Valence Regulation of Ultrathin Cerium Vanadate Nanosheets for Enhanced Photocatalytic CO 2 Reduction to CO

: Atomic valence state regulation is an advantageous approach for improving photocatalytic efﬁciency and product selectivity. However, it is difﬁcult to precisely control the ratio of the different valence states on the surface and the relationship between the surface valence change and catalytic efﬁciency in the photocatalytic reaction process is unclear. Herein, CeVO 4 ultrathin nanosheets were fabricated by one-step solvothermal method with ethanolamine (MEA) as the structure-directing agent. The ratio of the concentrations of intrinsic Ce 4+ and Ce 3+ ions is precisely modulated from 19.82:100 to 13.33:100 changed by the volume of MEA added without morphology modiﬁcation. The photocatalytic efﬁciency increases as the concentrations of intrinsic Ce 4+ and Ce 3+ ions decrease and CV3 (prepared with 3 mL of MEA) shows the highest CO generation rate approximately 6 and 14 times larger than CV (prepared without MEA) and CV1 (prepared with 1 mL of MEA), respectively, in the photocatalytic CO 2 reduction. Interestingly, about 6.8% photo-induced Ce 4+ ions were generated on the surface of the catalysts during the photocatalytic CO 2 reduction without any phase and morphology changes for CV3. The photocatalytic reaction mechanism is proposed considering the intrinsic and photo-induced Ce 4+ ions to obtain efﬁcient photocatalysts.


Introduction
Photocatalytic conversion of CO 2 to a specific product is highly important for the use of solar-driven deoxidation and hydrogenation reaction of carbon dioxide to lower CO 2 concentration in an environmentally friendly manner [1][2][3]. Recently, various vanadate compounds have been investigated as photocatalysts for CO 2 reduction, including InVO 4 (E g : 2.7 eV, CO generation rate: 23.8 µmol g −1 h −1 ) [4], Co-doped BiVO 4 (E g :~2 eV, CO generation rate: 18.28 µmol g −1 h −1 ) [5], and LaVO 4 (E g : 2.07 eV, C 2 H 5 OH generation rate: 12.7 µmol g −1 h −1 ) [6]. Among these vanadates, CeVO 4 composed of the rareearth Ce element and ortho-vanadate group is a promising photocatalytic candidate in heterogeneous reaction. In particular, tetragonal CeVO 4 has a zircon structure in which [CeO 9 ] polyhedra are connected through edge sharing to the alternating [VO 4 ] tetrahedra along the c axis, and has a low surface energy that facilitates the natural formation of bulk nanostructured materials [7]. As a ternary metal oxide catalyst, CeVO 4 shows superior redox and optical properties due to the 4f electronic structure and abundant electronic transitions of its constituent lanthanide metal. With these promising characteristics, the photocatalytic efficiency of CeVO 4 nanostructures reported is supposed to be improved for application to CO 2 conversion.
Cerium orthovanadate was first reported in the application for photocatalysis in 2007 by Row et al. and it can be applied to photocatalytic reaction because of its advantageous bandgap compared to PrVO 4 and NdVO 4 [8]. The conduction band of CeVO 4 is lowered by excitations from the partially filled 4f orbitals to obtain an effective bandgap of 1.8 eV (band edge:~680 nm) [9]. The broad light absorption enabled by the low band gap favors the photocatalytic reaction driven by absorption of UV and visible light [10,11]. Ce 4+ and Ce 3+ ions coexisting on the surface of CeVO 4 nanostructures act as the main catalytic centers, and play the key role in the enhanced photocatalytic activity, as demonstrated by theory and experiments [12,13]. Peng et al. reported that Ce 3+ ions can be oxidized into Ce 4+ upon the substitution of Ca 2+ into CeVO 4 and this chemically driven charge transfer is irreversible due to the accompanying lattice contraction [14]. The intrinsic Ce 4+ ions act as recombination centers of photogenerated charge carriers and have a deleterious effect on the catalytic reaction; however, Ce 4+ ions can be changed into Ce 3+ ions by a post-treatment process. It was found that conversion between Ce 3+ and Ce 4+ (denoted as Ce p 3+ /Ce p 4+ ) driven by solar light or reactive intermediates is reversible [15] but the charge transfer mechanism is unclear. The relationship between the Ce p 3+ /Ce p 4+ redox states and the catalytic activity must be understood for the further improvement of the photocatalytic performance of CeVO 4 nanostructures; however, this relationship has rarely been examined in the studies reported in the literature.
