N₄ were also investigated for their photocatalytic activity. ZnV₂O₆@WO₃ resulted in 492 μmol g⁻¹, 137 μmol g⁻¹, and 27% yields of CO, CH₄, and imine, respectively (entry 6). The ZnV₂O₆@g-C₃N₄ composite afforded 728 μmol g⁻¹, 194 μmol g⁻¹, and 9% yields of CO, CH₄, and imine, respectively (entry 7). This indicates that the binary composites exhibited moderate photocatalytic activity for the redox reaction. Furthermore, the WCN nanocomposite was modified by loading ZnV₂O₆ as it has suitable CO₂ reduction and oxidation potentials. ZWC nanocomposites with different ZnV₂O₆ concentrations were screened for this reaction. ZWC-05 with 5 wt% loading of ZnV₂O₆ on the WCN composite showed a 29% N-benzylidene benzylamine yield. It was also found that the yields of CO and CH₄ were 218 and 741 μmol g⁻¹ (entry 8). The ZWC-10 nanocomposite showed an enhanced yield of 35% of N-benzylidene benzylamine. The yields of CO and CH₄ were 236 and 455 μmol g⁻¹ (entry 9). Increasing ZnV₂O₆ on WCN, i.e., ZWC-20 nanocomposite, enhanced the products' yields. The highest yield for N-benzylidene benzylamine was 55% within 6 h of the photocatalytic reaction. Also, the yield of CO and CH₄ was 1674 μmol g⁻¹ and 683 μmol g⁻¹, respectively (entry 10). This indicates that increasing the ZnV₂O₆ content on the WCN composite dramatically enhanced the yields of imine and CO₂ reduction products. Higher concentrations of ZnV₂O₆, ZWC-30, and ZWC-40 nanocomposites did not show improved yields. Both ZWC-30 and ZWC-40 resulted in a 52% yield of N-benzylidene benzylamine (entry 11). However, the CO yield was slightly higher (754 μmol g⁻¹) with ZWC-30 than with ZWC-40 (580 μmol g⁻¹) but less than half that achieved with ZWC-20. Similarly, the CH₄ yield was reduced from 513 to 433 μmol g⁻¹ for ZWC-30 and ZWC-40 nanocomposites (entry 12). Thus, the ZWC-20 nanocomposite was selected as an optimum photocatalyst for simultaneous photoreduction of CO₂ with benzylamine oxidation under visible light irradiation.

Further optimization of the reaction conditions was also carried out extensively. In this context, the effect of solvent plays a crucial role in the photocatalyst's activity. Thus, we selected five different solvents: water, tetrahydrofuran (THF), 1,4-dioxane, N,N-dimethylformamide (DMF), and acetonitrile, based on the previously reported literature for CO₂ photoreduction and amine coupling (Fig. 5c). In water, no imine formation occurred, while CO and CH₄ yields were 633 and 477 μmol g⁻¹, respectively (Table 1, entry 13). While using THF as a solvent, the yield of N-benzylidene benzylamine was improved by up to 10%, but CO and CH₄ yields were minimized to 102 and 93 μmol g⁻¹, respectively, which is not a significant increase in yield (entry 14). In 1,4-dioxane, imine's yield was 38%, with 500 and 433 μmol g⁻¹ yields of CO and CH₄, respectively (entry 15). When using DMF, N-benzylidene benzylamine crossed 50% yield, which means that DMF would be a good solvent for this reaction (51% yield). However, the yields of CO and CH₄ were not improved much, i.e., CO and CH₄ obtained were 705 and 590 μmol g⁻¹, respectively (entry 16). The highest yields were achieved with acetonitrile as the solvent. The N-benzylidene benzylamine yield was 55% in acetonitrile, while CO and CH₄ yields were improved to 1674 and 683 μmol g⁻¹. This indicates that benzylamine oxidation was effectively achieved in dioxane, DMF, and acetonitrile, while the CO yield was much higher in acetonitrile than in the other solvents. Furthermore, the reaction was prolonged to 24 h and achieved a maximum of 68% yield of N-benzylidene benzylamine with 2424 and 1598 μmol g⁻¹ of CO and CH₄, respectively (Fig. 5d). This concluded that prolonging the reaction did not provide significant results and a photocatalytic reaction time of 6 h is quite enough to provide significant activity.

