As a green and sustainable technology, photocatalysis has been increasingly studied due to its great potential in solving environmental pollution and energy crises [1-3]. In photocatalytic processes, photoactive materials, such as TiO₂ [4], CeO₂ [5], ZnCo₂O₄/Bi₂O₃ [6], g-C₃N₄/Ag₃PO₄ [7], SnO₂ [8], BiOBr/TiO₂/Ti₃C₂T_X [9], GaN [10], CdS [11], and BiVO₄/Znln₂S₄ [12] play decisive roles in the redox/charge-transfer reactions due to their unique electronic structures with a filled valence band (VB) and an empty conduction band (CB) [1]. Among them, CeO₂ has been widely used in industry because Ce, as a rare-earth element, has a low price and relatively high abundance in the Earth's crust. As an n-type semiconductor (with a band gap similar to that of TiO₂), it exhibits potential photocatalytic activity. However, the wide band gap restricts its photocatalytic efficiency [4,13]. Considering this, other components, including semiconductors and metals, have been introduced to improve UV and/or visible light absorption and prevent hole-electron recombination [14-17]. Moreover, the modification of CeO₂ by controlling its size, morphology, and number of defects will be significant for improving its intrinsic photocatalytic properties [18,19].

Nanoscale CeO₂, e.g., nanoparticles, nanosheets, and nanorods, can effectively enhance UV and/or visible light absorption and promote the adsorption of reactants due to its high specific surface area, thereby facilitating photocatalytic reactions [20]. Previous studies have mainly focused on the comparison of photocatalytic performance with different morphologies of CeO₂ [21,22]. However, the intrinsic photocatalytic performance of CeO₂ still needs to be further studied, especially the effect of defects on its catalytic performance, which needs to be studied with the same morphology or even uniform size. One-dimensional (1D) CeO₂ nanorods with well-defined {100} and {110} planes show better photocatalytic performances than other stable polyhedral-structured CeO₂ [13]. Additionally, the 1D nanorod structure dramatically improves the redox properties because of its more efficient and fast transport of photogenerated e⁻ and h⁺ along the nanorod surface [23-26].

Defects in catalysts are crucial factors affecting photocatalytic performance. The defect formation energies in nanomaterials are lower due to the existence of a high-density interface, which leads to an increase in nonstoichiometric defects and the generation of electron carriers [27]. In addition, Ce ions usually show two valences (Ce³⁺ and Ce⁴⁺) due to their special outer electron structure of [Xe] 4f²6s², thus tending to form multivalent oxides of CeO_2−x (x = 0−0.5) [13]. Therefore, a large number of oxygen vacancies (V_o) are easily formed on/in the surface/bulk of CeO₂. Many strategies have been developed for the introduction of oxygen vacancies to CeO₂ by using reductant including H₂, CO, NH₃, and Li [28-30]. These V_o are present both on the surface and in the bulk phase. The surface oxygen vacancies (V_o-s) can promote the increase in the Fermi level attributed to the generation of defect energy levels (near the bottom of the CB in CeO₂) in the band gap, leading to a narrowed band gap and enhanced light absorption [31]. In addition, coordinative-unsaturated metal atoms (surrounding V_o) can act as strong binding sites for organic dyes, thus accelerating photocatalytic reactions [32,33]. Moreover, bulk oxygen vacancies (V_o-b) may act as recombination centers of photogenerated e⁻ coming from the surface, resulting in a low utilization of photogenerated e⁻ [34]. With an increasing concentration ratio (from V_o-s to V_o-b), the utilization of photogenerated e⁻ for photodegradation can be improved, thus leading to improved photocatalytic performance [35-37]. Therefore, the precise regulation of V_o is of great significance to improve the intrinsic photocatalytic performance of CeO₂ when applied to the degradation of organic dyes.

