Introduction

Selective oxidation of saturated hydrocarbons and aromatic hydrocarbons, such as cyclohexane, toluene, and p-xylene, is important for producing valuable oxygenates in the petrochemical industry1. Through liquid-phase partial oxidation, cyclohexane can be converted into a cyclohexanone and cyclohexanol mixture (KA oil), which can subsequently be dehydrogenated into cyclohexanone for a wide range of applications in medicine, coatings, dyes, caprolactam production, and adipic acid production2. Safety, selectivity, and environmental impacts are the major concerns in selective oxidation processes. Most of the existing commercial selective oxidation processes are liquid-phase reactions using either oxygen with/without inert gas dilution or H2O2 as the oxidant in the presence of soluble cobalt and/or manganese salts3,4. The reaction conditions are usually 170 ℃ to 230 ℃ and 1.0 MPa to 2.0 Mpa, harsh and expensive5. To ensure high selectivity and safety, hydrocarbon conversion is minimized by preventing rapid and intensive heat release6,7. Meanwhile, huge volumes of inert gas, such as nitrogen, are employed as sweeping gas to keep the oxygen concentration in the bulk gas phase below 5%, thus ensuring the safety of the process8. Lowering the reaction temperature will reduce the hydrocarbon vapor pressure and the sweeping gas demand. However, activating the C-H bond in cyclohexane at lower temperatures is critical but challenging for developing a safer oxidation process.

Photocatalysis is a promising advanced oxidation technology that can alleviate water pollution problems9. Photocatalysis promotes selective hydrocarbon oxidation under ambient conditions10. Giannott et al.11 discovered that TiO2, WO3, and ZnO promote the oxidation of cyclohexane into cyclohexanone by dissolved oxygen under UV light irradiation at ambient temperature. The photocatalytic cyclohexane oxidation is initiated by the photogenerated holes and ·O2− subtracting hydrogen from cyclohexane, forming alkyl radicals and propagating the oxidation12,13,14. However, TiO2 can only be excited by UV light, while a highly active and visible light excitable photocatalyst is desired. WO3, C3N415, and BiOBr16 have been identified as catalysts for cyclohexane oxidation into KA oil under visible light irradiation. Despite the strong oxidation capacity of WO3 benefiting from the high oxidation potential (3.2 eV) of its valence band, its insufficient reduction capacity due to a poor reduction potential (0.4 eV) of the conduction band limits the reduction reaction. As a result, the photogenerated holes and electrons on WO3 easily recombine with each other, thus greatly hindering its application. Introducing Pt nanoparticles onto WO3 can promote the consumption of electrons accumulated on the WO3 conduction band by facilitating the conversion of O2 to H2O2 on Pt particles17. As a result, the KA oil accumulation rate on Pt/WO3 is about 5 times higher than that on WO3. Although C3N4 has many fascinating properties, such as chemical stability and low cost, it also features poor crystallinity, visible light trapping edge, easy charge pair recombination, small surface area, and slow charge migration18. Catalysts with narrow band gaps can be excited by visible light, but their redox capability is usually insufficient due to the quick recombination of the photogenerated carriers19. Huang et al.20 prepared a novel WO3/g-C3N4 composite photocatalyst, which showed higher photocatalytic activity than pure WO3 and pure g-C3N4. Li et al.21 believed that the composition of heterojunction was conducive to charge separation and transfer, which was an effective strategy to improve the photoelectrochemical performance of the WO3 photoanode. Forming Z-type heterojunctions by coupling two semiconductors with suitable band structures facilitates the separation of the photogenerated carriers and the redox sites while maintaining a reasonable energy band structure, thus enhancing the redox reaction on both sides22. Combining the highly reductive C3N4 with the highly oxidative WO3 particles could combine the holes in the C3N4 valence band and the electrons in the WO3 conduction band via the formation of a Z-type heterojunction, thus endowing the compound catalyst with excellent redox capability23,24. Thus, the resultant WO3/C3N4 heterojunction catalyst could promote the photocatalytic oxidation of organics as it oxidizes H2O into ·OH and reduces O2 to ·O2- under visible light irradiation due to its high valence band energy (> 2.7 eV)25.

Yang et al. found that the presence of water could greatly enhance photocatalytic cyclohexane conversion over V2O5@TiO2 under visible light irradiation and 2 MPa O2 pressure26. The preferential adsorption of the polar oxygenates, such as cyclohexanol and cyclohexanone, instead of cyclohexane on the photoexcited TiO2 (P25) surface limits the selectivity of KA oil to about 55.5%. Doping TiO2 with other metal oxides, such as V2O5, could promote cyclohexane adsorption and facilitate oxygenate desorption, raising the KA oil selectivity to 80%27. Polar solvents, such as acetonitrile and dichloromethane, can also improve the selectivity for intermediate oxygenates due to the strong adsorption of solvent molecules26,28. Introducing H2O into the cyclohexane photooxidation system could promote either ·OH generation or KA oil separation from the catalyst, thus facilitating environmentally friendly KA oil production.