In addition, we performed experiments with different substrate concentrations to investigate the optimum and effective concentration under identical conditions (Fig. S12†). We used 0.02 mmol and 0.05 mmol of benzylamine and found 61% and 59% yields of N-benzylidene benzylamine within six hours. In comparison, 0.1 mmol of benzylamine afforded a 55% yield of N-benzylidene benzylamine. However, a higher amine concentration (0.15 mmol) resulted in ∼54% yield. This indicates that there was no significant enhancement in the yield by varying the concentrations of benzylamine in the photocatalytic experiments. Thus, 0.1 mmol of benzylamine was a reasonable and acceptable concentration of the substrate for the desired redox photocatalytic reaction.

3.3 Scavenger and stability experiments

The effect of electron and hole scavengers could also be studied for the photoreduction of CO₂ with benzylamine oxidation (Fig. 6a). We used silver nitrate and ammonium oxalate as electron and hole scavengers, respectively. When 0.025 mmol of silver nitrate was used, the yields of CO and CH₄ were diminished to 468 and 214 μmol g⁻¹, while the yield of N-benzylidene benzylamine was 54%. This indicates that silver nitrate quenched the electrons from benzylamine donors, which were readily available for the reduction of CO₂; thus, the yields of CO and CH₄ were diminished. Moreover, when 0.025 mmol of ammonium oxalate was employed, the holes were quenched, resulting in a very low 7% yield of N-benzylidene benzylamine; however, the yields of CO and CH₄ were nearly similar under the optimized conditions, i.e., 1616 and 607 μmol g⁻¹, respectively. Thus, the photocatalytic reduction of CO₂ and oxidative coupling of amines could be quenched by silver nitrate and ammonium oxalate, consequently diminishing the corresponding oxidation and reduction products' yields.

Fig. 6 (a) Scavenger study, (b) recycling experiments, (c) XRD (# for WO₃ and ♦ for ZnV₂O₆) and (d) FTIR spectra of fresh and spent ZWC-20 nanocomposites after four cycles.

The applicability of the photocatalyst depends on its reusability and how it retains its photocatalyst activity. So, the reusability of the photocatalyst needs to be determined to check its durability. The photocatalytic reaction of CO₂ photoreduction and amine oxidation was performed with the ZWC-20 photocatalyst. The photocatalyst was recovered by washing with ethanol, dried in an oven at 60 °C for 24 h, and reused for successive photocatalytic reactions. The recycling experiments were performed for four cycles and achieved similar CO, CH₄, and N-benzylidene benzylamine yields (Fig. 6b). This indicates that the ZWC-20 photocatalyst exhibited good photocatalytic activity and was stable enough for the present reaction conditions.

Furthermore, the recovered photocatalyst was characterized using XRD and FTIR spectroscopy. After four cycles, the diffraction pattern of the spent ZWC-20 nanocomposite was similar to that of the fresh ZWC-20 nanocomposite (Fig. 6c). It showed the diffraction peaks of WO₃, ZnV₂O₆, and g-C₃N₄ with a minor change in intensity. This indicates that the ZWC-20 nanocomposite remained intact after four cycles of the photocatalytic experiments. Also, FTIR spectra were recorded for the spent ZWC-20 nanocomposite (Fig. 6d). Similarly, there was no change in the pattern of tri-s-triazine, C–N, CN, and –NH₂ groups' peaks of the spent ZWC-20 nanocomposite. In addition, the VO₄³⁻ and V–O–Zn bindings were relatively similar to those of the fresh ZWC-20 nanocomposite. Thus, the analysis of the recovered ZWC-20 photocatalyst confirmed that the photocatalyst was stable enough for the photocatalytic reduction of CO₂ with the oxidative coupling of amines.

In addition, the recovered ZWC-20 nanocomposite was analyzed using HR-XPS analysis (Fig. 7a and b, and S13†). The recovered ZWC-20 nanocomposite exhibited binding energy deviation of the elements C-1s, N-1s, O-1s, and Zn-2p (except W-4f and V-2p) compared to the fresh catalyst. This deviation in C-1s, N-1s, and O-1s might be due to the significant interaction between the substrate molecules and the nanocomposite's surface. Also, HR-TEM analysis found a similar morphology to the ZWC-20 nanocomposite (Fig. 7c–f), where ZnV₂O₆ had a nanorod-like morphology wrapped in g-C₃N₄ nanosheets. The SAED pattern reveals that the nanocomposite retained the high crystallinity after four cycles of consecutive runs of the recovered nanocomposite photocatalyst, demonstrating high stability under reaction conditions. In addition, the recovered photocatalyst's EDX pattern confirmed that there is no marginal leaching of Zn, V, and W metals compared to the fresh one, indicating high stability up to four cycles of photocatalytic experiments (Fig. S14†).

