Preparation and characterizations of BiVO4 photocatalysts. We hypothesize that the low photocatalytic activity of commonly-used submicro-sized BiVO4 particles (s-BiVO4) originates from their rather narrow bandgap and small surface-to-volume ratio. Therefore, a hard-template method was developed for preparing q-BiVO4 with shrunk size, which would have an increased bandgap and larger surface-to-volume ratio. As shown in Fig. 2a, homogeneous blending of silica gels with Bi(NO3)3 and NH4ViO3 precursors followed by calcination resulted in a composite structure with silica template encapsulated by BiVO4 (Fig. S1). Subsequent high-pressure hydrothermal reaction made silica template fully dissolve, which broke BiVO4 shell into quantum sized particles. Transmission electron microscopy (TEM) images reveal that the spherical aggregates (Fig. 2b, c) are composed of point-like q-BiVO4 with an average particle size of ~ 4.5 nm (Fig. 2d, e). It is known that the Bohr radius of BiVO4 is about 2 nm18, 19 which is close to the radius of q-BiVO4 (2.25 nm). Thus, q-BiVO4 should exhibit a strong quantum confinement effect. X-ray diffraction (XRD) pattern indicates that as-synthesized q-BiVO4 is of monoclinic scheelite structure without any impurity crystal phase (Fig. 2g and Fig. S2). Note that the absence of broad XRD shoulder corresponding to amorphous silica manifests the successful removal of template (Fig. S2). The characteristic XRD peaks are also used to analyze the average aggregation size of q-BiVO4 (10.5 nm) within Debye-Scherrer method (Fig. S3). The clear lattice fringe in high-resolution TEM (HRTEM, Fig. 2f) and the selected area electron diffraction (SAED) pattern both confirm the high crystalline nature of q-BiVO4. Element mapping images based on high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM, Fig. S4) display that all the Bi, V and O elements are homogeneously distributed over q-BiVO4, and the quantitative estimation by energy dispersive X-ray (EDX) spectroscopy presents the stoichiometric composition of Bi, V and O elements in q-BiVO4 (Fig. S5).
Photocatalytic performance. The photocatalytic CH4 oxidation was investigated in a high-pressure stainless-steel vessel with a transmittance quartz glass window on the top plate (Fig. S6). The adoption of high-pressure reactor is based on below two considerations. One is to economically utilize the available pressure energy from the practical transportation and storage condition of natural gas, where the pressure is as high as 70-200 bar in main pipelines and tanks20, 21. The specific analysis is briefly summarized in supplementary information (Fig. S7). The other is to promote the yield of desired products through dissolving more CH4 reactants in the solvent at high pressure. In a typical reaction, 10 mg q-BiVO4 photocatalysts were dispersed in certain amount of H2O, while the total pressure of CH4 and O2 mixed gas was fixed to be 20 bar. After irradiation with Hg lamp (excitation wavelength of 300-600 nm) at room temperature (surface temperature of catalyst = 28.5±0.5oC, Fig. S8), the yield of liquid organic products such as CH3OH and C2H5OH was evaluated by nuclear magnetic resonance spectroscopy (NMR) and gas chromatography (GC) with flame ionization detector. Note that liquid HCHO is known to exist in the form of methanediol (HOCH2OH) in aqueous solution22, which 1H NMR signal overlaps with the broad peak of water solvent. So, the quantity of HCHO was determined by acetylacetone color-developing method (Fig. S9, S10, S11). Other gaseous products (CO2, C2H6, O2 and H2) were monitored by GC spectra with flame ionization and thermal conductivity detectors.