Both ·OH and ·O2− are highly active oxidizing species in photocatalytic oxidation, which can be employed in the photocatalytic synthesis of KA oil from cyclohexane. This work aims to synthesize a visible light efficient photocatalyst by combining C3N4 with WO3 and develop a novel photocatalytic cyclohexane partial oxidation system with C3N4/WO3 as the catalyst and water as the promotor. The proposed composite catalyst can produce the key active species for cyclohexane oxidation in the H2O–O2 system. Cyclohexane was converted into cyclohexyl radical under the action of ·OH, and ·O2- converted most products into cyclohexanone. The catalyst can be recycled under optimized process conditions while reaching a KA oil yield of 139.73 μmol g−1 h−1 and a selectivity of 93.1%. The structure and catalytic mechanism of the proposed catalyst were discussed. The graphical abstract is shown in Fig. 1.

Figure 1
figure 1

A visible light efficient photocatalyst by combining C3N4 with WO3 was synthesized and a novel photocatalytic cyclohexane partial oxidation system with C3N4/WO3 as the catalyst and water as the promoter was developed.

Full size image

Experimental

C3N4 preparation

Adding 3 g of melamine into a 50 mL crucible with a lid and calcined in a muffle furnace at 520 °C for 2 h with the temperature increased by 5 °C min−1. After cooling naturally to room temperature, the temperature was raised to 500 °C at the same rate and kept constant for 2 h. The yellow solid product was collected and ground into powder.

WO3 and C3N4/WO3 preparation

First, 0.5 g of Na2WO4·2H2O was dissolved in 50 mL deionized water and mixed with 1 mL lactic acid under vigorous stirring for 15 min to yield a faint yellow solution. Then, a given amount of C3N4 (x = 0.1, 0.2, 0.3, 0.4, and 0.5 g) was added to the solution under vigorous stirring for another 15 min, followed by ultrasonication for 20 min. The pH of the slurry was adjusted to 1 with a 1 mol L−1 HCl solution. The mixture was transferred to a 100 mL Teflon-lined autoclave and hydrothermally treated at 180 °C for 24 h. C3N4/WO3 composite was obtained after repeated rinsing with deionized water and ethanol. The resultant composites were dried, ground, and labeled as xC3N4/WO3 according to their C3N4 dosage. Pure WO3 was prepared following the same procedure without adding C3N4.

Characterization

The X-ray diffraction (XRD) spectra were recorded on a Powder X-ray diffractometer (DX2700, Haoyuan Instrument Co., Ltd.) with Cu Kα radiation (k = 0.15418 nm) in a 2θ range of 5° to 80°. The scanning step was 0.03°, and the accumulation time was 0.5 s. The Fourier transform infrared (FT-IR) spectra of the samples were recorded on an FT-IR spectrometer (Spectrum Two Li10014, Perkin Elmer). Micromorphology and surface element distribution of the samples were observed via field-emission scanning electron microscopy (SEM, JSM 7610F, JEOL Ltd.) with energy-dispersive X-ray spectrometry (EDX). Nanostructure analysis was performed on a field-emission transmission electron microscope (TEM, JEM-F200, JEOL Ltd.). The UV–vis diffuse reflectance spectra (UV–vis-DRS) were recorded on a UV–vis-DRS spectrometer (Lambda750S, PerkinElmer). X-ray photoelectron spectra (XPS) were obtained using an ESCALAB 250Xi XPS spectrometer with an Al Ka monochromatized X-ray source. The Ultraviolet Photoelectron Spectroscopy (UPS) of the sample was recorded on the X-ray photoelectron spectrometer (AXIS SUPRA+, Shimadzu Corp.). The photoluminescence (PL) spectra were obtained using an FLS1000 spectrophotometer (Edinburgh Instruments). The electron spin resonance (ESR) measurements were obtained using an electron paramagnetic resonance spectrometer (MS-5000X, Magnettech) with 5,5-dimethyl-l-pyrrole-n-oxide (DMPO) as the capturing agent. The photoelectrochemical experiments are detailed in S1.

Photocatalytic activity test

The photocatalytic cyclohexane oxidation was conducted in a 100 mL quartz reactor with two openings sealed using a silicone rubber septum. A 300 W xenon lamp with a UV cutoff filter (λ > 380 nm) served as the light source at the top of the reactor. The reactor was immersed in a 15 °C water bath to avoid heating the reaction mixture and excessive cyclohexane evaporation. Typically, 2 mL cyclohexane, 8 mL ultra-pure water, and 0.05 g catalyst were mixed in the quartz reactor with stirring and visible light irradiation for 6 h. Oxygen was introduced to the reactor by bubbling through the reaction mixture at 6 mL min−1. After the reaction, liquid samples were collected using a syringe with a 0.22 μm filter to separate the liquid mixture from the catalyst.