Herein, we synthesize and report a 1D CeO₂ nanorod with uniform size and morphology. The hydrothermal method is used to tune the concentration ratio of surface/bulk oxygen vacancies in CeO₂ nanorods. With the highest concentration of V_o-s to V_o-b, the CeO₂ nanorod presents the narrowest band gap and the highest utilization of photogenerated e⁻. In regard to the photodegradation of rhodamine B, the percentage of removal reaches 70.2% after 120 min of ultraviolet−visible light irradiation, exhibiting potential application in wastewater purification. This tuning strategy of V_o also provides guidance for the development of other advanced metal-oxide semiconductor photocatalysts for the photodegradation of organic dyes.

The X-ray powder diffraction (XRD) patterns of the as-prepared CeO₂-1, CeO₂-2, and CeO₂-3 are shown in Fig. 1a. The diffraction peaks of the catalysts were mainly assigned to the (111), (200), (220), (311), (222), (400), and (331) planes, which were indexed to the cubic fluorite structure of CeO₂ crystals (JCPDS No. 34-0394). The field-emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images of CeO₂-1, CeO₂-2, and CeO₂-3 are shown in Figs. 1b and c and Fig. S1 (Supporting information), respectively. All the samples are 1D nanorod structures. The shorter and narrower CeO₂-1 and CeO₂-2 nanorods loaded with small nanoparticles led to irregular surfaces with rough structures (Fig. 1c and Fig. S1). CeO₂-3 exhibited longer and broader nanorods with smooth surfaces (Fig. S1d). Furthermore, two types of interference fringes of 0.19 and 0.31 nm were observed at the outer and inner nanorods from the high resolution transmission electron microscopy (HRTEM) image, which could be assigned to the (220) and (111) planes of the cubic fluorite structure CeO₂, respectively (Fig. 1d). This result suggests that the preferred orientation growth of the individual CeO₂-2 nanorods was along the [110] direction and that the main exposed crystal surface was the (220) plane, which had a higher photocatalytic activity than (111). In addition, an ~5 nm CeO₂ nanoparticle on the CeO₂-2 surface could be clearly seen, which likely formed during the initial crystal growth processes after the distribution of nucleation sites. Moreover, CeO₂-2 exhibited a Brunauer-Emmett-Teller (BET) surface area of 137 m²/g, which was larger than those of CeO₂-1 and CeO₂-3 (52 and 47 m²/g, respectively) (Fig. S2 in Supporting information). Clearly, this large BET specific surface area was mainly attributed to its rough surface (Fig. 1c), thus providing abundant active sites and more V_o-s. These V_o-s were conducive to enhanced light absorption and reactant adsorption; additionally, the exposed (220) facets contributed to the high reactive activity, and both factors improved the photocatalytic performance of CeO₂-2 nanorods.

At present, several detection technologies can be used to resolve the V_o in metal oxides, such as, Raman spectrum, extended X-ray absorption fine structure spectroscopy (EXAFS), positron annihilation lifetime spectroscopy (PALS), electron paramagnetic resonance spectrum (EPR), aberration-corrected transmission electron microscopy (ACTEM), electron energy loss spectroscopy (EELS), and X-ray photoelectron spectroscopy (XPS) [38, 39]. Among those detection method, Raman, EXAFS, PALS, and EPR are often used to analyze the total V_o information in catalyst, while the ACTEM, EELS, and XPS are often used to analyze V_o in the local or surface of catalyst. In this paper, we combined Raman and XPS to semi-quantitatively analyze the concentration ratio of surface/bulk V_o in CeO₂.

