Cu/ZnV2O4 Heterojunction Interface Promoted Methanol and Ethanol Generation from CO2 and H2O under UV-Vis Light Irradiation.

Adopting the concurrent reduction of Cu2O during hydrothermal preparation of ZnV2O4, metal-semiconductor heterojunction Cu/ZnV2O4 nanorods were synthesized and applied to the catalytic generation of methanol and ethanol from CO2 aerated water under UV-vis light irradiation. 10Cu/ZnV2O4 obtained from 10 wt % composite amount of Cu2O exhibited a total carbon yield of 6.49 μmol·g-1·h-1. The yield of CH3OH and C2H5OH reached 3.30 and 0.86 μmol·g-1·h-1, respectively. 2.5Cu/ZnV2O4 displayed the highest ethanol yield of 1.58 μmol·g-1·h-1 due to the strong absorption in the visible light. Cu/ZnV2O4 was characterized using X-ray diffraction (XRD), scanning transmission electron microscopy (STEM), X-ray photoelectron spectroscopy (XPS), ultraviolet-visible (UV-vis) spectra, photoluminescence (PL) spectra, transient photocurrent response, and electrochemical impedance spectroscopy (EIS). Results showed that composite Cu0-ZnV2O4 increased the surface area and tuned the energy band position, which matches the reaction potential toward methanol and ethanol. The photocatalytic activity toward CH3OH and C2H5OH on Cu/ZnV2O4 is attributed to faster transmission and a slow recombination rate of photogenerated carriers at the heterojunction interface. Multielectron reactions for the production of CH3OH and C2H5OH are promoted. Free radical capture experiments indicated that the active species boost the reaction in the order of •OH > e- > h+.

Introduction

Photocatalytic transformation of CO2 and H2O to methanol and ethanol is a desired reaction. On the one hand, the excessive emissions of atherogenic CO2 urgently need reduction/circulation to curb the greenhouse effect. On the other hand, methanol can be a viable energy carrier[1] powering the future with liquid sunshine.[2]

However, quantum yield and product selectivity in the photocatalytic reduction of CO2 remain challenging. To harvest as much sunlight as possible and slow down recombination of photogenerated carriers, several strategies for catalyst construction have been developed, for instance, crystal-facet engineering,[3] cocatalyst modification,[4] and heterostructure engineering.[5−7] It has been recognized that the interfacial effect among semiconductors and cocatalysts can accelerate the carrier separation.[8]

Due to suitable band structure and availability, zinc vanadate has been employed in photodegradation,[9,10] batteries,[11,12] and CO2 reduction.[13] ZnV2O6/g-C3N4 heterojunction with a 2D/2D interface increased CH3OH formation[13] due to g-C3N4 being used as a mediator. ZnV2O7 as the hole reaction site in TiO2/vanadate suppressed the recombination and increased the catalytic activity toward CH4.[14]

Spinel ZnV2O4, an n-type semiconductor, is composed of a ZnO4 tetrahedron and a VO6 octahedron. 3D VO2/ZnV2O4 with favorable morphology and hierarchical pores promoted photogeneration of CH3OH, as well as CO and CH4 from CO2 and H2O, in the gas–solid reaction condition, due to efficient separation of carriers.[15] Nonetheless, the preparation process of ZnV2O4 is difficult to control because of the multiple oxidations of VO3–, and the VO or ZnO impurity is easily formed.

As far as the metal cocatalysts are concerned, copper was reported to be active for photoreduction formation of CH3OH from CO2 and H2O.[16] The valence state of the Cu active site has been studied. A study reported that Cu1+ is active in enhancing a multielectron photoreaction.[17] The Cu2O (110) crystal plane enabled CO2 to be converted to •CO2, which increases the photoreduction efficiency, whereas the Cu2O (100) plane was inert.[18] Supported Cu2O promoted photoinduction efficiency under visible light.[19−22] Cu-modified TiO2 exhibited light olefin selectivity of 60.4% at 150 °C, attributed to the Cu+ species for C–C coupling.[23] Another study found that there exists a synergistic effect between Cu1+ and Cu2+. Z-scheme Na2Ti6O13/CuO/Cu2O with more Cu2O favored H2 evolution, whereas that with more CuO favored CH2O and CH3OH formation due to the coupled band gaps.[24] Furthermore, the synergy between the outside Cu and the inside Cu+ in Cu/Cu+@TiO2 enriched electrons for CO2 reduction to CO and CH4.[25] Cu on TiO2 nanosheet could activate CO2 when Cu was oxidized.[26] Cu in Pd matrix forming Cu-Pd sites improved the CO2 activation.[27] But excess Cu became new recombination centers of electron–holes.[28]

Although metal can enrich electrons in metal–semiconductor heterojunctions and become the active sites for CO2 reaction,[29] the performance of Cu/ZnV2O4 for the catalytic generation of methanol and ethanol from an aqueous solution of CO2 under UV–vis light irradiation has rarely been probed to our knowledge.