Cyclohexane, cyclohexanol, and cyclohexanone were analyzed through gas chromatography (GC 9790, Zhejiang Fuli Analytical Instruments Co., Ltd), with a hydrogen flame ionization detector (FID) and a KB-Wax capillary column (30 m × 0.32 mm × 0.50 μm) using n-butanol as the internal standard. The injector and detector temperatures were 190 °C and 200 °C, respectively. The oven temperature was initially kept at 80 °C for 1 min. Then, the temperature was ramped to 110 °C at a heating rate of 20 °C min−1 and held for 1 min before ramped to 180 °C at 10 °C min−1. The products were confirmed using a gas chromatograph-mass spectrometer. The product identification results are shown in Fig. S1. The amount of CO2 in the gas phase was determined through the barium hydroxide method29. Oxygen was bubbled into the reactor, and the vent gas was introduced to a saturated Ba(OH)2 solution to absorb the CO2. The KA oil selectivity is calculated as follows30:

$$\begin{array}{c}{S}_{KA oil} = \frac{{n}_{cyclohexanol}+{n}_{cyclohexanone} }{{n}_{cyclohexanol}+{n}_{cyclohexanone}+\frac{1}{6}{n}_{{CO}_{2}}}\end{array}$$
(1)

Moreover, a 10 mmol L−1 isopropanol (IPA), triethanolamine (TEOA), ascorbic acid, and KBrO3 were employed as trapping agents for hydroxyl radical (·OH), photogenerated hole (h+), superoxide radical (·O2−), and photogenerated electron (e−), respectively.

DFT calculations

The space group and unit cell parameters of monoclinic WO3 and lamellar C3N4 have been adopted in this study31,32. The C3N4/WO3 catalyst composed of lamellar C3N4 and WO3 (2 0 0) crystal face was constructed. A 2 nm vacuum layer was set up to avoid between-layer interaction. Geometry structure optimization and average potentials of WO3, C3N4, and C3N4/WO3 were calculated with the density functional theory (DFT) based on the CASTEP code and plane-wave pseudopotential method and performed with the periodic super-cells under generalized gradient approximation (GGA). The Perdew-Burke-Ernzerhof (PBE) functional exchange–correlation function was used. The interaction between the valence electron and the ionic core was expressed by the ultrasoft pseudopotential. The HSE06 hybrid function was adopted to obtain the electronic densities and band structure of C3N4, WO3, and C3N4/WO3 heterojunctions. The Brillouin zones were sampled with a 3 × 3 × 3 k- point grid. The converge criteria for the energy, maximum force, pressure, and displacement are 1 × 10−5 eV atom−1, 3 × 10−1 eV nm−1, 0.05 GPa, and 1 × 10−4 nm, respectively.

The transition states (TS) and the oxygenate intermediates were optimized using the Dmol3 program package. The GGA with the BP exchange–correlation functional was adopted. The converge criterion judged by the energy, force, and displacement were 2 × 10−5 Ha, 4 × 10−2 Ha nm−1, and 5 × 10−4 nm, respectively. The density of state (DOS) and partial DOS (PDOS) of WO3, C3N4, and C3N4/WO3 are detailed in S7.

Results and discussion

Catalyst structure and morphology

Figure 2a shows the XRD patterns of pure WO3, C3N4, and xC3N4/WO3 composites. The diffraction peaks at 13.1° and 27.6° observed on C3N4 and xC3N4/WO3 correspond to the diffraction of the (1 0 0) and (0 0 2) lattice plane of graphitic C3N4 (JCPDS 87-1526). The diffraction peaks at 23.1°, 23.6°, and 24.3° observed on WO3 and xC3N4/WO3 correspond to the diffraction on the (0 0 2), (0 2 0), and (2 0 0) lattice plane of monoclinic WO3 (JCPDS 83-0951), respectively. The 27.6° diffraction peaks corresponding to C3N4 were observed in xC3N4/WO3 (x = 0.3, 0.4, 0.5), while the 27.6° and 13.1° diffraction peaks of C3N4 were not observed in xC3N4/WO3 (x = 0.1, 0.2), which may be due to the low content of C3N4. Fig. S3a exhibits the XRD spectra of catalysts prepared with different lactic acid dosages.

Figure 2
figure 2

Catalyst composition and structure characterization: (a) XRD patterns; (b) FT-IR spectra.

Full size image

The FT-IR spectra (Fig. 2b) of the catalyst also show the co-presence of WO3 and C3N4 on the C3N4/WO3 composite. The absorption peaks of WO3 and C3N4/WO3 composites at 763 cm−1 correspond to the stretching vibration of W–O–W. In the spectra of C3N4 and C3N4/WO3 composites, the absorption peaks at 1243 cm−1, 1325 cm−1, and 1411 cm−1 correspond to the stretching vibration of the triazine aromatic structure C–N in C3N4, and the absorption peaks at 1577 cm−1 and 1632 cm−1 are attributed to the stretching vibration of the C–N bond. The peak at 810 cm−1 corresponds to condensed C-N heterocyclic rings. The absorption of water molecules and N–H vibration of the uncondensed amine group result in a wide absorption peak near 3400 cm−1. As the amount of C3N4 increases, xC3N4/WO3 shifts to the right at the peak position of 810 cm−1, indicating that C3N4 and WO3 have a strong coupling in nanocomposites33. All C3N4/WO3 samples have obvious characteristic peaks of C3N4, with obvious increases in the proportion of characteristic peak intensity. The results show that WO3 and C3N4 can be successfully prepared as composite catalysts.