Recent reports have demonstrated that the Ce oxidation state is closely correlated with the local structure and chemical environment of the Ce atom. Localization of Ce 4f electrons and the energy levels difference between the V 3d and Ce 4f orbitals generally tends to reduce Ce 4+ to Ce 3+ [16,17]. The V 5+ species (3d states) remain stable in this redox cycle most likely due to its higher 3d energy levels that are form the bandgap states together with the O 2p states of the VO 4 3− ions [18,19]. Based on the Ce p 3+ /Ce p 4+ pairs, CeVO 4 is advantageous for combination with other material to improve the efficiency of photocatalysis. For example, Huo et al. reported that the photocatalytic performance of CeVO 4 was greatly improved by the La 2 O 3 media combination via electron transfer enhancement [20]. It was also reported that Ag + ions can be directly reduced by Ce 3+ ions to form CeVO 4 /Ag heterojunction nanocrystals (NCs) with a high photothermal conversion efficiency. Moreover, cerium vanadate nanostructures were used to detect clioquinol through the evaluation of the redox reaction on the electrode surface [21]. Since modulation of the surface atomic valence state can strongly alter the catalytic efficiency and product selectivity, it is important to study the mechanism and the effects of the Ce valence change for CeVO 4 photocatalysts.
Herein, CeVO 4 ultrathin nanosheets were synthesized by one-step solvothermal method with ethanolamine (MEA) as the structure-directing agent and then applied to photocatalytic CO 2 reduction. The photocatalytic efficiency of CO 2 conversion for the CeVO 4 nanosheets (referred to CV1-4, corresponding to the different volumes of MEA ranging from 1 to 4 mL) is improved compared to the CV prepared without MEA added. In particular, CV3 shows the maximum CO selectivity (40.97 µmol g −1 h −1 ) that is 6 and 14 times larger than those of CV and CV1, respectively, which has the lowest concentration of the intrinsic Ce 4+ ions. Moreover, the CO generation rate increases with decreasing intrinsic Ce 4+ /Ce 3+ ions ratio that can be modulated by MEA without affecting phase and morphology of the CeVO 4 . Furthermore, it was found that approximately 6.8% of Ce p 4+ /Ce p 3+ ion pairs are generated in the photocatalytic reaction process and tend to disappear after thermal treatment. As reactive centers, the Ce p 4+ /Ce p 3+ ion pairs are advantageous for the photocatalytic CO 2 conversion to CO by promoting the molecular adsorption of the otherwise inert CO 2 . Supported by optical and electrochemical characterization, the possible reactive mechanism is proposed to explain the efficiency and selectivity of the obtained CeVO 4 photocatalyst.

Morphology and Phase Characterization
Large CeVO 4 crystallites with different sizes (>500 nm) were formed using the hydrothermal method without any structure-directing agents ( Figure 1a). Comparatively, the as-prepared samples grew into uniformly nanoscale structures after MEA addition as observed from the scanning electron microscopy (SEM) images (Figure 1b-e). A detailed examination of the CeVO 4 samples showed that most of the as-prepared CeVO 4 (>99.9%) crystallized into translucent nanosheets in the MEA-assisted alkaline environment as shown in the TEM images. These ultrathin nanosheets are monodispersed and the average diameter of the oval-shaped nanosheets is approximately 10 nm ( Figure 2). The lattice fringes with d-spacing values of 0.32 nm and 0.27 nm correspond to the (002) and (112) planes of the tetragonal CeVO 4 phase, respectively, and their intersection angle is 63 • , in agreement with the theoretical value (Figure 3a,b). Examination of the high-resolution TEM (HRTEM) images of crimpled nanosheets shows that they are approximately 1.7 nm thick (Figure 3c), which is also certified by AFM characterization with non-contact mode ( Figure 4d) and their preferentially vertical growth is attributed to the high MEA absorption on the surface of the nanosheets [22][23][24]. The strongly adhesion of MEA blocks the epitaxial growth of CeVO 4 nanosheet toward the longitudinal. The crystal structure of the CeVO 4 nanosheet is schematically illustrated in Figure 3d. The vanadium atom environment is quite symmetric with four equivalent V-O bonds. Each A-site cation (Ce 3+ ) has D 2d site symmetry and is surrounded by eight oxygen atoms forming an antiprism with two different sets of A-O bonds of almost equal length [14]. The advantageous crystal symmetry and abundant atom species is listed as a potential candidate for photocatalytic application.