Fig. 7 C-1s and Zn-2p XPS spectra, HR-TEM images, and SAED pattern of the spent ZWC-20 nanocomposite.

3.4 Comparison with reported photocatalysts

The photoreduction of CO₂ coupled with the oxidation of benzylamine is seldom reported. Compared to the previously reported literature, we found that the reported system, ZWC-20, achieved good conversion and yields of amines (55%) and C₁ (1674 μmol g⁻¹ of CO and 683 μmol g⁻¹ of CH₄) products, respectively, within a shorter reaction time (6 h). However, a prolonged reaction time afforded 68% yield of imine and 2424 and 1598 μmol g⁻¹ of CO and CH₄, respectively. This shows the supremacy of the developed ZWC-20 heterojunction photocatalyst (Table 2).

Table 2 Comparison of ZnV₂O₆@WO₃–g–₃N₄ photocatalyst's activity with previously reported literature

Entry	Photocatalyst	Time (h)	YieldImine (%)	C1 Yields (μmol g^−1)	Ref
1	CuMoxW(1−x)O4 (5 bar)	24	82.1	CH3OH (1017)	16
2	Cu/TiO2 (UV light)	10	88.7	CH3OH (931)	61
3	C3N4	24	34	CO2 as oxidant	17
4	Cu/Cu2O@C3N4	16	54	CH3OH (437)	62
5	ZnV2O6@WO3–g–3N4	6	55	CO (1674) & CH4 (683)	This work

---

3.5 Product and intermediate identification

The products of benzylamine oxidative coupling with CO₂ photoreduction were confirmed using GC-MS and ¹H and ¹³C NMR spectra. The GC-MS spectra of N-benzylidene benzylamine showed a molecular ion peak at an m/z value of 91, corresponding to (Fig. S15†). Also, the ¹H NMR spectra of N-benzylidene benzylamine showed a de-shielded imine proton at an 8.34 ppm chemical shift (Fig. S16a†). In addition, the de-shielded imine carbon peak was found at 162 ppm in ¹³C NMR spectra, confirming the product formation (Fig. S16b†). The CO₂ photoreduction products, i.e., CO and CH₄, were identified using a calibrated GC-FID equipped with a capillary column. In addition, CO and CH₄ were identified using GC-MS (Shimadzu, SH-Rt-Msieve 5A, 30 m, 0.32 mm, 30 μm, and temp. 30–250 °C), which showed fragmentation at the corresponding m/z values. Two m/z values, 16 and 28, were observed for CO, while CH₄ exhibited a base peak as a molecular ion peak (m/z = 16) (Fig. S17†).

Furthermore, several experiments were performed to understand the mechanistic route, and the intermediates were detected using GC-MS and NMR studies. Benzylamine has converted into N-benzylidene benzylamine via benzaldehyde intermediate formation, which GC-MS detected. Within 2 h of the reaction, the benzaldehyde intermediate was found at a retention time of 4.3 min in the total ion chromatogram (TIC) (Fig. S18†). The fragmentation pattern confirmed the benzaldehyde intermediate formation in the reaction system. This indicates the reaction pathway where a certain amount of benzylamine got oxidized to benzaldehyde. Then the remaining benzylamine is coupled with the benzaldehyde intermediate in a dehydrogenated manner to produce N-benzylidene benzylamine as the desired product. This was also observed in NMR studies where the aldehydic protons were found at a chemical shift of 9.85 ppm in ¹H NMR, while the aldehydic carbon was detected at a 192 ppm chemical shift (Fig. S19a and b†). This confirmed the benzaldehyde intermediate formation during benzylamine coupling with the CO₂ photoreduction reaction.