The effect of photocatalyst type, O2 and H2O amount, reaction time, irradiation wavelength as well as light intensity on CH4 conversion performance is investigated in detail. First of all, photocatalytic activity of q-BiVO4 (4.5 nm) is compared with conventional s-BiVO4 (454.3 nm, Fig. S12). As shown in Fig. 3a, about 4 folds in the oxygenated products by q-BiVO4 (2.3 mmol g-1 CH3OH, 1.9 mmol g-1 HCHO and 0.3 mmol g-1 CO2) are acquired with respect to s-BiVO4 (0.5 mmol g-1 CH3OH, 0.5 mmol g-1 HCHO and 0.1 mmol g-1 CO2). The corresponding increment is also distinguished in conversion of CH4 and O2 reactants (Fig. S13a). These enhancements by q-BiVO4 are ascribed to both the raised charge carrier kinetics (Fig. S14) and the improved Brunauer-Emmett-Teller (BET) specific surface area (226.9 m2 g-1, Fig. S15). This deduction is supported by comparing the BET surface area normalized CH4 oxidation activity (Table S1), and q-BiVO4 still exhibits higher performance than s-BiVO4. Except for s-BiVO4, commercial TiO2 nanoparticle (P25) is also employed as the contrast sample and tested under the same reaction condition, and only HCHO (48.9 μmol g-1) and CO2 (48.9 μmol g-1) are found in the products (Fig. S16).
Optimizing the yield of both CH3OH and HCHO is carried out through varying O2 amount (Fig. 3b)12, 23. The highest productivity of CH3OH and HCHO is 2.3 and 1.9 mmol g-1 at the O2 pressure of 10 bar, respectively, and the corresponding conversion percentage of O2 is 0.23% (Fig. S13b). The gradual increase of oxygenates at the O2 partial pressure of less than 10 bar implies that O2 involves the rate-determining step of CH4 oxidation. When the O2 pressure surpasses 10 bar, the reduced productivity of CH3OH and HCHO is ascribed to the lowered CH4 partial pressure and their overoxidation to CO2. Hence, the partial pressure of O2 is controlled to be 10 bar.
The inspection of reaction time discloses that 3 h is the best for the yield of CH3OH (2.3 mmol g-1) though its selectivity continuously decreases (Fig. 3c). It is noticed that when the reaction time arrives 7 h, the productivity of HCHO ascends to 5.6 mmol g-1 with 69.8% selectivity, demonstrating that the prolonged reaction promotes oxidation of both CH4 reactants and CH3OH intermediates. The corresponding conversion percentage of CH4 at 7 h is 0.40% (Fig. S13c) with turnover number (TON) of 2.6. Evidently, extending the reaction time enables improving the HCHO selectivity simultaneously ensuring its output. The balance between the consumption rate of feed gas (CH4 and O2) and the formation rate of products (CH3OH, HCHO and CO2) is explored by Fig. S17 and S18.
H2O acts as the solvent for CH4 oxidation. Fig. 3d shows that with the H2O volume increase from 10 to 80 mL, conversion of CH4 and O2 synchronously improves from 0.23% (TON = 1.5) to 0.54% (TON = 3.5) (Fig. S13d), while the yield of CH3OH and HCHO gets to 6.8 and 4.0 mmol g-1, respectively. More CH4 molecules are dissolved in aqueous solution with H2O amount increasing, leading to the rise of CH3OH product. Moreover, the concentration of CH3OH is diluted in a large volume of H2O, alleviating overoxidation to HCHO and CO2 and causing enhancement of the selectivity towards CH3OH. It is noted that owing to the possible test error in low concentration of CH3OH solution (Fig. S19), further increase of H2O volume is not tried. Nevertheless, augmenting the solvent volume is an efficient method to boost the selectivity of CH3OH with high yield.
The relationship between photocatalytic performance and input photon number or light energy is also examined. As indicated in Fig. 3e, along with the increase of light intensity of Hg lamp, the yield of CH3OH, HCHO and CO2 all raises from 1.0 (CH3OH, 100 mW cm-2), 0.5 (HCHO, 100 mW cm-2) and 0.2 (CO2, 100 mW cm-2) to 2.3 (CH3OH, 170 mW cm-2), 1.9 (HCHO, 170 mW cm-2) and 0.3 mmol g-1 (CO2, 170 mW cm-2), respectively. This result manifests that the photocatalytic CH4 conversion largely depends on the inputted photon number. Fig. 3f and Table S2 further summarize the photocatalytic efficiency under irradiation of monochromatic light with different energy. In good agreement with change in the diffuse reflectance spectrum, the quantum efficiency (Q.E.) value of q-BiVO4 (0, 0, 0.47, 0.82 and 3.07%) increases with the reduction of irradiation light wavelength (630, 535, 470, 420 and 365 nm). The prominent photon utilization efficiency of 3.07% at 365 nm demonstrates that irradiation with the short-wavelength and high-energy light would promote the oxidation conversion of CH4. The quantitative conversion percentage of CH4 and O2 is outlined in Fig. S13e and S13f.