the possible reactive mechanism is proposed to explain the efficiency and selectivity of the obtained CeVO4 photocatalyst.

Morphology and Phase Characterization
Large CeVO4 crystallites with different sizes (>500 nm) were formed using the hydrothermal method without any structure-directing agents ( Figure 1a). Comparatively, the as-prepared samples grew into uniformly nanoscale structures after MEA addition as observed from the scanning electron microscopy (SEM) images (Figure 1b-e). A detailed examination of the CeVO4 samples showed that most of the as-prepared CeVO4 (>99.9%) crystallized into translucent nanosheets in the MEA-assisted alkaline environment as shown in the TEM images. These ultrathin nanosheets are monodispersed and the average diameter of the oval-shaped nanosheets is approximately 10 nm (Figure 2). The lattice fringes with d-spacing values of 0.32 nm and 0.27 nm correspond to the (002) and (112) planes of the tetragonal CeVO4 phase, respectively, and their intersection angle is 63°, in agreement with the theoretical value (Figure 3a,b). Examination of the high-resolution TEM (HRTEM) images of crimpled nanosheets shows that they are approximately 1.7 nm thick (Figure 3c), which is also certified by AFM characterization with non-contact mode ( Figure 4d) and their preferentially vertical growth is attributed to the high MEA absorption on the surface of the nanosheets [22][23][24]. The strongly adhesion of MEA blocks the epitaxial growth of CeVO4 nanosheet toward the longitudinal. The crystal structure of the CeVO4 nanosheet is schematically illustrated in Figure 3d. The vanadium atom environment is quite symmetric with four equivalent V-O bonds. Each A-site cation (Ce 3+ ) has D 2d site symmetry and is surrounded by eight oxygen atoms forming an antiprism with two different sets of A-O bonds of almost equal length [14]. The advantageous crystal symmetry and abundant atom species is listed as a potential candidate for photocatalytic application.  of the as-prepared samples (Figure 1f). Compared to the CV prepared without MEA, the light absorption of the CV1-4 samples is much stronger and extends much further into the visible range (ca. 700 nm) with all samples showing similar bandgaps of approximately 1.76 eV, which is slightly smaller than the previously reported [9,26].   mately 1.76 eV, which is slightly smaller than the previously reported [9,26].   ucts were detected in the absence of either simulated solar radiation or catalysts. Moreover, isotopic substitution experiments showed that for CV3, the obtained 13 CO product definitely originates from the reduction in gaseous 13 CO2 rather than from other C-containing compounds (Figure 4f). The CO generation rate of the CV3 sample is higher than those for the other reported vanadate photocatalysts [2][3][4]. which shows adverse interfacially electronic structure [27].  The purity and crystallinity of the prepared samples were characterized by X-ray diffraction (XRD). As shown in Figure 3e, no impurity peaks were detected and all observed diffraction peaks are attributed to the pure tetragonal phase of CeVO 4 (a= b = 7.399 Å, c = 6.496 Å, I4 1 /amd). Comparison of the XRD spectra shows that the full width at half maximum (FWHM) values of the stongest peak increase for the CV1-4 samples with more MEA added, indicating that the longitudinal dimensions are effectively reduced relative to CV to obtain ultrathin nanosheets. Moreover, the strongest diffraction peak at 24.03 • for (200) facets most likely characterizes the main horizontally exposed planes of the CV1-4 nanosheets [20,25]. UV-vis diffuse reflectance spectra show the light absorbing ability of the as-prepared samples (Figure 1f). Compared to the CV prepared without MEA, the light absorption of the CV1-4 samples is much stronger and extends much further into the visible range (ca. 700 nm) with all samples showing similar bandgaps of approximately 1.76 eV, which is slightly smaller than the previously reported [9,26].

