Photocatalysis of Cr- and Fe-Doped CeO2 Nanoparticles to Selective Oxidation of 5-Hydroxymethylfurfural

Oxygen vacancies (Vo) present in CeO2 nanoparticles (NPs) can effectively boost their photocatalytic activity under ultraviolet (UV) light. To improve photocatalytic performance, Cr- and Fe-doped CeO2 NPs with increased Vo were prepared using a simple method of doping Cr and Fe ions into CeO2 NPs, which was confirmed by an in-depth analysis of the structural and electronic changes. Through photocatalytic degradation (PCD) experiments with 5-hydroxymethylfurfural (HMF), we found that the PCD rates of the two doped CeO2 NPs were faster than that of the CeO2 NPs. In addition, the conversion of HMF to 2,5-furandicarboxylic acid (FDCA) using the doped CeO2 NPs occurred only through the mechanism of the selective oxidation to 5-hydroxymethyl-2-furancarboxylic acid (HMFCA), exhibiting better efficiency than using CeO2 NPs.


Introduction
Over the last few decades, metal oxides, such as CeO 2 , ZnO, and TiO 2 , have been studied for applications in various commercial fields [1][2][3]. Among metal oxides, CeO 2 is considerably attractive because it has the unique characteristics of a reversible valence change between Ce 4+ and Ce 3+ and oxygen vacancies (V o ) [4], leading to the engineering of V o to improve its photocatalytic performance [5][6][7]. With the tunable photocatalytic properties of CeO 2 , photocatalysis by its action has been highlighted as a promising method for solving energy issues with plentiful sunlight [8,9].
Among various energy problems, the depletion of petroleum-based fuels is primary tasks to be solved because they directly affect the welfare of humanity. In this regard, cheap and abundant biomass is being considered as a potential alternative to petroleum-based fuels. The effective utilization of biomass can play a crucial role in solving the current energy crisis [10,11]. To this end, a facile strategy involves the oxidation of biomass and its derivatives using CeO 2 nanoparticles (NPs) as photocatalysts, and numerous efforts have been made to understand the mechanism of such reactions for various practical applications [12][13][14]. One of the most widely explored molecules as a biomass platform compound is 5-hydroxymethylfurfural (HMF), which is capable of being selectively oxidized through its two different functional groups: an aldehyde and alcohol [15,16]. To be precise, HMF can be converted to 2,5-furandicarboxylic acid (FDCA) through the production of 5-formyl-2-furancarboxylic acid (FFCA) by two distinct pathways; one is through 5-hydroxymethyl-2-furancarboxylic acid (HMFCA) by the preferential oxidation of the aldehyde group of HMF and the other is via 2,5-diformethylfuran (DFF) by that of its alcohol group (Scheme 1) [12,17,18]. HMFCA and FDCA produced by HMF are considered 5-hydroxymethyl-2-furancarboxylic acid (HMFCA) by the preferential oxidation of the aldehyde group of HMF and the other is via 2,5-diformethylfuran (DFF) by that of its alcohol group (Scheme 1) [12,17,18]. HMFCA and FDCA produced by HMF are considered suitable replacements of furanic polyester and phthalic acid obtained from fossil resources, respectively [19][20][21]. In particular, FDCA is recognized by the U.S. Department of Energy as one of the top 12 bio-based chemicals [22,23]. Therefore, the selective production of HMFCA and FDCA from HMF and the increase in conversion efficiency using CeO2 NPs as photocatalysts have become a research hotspot [24]. Scheme 1. Two reaction pathways for the conversion from HMF to FDCA.
In pursuit of highly selective and efficient production of HMFCA/FDCA, we introduced simple transition metal (TM)-ion doping into CeO2 NPs. It is known that the amount of Vo in CeO2 NPs increases with the modification of the CeO2 surface from Crand Fe-ion doping [9,25]. Therefore, we explored the photocatalytic activity of three distinct CeO2, Cr-doped CeO2 (Cr@CeO2), and Fe-doped CeO2 (Fe@CeO2) NPs through photocatalytic experiments with HMF. In addition, the structural and electronic changes between CeO2 and the doped CeO2 NPs were thoroughly analyzed using a combination of high-resolution transmission electron microscopy (HRTEM), scanning transmission electron microscopy with energy dispersive X-ray spectroscopy (STEM-EDS), X-ray diffraction (XRD), Raman spectroscopy, ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS), and X-ray absorption spectroscopy (XAS).