Based on the above results, we conclude the favorable reaction condition for production of HCHO and CH3OH, respectively. Since HCHO is the further oxidation product of CH3OH, increasing the oxidation capacity would be an efficient strategy to elevate the selectivity of HCHO. As demonstrated in Fig. 4a, utilization of short-wavelength UV irradiation (300-400 nm) with high intensity (170 mW cm-2) not only promotes the CH4 conversion but also accelerates the oxidation of CH3OH to HCHO. Significantly, upon a long time oxidation of 7 h, HCHO product is achieved with a good selectivity and yield of 86.7% and 13.1 mmol g-1 (TON = 4.7), respectively. The corresponding conversion of both CH4 and O2 reaches 0.73% (Fig. S13g). On the contrary, reducing the oxidation degree of reaction system could be conducive to production of CH3OH. Using visible Xenon lamp irradiation (400-780 nm), a remarkable 92.8% selectivity of CH3OH is achieved in 10 mL H2O. It needs to be pointed out that the weak oxidation capacity under visible light inevitably results in low productivity of oxygenates. Therefore, by increasing the dissolved CH4 in 80 mL H2O, the yield of CH3OH is promoted to 1.1 mmol g-1 with an exceptional selectivity of 96.6% (TON = 0.4, Fig. 4b, S20). The corresponding conversion of both CH4 and O2 is 0.06% (Fig. S13h). Evidently, the diluted concentration of CH3OH generated in large solvent volume gives rise to the enhanced selectivity from 92.8% to 96.6% through depressing the overoxidation (Fig. S21). The contrast experiments with absence of catalyst, CH4, O2 or H2O show no production of oxygenates under visible light irradiation (Table S3). To preclude the influence of temperature, thermocatalytic reaction (30oC) under similar condition is also conducted and no products is observed (Fig. S22). It deserves to be stressed that the highly selective generation of CH3OH or HCHO from CH4 oxidation over the single photocatalyst is unattained in previous works (Table S4).
Photocatalytic mechanism. Fig. 5A illustrates the proposed radical mechanism for CH4 oxidation on q-BiVO4 (Fig. 5). Under light irradiation, q-BiVO4 is excited to induce •OH generation via two routes: oxidation of H2O by valence band holes and reduction of O2 by conduction band electrons (step 1). The •OH cleaves C-H bond for producing the methyl radical (•CH3) (step 2), which is a rate-determining step. Also as an oxidant, O2 rapidly binds to •CH3 and then reacts with as-formed H+ and electron to generate methylhydroperoxide (CH3OOH, step 3), which will be reduced by electrons23, 24 or decomposed under UV irradiation to form CH3OH (step 4)25, 26. Upon oxidation by holes from valence band of q-BiVO4, as-formed CH3OH is further activated to •CH2OH that is combined with •OH to produce HCHO in the form of HOCH2OH (step 5). In the following parts, we validate this reaction mechanism step by step.