Pure WO3 is identified as nanosheets with a clear surface and a thickness of about 10 to 50 nm (Fig. 3a). Figure 3b shows that pure C3N4 has a typical layered structure with obvious wrinkles on the surface. Deposition of C3N4 on the surface of WO3 was observed on all the xC3N4/WO3 composites (Fig. 3c–g). The C3N4 accumulation on the WO3 surface is found to increase with the increase of C3N4 loading in xC3N4/WO3 composites. Fig. S2 displays the SEM and EDS images of 0.3C3N4/WO3 and the EDS mapping diagrams of four elements in the 0.3C3N4/WO3 catalyst.

Figure 3
figure 3

SEM images of (a) WO3, (b) C3N4, (c) 0.1C3N4/WO3, (d) 0.2C3N4/WO3, (e) 0.3C3N4/WO3, (f) 0.4C3N4/WO3, (g) 0.5C3N4/WO3, and (h) TEM image of 0.3C3N4/WO3, (i) HRTEM image of 0.3C3N4/WO3.

Full size image

The microstructure of 0.3C3N4/WO3 was characterized and analyzed through TEM and HRTEM. Figure 3h shows that the composite material comprises WO3 nanoplates and a small amount of layered C3N4. The C3N4 exists as relatively thin layers highly dispersed and in close contact with the WO3 interface. HRTEM images of the composite catalyst (Fig. 3i) clearly show the contact interface. On the C3N4 side, lattice fringes are not observed due to poor crystallinity. On the other side, the crystal lattice stripe with a width of 0.361 nm can be clearly identified, belonging to the (2 0 0) plane of monoclinic phase WO3. The intimate interfacial contact indicates the formation of a heterojunction.

The composite catalyst is prepared at high temperature in an acidic solution, during which C3N4 is stripped into highly dispersed and smaller scale nanoparticles, and some C3N4 is even protonated or decomposed. Therefore, the WO3 with strong oxidation performance forms the main body of the composite catalyst.

The XPS spectra of WO3, C3N4, and 0.3C3N4/WO3 samples were analyzed. Figure 4a shows the XPS spectra of WO3, C3N4, and 0.3C3N4/WO3 catalysts, where the corresponding peaks of their constituent elements can be found. The C 1s spectra of pure C3N4 show a binding energy of 288.2 eV for the sp2-bonded carbon in N–C=N groups (Fig. 4a). The binding energy of the sp2-bonded carbon in N–C=N groups was found shifting to 288.8 eV in the 0.3C3N4/WO3 composite. The different electrostatic potential of the intimate interfaces between WO3 and C3N4 indicates the strong interaction between the two components. The peaks at 286.2 eV and 290.1 eV are attributed to C–OH and –COOH, respectively, which may be derived from the lactic acid in the catalyst preparation process. The peaks at 398.7 eV, 400.1 eV, and 401.1 eV (Fig. 4d) are attributed to N 1s of sp2 N (N-C2), tertiary carbon–nitrogen bonds (N-C3), and NHX group in pure C3N4, which were found shifting to 399.4 eV, 400.9 eV, and 402.0 eV in 0.3C3N4/WO3, respectively34. The peak at 404.5 eV is attributed to the π → π* transition of C–N heterocycles. The outer electrons of the N atom in C3N4 are shifted towards the WO3 interface, causing the corresponding peak to shift in the direction of increasing binding energy. These results show strong interactions between the two components.

Figure 4
figure 4

The XPS spectra of the sample: (a) C 1s spectra, (b) W 4f spectra, (c) O 1s spectra, (d) N 1s spectra.

Full size image

The XPS peaks at binding energies of 35.6 eV and 37.7 eV (Fig. 4b) are attributed to W 4f7/2 and W 4f5/2 of WO3, respectively24. Moreover, these two peaks are identical in WO3 and 0.3C3N4/WO3. As shown in Fig. 4c, the peaks at 530.4 eV and 531.8 eV are ascribed to lattice oxygen and adsorbed hydroxyl groups in WO3, respectively35. Similarly, the peak position of the composite catalyst did not shift. The XPS results indicate that after WO3 and C3N4 are compounded, all the peaks of elements in C3N4 are shifted. Therefore, the composite catalyst formed a heterojunction with obvious interaction between the two components36. However, the peak position of elements in WO3 showed no significant change, confirming the strongly oxidative WO3 as the main body of the composite catalyst35. Figure 4d exhibits the N 1s spectra of the samples.

Photoelectric properties

The UV–Vis DRS of WO3, C3N4, and 0.3C3N4/WO3 (Fig. 5a) shows that the absorption intensity of 0.3C3N4/WO3 is in the range of 350 to 500 nm, which is somewhere between that of pure WO3 and C3N4. This result confirmed that 0.3C3N4/WO3 can absorb visible light.

Figure 5
figure 5

The photo absorption and charge carrier transfer properties of the catalysts. (a) UV–vis diffuse reflectance spectra of WO3, C3N4, and 0.3C3N4/WO3, (b) band gap energies of WO3 and C3N4, (c) PL spectra of WO3, C3N4, and 0.3C3N4/WO3, (d) Nyquist plots of WO3, C3N4, and 0.3C3N4/WO3.