Chemicals and NP Preparation
CeO2 NPs were synthesized using a modified thermal method [26]. CeCl3•7H2O (0.2 g; Sigma-Aldrich, St. Louis, MO, USA, 99.9%) and PVP (0.3 g; Sigma-Aldrich) were dissolved in 40 mL of deionized water under vigorous magnetic stirring. Then, 1 mL of formamide (Sigma-Aldrich, 99.9%) and 0.2 mL of H2O2 (Sigma-Aldrich) were added to the solution under continuous stirring at 90 rpm for 1 h. The as-formed yellow solution was treated by adding KOH (Sigma-Aldrich) until the pH was 13.0 and then held at that pH for 3 h. Next, the solution was transferred into a Teflon-lined autoclave and heated for 9 h at 220 °C. The resulting CeO2 NPs were filtered and washed with double-distilled water (DDW) to remove any impurities. The products (CeO2 NPs) were then washed with absolute ethanol and dried for 48 h at 95 °C. After fabricating the CeO2 NPs [26,27], two different TM ions (TM nitrate n-hydrate: Cr(NO3)3•9H2O and Fe(NO3)3•6H2O, Sigma-Aldrich, 99%) were added to the CeO2 NPs at 90 °C with moderate stirring until a homogeneous and transparent mixture was formed (approximately 3 h). The obtained solution was then heated at 220 °C for 9 h. The resulting TM@CeO2 NPs were filtered and washed with DDW to remove any impurities, resulting in TM@CeO2 NPs with 5 wt% Cr 3+ or Fe 3+ ions. The photocatalytic properties of CeO2, Cr@CeO2, and Fe@CeO2 NPs were determined using HMF (Sigma-Aldrich, 99%). In pursuit of highly selective and efficient production of HMFCA/FDCA, we introduced simple transition metal (TM)-ion doping into CeO 2 NPs. It is known that the amount of V o in CeO 2 NPs increases with the modification of the CeO 2 surface from Cr-and Fe-ion doping [9,25]. Therefore, we explored the photocatalytic activity of three distinct CeO 2 , Cr-doped CeO 2 (Cr@CeO 2 ), and Fe-doped CeO 2 (Fe@CeO 2 ) NPs through photocatalytic experiments with HMF. In addition, the structural and electronic changes between CeO 2 and the doped CeO 2 NPs were thoroughly analyzed using a combination of high-resolution transmission electron microscopy (HRTEM), scanning transmission electron microscopy with energy dispersive X-ray spectroscopy (STEM-EDS), X-ray diffraction (XRD), Raman spectroscopy, ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS), and X-ray absorption spectroscopy (XAS).

Chemicals and NP Preparation
CeO 2 NPs were synthesized using a modified thermal method [26]. CeCl 3 ·7H 2 O (0.2 g; Sigma-Aldrich, St. Louis, MO, USA, 99.9%) and PVP (0.3 g; Sigma-Aldrich) were dissolved in 40 mL of deionized water under vigorous magnetic stirring. Then, 1 mL of formamide (Sigma-Aldrich, 99.9%) and 0.2 mL of H 2 O 2 (Sigma-Aldrich) were added to the solution under continuous stirring at 90 rpm for 1 h. The as-formed yellow solution was treated by adding KOH (Sigma-Aldrich) until the pH was 13.0 and then held at that pH for 3 h. Next, the solution was transferred into a Teflon-lined autoclave and heated for 9 h at 220 • C. The resulting CeO 2 NPs were filtered and washed with double-distilled water (DDW) to remove any impurities. The products (CeO 2 NPs) were then washed with absolute ethanol and dried for 48 h at 95 • C. After fabricating the CeO 2 NPs [26,27], two different TM ions (TM nitrate n-hydrate: Cr(NO 3 ) 3 ·9H 2 O and Fe(NO 3 ) 3 ·6H 2 O, Sigma-Aldrich, 99%) were added to the CeO 2 NPs at 90 • C with moderate stirring until a homogeneous and transparent mixture was formed (approximately 3 h). The obtained solution was then heated at 220 • C for 9 h. The resulting TM@CeO 2 NPs were filtered and washed with DDW to remove any impurities, resulting in TM@CeO 2 NPs with 5 wt% Cr 3+ or Fe 3+ ions. The photocatalytic properties of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs were determined using HMF (Sigma-Aldrich, 99%).