3.6 Isotopic labelling (¹³CO₂)

To better understand the carbon source in the products CO and CH₄, photocatalytic oxidative coupling using labelled CO₂ (¹³CO₂) was performed. In the isotope labelling experiment, we used labelled CO₂ (¹³CO₂) to find the source of C₁ products. Under identical conditions in a ¹³CO₂ atmosphere, we detected two peaks of ¹³CO and ¹³CH₄ in the TIC and fragmentation patterns of ¹³CO and ¹³CH₄ with m/z values of 29 and 17, respectively (Fig. S20a–c†). This confirms that CO and CH₄ were produced from the photoreduction of CO₂.

3.7 Substrate scope

In order to prove the significance of this simultaneous redox reaction, various substituted benzylamines and other unsaturated amines at the β-position to the amine group were investigated (Table 3). Benzylamine as the model substrate resulted in a maximum of 55% yield with the highest CO (1674 μmol g⁻¹) and CH₄ (683 μmol g⁻¹) yields (entry 1). Benzylamines having electron-donating groups at the para position resulted in relatively higher yields than electron-withdrawing groups. The yields of 4,4′-dimethyl benzylidene benzylamine and 4,4′-dimethoxy benzylidene benzylamine were 53% and 49%, but CO and CH₄ achieved moderate yields (entries 2 & 3). 4,4′-Dichloro and 4,4′-difluoro benzylidene benzylamine achieved 49% and 48% yields. Furfuryl-derived amine was also investigated for this reaction under similar conditions. Furfuryl amine afforded a 51% yield of the coupled product. Also, we employed aniline as the substrate, where there was no imine formation and significantly less CO (107 μmol g⁻¹) and CH₄ (75 μmol g⁻¹) yields.

Table 3 Scope of amine substrates for the photoreduction of CO₂^a

Entry	Substrate	Imine	YieldImine^b (%)	YieldCO^c (μmol g^−1)	YieldCH4^c (μmol g^−1)
a Reaction conditions: amine substrate: 0.1 mmol; catalyst: 10 mg; acetonitrile: 2 mL; reaction time: 6 h; light source: 350 W xenon lamp. b Yield determined by GC-MS. c Yield determined by GC-FID.
1			55	1674	683
2			53	767	347
3			49	550	382
4			49	444	614
5			48	604	441
6			51	1531	594
7			—	107	75

---

3.8 Mechanism of the simultaneous photocatalytic redox reaction

As mentioned earlier, the band gaps of g-C₃N₄, WO₃, and ZnV₂O₆ were 2.68, 2.53, and 2.09 eV, respectively. Based on the band edge potentials calculated above in Section 3.1, the energy band structure for g-C₃N₄, WO₃, and ZnV₂O₆ is proposed (Scheme 3). The CB of g-C₃N₄ (−1.08 eV) and ZnV₂O₆ (−0.65 eV) calculated above was more negative than that of WO₃ (0.57 eV). Also, the E_fb of WO₃ (0.54 V vs. NHE) is lower than that of g-C₃N₄ (−0.10 V vs. NHE) and ZnV₂O₆ (0.16 V vs. NHE), and M–S plots confirmed that these three semiconductors are n-type semiconductors. The E_fb of g-C₃N₄ was higher than that of WO₃; this tends to transfer electrons from g-C₃N₄ to the WO₃ surface, making WO₃ highly electronegative and g-C₃N₄ highly electropositive.⁶³ Thus, at equilibrium, the induced electric field is directed from g-C₃N₄ to WO₃. When the photocatalytic system is irradiated with visible light, photogenerated electrons and holes are produced, and the electrons get excited to the CB. In contrast, the photogenerated holes are left in the VB of the semiconductors. The CB (0.54 V vs. NHE) of WO₃ is positive in energy, which does not favour the reduction of CO₂ to any C₁ products (CO₂/C₁ ≥ −0.24 eV, for CH₄). Also, when the photogenerated electrons from the CB of g-C₃N₄ transferred to the CB of WO₃, it could not result in any C₁ product. However, the highly positive VB (2.56 eV) of WO₃ favours the oxidation of amines under visible light irradiation. In addition, from CV, WO₃ showed strong oxidizing properties, which shows that the photogenerated holes in the VB of WO₃ are entirely involved in the oxidation process. Thus, the possibility of the photoreduction of CO₂ over WO₃ is ruled out.

Scheme 3 (a) Band energy diagram depicting the CB and VB in the ternary heterojunction and (b) representation of photocatalytic CO₂ reduction along with amine oxidation.