Full size image

Figure S4a shows the wide-scan XPS survey spectra of the samples. The valence band spectra of WO3 and C3N4 are 2.84 eV and 1.80 eV (Fig. S4b), respectively, which indicates the strong oxidation strength of WO3 and weak oxidation strength of C3N4. In order to better estimate the valence band, UPS characterization was conducted37. Fig. S5a shows the UPS spectra of WO3 and C3N4. The valence bands of WO3 and C3N4 are estimated to be 1.80 eV and 2.84 eV, respectively.

The band gap width of the semiconductor was obtained by treating the UV–Vis results of the sample with Equation S1. WO3 is a direct band-gap semiconductor, n = 1/2, and C3N4 is an indirect band-gap semiconductor, n = 2. As shown in Fig. 5b, the band gap widths of WO3 and C3N4 are 2.71 eV and 2.45 eV, respectively. Considering their valence band levels, the conduction band levels of WO3 and C3N4 are 0.13 eV and −0.65 eV, respectively. These results confirm that the surface of C3N4 can generate O2/O2− (−0.33 eV, vs NHE), and the WO3 surface can produce H2O/·OH (2.69 eV, vs NHE).

The PL spectra obtained with the photocatalysts at the excitation wavelength of 350 nm are shown in Fig. 5c. The rapid recombination of surface photoelectric charge and holes grants C3N4 with stronger PL strength. After adding WO3, the peak strength of the 0.3C3N4/WO3 composite was weakened, indicating inhibited photogenerated electron–hole recombination. The peak intensity of the PL spectrum of WO3 is almost zero and is unrelated to charge separation. In 0.3C3N4/WO3, the photogenerated electrons of the WO3 conduction band recombine rapidly with the photogenerated holes of the C3N4 valence band, accompanied by low energy fluorescence generation. Thus, the holes with strong oxidizing properties of WO3 and the photogenerated electrons with strong reducing properties of C3N4 are separated and retained. This result is consistent with the carrier transport theory of Z-type heterojunction and is similar to a previous report38.

To explore the influence of WO3 and C3N4 composite on the charge transfer performance, electrochemical impedance spectroscopy (EIS) is performed on WO3, C3N4, and 0.3C3N4/WO3, as shown in Fig. 5d. Comparing with pure WO3 and C3N4, the Nyquist plot radius of 0.3C3N4/WO3 is smaller, indicating that the composite catalyst has lower resistance due to the higher space charge separation and migration rate. Although the recombination of some holes and electrons at the composite interface accelerated, strong oxidizing holes and strong reducing electrons were retained, which promoted their effective separation and migration. The formation of Z-type heterojunction improves the photocatalytic performance of the composites.

DFT simulation of the photocatalyst band structure

DFT calculations were employed to understand the energy band structure of the catalysts and the work function of each component. The charge transfer between WO3 and C3N4 significantly affects the interfacial energy of the heterojunctions. The work functions of WO3 and C3N4 were clarified through DFT calculations. The results are presented in Fig. 6a, b. The work functions of WO3 and C3N4 are 5.76 eV and 4.69 eV, respectively, indicating that WO3 is more capable of binding electrons. The work function difference between the two materials is |ΔW1,2 |= 1.07 eV, and the delocalized π bonds of C3N4 act as electron donors at the interface of the two components, indicating easy charge transfer between WO3 and C3N439. The Z-type heterojunction formation mechanism of WO3 and C3N4 composite is shown in Fig. 6c. As the work function of WO3 is higher than that of C3N4, it forms an internal electric field in the composite. The electrons of C3N4 transfer to WO3 at the interface, leading to the identical Fermi energy level of the two materials. The WO3 surface is partially negatively charged, and the electron accumulation leads to the downward bending of the conduction and valence bands. As a result, the C3N4 surface is positively charged, leading to the upward bending of the conduction and valence bands. The work function difference between the two components (|ΔW1,2 |= 1.07 eV) is larger than the conduction band energy level difference (|ΔECB1,2 |= 0.78 eV) and valence band energy level difference (|ΔEVB1,2 |= 1.04 eV) between the two components, further confirming the formation of Z-type heterojunction between WO3 and C3N4. WO3 has a valence bond of 2.84 eV and a conduction band of only 0.13 eV. While C3N4 is a reduced photocatalyst, its valence bond position is only 1.80 eV, and its conduction band energy level reaches −0.65 eV. When the composites are excited by light, the electrons of the valence bonds in WO3 and C3N4 will jump into their conduction bands, respectively. Under the internal electric field and band bending, the recombination of the holes in the conduction band (CB) of WO3 and the photogenerated electrons in the valence bond of C3N4 are promoted. In turn, the recombination of holes in the valence bond of WO3 and the photogenerated electrons in the conduction band of C3N4 is inhibited. Therefore, Z-type heterojunction photocatalyst can maintain the strong oxidation strength of WO3 and the strong reduction strength of C3N4 under visible light irradiation. On the surface of this Z-type heterojunction, water can be oxidized on the WO3 sites, and oxygen can be reduced on the C3N4 sites, leading to the formation of ·OH and ·O2−.