In this work, a two-step synthesis strategy was employed to obtain Cu/ZnV2O4. Cu2O was first synthesized and then added into the hydrothermal synthesis liquid of ZnV2O4. The Cu0 was obtained by the reduction of vanadate. Characterizations indicated that a metal–semiconductor heterojunction was formed at the Cu0-ZnV2O4 interface and the as-constructed interface was active toward generating methanol and ethanol from an aqueous solution of CO2 under UV–vis irradiation. The photocatalytic functionality was discussed.

Results and Discussion

Photocatalytic Activity

The carbon product distribution of the photocatalytic CO2 reaction with H2O is shown in Table and Figure . The products included CH4, CO, CH3OH, and C2H5OH. O2 was detected but was not quantified. From Table , the total carbon (TC) yields on the composite samples were higher than those on ZnV2O4. 10Cu/ZnV2O4 showed the highest TC and CH3OH yields. 2.5Cu/ZnV2O4 displayed the highest ethanol yield. The results indicated that Cu/ZnV2O4 promotes the generation of methanol and ethanol, particularly the ethanol yield with the C–C bond formation, and this promotion is affected by the Cu amount and the UV–vis light response (see Figure ). The methanol and ethanol selectivity showed a slight decrease because of the increase in CO (Figure ). This can be attributed to its light capture of a wider response, higher surface area, and higher conduct band position than E(COθ and E(COθ. Composite Cu/ZnV2O4 formed a Mott–Schottky heterojunction interface, promoting photogenerated carrier separation and trapping photogenerated electrons to improve the 6e and 12e transfer reactions to obtain methanol and ethanol (see subsequent characterizations). The generation of methanol and ethanol can be facilitated by electron enrichment on the Cu surface through CO2•– and *CO formation, *CO polymerization, and formation of the C–C bond.[30,31] The combination of an appropriate amount of Cu with ZnV2O4 can capture and migrate photogenerated electrons and increase active sites. But excessive metals form new recombination centers inhibit the interfacial charge transfer and abate the activity.[28,32]

Table 1

Photocatalytic Activity on Synthesized Samplesa

	yield (μmol·g–1·h–1)
samples	TC	CH3OH	C2H5OH	CH3OH + C2H5OH sel. (%)
ZnV2O4	3.23	2.01	0.46	77.18
1.25Cu/ZnV2O4	5.08	2.20	1.31	69.09
2.5Cu/ZnV2O4	5.76	2.80	1.58	76.04
5Cu/ZnV2O4	5.81	2.78	0.69	59.72
10Cu/ZnV2O4	6.49	3.30	0.86	64.10
20Cu/ZnV2O4	3.37	2.19	0.39	76.56

Testing conditions: 40 mg of catalyst, 80 °C, 0.2 MPa, and 4 h of irradiation.

Figure 2

Change in alcohol yield with irradiation time. Testing conditions: 40 mg of 10Cu/ZnV2O4, 80 °C, 0.2 MPa, and separate irradiation time.

Figure 8

Diffuse reflectance UV–vis spectra of synthesized samples.

Figure 1

Gas products on Cu/ZnV2O4 with different Cu contents. Testing conditions: 40 mg of catalyst, 80 °C, 0.2 MPa, and 4 h of irradiation.

Figure shows the gas product distribution on Cu/ZnV2O4 in varied composite contents of Cu2O. As the Cu content was increased, the CH4 yield decreased whereas the CO and H2 yields were maximum and minimum, respectively, implying the competition of electrons between CO2 reduction and H2 evolution during the reaction. When the Cu content reaches 20 wt %, Cu on the surface of ZnV2O4 may form a new recombination center due to excessive combination. In this case, it is difficult for the multielectron reduction of CO2 to occur. The 2e reaction dominates, and the H2 yield increases.