Photocatalytic CO 2 Reduction Activity
The photocatalytic CO 2 reduction activity of the as-prepared CeVO 4 nanosheets was studied in the gas-phase reaction system (water vapor, CO 2 gas) under simulated solar radiation. During the photocatalytic reaction, carbon monoxide (CO) was detected as the main product and methane (CH 4 ) was detected as the secondary C1 product (Figure 4a,b). With longer time, the rate of CO generation increased linearly except for the CV2 and CV3 samples for which the products cannot be completely detected due to either light absorption by reactive intermediates or the transformation of multielectron-involved products accompanying the high reaction rate. In the photocatalytic CO 2 reaction, the rate of CO production first increased and then decreased with the increasing amount of the MEA additive in the preparation process, and the maximum CO generation rate reached 40.97 µmol g −1 h −1 for CV3. This value is 6 and 14 times larger than the corresponding values for CV and CV1, respectively (Figures 4c and 5a,b). CV2 showed superior CO generation (28.94 µmol g −1 h −1 ) compared to CV1 with a much greater increase compared to CV, demonstrating that for these materials, the interfacial effect modulated by MEA plays a more important role in the photocatalytic process than the specific surface area that is largely controlled by the nanomaterial size. Moreover, the catalytic selectivity was calculated as the ratio of the CO and CH 4 production rates and the results are shown in Table 1 and Figure 4d. The catalytic selectivity results show the same trend as that for the CO generation rate and the improved selectivity may be correlated with surface properties controlled by the MEA agent, with CV3 showing the highest CO selectivity with a CO/CH 4 ratio of 301.92. In addition, tiny amounts of C2 products were also detected in the reaction system (Figures 4e and 5c). Comparison of the XRD spectra of the as-prepared sample and the sample after the reaction reveals no changes (Figure 5d), indicating that the phase of these nanosheets remain stable during the photocatalytic CO 2 reaction process. No products were detected in the absence of either simulated solar radiation or catalysts. Moreover, isotopic substitution experiments showed that for CV3, the obtained 13 CO product definitely originates from the reduction in gaseous 13 CO 2 rather than from other C-containing compounds (Figure 4f). The CO generation rate of the CV3 sample is higher than those for the other reported vanadate photocatalysts [2][3][4]. which shows adverse interfacially electronic structure [27].

Spectroscopy Analysis
To investigate the electronic state of the surfaces of the as-prepared samples, X-ray photoelectron spectroscopy (XPS) characterizations were carried out using the C1s peak (294.8 eV) for calibration and Ce 3d, V 2p, O 1s peaks for analysis (Figure 6a). The ratio of Ce and V element is in accordance with the stoichiometric ratio of CeVO4. The Ce 3d XPS spectra of the as-prepared samples display four peaks at the binding energies of 881.6,

Spectroscopy Analysis
To investigate the electronic state of the surfaces of the as-prepared samples, X-ray photoelectron spectroscopy (XPS) characterizations were carried out using the C1s peak (294.8 eV) for calibration and Ce 3d, V 2p, O 1s peaks for analysis (Figure 6a). The ratio of Ce and V element is in accordance with the stoichiometric ratio of CeVO 4 . The Ce 3d XPS spectra of the as-prepared samples display four peaks at the binding energies of 881.6, 885.3, 900.1 and 903.5 eV labeled as V 0 , V, u 0 and u, respectively, corresponding to the Ce 3+ state (Figure 6b). The u' and u" signature peaks or Ce 4+ at the binding energies of 905.7 and 914.3 eV [13,28] are also observed in the XPS spectra. Generally, the integrated areas of the XPS spectra and the Ce 4+ /Ce 3+ molar concentration ratio are positively correlated. Therefore, the XPS spectra were deconvoluted into the contributions of the two different Ce 3d valence states in order to obtain the Ce 4+ /Ce 3+ molar concentration ratio according to the following formula [29]: where A i is the total integrated area of fitting peak i. The results of the calculations are shown in Figure 6c. It is observed that as the amount of the added MEA increases, the Ce 4+ /Ce 3+ molar concentration ratio gradually decreases for the CV1-3 samples, and CV3 shows the lowest Ce 4+ /Ce 3+ proportion (13.33:100). This change can be understood based on the molecular formula of MEA. MEA contains reducing amino groups and oxidizing hydroxyl groups ( Figure 6d); when MEA is used as an addictive in small amounts, the amino groups absorbing on the surface of Ce 3+ act as stabilizers and protect the Ce 3+ ions from oxidation in the solvent. However, when excessive MEA is introduced in the reactor, hydroxyl groups oxidize Ce 3+ to the Ce 4+ state [30][31][32], increasing the concentration of the Ce 4+ ions in the CV4 sample.