The Raman spectra providing the V_o information of the total crystal, including the bulk and surface, are shown in Fig. 2a. The sharp Raman peaks of CeO₂-1, CeO₂-2, and CeO₂-3 were located at frequencies of 459, 459, and 461 cm⁻¹, respectively, and these were attributed to the first-order vibrational mode with F_2g vibration (symmetrical stretching mode of the Ce−O⁸ vibration unit) in a fluorite cubic structure. Compared with the reported value of 466 cm⁻¹, the redshifts of the F_2g vibration peaks were ascribed to the nanocrystalline nature of CeO₂ [40]. The peak shifts of CeO₂-1 and CeO₂-2 were larger than those of CeO₂-3, further indicating the smaller crystal grains of CeO₂-1 and CeO₂-2. The broader peaks at approximately 270 and 597 cm⁻¹ corresponded to the second-order transverse acoustic mode and the defect (D) vibration (i.e., V_o), respectively [17,41]. TheI(D)/I(F_2g) ratios of CeO₂-1, CeO₂-2, and CeO₂-3 were estimated to be 0.048, 0.047 and 0.043, respectively, indicating all three had similar total V_o concentrations (Fig. S3 and Table S1 in Supporting information).

The Ce 3d XPS spectra presented the chemical state of elemental Ce (Fig. 2b). All the peaks could be mainly attributed to the spin−orbital doublet peaks of Ce 3d_5/2 and 3d_3/2, which could be deconvolved into four pairs, i.e., v1 (882.7 eV)/u1 (901.0 eV), v2 (885.4 eV)/u2 (903.7 eV), v3 (889.2 eV)/u3 (907.5 eV), and v4 (898.4 eV)/u4 (916.7 eV) [42, 43]. Among them, the v1/u1, v3/u3, and v4/u4 peaks referred to the Ce⁴⁺ 3d states, while the v2/u2 peak corresponded to the Ce³⁺ 3d states, indicating the composition of Ce⁴⁺ and Ce³⁺ in these CeO₂. This absence of Ce³⁺ provided evidence of lattice oxygen depletion, further illustrating the existence of V_o on CeO₂ surfaces. The surface oxygen vacancies, i.e., V_o-s, could decrease the band gap, thus enhancing light absorption and promoting photogenerated e⁻ for photocatalysis. The Ce³⁺/Ce⁴⁺ ratios of CeO₂-1, CeO₂-2, and CeO₂-3 were estimated to be 4.1%, 8.0%, and 3.8%, respectively, indicating that the order of V_o-s concentration was CeO₂-2 > CeO₂-1 > CeO₂-3. Combining the Raman analysis (the total V_o concentrations in the order of CeO₂-2 ≈ CeO₂-1 ≈ CeO₂-3) with the XPS results (the V_o-s concentrations in the order of CeO₂-2 > CeO₂-1 > CeO₂-3), the order of V_o-b concentrations should be CeO₂-2 < CeO₂-1 < CeO₂-3. Therefore, the concentration ratio of V_o-s to V_o-b was in the order of CeO₂-2 > CeO₂-1 > CeO₂-3 (Table S1 in Supporting information). Fig. 2c illustrates the action mechanism of photodegradation by V_o under different concentration ratios of V_o-s to V_o-b. Briefly, h⁺/e⁻ pairs would be generated under irradiation and migrate to the surface of CeO₂, followed by their participation in catalytic reactions. During the migration process, a fair number of photogenerated e⁻ may be captured by V_o-b and become invalid [35]. A higher concentration ratio of V_o-s to V_o-b led to more photogenerated h⁺/e⁻ pairs and less recombination, which would improve the utilization of photogenerated e⁻. In contrast, in the case of a low ratio of V_o-s to V_o-b, most photogenerated e⁻ would be depleted by recombination with more V_o-b, which would deteriorate the photodegradation activity.