To solidly prove that the oxygenated products results from the conversion of CH4, 13CH4 was used as reactant instead of 12CH4. Gas chromatography-mass spectrometry (GC-MS) with isotopically labelled 13CH4 discloses that CH4 is the carbon source of CH3OH with the appearance of 13CH3OH peak at m/z = 33 (red curve in Fig. 6a). The obvious 13C NMR peaks of CH3OH and HOCH2OH (HCHO) also verify that the C1 oxygenated products are derived from CH4 (black curve in Fig. S23a). The carbon type of C1 oxygenated products is identified from the 13C DEPT-135 (distortionless enhancement by polarization transfer) spectrum, where up and down signals represent -CH3 and -CH2 groups from CH3OH and HOCH2OH (HCHO), respectively (red curve in Fig. S23a). Furthermore, as displayed in Fig. S23b, satellite peaks of 13CH3OH (δ = 3.12 and 3.40 ppm) are discerned in 1H NMR spectrum without 12CH3OH (δ = 3.28 ppm) signal23, indicating that CH4 is the 100% carbon source for CH3OH product. Similarly, 2D 1H-13C HMQC (heteronuclear multiple-quantum correlation, Fig. S24) and 2D 1H-13C HMBC (heteronuclear multiple-bond correlation, Fig. S25) spectra indicate that 13CH4 is the feedstock for HCHO (HOCH2OH) generation.
According to the mechanism illustration, •OH is responsible for removing the H atom from CH4 molecule, which is the key step for CH4 activation. To explore the ability of q-BiVO4 towards the generation of •OH in step 1, its band structure is established through UV-Vis diffuse reflectance spectrum27 (Fig. S26a), transformed Kubelka-Munk function plot (Fig. S26b)28, Mott-Schottky plot (Fig. S26c, S27)29 and ultraviolet photoelectron spectroscopy (Fig. S28). As shown in Fig. 6b, the conduction and valence bands of q-BiVO4 are tested to be 0.075 and 2.555 V vs. normal hydrogen electrode (NHE) at pH=0, respectively. Therefore, the •OH can be produced through two approaches: H2O oxidation (•OH/H2O: 2.380 eV vs. NHE at pH=0)30 and two-electron O2 reduction (O2/•OH: 0.695 eV vs. NHE at pH=0)31. In order to assess the generation of •OH from both hole oxidation and electron reduction of q-BiVO4, we carried out the fluorescence detection of coumarin solution without or with O2 (Fig. S29). In the absence of O2 (Fig. S29a), an enhanced fluorescence signal is achieved with q-BiVO4 photocatalyst under visible light, indicating that the holes from valence band can successfully oxidize H2O into •OH. After the solution is saturated with O2 (Fig. S29b), the fluorescence intensity with q-BiVO4 exhibits 2.3 times increase than that in absence of O2, confirming that O2 also greatly contributes to the generation of •OH. Note that the quantity of O2 solely generated from H2O decomposition by q-BiVO4 is too low to make a detectable yield of oxygenates (Fig. S30). Moreover, the energy band structure of s-BiVO4 is likewise constructed (Fig. S31 and S32) with the conduction and valence bands at 0.084 and 2.474 V vs. NHE at pH=0, respectively. With respect to the energy band structure of s-BiVO4, the more negative conduction band and positive valence band of q-BiVO4 lead to more •OH species for CH4 oxidation (Fig. S33). We notice that such fluorescence signal comes from 7-hydroxycoumarin that is the •OH trapping product of coumarin (Fig. S34 and S35).
Step 2 refers to the activation step of CH4 by •OH, which is recognized as the rate-determining step in this work. To verify that •OH is the initiator for CH4 activation on the surface of q-BiVO4 rather than the photoinduced hole, we performed a thermocatalytic reaction containing 10 mg q-BiVO4, 5 mL H2O2, 5 mL H2O, 10 bar O2 and 10 bar CH4 at 60oC in absence of light irradiation. The H2O2 is decomposed to provide •OH at 60oC, and the dark condition prevents the hole formation on the surface of q-BiVO4. After reaction, CH4 is found to be oxidized to CH3OOH, which is not further reduced to CH3OH due to the lack of photogenerated electrons under dark condition (Fig. 6c). In absence of H2O2 (Fig. S36) or q-BiVO4 (Fig. 6d), no oxygenated product is distinguished. Thus, we deduce that •OH activates CH4 for oxygenate generation on the surface of q-BiVO4. Besides, the solid evidence that CH4 is not directly oxidized by photoinduced holes from q-BiVO4 is supported by electron spin resonance (ESR) spectroscopy (Fig. 6e). Under Xenon lamp irradiation in argon (Ar) atmosphere, the ESR spectrum of q-BiVO4 shows a g value of 2.00632, corresponding to the active hole center O- (Fig. 6e, blue curve); subsequently, atmosphere is switched from Ar to CH4, no distinct intensity decrease is discerned on O- signal (Fig. 6e, green curve), indicating no occurrence of reaction between O- hole and CH4 molecule33; on the contrary, upon continuous irradiation for 15 min in CH4 atmosphere, the intensity of O- signal increases (Fig. 6e, red curve), again denying that the O- hole can directly oxidize CH4. The fact that C-H bond breakage is rate-determining step of CH4 oxidation was verified through kinetic isotope effect (KIE)34. Taking 420 nm monochromatic light for irradiation, the KIE value is estimated to be 8.3 (larger than 6)35 based on the reaction rate ratio (kH / kD) using CH4 or CD4 as reactants, respectively (Fig. 6f and S37). This large value reveals the C-H bond cleavage in the rate-determining step.