Examination of the XPS results shows that there is no variation in the integrated areas of the V 2p and O 1s XPS spectra of the as-prepared samples (Figure 7a,b). Two V 5+ peaks located at 517.2 and 524.8 eV are due to V 2p 3/2 and V 2p 1/2 , respectively, and are accompanied by a small peak at 523.1 eV corresponding to the V 2p 1/2 state of V 3+ . Only two types of surface oxygen (O lattice and O adsorbate ) are observed in the XPS spectra for these samples, corresponding to the two peaks at 529.8 and 531.2 eV, respectively. Compared to CV, the peak position of CV1-4 gradually moves to the lower binding energies, which is likely due to the increasing electron enrichment of the O atoms in these samples. The positive slope of the Mott-Schottky (MS) curve further shows that the as-prepared CeVO 4 is a characteristic n-type semiconductor (Figure 7c). The tangent of the MS curve has an x-axis intercept of −0.757 V corresponding to the flat charged potential (vs. Ag/AgCl, pH = 7). Generally, the CB potential of an n-type semiconductor is approximately equal to the flat charging potential. According to the conversion formula [33]: where E (Ag/AgCl) = 0.197 V at pH = 7. The conduction band (CB) potential of CV3 is determined to be −0.54 V (vs. NHE, pH = 7) and the corresponding valence band (VB) is 1.2 V (vs. NHE, pH = 7), effectively driving the photocatalytic conversion from CO 2 to the CO, CH 4 and C2 products kinetically. Moreover, the reactive barrier for catalytic reduction can be lowered with the intrinsic Ce 4+ defective states, as confirmed by the LSV curves of the as-prepared samples (Figure 8a). The Tafel slope value for the CV3 nanosheets is smaller than that for the CV2 and other samples (363, 388, 311, and 371 mV decade −1 for CV1, CV2, CV3, and CV4, respectively) due to its lower energy input in the catalytic reaction. In addition, the Ce p 4+ /Ce p 3+ ratio increased by 6.78:100 from 13.33:100 to 20.11:100 for CV3 after the photocatalytic reaction (Figure 7d), as is also confirmed by the results of Raman spectroscopy analysis (Figure 8b,c) carried out to explore the interactions between various groups ( Table 2). The characteristic Raman peaks of the symmetrical stretching (A 1g ) and the antisymmetric stretching (B 1g ) of the VO 4 3− tetrahedron are observed at 831 and 760 cm −1 , respectively. The Raman peaks at 446 and 354 cm −1 are assigned to the B 1g and A 1g bending modes of VO 4 3− . The external mode of Ce-VO 4 vibration appearing below 250 cm −1 confirms the formation of vanadate (Figure 8b) [10,34]. There is no change in the positions of the characteristic peaks of these groups after the photocatalytic reaction (Figure 8c). By contrast, the Raman peaks of the B 1g bending modes of VO 4 3− (448 cm −1 ) are clearly redshifted, particularly for the CV3 and CV2 samples that show much higher photocatalytic activities. The red-shift corresponding to stronger Raman scattering originates from the greater energy capture that is likely to be related to the emergence of Ce 4+ that lowers the geometric symmetry of the CeVO 4 crystal structure. Moreover, the lower photoluminescence (PL) intensity of CV3 compared to CV2 demonstrates its superb photogenerated charge carrier transport behavior that is advantageous for the photocatalytic CO 2 reduction (Figure 8d). The PL intensity change after the photocatalytic reaction further confirm presence of photo-induced Ce 4+ ions that then reduced into Ce 3+ after the post-treatment at 400 • C (Figures 8e and 7d). Additionally, the CV1-4 samples display similar specific surface areas of 22.5332, 23.8444, 21.2346 and 22.9847 m 2 g −1 (Figure 8f), respectively, as evaluated by the Brunauer-Emmet-Teller (BET) measurements, showing that the photocatalytic activity difference is not attributed to the surface area.