The change in the alcohol yield with the irradiation time of 8 h is shown in Figure . The CH3OH yield increased upon extending the irradiation time. The C2H5OH yield decreased after an increase. Methanol and ethanol generation became slow after 4 h at the set experimental conditions.

To understand the contribution of light, an activity test under visible light irradiation (λ ≥ 420 nm) was conducted, and the result is shown in Figure S2. 10Cu/ZnV2O4 showed photocatalytic activity for the objective reaction under visible light irradiation, with the TC yield of 4.5 μmol·g–1, indicating the response of Cu/ZnV2O4 to visible light. These results demonstrated that composite Cu/ZnV2O4 enhanced alcohol generation from CO2 and H2O under UV–vis irradiation.

Structure, Morphology, and Surface Area

In Figure , the diffraction peaks at 2θ of 18.3, 30.2, 35.7, 56.7, and 62.4° were observed, corresponding to the (111), (220), (311), (400), (422), and (440) crystal planes of ZnV2O4, respectively. In 2.5Cu/ZnV2O4, as the Cu amount was increased, Cu/ZnV2O4 started to produce diffraction peaks at 2θ of 43.2, 50.4, and 74.1° ascribed to (111), (200), and (220) planes of crystalline Cu, respectively (PDF# 99-0034). The intensity of Cu peaks was gradually increased. Composite metal–semiconductor Cu/ZnV2O4 was obtained using a two-step procedure.

Figure 3

XRD patterns of xCu/ZnV2O4.

Synthesized ZnV2O4 emerged as well-distributed nanostrips with a diameter of ca. 150 nm and rough steplike surfaces (Figure a).[13] 2.5Cu/ZnV2O4 was shaped as uniform nanorods (Figure b) with Cu dots (Figure c), indicating that Cu disperses on the surface of ZnV2O4 nanorods uniformly. The diffraction bright spots in the SAED photographs (Figure d) revealed the composite polycrystalline structure. The d-spacings of ca. 0.21 and 0.48 nm were ascribed to the (400) and (111) planes of ZnV2O4,[12] respectively. The d-spacing of ca. 0.22 nm was ascribed to the Cu (004) plane.[30] The heterojunction interface in Cu-ZnV2O4 was constructed successfully.

Figure 4

Scanning electron microscopy (SEM) and TEM photographs of synthesized samples (a) ZnV2O4 and (b–d) 2.5Cu/ZnV2O4.

The high-angle annular dark-field (HAADF) TEM photographs and the EDX mapping in Figure further exhibited the nanorod morphology of 2.5Cu/ZnV2O4 and the Zn, O, V, and Cu distribution.

Figure 5

TEM photographs and elemental distribution for 2.5Cu/ZnV2O4: (a) HAADF image, (b–e) EDX mapping, and (f) EDX spectrum of 2.5Cu/ZnV2O4.

Synthesized samples presented N2 adsorption–desorption isotherms of type IV (Figure ). From P/P0 > 0.8, the adsorption volume increased due to the capillary coalescence and the appearance of the hysteresis loop of H3-type, indicating the mesopores of 2–20 nm (Figure , inset). The surface area and pore volume of 10Cu/ZnV2O4 increased by 2.39 and 1.73 times compared to those of ZnV2O4, respectively.

Figure 6

N2 adsorption–desorption isotherms of synthesized samples.

Optical Properties

The surface elemental composition and valence state of 2.5Cu/ZnV2O4 were analyzed using XPS. The full spectrum revealed the electron binding energy (B.E.) of Zn 2p, O 1s, and V 2p (Figure S3). As shown in Figure a, the B.E. peaks at 933.27 and 932.2 eV are attributed to 2p3/2 of Cu0 and Cu1+, respectively (due to part oxidation in air).[33,34] The four B.E. peaks of 523.96 and 522.80 eV and 516.80 and 515.70 eV in Figure b are attributed to V 2p1/2 and V 2p3/2, respectively. The B.E. peaks of V 2p3/2 at 516.80 and 515.70 eV are ascribed to V5+ and V3+, respectively,[9,12] indicating that the mixed state of V5+ and V3+ exists.[35,36] The B.E. peaks at 1044.98 and 1021.89 eV are attributed to Zn 2p1/2 and Zn 2p3/2, respectively, with a spin–orbit coupling of 23.15 eV, demonstrating that Zn exists as Zn2+ in the Cu/ZnV2O4 heterostructure; 531.83 and 529.67 eV are attributed to O 1s of O2–, corresponding to the adsorbed oxygen and lattice oxygen, respectively. Compared with ZnV2O4, the B.E. of Zn, V, and O in 2.5Cu/ZnV2O4 exhibited a slight shift suggesting strong interaction between Cu and ZnV2O4 and different coordination environments associated with the changed Fermi energy level and orbital electron energy.[37]

Figure 7

XPS spectra of synthesized samples (a) Cu 2p and (b) V 2p.