In another way, the photogenerated electrons in the CB of WO₃ tend to migrate to the VB of g-C₃N₄, where they recombine with the photogenerated holes and thus result in photogenerated holes in the VB of WO₃ with strong oxidation ability, suitable for the oxidative coupling of amines into the corresponding imines. The excited photogenerated electrons in the CB of g-C₃N₄ are readily available for reduction and might show high activity in visible light. In PL spectra, it was also confirmed that the WCN heterojunction had a high separation of charge carriers and favours better charge separation than the individual semiconductor components. In addition, WO₃ and g-C₃N₄ were n-type semiconductors; thus, this photocatalytic system showed electron transfer in a step manner and favoured the S-scheme mechanism.⁶⁴ The readily available photogenerated electrons in the CB of g-C₃N₄ transferred to the lower energy CB of ZnV₂O₆, where the available excess of electrons reduced the adsorbed CO₂ into CO and CH₄ using benzylamine as a sacrificial donor and hole scavenger.²⁸ Here, benzylamine (sacrificial donor and proton source) boosted the yields of CO and CH₄ from the photoreduction of CO₂. Thus, the substrate molecules utilized simultaneous oxidation and reduction (redox) reactions more sustainably, making the entire redox photocatalytic process atom economical. This detailed mechanistic study using various techniques on this heterojunction system provides more understanding of the advantage of individual components and as a hetero-composite for applying this system in various other challenging redox reactions to be carried out, e.g., for the biomass component conversion strategy.

4 Conclusion

The present work demonstrated the use of a cost-effective and readily available carbon waste CO₂ to oxidize amines to the corresponding imines with simultaneous reduction of CO₂ to value-added C₁ products under simulated solar light irradiation. A ternary ZnV₂O₆@WO₃–g–C₃N₄ nanocomposite heterojunction had an excellent ability for the photoreduction of CO₂ and oxidative coupling of amines into CO and CH₄ and imines, respectively. The present photocatalytic system provided a good yield of CO (1674 μmol g⁻¹) and CH₄ (683 μmol g⁻¹), along with a 55% yield of N-benzylidene benzylamine within six hours of light irradiation. In addition, the recovered photocatalyst exhibited reusability without significant change in the activity up to four consecutive runs of the photocatalytic reaction. The present methodology developed a one-pot system with added benefits of atom economy via simultaneous oxidation and reduction reactions.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

The authors are thankful to the Institute of Nano Science and Technology (INST), Mohali, for providing instrumental facilities to carry out the scientific work.