Characterization
The morphology of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs was analyzed by HRTEM (JEM-ARM200CF, JEOL Ltd., Tokyo, Japan) using an accelerating voltage of 200 kV. In addition, the distribution of the constituent elements within the NPs at the nanoscale was mapped using STEM-EDS (JED-2300T, JEOL Ltd., Japan). The XRD data were recorded in the range 20-100 • in a scanning step of 0.02 • for 0.3 s using a MiniFlex600 system (Rigaku, Tokyo, Japan) with Cu K α radiation operated at 15 mA and 40 kV. Crystallite sizes and lattice parameters were calculated by Pawley refinement of the corresponding diffraction patterns using TOPAS software (Version 4.2, Bruker, Rheinstetten, Germany). Raman spectra were obtained using an XploRA Raman spectrometer (HORIBA, Kyoto, Japan) with a diode-pumped solid-state laser of 532 nm wavelength operating at 10 mW. The UV-Vis DRS experiments were performed on a UV-2600i UV-Vis spectrophotometer (Shimadzu, Kyoto, Japan) equipped with integrated spheres. To compare the electronic structure of CeO 2 and TM@CeO 2 NPs, we obtained Ce M-edge, O K-edge, Cr L-edge, and Fe L-edge spectra using XAS in the 8A1 beamline at the Pohang Accelerator Laboratory with a PHI-3057 electron analyzer (Physical Electronics, Chanhassen, MN, USA) under the base pressure of 2.0 × 10 −9 Torr.

Photocatalytic Experiments
To evaluate the photocatalytic degradation (PCD) activity, suspensions of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs (0.5 g/L) containing HMF (25 mM) were stirred in the dark for 2 h until adsorption equilibrium was attained. A 30 mL volume reactor with a solution containing each NP was placed in front of a blue light-emitting diode (LED) (λ = 365 nm, output power = 6 W; Thorlabs, Newton, NJ, USA) at a distance of 10 cm and magnetically stirred at 80 rpm with and without 5,5-dimethyl-1-pyrroline N-oxide (DMPO; 50 mM). At the distance of 10 cm from the blue LED along the irradiation axis, the light intensity was determined as 10 ± 0.7 mW/cm 2 . The HMF oxidation products were analyzed using high-performance liquid chromatography-mass spectrometry (HPLC-MS; ULTIMATE 3000 RSLC SYSTEM, Thermo Fisher Scientific and Q-EXACTIVE ORBITRAP PLUS MS, Thermo Fisher Scientific, Waltham, MA, USA).