Incorporation of O2 with •CH3 to form CH3OH in step 3 was proved by 18O2 and H218O isotope tests as well as the reusability of q-BiVO4 photocatalyst, considering that these three oxygen sources possibly involved the catalytic reaction. GC-MS analyses (Fig. 6g) demonstrate that CH318OH fragment occupies 100% intensity by using 18O2 and H216O, whereas the CH316OH fragment contains 100% intensity by using 16O2 and H218O. Therefore, one can deduce that the O element in H2O is not incorporated into CH3OH formation during the photocatalytic CH4 oxidation. To exclude the possibility of lattice O in q-BiVO4 as oxygen source, its stability was evaluated. After five photocatalytic cycles, q-BiVO4 almost preserves its initial catalytic activity (Fig. S38). Therefore, O2 is the mere oxygen source for CH3OH formation (step 3). TEM images (Fig. S39), XRD patterns (Fig. S40), XPS (Fig. S41), Raman (Fig. S42a), UV-Vis diffuse reflectance spectra (Fig. S42b) and inductively coupled plasma optical emission spectrometer tests (Fig. S43) were carried out to prove only slight change in the morphologies and chemical states of q-BiVO4 before and after photocatalytic reaction.
According to step 4, CH3OOH is first formed and then reduced to CH3OH. Since this reduction process easily happens under high photoinduced electron density36, the NMR signal of CH3OOH is difficult to be detected. Therefore, we increased the amount of q-BiVO4 (50 mg) and reduced the amount of H2O solvent (0.5 mL) in order to increase the concentration of CH3OOH intermediates. By prolonging the reaction time to 7 h, a weak CH3OOH peak appeared at 3.78 ppm, providing the clear evidence on CH3OOH production (Fig. 6h). As-formed CH3OH would be further oxidized to HCHO (HOCH2OH) in step 5. Based on the Gibbs free energy values (Fig. 1)22, 37, 38, oxidation of CH3OH to HCHO is thermodynamically favorable. To confirm this transformation process from the experimental aspect, CH3OH was taken as reactants in a similar photocatalytic system with argon (Ar) replacing CH4. The 13C NMR spectrum of CH3OH oxidation products verifies that CH3OH can be oxidized to HOCH2OH using q-BiVO4 as catalysts under light illumination (Fig. S6i).
Summary
Through regulation of reaction time, irradiation wavelength, light intensity and the adding amount of O2 as well as H2O, selective oxidation of CH4 into CH3OH and HCHO with satisfied yield is successfully realized over q-BiVO4 photocatalyst at room temperature under the aerobic condition. The quantum sized q-BiVO4 with an appropriate energy band structure enables simultaneous H2O oxidation via photogenerated holes and O2 reduction via photogenerated electrons, greatly promoting the production of •OH. As-formed •OH causes the cleavage of C-H bond in CH4 and gives to the generation of •CH3, which is proved to be a rate-determining step. Except for participating into •OH production, O2 is found to incorporate with •CH3, resulting in the oxygenated products. This work not only provides a comprehensive analysis for CH4 oxidation, but also broadens the avenue toward selective conversion of CH4 into valuable products in sustainable way.