References

1. C. Higman and S. Tam, Chem. Rev., 2014, 114, 1673–1708.
2. S. I. Murahashi, Angew. Chem., Int. Ed. Engl., 1995, 34, 2443–2465.
3. T. Mukaiyama, A. Kawana, Y. Fukuda and J.-i. Matsuo, Chem. Lett., 2001, 30, 390–391.
4. K. Nicolaou, C. J. Mathison and T. Montagnon, J. Am. Chem. Soc., 2004, 126, 5192–5201.
5. X. Li, Q. Su, Z. Liu, K. Luo, G. Li and Q. Wu, Chem. Res. Chin. Univ., 2020, 36, 1017–1023.
6. J. Yang and C.-Y. Mou, Appl. Catal. B, 2018, 231, 283–291.
7. A. Khampuanbut, S. Santalelat, A. Pankiew, D. Channei, S. Pornsuwan, K. Faungnawakij, S. Phanichphant and B. Inceesungvorn, J. Colloid Interface Sci., 2020, 560, 213–224.
8. C. Xu, H. Liu, D. Li, J.-H. Su and H.-L. Jiang, Chem. Sci., 2018, 9, 3152–3158.
9. V. R. Battula, H. Singh, S. Kumar, I. Bala, S. K. Pal and K. Kailasam, ACS Catal., 2018, 8, 6751–6759.
10. S. I. Seneviratne, M. G. Donat, A. J. Pitman, R. Knutti and R. L. Wilby, Nature, 2016, 529, 477–483.
11. G. Chen, G. I. Waterhouse, R. Shi, J. Zhao, Z. Li, L. Z. Wu, C. H. Tung and T. Zhang, Angew. Chem., Int. Ed., 2019, 58, 17528–17551.
12. P. K. Prajapati, A. Kumar and S. L. Jain, ACS Sustainable Chem. Eng., 2018, 6, 7799–7809.
13. S. Saini, H. Singh, P. K. Prajapati, A. K. Sinha and S. L. Jain, ACS Sustainable Chem. Eng., 2019, 7, 11313–11322.
14. M.-Y. Qi, Q. Lin, Z.-R. Tang and Y.-J. Xu, Appl. Catal. B, 2022, 307, 121158.
15. P. K. Prajapati, S. Saini, N. Nandal and S. L. Jain, J. CO₂ Util., 2021, 45, 101402.
16. C. Luo, T. Yang, Q. Huang, X. Liu, H. Ling, Y. Zhu, G. Xia, W. Zou and H. Wang, Nanomaterials, 2020, 10, 1303.
17. Y. Markushyna, P. Lamagni, J. Catalano, N. Lock, G. Zhang, M. Antonietti and A. Savateev, ACS Catal., 2020, 10, 7336–7342.
18. F. Wang, T. Hou, X. Zhao, W. Yao, R. Fang, K. Shen and Y. Li, Adv. Mater., 2021, 33, 2102690.
19. G. Wang, T. Zhang, W. Yu, Z. Sun, X. Nie, R. Si, Y. Liu and Z. Zhao, CCS Chem., 2022, 4, 2793–2805.
20. Q. Xu, Z. Xia, J. Zhang, Z. Wei, Q. Guo, H. Jin, H. Tang, S. Li, X. Pan and Z. Su, Carbon Energy, 2022, 1–43.
21. H. Shi, G. Chen, C. Zhang and Z. Zou, ACS Catal., 2014, 4, 3637–3643.
22. H. Yu, R. Shi, Y. Zhao, T. Bian, Y. Zhao, C. Zhou, G. I. Waterhouse, L. Z. Wu, C. H. Tung and T. Zhang, Adv. Mater., 2017, 29, 1605148.
23. K. Chu, Q.-q. Li, Y.-p. Liu, J. Wang and Y.-h. Cheng, Appl. Catal. B, 2020, 267, 118693.
24. T. Ohno, N. Murakami, T. Koyanagi and Y. Yang, J. CO₂ Util., 2014, 6, 17–25.
25. L.-f. Hong, R.-t. Guo, Y. Yuan, X.-y. Ji, Z.-d. Lin, J.-w. Gu and W.-g. Pan, Energy Fuels, 2021, 35, 11468–11478.
26. L. Wang, X. Shi, Y. Jia, H. Cheng, L. Wang and Q. Wang, Chin. Chem. Lett., 2021, 32, 1869–1878.
27. M. Du, Y. Chen, W. Wang, X. Xu, Y. Li, Y. Zhang, Z. Li and Z. Zou, Appl. Catal. B, 2022, 317, 121722.
28. A. Bafaqeer, M. Tahir and N. A. S. Amin, Appl. Catal. B, 2019, 242, 312–326.
29. A. Bafaqeer, M. Tahir, A. Ali Khan and N. A. Saidina Amin, Ind. Eng. Chem. Res., 2019, 58, 8612–8624.
30. A. Kumar, P. K. Prajapati, M. Aathira, A. Bansiwal, R. Boukherroub and S. L. Jain, J. Colloid Interface Sci., 2019, 543, 201–213.
31. P. Wang, S. M. Zakeeruddin, P. Comte, R. Charvet, R. Humphry-Baker and M. Grätzel, J. Phys. Chem. B, 2003, 107, 14336–14341.