Characterization of TM@CeO 2 NPs
HRTEM images of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs are shown in Figure 1a-c and Figure S1. As shown in the HRTEM images and the inserted fast Fourier transform (FFT) forms, there is no principal structural change between the three NPs (Figure 1a-c). In addition, on the basis of the STEM images and elemental distribution in the STEM-EDS data ( Figure S2), we confirm that the TM ions are homogeneously distributed in the TM@CeO 2 NPs. To obtain additional structural information on the particle regions by closer inspection, the line profiles were analyzed at regular intervals on the sub-nanometer scale for the three NPs, as shown in Figure 1d. Compared with CeO 2 NPs, Cr@CeO 2 and Fe@CeO 2 NPs exhibited a decrease of 10.8 and 9.8 pm per lattice fringe in the interlayer d-spacing of the [111] plane, respectively. In general, compared to CeO 2 NPs, the smaller interlayer d-spacing in TM@CeO 2 NPs reflects the higher density of V o [25,28], suggesting the successful fabrication of highly defective TM@CeO 2 NPs by Cr and Fe ion doping. Therefore, we can predict from the interlayer d-spacing values that the photocatalytic properties of the TM@CeO 2 NPs will be better than those of the CeO 2 NPs.
Next, XRD experiments were performed to investigate the changes in crystal structure of the TM@CeO 2 NPs by doping. In addition, the crystallite sizes and lattice parameters were determined through Pawley refinement of the obtained XRD spectra. As shown in Figure 1e, all the XRD spectra correspond to the diffraction patterns associated with the cubic fluorite CeO 2 crystal structure with the space group Fm-3m, indicating that no principal structural change occurred by TM ion doping, in accordance with the HRTEM results. Moreover, peak broadening for TM@CeO 2 NPs compared to CeO 2 NPs is not clearly recognized in the XRD spectra, which indicates that their grain sizes are similar [9]. The grains in nanoparticles are known to consist of crystallites [29]. Therefore, if we estimate the crystallite size from the XRD data, we can infer the grain size from the crystallite size, which is defined as the coherent diffraction domain. As shown in Table 1, we can clearly confirm that the crystallite sizes of the TM@CeO 2 NPs are similar to those of the CeO 2 NPs within an error range, indicating that the effect of the size of the NPs does not need to be considered in our results. Furthermore, the a-axis lattice parameter of the cubic fluorite CeO 2 crystal structure for TM@CeO 2 NPs is slightly smaller than that for CeO 2 NPs, showing a trend similar to the results of the line profiles measured from the HRTEM images shown in Figure 1d. It can be inferred that the shrinkage of the lattice constant for TM@CeO 2 NPs is due to the introduction of dopants into CeO 2 NPs, leading to an increase in V o , as reported in the literature [30]. Therefore, we expect that the photocatalytic abilities of the TM@CeO 2 NPs will be superior to those of the CeO 2 NPs, in agreement with the HRTEM data. Next, XRD experiments were performed to investigate the changes in crystal structure of the TM@CeO2 NPs by doping. In addition, the crystallite sizes and lattice parameters were determined through Pawley refinement of the obtained XRD spectra. As shown in Figure 1e, all the XRD spectra correspond to the diffraction patterns associated with the cubic fluorite CeO2 crystal structure with the space group Fm-3m, indicating that no principal structural change occurred by TM ion doping, in accordance with the HRTEM results. Moreover, peak broadening for TM@CeO2 NPs compared to CeO2 NPs is not clearly recognized in the XRD spectra, which indicates that their grain sizes are similar [9]. The grains in nanoparticles are known to consist of crystallites [29]. Therefore, if we estimate the crystallite size from the XRD data, we can infer the grain size from the crystallite size, which is defined as the coherent diffraction domain. As shown in Table 1, we can clearly confirm that the crystallite sizes of the TM@CeO2 NPs are similar to those of the CeO2 NPs within an error range, indicating that the effect of the size of the NPs does not need to be considered in our results. Furthermore, the a-axis lattice parameter of the cubic fluorite CeO2 crystal structure for TM@CeO2 NPs is slightly smaller than that for CeO2 NPs, showing a trend similar to the results of the line profiles measured from the HRTEM images shown in Figure 1d. It can be inferred that the shrinkage of the lattice constant for TM@CeO2 NPs is due to the introduction of dopants into CeO2 NPs, leading to an increase in Vo, as reported in the literature [30]. Therefore, we expect that the photocatalytic abilities of the TM@CeO2 NPs will be superior to those of the CeO2 NPs, in agreement with the HRTEM data. Table 1. Crystallite sizes and lattice constants of the cubic fluorite CeO2 crystal structure estimated through Pawley refinement for XRD spectra acquired from CeO2, Cr@CeO2, and Fe@CeO2 NPs. The error rates indicate the 95% confidence interval.  Table 1. Crystallite sizes and lattice constants of the cubic fluorite CeO 2 crystal structure estimated through Pawley refinement for XRD spectra acquired from CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs. The error rates indicate the 95% confidence interval.

Type of NPs Crystallite Size (nm) a-Axis Lattice Parameter (Å)
CeO 2 19.1 ± 1.0 5.4150 ± 0.0008 Cr@CeO 2 18.3 ± 1.1 5.4123 ± 0.0004 Fe@CeO 2 19.2 ± 0.5 5.4136 ± 0.0008 Figure 1f shows the Raman spectra of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs, where a strong band (F 2g ) appears at approximately 467 cm −1 due to the symmetric breathing mode of the oxygen ions coordinated with each Ce 4+ ion in the cubic fluorite CeO 2 structure [31]. Through the Raman profiles, we also confirm that TM ion doping into the CeO 2 NPs barely influences the CeO 2 structure. In addition to the F 2g peak, two weak bands related to the transverse acoustic mode and the non-degenerate longitudinal optical mode are observed at approximately 250 and 600 cm −1 in Figure 1f, respectively [32,33]. These two peaks for the TM@CeO 2 NPs are stronger than those for the CeO 2 NPs. According to the literature, these bands can be attributed to the presence of defects dominantly created by V o [32,33]. In particular, the peak at approximately 600 cm −1 arises from V o caused by the reduced Ce 3+ ions [34]. Hence, we anticipate that TM@CeO 2 NPs will have better photocatalytic properties than those of CeO 2 NPs, owing to the increase in the number of defects attributed to V o .