Figure 6
figure 6

(a,b) Calculation results of WO3 and C3N4 work functions. (c) Energy band structure of WO3, C3N4, and 0.3C3N4/WO3, and the formation of Z-type heterojunctions with the photogenic carrier transfer mechanism.

Full size image

Photocatalytic activity of xC3N4/WO3

The photocatalytic activities of the catalysts were evaluated during selective cyclohexane oxidation. The results (Fig. 7) show that the KA oil production rate over C3N4 is almost negligible, while that over WO3 is about 23.60 μmol g−1 h−1. The combination of WO3 and C3N4 significantly increases the KA oil production rate (Fig. 7). Initially, the KA oil production rate increases with the increase of C3N4 in the composite catalyst, and the highest KA oil production rate of 118.08 μmol g−1 h−1 was achieved with 0.3C3N4/WO3. Further increased C3N4 dosages in 0.5C3N4/WO3 reduce the KA oil production rate. This is likely due to the change of C3N4 coverage over WO3 as observed by SEM (Fig. 3), and the simultaneous exposure of C3N4 and WO3 to the reaction media is critical to promote cyclohexane photooxidation. An optimal C3N4 coverage over WO3 can be found. These findings suggest strong interactions between C3N4 and WO3, most likely in the way of forming a Z-type heterojunction, which leads to much higher photocatalytic activity.

Figure 7
figure 7

Catalyst performance in reactions with 8 mL ultra-pure water, 2 mL cyclohexane, O2 at 6 mL min−1, 50 mg catalyst, and irradiation for 6 h. (a) C3N4 and xC3N4/WO3 prepared with 0.2 mol L−1 lactic acid (x = 0, 0.1, 0.2, 0.3, 0.4, 0.5), (b) C3N4 and xC3N4/WO3 prepared with 0.1 mol L−1 lactic acid (x = 0, 0.05, 0.1, 0.2, 0.3, 0.4).

Full size image

Figure 7a displays the catalyst performance in reactions with 8 mL ultra-pure water, 2 mL cyclohexane, and O2 at 6 mL min−1. Figure 7b shows the performance test results of C3N4/WO3 with lactic acid dosage reduced to 0.1 mol L−1 during the preparation process. The variation trend of KA oil production was similar to the C3N4/WO3 prepared with 0.2 mol L−1 lactic acid, but the maximum yield was only 79.91 μmol g−1 h−1. S4 illustrates the photocatalytic activity affected by the crystal phase of WO3 controlled via the amount of lactic acid in catalyst preparation. The conclusion is that adding lactic acid enables the complete transformation of WO3 into the monoclinic phase. This crystalline WO3 and its composite catalyst have better photocatalytic cyclohexane oxidation activity. Fig. S3b exhibits the performance comparison of catalysts prepared with different lactic acid dosages.

Effects of the reaction conditions on the KA oil production rate

The effects of oxygen flow rate on cyclohexane oxidation are shown in Fig. 8a. When purged with nitrogen, only a negligible amount of cyclohexanol and cyclohexanone is detected in the system. The KA oil production rate burst to about 67 μmol g−1 h−1 as oxygen was introduced at a flow rate of 3 mL min−1, indicating that O2 is critical for photogenerated electron consumption, and the photocatalytic H2O oxidation cannot proceed significantly effectively in the absence of O2 due to the difficulty of N2 reduction by the photogenerated electrons. Meanwhile, ·O2− generated from the O2 reduction by photoelectrons can also promote the radical oxidation of cyclohexane. Increasing the oxygen flow rate to 6 mL min−1 leads to a KA oil production rate of 113.33 μmol g−1 h−1, likely due to the enhancement of gas-to-liquid mass transfer. However, O2 reaches saturation as the flow rate goes up, leading to the level off of the reaction rate. The KA oil selectivity decreased slightly as the oxygen flow rate increased due to the higher activity of the intermediates.

Figure 8
figure 8

(a) Effects of different oxygen flow rates on the performance of reactions with 8 mL ultra-pure water, 2 mL cyclohexane, 50 mg 0.3C3N4/WO3, and irradiation for 6 h, (b) effects of different volume fractions of water in water/acetonitrile mixtures on the performance of reactions with 8 mL water/acetonitrile mixed liquor, 2 mL cyclohexane, O2 at 6 mL min−1, 50 mg 0.3C3N4/WO3, and irradiation for 6 h.

Full size image

The effects of H2O in the reaction system were investigated by varying the water content in the water-acetonitrile solution. As shown in Fig. 8b, the KA oil production rate of the system is negligible without water but increases significantly as the volume fractions of water in the water/acetonitrile mixtures increase. A higher H2O concentration means a higher ·OH concentration under light irradiation, which accelerates cyclohexane oxidation, suggesting that the direct cyclohexane oxidation by the photogenerated holes is negligible, and the photooxidation of water on the photocatalyst is critical for the cyclohexane oxidation. The KA oil selectivity is above 90% in all cases.