Fig. 3a shows the photocatalytic performances of the as-synthesized CeO₂ nanorods for rhodamine B photodegradation under UV–vis light irradiation. The results showed that 29.8% and 67.4% of rhodamine B was discolored by CeO₂-1 and CeO₂-2, respectively. CeO₂-2 displayed more efficient photocatalytic performance, which was approximately 2.3 times higher than that of CeO₂-1 after 120 min of light irradiation. CeO₂-3 demonstrated little activity for rhodamine B photodegradation, as shown in Fig. 3a. To quantitatively clarify the photocatalytic activities of different catalysts, rhodamine B photodegradation was studied based on the plots of −ln(C_t/C₀) versus time (Fig. 3b) corresponding to the first-order kinetics model with a linear regression mechanism. As shown, the reaction rate constant k value of CeO₂-2 was 0.0093 min⁻¹, which was approximately 2.5 times that of CeO₂-1 (0.0038 min⁻¹). Accordingly, rhodamine B could be easily adsorbed on the CeO₂-2 surface with abundant V_o-s by a first-order kinetics model. In contrast, the rhodamine B hardly degraded without photocatalyst under UV–vis light. These experimental results further confirmed that V_o-s played a crucial role in improving photocatalytic activity. Next, we explored the reusability of the catalyst (Fig. 3c). The photocatalytic cycling performance of CeO₂-2 over six cycles slightly decreased for each 120 min run of rhodamine B degradation. These results suggest that CeO₂-2 (with abundant V_o-s) was stable and could be reused for rhodamine B photodegradation. For congo red and methyl orange, CeO₂-2 sample also has excellent degradation performance, indicating its application prospect in sewage treatment (Fig. S4 in Supporting information).

The photocurrent responses of the catalysts are shown in Fig. 3d, while the order of photocurrent response was CeO₂-2 > CeO₂-1 > CeO₂-3 under alternating irradiation (lights on and off), which was consistent with the order of the concentration ratio of V_o-s to V_o-b. The photocurrent densities based on the BET surface area of three CeO₂ nanorods are similar after 400s, indicating a negligible effect of surface area on the photocatalytic efficiency in the total photocatalytic processes (Fig. S5 in Supporting information). In regard to the CeO₂-2 sample, its high concentration ratio of V_o-s to V_o-b could ensure that plenty of photogenerated e⁻ were consumed during photodegradation, corresponding to the largest photocurrent intensity. Moreover, the photocurrent intensities of CeO₂-2 exhibited a very slight decrease during the initial cycles and tended to be flat with increasing time, indicating a stable photoelectric characteristic.

The optical properties and band structures of CeO₂-1, CeO₂-2, and CeO₂-3 were analyzed by UV–vis diffuse reflectance spectroscopy (DRS). Fig. 4a shows that the absorption band edges of CeO₂-1, CeO₂-2, and CeO₂-3 were 428, 436, and 425 nm, respectively, which were higher than that reported for CeO₂ (approximately 420 nm). These redshifts in the absorption bands indicated the quantum confinement effect of 1D-nanostructured CeO₂ nanorods. Furthermore, the absorption band edge of CeO₂-2 exceeded that of other CeO₂ nanorods, suggesting an expanded UV light absorption range and enhanced UV-light-response ability due to its abundant V_o-s, which was the reason for its higher photocatalytic efficiency. The band gap energy (E_g) of three samples could be calculated based on the Tauc plot (Eq. 1):

(1)

where α, hν, A, and E_g signify the absorption coefficient, photoenergy, proportionality constant, and band gap energy, respectively. The constant n here equaled 1/2 because CeO₂ is a direct-gap semiconductor. As shown in Fig. 4b, the E_g of CeO₂-1, CeO₂-2, and CeO₂-3 was estimated to be 2.88, 2.82, and 2.92 eV, respectively, exhibiting a lower value compared with that reported for CeO₂ (3.2 eV) [44]. This result could be attributed to the formation of localized states within the band gap due to the V_o and increased Ce³⁺ concentration. Previously, it has been shown that the presence of Ce³⁺ reduces the band gap, leading to an expanded UV light absorption range [45, 46]. CeO₂-2 exhibited the narrowest E_g and generated more light-excited carriers under UV irradiation. The evaluated valance band potentials (E_VB) of CeO₂-1, CeO₂-2, and CeO₂-3 were +2.50, +2.47, and +2.52 eV, respectively [47]. They were smaller than the required potential for OH radical formation by OH⁻ (^∙OH/H₂O pair, +2.68 eV) but larger than that by H₂O (^∙OH/OH⁻ pair, +1.99 eV), indicating that the holes in the valance band could directly oxidize rhodamine B but could not oxidize water. The conduction band edge position could be calculated by the following equation (Eq. 2):