Photocatalytic Mechanism
These two Ce 4+ ion species play important roles in the photocatalytic reaction process. The intrinsic Ce 4+ ions act as recombination centers and are unfavorable for photocatalytic CO 2 reduction, in which the CO product selectivity is inversely proportional to the intrinsic Ce 4+ /Ce 3+ ratio. By contrast, the Ce p 4+ are active and inferred to improve the selective photocatalytic CO 2 reduction to CO with the valence change. The pathway of photocatalytic CO 2 reduction with H 2 O vapor is illustrated in Figure 9. First, the photogenerated carriers (holes and electrons, denoted as h + and e − , respectively) are generated on the surface of CeVO 4 through the absorption of photon energy (hV). Simultaneously, a large number of Ce p 4+ ions appear on the surface of the photocatalysts under solar light irradiation. Next, a Ce p 4+ ion can be reduced to a Ce p 3+ ion by capturing the photogenerated electrons and generating an oxygen vacancy (Vo), providing the active site for CO 2 adsorption as described by hydroxyl groups (Figure 6d); when MEA is used as an addictive in small amounts, the amino groups absorbing on the surface of Ce 3+ act as stabilizers and protect the Ce 3+ ions from oxidation in the solvent. However, when excessive MEA is introduced in the reactor, hydroxyl groups oxidize Ce 3+ to the Ce 4+ state [30][31][32], increasing the concentration of the Ce 4+ ions in the CV4 sample. Examination of the XPS results shows that there is no variation in the integrated areas of the V 2p and O 1s XPS spectra of the as-prepared samples (Figure 7a,b). Two V 5+ peaks located at 517.2 and 524.8 eV are due to V 2p3/2 and V 2p1/2, respectively, and are accompanied by a small peak at 523.1 eV corresponding to the V 2p1/2 state of V 3+ . Only two types of surface oxygen (Olattice and Oadsorbate) are observed in the XPS spectra for these samples, corresponding to the two peaks at 529.8 and 531.2 eV, respectively. Compared to CV, the peak position of CV1-4 gradually moves to the lower binding energies, which is likely Then, the Ce p 3+ ions capture adsorbed CO 2 with Ce p 4+ and the CO product is generated with assistance of H + [27,35] as described by The conduction band position (−0.54 eV vs. NHE for CV3) is more negative than the standard redox potentials of CO and CH 4 . The existence of the Ce p 3+ /Ce p 4+ pairs not only improves the effective separation of the photogenerated electrons and holes, but also promotes the absorption and extensive activation of CO 2 . CV1, CV2, CV3, and CV4, respectively) due to its lower energy input in the catalytic reaction. In addition, the Cep 4+ /Cep 3+ ratio increased by 6.78:100 from 13.33:100 to 20.11:100 for CV3 after the photocatalytic reaction (Figure 7d), as is also confirmed by the results of Raman spectroscopy analysis (Figure 8b, c) carried out to explore the interactions between various groups (Table 2).     improves the effective separation of the photogenerated electrons and holes, but also promotes the absorption and extensive activation of CO2. Figure 9. Schematic illustration of the catalytic pathways for the photocatalytic CO2 reduction with H2O vapor.