The effect of 0.3C3N4/WO3 dosage on photocatalytic cyclohexane oxidation was investigated. As shown in Fig. 9a, when the catalyst dosage is increased to 2 g L−1, the KA oil yield per unit catalyst mass increases to 124.86 μmol g−1 h−1. With the increase in dosage, the catalyst concentration increases between the oil and water in the upper layer, as does the concentration of ·OH and ·O2− generated near the oil phase, rendering it easier for the diffusion to the oil phase to oxidize cyclohexane. As the catalyst dosage is further increased to 5 g L−1, the KA oil yield per unit catalyst mass shows no significant change, but the cyclohexanone to cyclohexanol ratio increases from 2.15 to 10.63. Therefore, increasing the catalyst dosage improves the oxidation of cyclohexane and cyclohexanol. as evidenced by the increased cyclohexanone and cyclohexanol production rates and the ketone-to-alcohol ratio (Fig. 9a). Further increasing the catalyst dosage blocks more light, inconducive to photocatalytic cyclohexane oxidation. The spent 0.3C3N4/WO3 catalyst was recycled by filtration and washing, which showed reasonable stability (Fig. 9b). As shown in Fig. S5b, FTIR spectra of the WO3/C3N4 composite material before and after five cycles, and FTIR spectra of the photocatalyst show no significant change, indicating that the photocatalyst is stable and can be used for multiple cycles.

Figure 9
figure 9

(a) Effects of different 0.3C3N4/WO3 dosages on the performance of reactions with 8 mL ultra-pure water, 2 mL cyclohexane, O2 at 6 mL min−1, and irradiation for 6 h, (b) experimental results of five cycles.

Full size image

Photocatalytic cyclohexane oxidation mechanism over C3N4/WO3

Various radical trapping agents and ESR were employed to understand the photocatalytic cyclohexane oxidation mechanism over 0.3C3N4/WO3. The ESR results show ·OH radical generation, as evidenced by the symmetrical quadruple peaks (Fig. 10a). The ·OH radicals are most likely generated from H2O over WO3 as only the valence band (2.71 eV) of WO3 is high enough to oxidize H2O into ·OH. With the same test conditions and catalyst dosage, the radical signals of 0.3C3N4/WO3 are much higher than those of C3N4 and WO3, suggesting a higher ·OH generation frequency on 0.3C3N4/WO3. These findings reconfirm the strong interaction between C3N4 and WO3, which enhances the photocatalytic activity. The signal over C3N4 is the weakest, indicating its weakest photooxidation capability, which agrees well with the theoretic valence band analysis. The signal over C3N4/WO3 is the strongest, suggesting the interaction between the two components.

Figure 10
figure 10

ESR spectra of WO3, C3N4, and 0.3C3N4/WO3: (a) DMPO-⋅OH in aqueous, (b) DMPO-⋅O2− in ethyl alcohol, (c) Koutecky-Levich plots of data obtained by rotating disk electrode (RDE) analysis of C3N4 and 0.3C3N4/WO3, (d) radical trapping experiment results with 8 mL ultra-pure water, 2 mL cyclohexane, O2 at 6 mL min−1, 50 mg catalyst, irradiation for 6 h, and 10 mmol L−1 trapping agent.

Full size image

The ESR spectra of 0.3C3N4/WO3 and C3N4 in Fig. 10b clearly show a typical six-fold peak of ·O2−, and the results of WO3 show its difficulty in generating ·O2−. C3N4 is a common photocatalyst capable of producing ·O2−. The results show that the 0.3C3N4/WO3 composite can retain the strong reducing ability of C3N4 and reduce O2 into ·O2−.

These results once again prove that the composite catalyst has the properties of Z-type heterojunction, i.e., strong redox ability and capacity to produce more oxidizing active species. Figure 10c shows the test results of electron transfer number during oxygen reduction on C3N4 and 0.3C3N4/WO3 surfaces, demonstrating that the conversion of O2 to H2O2 occurs on the C3N4 surface via two successive single-electron reaction steps, while the conversion of O2 to ·O2− occurs on the 0.3C3N4/WO3 surface via single-electron transfer.

Figure 10d shows the effects of adding isopropanol, TEOA, ascorbic acid, and KBrO3 on the photooxidation of cyclohexane to KA oil. The presence of isopropanol and TEOA almost completely diminished cyclohexane photooxidation as isopropanol is known to trap ·OH, while TEOA consumes the photogenerated holes. These results suggest that ·OH is critical for cyclohexane oxidation, and photogenerated holes are critical for ·OH generation. The cyclohexanone and cyclohexanol production rates are largely suppressed , with ascorbic acid added to selectively eliminate ·O2−, suggesting ·O2− as one of the major sources of ·OH. However, the results after adding KBrO3 suggest that ·O2− radicals are not generated directly by O2 photoreduction since the consumption of photogenerated electrons by KBrO3 does not significantly affect the cyclohexane oxidation.