(2)

As a result, the conduction band potentials (E_CB) of CeO₂-1, CeO₂-2, and CeO₂-3 were calculated to be −0.38, −0.35, and −0.40 eV, respectively, which were larger than the reduction potential for the reduction of molecular oxygen to superoxide anion radicals with strong oxidation (O₂/^∙O₂⁻ pair, −0.33 eV), which energetically satisfied the requirements for the degradation of rhodamine B.

Based on the E_g and band edge positions of the CeO₂ nanorods, a photodegradation mechanism was proposed and is shown in Fig. 4c. Excitations (e⁻/h⁺) were produced when UV–vis light irradiated the CeO₂ nanorods. These charge carriers rapidly transferred to the CeO₂ surface due to the short transfer length in the 1D nanorod structure. The h⁺ in the valance band could directly oxidize rhodamine B; simultaneously, the e⁻ in the conduction band could reduce molecular oxygen to superoxide anion radicals. In regard to the CeO₂-2 sample, the narrow band gap ensured the generation of a large number of photogenerated e⁻. The higher concentration ratio of V_o-s to V_o-b improved the utilization of photogenerated e⁻ because of the fast transfer of photogenerated e⁻ for photodegradation on the surface and the decreased probability of photogenerated e⁻ and V_o-b recombination in bulk. Additionally, the high specific surface area of CeO₂-2 provided a large number of adsorption sites for the photodegradation of rhodamine B. Based on the above reasons, the CeO₂-2 sample exhibited the best photodegradation performance.

In summary, we successfully prepared 1D CeO₂ nanorods with a uniform size and morphology. By an easy hydrothermal method, we tuned the concentration of surface/bulk oxygen vacancies in CeO₂ nanorods. With the highest concentration ratio (from V_o-s to V_o-b), the CeO₂ nanorod presented more efficient utilization of photogenerated e⁻ due to the fast transfer of photogenerated e⁻ for photodegradation on the surface and the decreased probability of photogenerated e⁻ and V_o-b recombination in bulk. In regard to the photodegradation of rhodamine B, the removal percentage reached 70.2% after 120 min of UV–vis light irradiation, exhibiting potential applications for wastewater purification. This V_o tuning strategy provides guidance for the development of other advanced metal-oxide semiconductor photocatalysts for the photodegradation of organic dyes.

Declaration of competing interest

The authors report no declarations of interest.

Acknowledgments

This work was financially supported by the National Natural Science Foundation of China (Nos. 21404054, 21667019, 22066017, 52000163), the Key Project of the Natural Science Foundation of Jiangxi Province (No. 20171ACB20016), the Jiangxi Province Major Academic and Technical Leaders Cultivating Object Program (No. 20172BCB22014), the Science and Technology Department of Jiangxi Province (Nos. 20181BCB18003, 20181BAB216012, 20181ACG70025, and 20192BAB206013), the Key Laboratory of Photochemical Conversion and Optoelectronic Materials, TIPC, CSA (No. PCOM201906), and the Key Project of Science and Technology Research of the Jiangxi Provincial Department of Education (Nos. DA201602063, GJJ13508, and GJJ191044), the Aviation Science Foundation of China (No. 2017ZF56020), the Fujian Key Laboratory of Measurement and Control System for Shore Environment (No. S1-KF1703), and the Program B for Outstanding PhD Candidate of Nanjing University (No. 202002B076).

Appendix A. Supplementary data

Supplementary material related to this article can befound, in the online version, at doi: https://doi.org/10.1016/j.cclet.2021.01.044.