The diffuse reflectance UV–vis absorption spectra of synthesized samples are shown in Figure . Strong absorption in the visible region (>420 nm) was observed. No drop absorption edge from about 700 nm was found for ZnV2O4. Cu/ZnV2O4 showed evidently an absorption edge from about 700 nm ascribed to the Schottky effect. The Mott–Schottky heterojunction interface can accelerate charge transfer, improving the catalytic activity.[29] The enhanced response of Cu/ZnV2O4 to visible light with an increased Cu amount is attributed to the rearrangement of the electron cloud caused by the sp hybridization at the Cu-ZnV2O4 heterojunction interface, which changes the Fermi energy level and narrows the apparent band gap.[24] 1.25Cu/ZnV2O4 and 2.5Cu/ZnV2O4 showed strong absorption intensity in the visible light region promoting the ethanol yield, implying that the light response is partly a factor for catalytic activity.

As shown in Figure , the band gap energy (Eg) for ZnV2O4 was 2.95 eV, similar to that reported.[37] The Eg of 2.5Cu/ZnV2O4 was decreased by 0.21 eV compared with that of ZnV2O4, attributed to the change in the Fermi energy level of composite Cu-ZnV2O4.

Figure 9

Fitted curves for UV–vis spectra.

Figure shows the valence band (VB) position of ZnV2O4 measured by the XPS valence band spectrum.[13] The VB top position was 2.51 eV, and the conduction band (CB) bottom position was calculated as −0.44 eV from Eg = EVB – ECB.

Figure 10

XPS VB spectra of ZnV2O4.

Optoelectronic Performance

The photoluminescence (PL) spectra of synthesized samples are shown in Figure . ZnV2O4 produced PL peaks at 380, 460, and 617 nm. Compared with ZnV2O4, Cu/ZnV2O4 showed a higher PL peak intensity at around 380 nm but a lower PL peak intensity at around 460 and 617 nm. As the Cu amount was varied, the PL intensity for Cu/ZnV2O4 changed. The Cu amount affected the photoelectron transmission evidently. 10Cu/ZnV2O4 showed a lower PL intensity than 1.25Cu/ZnV2O4. The PL intensity of 20Cu/ZnV2O4 increased again. The PL data suggested that the recombination of photogenerated electron–holes at around 460 and 617 nm is abated due to the Cu-ZnV2O4 m–s heterojunction. An appropriate Cu amount benefits the photopromotion activity toward methanol and ethanol.[17,28]

Figure 11

Photoluminescence spectra of synthesized samples.

The heterojunction interface separates the photogenerated electron–holes effectively and prolongs their lifetime, in accordance with the strong photocurrent response in Figure . No peak at 617 nm was observed for Cu/ZnV2O4 owing to the fast separation of photogenerated electron–holes under visible light irradiation. 10Cu/ZnV2O4 presented the lowest recombination rate of photogenerated electron–holes, which provides more photogenerated electrons and facilitates multielectron reactions for methanol and ethanol generation.

Figure 12

Transient photocurrent response curves of samples (a) 5Cu/ZnV2O4 and (b) ZnV2O4.

Transient photocurrent response was used to characterize the photogenerated carrier density. A higher photocurrent density indicates faster migration of carriers. As shown in Figure , 5Cu/ZnV2O4 presented a current density 50 times higher than that of ZnV2O4, attributed to the Cu0-ZnV2O4 interface, which accelerates the migration of photogenerated electrons, resulting in more photogenerated electrons. When light irradiation was turned on again, the photocurrent response increased rapidly and the photogenerated carriers emerged again. The recombination rate of photogenerated carriers for Cu/ZnV2O4 was lower than that of ZnV2O4, demonstrating that composite Cu/ZnV2O4 can effectively suppress the recombination rate. In addition, the decrease in photocurrent density suggested the instability in the photoperformance of the catalyst.