With ·O2− eliminated by ascorbic acid, KA oil production was decreased. For one thing, ascorbic acid contributes to cyclohexane oxidation; for another, it partially consumes other oxidized species. The capture of photogenerated electrons by KBrO3 also causes the failure to produce ·O2−. The KA oil yield did not decrease, but the cyclohexanone to cyclohexanol ratio decreased from 10.63 to 2.24, suggesting that the presence of ·O2− facilitates cyclohexanol oxidization to cyclohexanone.

The radical trapping experiments showed that ·OH and ·O2− are the key oxidizing species in cyclohexane oxidation. Figures S6 and S7 displayed the DOS and PDOS calculation results of WO3, C3N4 and C3N4/WO3. DFT calculation results of the transition state search for different reaction paths are shown in Fig. 11. The energy barrier of direct cyclohexane oxidation by molecular oxygen is about 86.66 kcal mol−1 due to the difficulty of activating molecular oxygen.

Figure 11
figure 11

Calculation results of transition state search for different reaction paths: (a) the cyclohexane oxidation path, (b) the product transformation path.

Full size image

The energy barrier of the reaction between ·O2− and cyclohexane to form cyclohexyl hydrogen peroxide is 45.82 kcal mol−1, far lower than the energy required for direct cyclohexane activation with molecular oxygen. When ·OH reacts with cyclohexane molecules, relaxation after a C–H bond length of cyclohexane growth from 0.110 to 0.298 nm, the · OH and cyclohexane reaction energy barrier is 32.40 kcal mol−1, and the reaction transition state is C–H after the fracture of cyclohexyl free radicals. These results confirm that the composite catalyst can use the generated free radicals for photocatalytic cyclohexane oxidation in systems with low energy consumption.

The possible reactions between products and by-products are examined. The energy barrier for continued cyclohexanol oxidation by ·OH was 41.29 kcal mol−1. Thus, the oxidation capacity of ·OH for the reactant cyclohexane was higher than that for the product cyclohexanol, which would not lead to excessive cyclohexanol oxidation. However, the energy barrier of cyclohexanol oxidation by ·O2− is −24.59 kcal mol−1, and the energy of the cyclohexanone product system is 26.32 kcal mol−1 below that of the reactant system. Hence, ·O2− and cyclohexanol react spontaneously, and the product cyclohexanone can exist relatively stably. This conclusion was also confirmed by the much higher product cyclohexanone ratio than that of cyclohexanol, and the remaining small amount of cyclohexanol was due to the timely desorption from the catalyst surface to the reaction solution to avoid being retained by ·O2− oxidation with limited lifetime.

The above experimental result analysis and theoretical calculations suggest the following reaction path in the photocatalytic cyclohexane oxidation system. First, as expressed in Eqs. (2) and (3), H2O and O2 are converted to ·OH and ·O2− on the catalyst surface, respectively. Subsequently, as expressed in Eqs. (4) and (5), cyclohexane is oxidized to cyclohexanol and cyclohexyl hydrogen peroxide in the presence of ·OH and ·O2−, respectively. Finally, as expressed in Eqs. (6) and (7), cyclohexyl hydrogen peroxide is decomposed into cyclohexanol and cyclohexanone, and most of the cyclohexanol in the reaction solution is oxidized to cyclohexanone by ·O2−.

$$\begin{array}{c}{\text{H}}_{2}O+{\text{h}}^{+}\to \cdot OH\end{array}$$
(2)
$$\begin{array}{c}{\text{O}}_{2}+{\text{e}}^{-}\to \cdot {\text{O}}_{2}^{-}\end{array}$$
(3)
$$\begin{array}{c}{\text{C}}_{6}{\text{H}}_{12}+ \cdot OH\to {\text{C}}_{6}{\text{H}}_{12}O+{\text{H}}^{+}\end{array}$$
(4)
$$\begin{array}{c}{\text{C}}_{6}{\text{H}}_{12}+ \cdot {\text{O}}_{2}^{-}\to {\text{C}}_{6}{\text{H}}_{12}OO\end{array}$$
(5)
$$\begin{array}{c}{\text{C}}_{6}{\text{H}}_{12}OO\to {\text{C}}_{6}{\text{H}}_{12}O+{\text{C}}_{6}{\text{H}}_{10}O\end{array}$$
(6)
$$\begin{array}{c}{\text{C}}_{6}{\text{H}}_{12}O+ \cdot {\text{O}}_{2}^{-}\to {\text{C}}_{6}{\text{H}}_{10}O\end{array}$$
(7)

Conclusions

In this study, WO3 and C3N4 were successfully combined to form a Z-type heterojunction photocatalyst with strong redox performance and carrier separation ability. As the main component of the composite catalyst, monoclinic WO3 is more suitable for the photocatalytic cyclohexane oxidation system. The new photocatalytic cyclohexane oxidation system with H2O and O2 as oxidation media achieved efficient conversion into KA oil under the action of ·OH and ·O2- generated on the C3N4/WO3 surface. ·OH is the main active substance in cyclohexane oxidation, while ·O2- produced from O2 reduction is involved in cyclohexane oxidation and oxidizes most cyclohexanol to cyclohexanone. The reaction mechanisms and conditions were explored and optimized, which could guide the practical promotion of photocatalytic cyclohexane oxidation.