32. L. Shi, L. Liang, F. Wang, M. Liu, K. Chen, K. Sun, N. Zhang and J. Sun, ACS Sustainable Chem. Eng., 2015, 3, 3412–3419.
33. S. Wei, J. Zhao, B. Hu, K. Wu, W. Du and M. Zhou, Ceram. Int., 2017, 43, 2579–2585.
34. Y. Bao and K. Chen, Nano-Micro Lett., 2016, 8, 182–192.
35. B. Paul, B. Bhuyan, D. D. Purkayastha and S. S. Dhar, Catal. Commun., 2015, 69, 48–54.
36. A. Malik, P. K. Prajapati, B. M. Abraham, S. Bhatt, P. Basyach and S. L. Jain, Catal. Sci. Technol., 2022, 12, 2688–2702.
37. S. Le, T. Jiang, Q. Zhao, X. Liu, Y. Li, B. Fang and M. Gong, RSC Adv., 2016, 6, 38811–38819.
38. F. Kröger and H. Vink, in Solid State Physics, Elsevier, 1956, vol. 3, pp. 307–435.
39. S. Hong, S. J. Burkhow, R. M. Doughty, Y. Cheng, B. J. Ryan, A. Mantravadi, L. T. Roling, M. G. Panthani, F. E. Osterloh and E. A. Smith, Chem. Mater., 2021, 33, 1667–1682.
40. Y. Yu, P. Zhang, Y. Kuang, Y. Ding, J. Yao, J. Xu and Y. Cao, J. Phys. Chem. C, 2014, 118, 20982–20988.
41. T. Tachikawa, T. Ochi and Y. Kobori, ACS Catal., 2016, 6, 2250–2256.
42. L. Wang, Y. Dong, J. Zhang, F. Tao and J. Xu, J. Solid State Chem., 2022, 308, 122878.
43. J. Hao, B. Qi, J. Wei, D. Li and F. Zeng, Fuel, 2021, 287, 119439.
44. M. Aslam, I. M. Ismail, T. Almeelbi, N. Salah, S. Chandrasekaran and A. Hameed, Chemosphere, 2014, 117, 115–123.
45. M. Alenizi, R. Kumar, M. Aslam, F. Alseroury and M. Barakat, Sci. Rep., 2019, 9, 1–8.
46. X. Han, D. Xu, L. An, C. Hou, Y. Li, Q. Zhang and H. Wang, Int. J. Hydrogen Energy, 2018, 43, 4845–4855.
47. L. Lin and S. Zhang, Chem. Commun., 2012, 48, 10177–10179.
48. G. Rajender, B. Choudhury and P. Giri, Nanotechnology, 2017, 28, 395703.
49. C. Rajkumar, P. Veerakumar, S.-M. Chen, B. Thirumalraj and K.-C. Lin, ACS Sustainable Chem. Eng., 2018, 6, 16021–16031.
50. S. Siwal, N. Devi, V. K. Perla, S. K. Ghosh and K. Mallick, Catal. Struct. React., 2019, 5, 1–9.
51. Y. Yin, C. Lan, H. Guo and C. Li, ACS Appl. Mater. Interfaces, 2016, 8, 3861–3867.
52. W. Li, Q. Chen and Q. Zhong, J. Mater. Sci., 2020, 55, 10712–10724.
53. S. C. Rasmussen, R. L. Schwiderski and M. E. Mulholland, Chem. Commun., 2011, 47, 11394–11410.
54. C. Ji, S.-N. Yin, S. Sun and S. Yang, Appl. Surf. Sci., 2018, 434, 1224–1231.
55. S. E. Arasi, P. Devendran, R. Ranjithkumar, S. Arunpandiyan and A. Arivarasan, Mater. Sci. Semicond. Process., 2020, 106, 104785.
56. M. M. Sajid, N. A. Shad, S. B. Khan, Z. Zhang and N. Amin, J. Alloys Compd., 2019, 775, 281–289.
57. S. Tonda, S. Kumar, S. Kandula and V. Shanker, J. Mater. Chem. A, 2014, 2, 6772–6780.
58. W.-J. Ong, L.-L. Tan, S.-P. Chai and S.-T. Yong, Dalton Trans., 2015, 44, 1249–1257.
59. D. Vernardou, H. Drosos, E. Spanakis, E. Koudoumas, C. Savvakis and N. Katsarakis, J. Mater. Chem., 2011, 21, 513–517.
60. A. Bafaqeer, M. Tahir and N. A. S. Amin, Appl. Surf. Sci., 2018, 435, 953–962.
61. T. Yang, Q. Yu and H. Wang, Catal. Lett., 2018, 148, 2382–2390.
62. P. K. Prajapati, D. Tripathi, M. K. Poddar, P. Gupta and S. L. Jain, Sustainable Energy Fuels, 2022, 6, 2996–3007.
63. M. Sun, Y. Zhou, T. Yu and J. Wang, Int. J. Hydrogen Energy, 2022, 47, 10261–10276.
64. Q. Xu, L. Zhang, B. Cheng, J. Fan and J. Yu, Chem, 2020, 6, 1543–1559.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2ta07642j

---