Materials and Methods
CeVO4 were synthesized by one-step solvothermal method. Firstly, Ce(NO3)3.6H2O 1.0 mmol (≥ 99.0%, Sinopharm, Shanghai, China) and NH4VO3 1.0 mmol (≥ 99.0%, Sinopharm, Shanghai, China) were added into 20 mL distilled water, respectively. Secondly, MEA (≥ 99.0%, Lingfeng, Shanghai, China) was dropped into the prepared solutions as structure-directing agent. The volume of MEA was varied from 1 to 4 mL with the sample of MEA scarcity for comparison. Then, the mixture was transferred to a Teflon-lined stainless-steel autoclave with a 30 mL inner volume after stirring for 2 h and put in an oven heated at 200 °C for 36 h. After the reaction, these samples cooled down to room temperature naturally. The bottom precipitates were centrifuged and washed for 3 times with ethanol and deionized water, respectively. The dark brown CeVO4 samples were successfully obtained after dry in freeze dryer.
The phases of these prepared powder samples were characterized by X-ray diffractometry (XRD; Rigaku Ultima III, Rigaku, Tokyo, Japan) at room temperature. Cu-Ka radiation (λ = 0.154178 nm) was used at 40 kV and 40 mA in a scan range of 10°-90° at a scan rate of 20° min −1 . The morphology of the prepared samples was characterized by scanning electron microscopy (SEM, JSM-7800F, JEOL, Tokyo, Japan). Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) images were taken by TEM (JEM 200CX, JEOL, Tokyo, Japan). UV-visible diffuse reflectance spectra were obtained at room temperature using an ultraviolet-visible spectrophotometer (UV-2550, Shimadzu, Kyoto, Japan) and directly converted to absorption spectra by using the Kubelka-Munk equation in software. AFM images were taken using a Park XE7 (Park SYSTEMS, South Korean) atomic force microscope in non-contact mode. The PL spectra were obtained by using (F-7000, Tianmei, China) FL Spectrophotometer (excitation wavelength 280 nm). Raman spectroscopy is the use of the model Raman spectrometer (WITec alpha 300 M + , WITec, German) under 532 nm laser. The chemical state was examined by X-ray photoelectron spectroscopy (XPS; K-Alpha, Thermo Fisher Scientific, American), and charge correction was performed based on the binding energy of the indefinite C1s peak at 284.8 eV. Using a electrochemical workstation (ZaHNER PP211, Zahner, German

Materials and Methods
CeVO 4 were synthesized by one-step solvothermal method. Firstly, Ce(NO 3 ) 3 .6H 2 O 1.0 mmol (≥99.0%, Sinopharm, Shanghai, China) and NH 4 VO 3 1.0 mmol (≥99.0%, Sinopharm, Shanghai, China) were added into 20 mL distilled water, respectively. Secondly, MEA (≥99.0%, Lingfeng, Shanghai, China) was dropped into the prepared solutions as structuredirecting agent. The volume of MEA was varied from 1 to 4 mL with the sample of MEA scarcity for comparison. Then, the mixture was transferred to a Teflon-lined stainless-steel autoclave with a 30 mL inner volume after stirring for 2 h and put in an oven heated at 200 • C for 36 h. After the reaction, these samples cooled down to room temperature naturally. The bottom precipitates were centrifuged and washed for 3 times with ethanol and deionized water, respectively. The dark brown CeVO 4 samples were successfully obtained after dry in freeze dryer.