EIS was used to characterize the resistance properties of synthesized samples. A small impedance semicircle radius stands for lower resistance to charge transfer.[38] From the Nyquist curve in Figure , the radius of the impedance semicircle of 5Cu/ZnV2O4 significantly became small, suggesting that the resistance to charge transfer in the Cu-ZnV2O4 interface was quite low. This result confirmed that composite Cu can reduce resistance and favor charge transfer, facilitating the CO2 reduction reaction, consistent with the activity in Table .

Figure 13

EIS plots of synthesized samples.

Mechanism

To understand the reaction process, free radical capture experiments were conducted for the reaction on 10Cu/ZnV2O4 under UV–vis light irradiation. Triethanolamine (TEOA), tert-butanol (TBA), and potassium dichromate (K2Cr2O7) were added into the reaction solution to capture photogenerated holes (h+), hydroxyl radicals (•OH), and photogenerated electrons (e–) during the reaction,[37] respectively. From Figure , addition of TEOA caused an increase in CH3OH yield by two times, implying that reduction of h+ conduces generation of CH3OH. No products were detected when TBA was added, indicating that •OH is essential to the reaction. In the case of adding K2Cr2O7, a low amount of gas products was detected without CH3OH, suggesting that reduction of e– is adverse to generation of CH3OH. These data indicated that the active species enhance the reaction in the order of •OH > e– > h+.

Figure 14

Free radical capture experimental conditions: 40 mg of 10Cu/ZnV2O4, 80 °C, 0.2 MPa, and 4 h of irradiation.

The aforementioned PL spectra, transient photocurrent response, and EIS characterizations confirm the formation of photogenerated carriers from composite Cu/ZnV2O4, as shown in eqs and 2.

The calculated CB position for ZnV2O4 (−0.44 eV) is lower than the reduction reaction potential of E(CO2/CH3OH) = −0.38 eV, E(CO2/C2H5OH) = −0.35 eV, and E(CO2/CH4) = −0.25 eV. The VB position of ZnV2O4 (2.51 eV) is higher than the oxidation reaction potential of E(H2O/•OH) = 2.31 eV, theoretically matching the redox requirement of CO2 or H2O.

Under UV–vis light irradiation, electrons transit from VB to CB, and the Cu-ZnV2O4 heterojunction interface promotes electron transport. Some excited electrons of high energy migrate through the Schottky barrier to the Cu surface and react with CO2 to generate the products. The energy band bending induces generation of CO (E(CO2/CO) = −0.52 eV). The occurrence of the photocatalytic reaction was proposed as shown in Figure .

Figure 15

Electron transfer and photoreaction on Cu-ZnV2O4.

The electrons (e–) in the CB of ZnV2O4 migrate to the Cu interface, reducing CO2 aerated H2O to CO, CH3OH, CH4, and C2H5OH.

The holes (h+) left in the VB of ZnV2O4 promote oxidation of H2O. The produced •OH and H+ participate in the CO2 reduction, as shown in eqs –7.

Conclusions

For photogeneration of methanol and ethanol from CO2 and H2O, composite metal–semiconductor Cu/ZnV2O4 heterojunction nanorods were successfully synthesized by a two-step hydrothermal reduction method. Cu0-ZnV2O4 improves the visible light harvest and surface area and results in a faster transport of photogenerated electrons, which can be tuned through varying the composite amount of Cu2O. The energy band position of the Cu0-ZnV2O4 heterojunction matches with the photoreaction to methanol and ethanol. Cu0 on the surface of ZnV2O4 not only increases the active sites but also accelerates electron transfer in the Cu0-ZnV2O4 heterojunction interface. An increased number of effective photogenerated electrons promotes multielectron reactions for methanol and ethanol generation. With 10Cu/ZnV2O4, the total carbon yield for the photocatalytic reduction of CO2 aerated H2O was 25.96 μmol·g–1 under UV–vis light irradiation for 4 h. The selectivity of CH3OH and C2H5OH reached 50.85 and 13.25%, respectively. 2.5Cu/ZnV2O4 showed the highest ethanol yield of 1.58 μmol·g–1·h–1. The promotion is affected by the Cu amount and the UV–vis light response. •OH and e– were proved to be active intermediate species during the formation of methanol and ethanol. This provides new insights into the heterojunction interface in Cu/ZnV2O4 for CO2 photoreduction in H2O to methanol and ethanol.