Electronic Properties of TM@CeO 2 NPs
The doping of CeO 2 NPs with Cr or Fe ions changes the properties of the material at the nanoscale. To understand the implication of this process on the photocatalytic activity, a deeper understanding of the effect of the dopants on the change in electronic properties within the material is necessary. To investigate it, we performed UV-Vis DRS experiments to measure the band gap of the CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs. As shown in Figure  S3, the band gap of CeO 2 NPs was determined to be 3.26 eV, which is almost identical to the reported value of 3.25 eV [35]. Glancing the change in band gap by the dopants, the band gap of Cr@CeO 2 and Fe@CeO 2 NPs slightly increases to 3.32 and 3.34 eV, respectively, suggesting the modulation of the band gap of the CeO 2 NPs by the dopants. To obtain information on the electronic structure of the unoccupied state at the NP surface, we acquired the XAS profiles of the Ce M-edge, O K-edge, Cr L-edge, and Fe L-edge for the CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs, as shown in Figure 2, where the spectra are normalized and overlaid vertically to identify changes in the electronic structure. In Figure 2a, the peaks corresponding to M 5 (marked as B) and M 4 (marked as C) are clearly observed, and additional satellite peaks are also found at higher photon energies for all NPs [36]. The B/C intensity ratio for the two TM@CeO 2 NPs is slightly higher than that of the CeO 2 NPs. This indicates that more Ce 3+ states generated by the oxygen defect structure exist in TM@CeO 2 NPs [37,38]. In addition, the pre-edge peak (marked as A) is directly related to the defect structure. As can be seen from the Ce M-edge spectra, the intensity of peak A for the two TM@CeO 2 NPs is slightly higher than that for the CeO 2 NPs. Therefore, the photocatalytic effects of the TM@CeO 2 NPs containing more defects are expected to be better than that of CeO 2 NPs. The peaks marked as D, E, and F in the O K-edge spectra (Figure 2b) correspond to transition-induced O 2p-Ce 4f, Ce 5d-e g , and Ce 5d-t 2g states, respectively [7,39]. Focusing on the relative intensity between peaks D and E, it can be seen that the D/E intensity ratio of the two TM@CeO 2 NPs is lower than that of the CeO 2 NPs. This indicates that the number of defects in TM@CeO 2 NPs is greater than in the CeO 2 NPs [40][41][42][43]. To confirm the oxidation state and electronic structure of the TMs formed on the surface of the TM@CeO 2 NPs, we measured the Cr L-edge spectrum for Cr@CeO 2 NPs and the Fe L-edge spectrum for Fe@CeO 2 NPs (Figure 2c-d). From these spectra, we confirm that the Cr and Fe L-edge profiles show the shapes of typical Cr 2 O 3 and Fe 2 O 3 composition, respectively [44][45][46][47]. From the XAS data, we envision that the formation of chromium or iron oxides on the surface of the TM@CeO 2 NPs can give rise to oxygen deficiencies, leading to the promotion of conversion of Ce 4+ to Ce 3+ ions to compensate for the charge valence. As a result, we propose that the concentration of V o in the TM@CeO 2 NPs is higher than that in the CeO 2 NPs, resulting in an increase in photocatalytic performance.

Selective Production of HMFCA and High Conversion Efficiency to FDCA by Photolysis of HMF in the Presence of TM@CeO2 NPs
Maximizing the amount of Vo by doping without altering the principal structure of CeO2 NPs is one of the easiest and most effective strategies for improving the efficiency of