The phases of these prepared powder samples were characterized by X-ray diffractometry (XRD; Rigaku Ultima III, Rigaku, Tokyo, Japan) at room temperature. Cu-Ka radiation (λ = 0.154178 nm) was used at 40 kV and 40 mA in a scan range of 10 • -90 • at a scan rate of 20 • min −1 . The morphology of the prepared samples was characterized by scanning electron microscopy (SEM, JSM-7800F, JEOL, Tokyo, Japan). Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) images were taken by TEM (JEM 200CX, JEOL, Tokyo, Japan). UV-visible diffuse reflectance spectra were obtained at room temperature using an ultraviolet-visible spectrophotometer (UV-2550, Shimadzu, Kyoto, Japan) and directly converted to absorption spectra by using the Kubelka-Munk equation in software. AFM images were taken using a Park XE7 (Park SYSTEMS, South Korean) atomic force microscope in non-contact mode. The PL spectra were obtained by using (F-7000, Tianmei, China) FL Spectrophotometer (excitation wavelength 280 nm). Raman spectroscopy is the use of the model Raman spectrometer (WITec alpha 300 M + , WITec, German) under 532 nm laser. The chemical state was examined by X-ray photoelectron spectroscopy (XPS; K-Alpha, Thermo Fisher Scientific, American), and charge correction was performed based on the binding energy of the indefinite C1s peak at 284.8 eV. Using a electrochemical workstation (ZaHNER PP211, Zahner, German), a 0.2 M Na 2 SO 4 solution was used as the electrolyte, Ag/AgCl was used as the reference electrode, and Pt was the counter electrode as the standard of three-electrode system. The Mott-Schotty test was carried out at a frequency of 5 kV, Scan rate is 5 mV/min, and voltage rage at −0.9-0.1 V. The Tafel test by using the other electrochemical workstation (CHI660E, Huachen, Shanghai, China). Scan rate is 5 mV/min, and voltage rage at 0-0.6 V. Working electrode was prepared by deposited catalyst on the F-doped SnO 2 conductive glass (FTO).
The photocatalytic reaction of carbon dioxide reduction was carried out at room temperature, and the 0.05 g sample of the prepared powder was evenly weighed on the top of a glass reactor with an area of 4.91 cm 2 . The bottom of the glass reactor was tied with a magnetic stirrer, it was guaranteed that during the reaction pure CO 2 gas is slowly and evenly injected until it fills the reactor. The excess gas was collected with NaHCO 3 solution, and finally, the glass reactor was put into a gas-tight reaction system with a volume of about 440 mL. 0.4 mL of pure water was injected from the silicone rubber diaphragm into the reactor through a micro syringe as a reducing agent and injected into the reaction system. After stirring the stirrer for 1 h, the adsorption of CO 2 -H 2 O atmosphere was balanced to ensure complete adsorption. A solar simulator (Micro-solar 300 W xenon lamp) was used to illuminate the vertical sample. From the start of the xenon lamp, the lighting time is 6 h. Draw about 1 mL of gas from the reaction cell. After injection, a gas chromatograph (GC-2014, Shimadzu Corporation, Kyoto, Japan) was used to analyze the subsequent reaction products. The isotope-labelled experiments were implemented using 13 CO 2 as the original reactant, and the products were analyzed using gas chromatograph-mass spectrometry (QP2010SE, Shimadzu Corporation, Kyoto, Japan).

Conclusions
In summary, ultrathin CeVO 4 nanosheets with a thickness of approximately 1.7 nm were synthesized by one-step hydrothermal method with MEA as the structure-directing agent. This photocatalyst has a narrow bandgap of approximately 1.74 eV, extending the optical absorbance edge to 700 nm, and exhibits superior light capture. The proportion of Ce 4+ ions can be decreased by MEA without either morphology or phase changes. The optimal photocatalytic activity is obtained when the Ce 4+ /Ce 3+ molar ratio decreases to 13.33:100 for the CV3 sample. The maximum yield rate of CO is 40.97 µmol g −1 h −1 for CV3, showing an enhancement by 6 times compared to CV. Moreover, the Ce 4+ /Ce 3+ ratio is negatively correlated with and modulates the C1 product selectivity. Finally, the likely reaction mechanism is proposed based on the Ce p 4+ /Ce p 3+ pairs that give rise to the photocatalytic reaction process and drive the selective photocatalytic CO 2 reduction to CO. The atomic valence regulation demonstrated in this work provides an effective alternative approach for the improvement of the product selectivity and efficiency enhancement in photocatalytic reactions.