Experimental Section

Catalyst Preparation

Cu2O was prepared using hydrothermal reduction in a N2 atmosphere. Briefly, 0.2 g of Cu(CH3COO)2·H2O was dissolved in 60 mL of deionized water in a three-neck flask under stirring for 30 min. Then, 0.87 g of sodium dodecyl dimethyl sulfate was added, and the solution was continuously stirred for 20 min. Next, 2.6 mL of NaOH solution (1 M) and 24 mL of NH2OH·HCl (0.1 M) were added rapidly and stirred vigorously for 1 h. The product was washed with deionized water to adjust the pH to 7 and dried at 80 °C in vacuum for 12 h. Brick-red Cu2O powder was obtained.

ZnV2O4 was synthesized using a solvothermal procedure. Briefly, 0.002 mol of NH4VO3 and 0.006 mol of H2C2O4·2H2O were dissolved in 40 mL of N,N-dimethylformamide and stirred for 30 min. Then, 0.001 mol of Zn(CH3COO)2·2H2O was added and continuously stirred for 30 min. The mixture was transferred to an 80 mL Teflon-lined autoclave, heated at 180 °C for 24 h, and then cooled to room temperature. The product was washed with deionized water and ethanol, dried at 80 °C in vacuum for 12 h, and calcined at 550 °C for 5 h. Black ZnV2O4 powder was obtained.

For Cu/ZnV2O4 preparation, Cu2O with the desired amount was dispersed in 40 mL of N,N-dimethylformamide, and the other steps were the same as those for the synthesis of ZnV2O4. The composite content of Cu2O was 1.25, 2.5, 5, 10, and 20 wt % of ZnV2O4, and the resulting samples were denoted as 1.25, 2.5, 5, 10, and 20Cu/ZnV2O4, respectively.

Characterization

The phase analysis was carried out on a D8 ADVANCE A25 with Cu Kα (λ = 0.1542 nm) at 40 kV/40 mA. Surface area and pore distribution were determined using N2 adsorption on a JWBK132F instrument. The pore distribution was calculated using the BJH model. Microscopic morphology was observed on a Zeiss Merlin Compact scanning electron microscope at 520 kV using gold sputtering samples or a Talos F200i FETEM at 200 kV using ethanol-dispersed samples. Surface species were analyzed on a Thermo Scientific photoelectron spectrometer with Al Kα and C 1s (284.6 eV) correction. The photoluminescence spectra were recorded on an Edinburgh FLS1000 at an excitation wavelength of 325 nm.

The diffuse reflectance absorption spectra were recorded on a Shimadzu UV-3600 UV–vis–NIR spectrophotometer. The band gap energy (Eg) was calculated according to the Kubelka–Munk equation (αhν)1/ = A(hν – Eg), where α is the absorption coefficient and n = 1/2 for Cu/ZnV2O4. Specifically, Eg was obtained from the intersection of the tangent of the starting curve with the hν axis.[20]

The transient photocurrent response and electrochemical impedance spectroscopy (EIS) were measured on a CHI 760E electrochemical workstation. Working electrode: 3 mg of the sample was placed in a centrifuge tube, and then 200 μL of ethanol and 10 μL of 0.5% Nafen solution were added and ultrasonically dispersed for 1 h. Ag/AgCl was used as the reference electrode. A Pt electrode was the counter electrode, and 0.5 mol/L Na2SO4 was the electrolyte solution.

Photocatalytic Reduction Activity Test

A reactor of 100 mL was used with a quartz window and a 300 W UV–vis xenon lamp on the top using a current of 15 A as shown in Figure .

Figure 16

Schematic of the photoreactor for CO2 and H2O.

Briefly, 40 mg of catalyst and 40 mL of an aqueous solution of NaOH (0.1 M) and Na2SO3 (0.1 M) were added to the reactor. The reactor was replaced by CO2 three times, switched to CO2 (20 mL/min) for 0.5 h, and then pressed to 0.2 MPa. Photoreduction proceeded under irradiation for set reaction times.

Gas products were analyzed on a GC9560 gas chromatograph with a 5A molecular sieve column (3 m × 3 mm). Liquid products were analyzed on another GC9560 with an FFAP column (30 m × 0.25 mm × 0.25 μm). The qualitative analysis of methanol and ethanol in the liquid products was performed using GC–MS spectra (Figure S1). The results were calculated as Yi (yield, μmol·g–1·h–1) = ni/m/t, where ni, m, and t are product amount (μmol), catalyst mass (g), and reaction time (h), respectively. TC (total carbon yield, μmol·g–1·h–1) = YCO + YCH + YCH + YC. Si (product selectivity, %) = (Yi/TC) × 100.