Selective Production of HMFCA and High Conversion Efficiency to FDCA by Photolysis of HMF in the Presence of TM@CeO 2 NPs
Maximizing the amount of V o by doping without altering the principal structure of CeO 2 NPs is one of the easiest and most effective strategies for improving the efficiency of the photocatalyst. To verify this, we investigated the photocatalytic activity of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs in the PCD reaction of HMF. As shown in Figure 3a, the PCD reaction rates of HMF in the presence of the two TM@CeO 2 NPs are higher than that of the CeO 2 NPs. Table 2 presents the photocatalytic efficiency of the three NPs for the PCD reaction of HMF for 24 h, and the PCD rates with the TM@CeO 2 NPs are approximately 1.6 times faster than that with CeO 2 NPs. We speculate that this is due to the enhanced photocatalytic performance of the TM@CeO 2 NPs compared to that of the CeO 2 NPs. As mentioned above, HMF, which contains aldehyde and alcohol groups, is converted to FDCA via pathways that produce HMFCA and/or DFF (Scheme 1). To trace the exact reaction mechanism of the photolysis of HMF, we performed a quantitative analysis of the HMFCA, DFF, and FDCA products formed over time using HPLC-MS during the photolysis of HMF with CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs, dividing it into two distinct phases as a function of the reaction time ( Figure 3b). As shown, in Phase A (0-6 h) only HMFCA is produced from the selective oxidation of HMF in all three NPs. Interestingly, DFF is not produced, which clearly suggests that only the aldehyde functional group within HMF is preferentially oxidized to the carboxyl functional group. From these results, we confirmed that only HMFCA, a suitable alternative to furanic polyesters obtained from fossil resources [19][20][21], is selectively produced from HMF. In Phase B (6-24 h), FDCA starts to form gradually over time with simultaneous decreases in the concentrations of HMF and HMFCA. This definitely demonstrates that the conversion reaction from HMF to FDCA occurs only through the formation of HMFCA and not through that of DFF, suggesting that selective extraction of HMFCA from HMF is possible. Furthermore, from the initial concentration of 25 mM HMF, 9.8 and 10.1 mM FDCA were obtained over 24 h for Cr@CeO 2 and Fe@CeO 2 NPs, respectively, which corresponds to approximately a 40% conversion yield, approximately 3.7 times better efficiency than the CeO 2 NPs (Table 2). We also conducted the repeated PCD experiments to check photocatalytic reusability ( Figure S4). As shown in Figure S4, the photocatalytic activity of the NPs was sustained for five consecutive cycles, indicating that they are robust even with repetitive use. As a result, we found that FDCA, a sustainable replacement for petrochemicals [19][20][21], is produced from HMF with a high conversion yield in TM@CeO 2 NPs prepared by the simple method of doping Cr and Fe ions into CeO 2 NPs.

Conclusions
Cr@CeO2 and Fe@CeO2 NPs were developed by introducing a simple method of doping Cr and Fe ions into CeO2 NPs. Based on a thorough analysis of the structural and electronic variation in CeO2, Cr@CeO2, and Fe@CeO2 NPs, we confirmed that the density of Vo in the two TM@CeO2 NPs was greater than that in the CeO2 NPs, suggesting an improvement in their photocatalytic performance. To prove this, we demonstrated the photocatalytic activity of CeO2, Cr@CeO2, and Fe@CeO2 NPs toward HMF oxidation. As expected, the PCD rates of HMF with TM@CeO2 NPs were approximately 1.6 times faster than that with CeO2 NPs. More importantly, a high conversion efficiency of 40% to FDCA through a selective pathway to produce HMFCA was achieved from the photolysis of HMF in the presence of TM@CeO2 NPs, where HMFCA and FDCA are known to be possible starting materials for the production of sustainable alternatives to petrochemical substances.

Conclusions
Cr@CeO 2 and Fe@CeO 2 NPs were developed by introducing a simple method of doping Cr and Fe ions into CeO 2 NPs. Based on a thorough analysis of the structural and electronic variation in CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs, we confirmed that the density of V o in the two TM@CeO 2 NPs was greater than that in the CeO 2 NPs, suggesting an improvement in their photocatalytic performance. To prove this, we demonstrated the photocatalytic activity of CeO 2 , Cr@CeO 2 , and Fe@CeO 2 NPs toward HMF oxidation. As expected, the PCD rates of HMF with TM@CeO 2 NPs were approximately 1.6 times faster than that with CeO 2 NPs. More importantly, a high conversion efficiency of 40% to FDCA through a selective pathway to produce HMFCA was achieved from the photolysis of HMF in the presence of TM@CeO 2 NPs, where HMFCA and FDCA are known to be possible starting materials for the production of sustainable alternatives to petrochemical substances.