Fabrication of CeO2–MOx (M = Cu, Co, Ni) composite yolk–shell nanospheres with enhanced catalytic properties for CO oxidation

CeO2–MOx (M = Cu, Co, Ni) composite yolk–shell nanospheres with uniform size were fabricated by a general wet-chemical approach. It involved a non-equilibrium heat-treatment of Ce coordination polymer colloidal spheres (Ce-CPCSs) with a proper heating rate to produce CeO2 yolk–shell nanospheres, followed by a solvothermal treatment of as-synthesized CeO2 with M(CH3COO)2 in ethanol solution. During the solvothermal process, highly dispersed MOx species were decorated on the surface of CeO2 yolk–shell nanospheres to form CeO2–MOx composites. As a CO oxidation catalyst, the CeO2–MOx composite yolk–shell nanospheres showed strikingly higher catalytic activity than naked CeO2 due to the strong synergistic interaction at the interface sites between MOx and CeO2. Cycling tests demonstrate the good cycle stability of these yolk–shell nanospheres. The initial concentration of M(CH3COO)2·xH2O in the synthesis process played a significant role in catalytic performance for CO oxidation. Impressively, complete CO conversion as reached at a relatively low temperature of 145 °C over the CeO2–CuOx-2 sample. Furthermore, the CeO2–CuOx catalyst is more active than the CeO2–CoOx and CeO2–NiO catalysts, indicating that the catalytic activity is correlates with the metal oxide. Additionally, this versatile synthesis approach can be expected to create other ceria-based composite oxide systems with various structures for a broad range of technical applications.


Introduction
As one of the most important rare-earth oxides, ceria (CeO 2 ) has attracted a great deal of research attention due to its high oxygen storage capacity (OSC) and good redox properties [1][2][3].
Because of these unique characteristics, CeO 2 has been widely used as environmental catalysts for the removal of harmful pollutants from exhaust gases, such as oxidation of low-concen-tration of CO [4], selective reduction of NO x with NH 3 [5], and oxidation of volatile organic compounds (VOC) [6,7]. The catalytic activity of CeO 2 is believed to originate from the reversible transformation between Ce 4+ and Ce 3+ and affected by various structural factors [8][9][10], including high surface area, preferential exposure of reactive facets and oxygen vacancy defects. However, pure CeO 2 deactivates seriously at an elevated temperature due to the decline of surface area and OSC. In order to maintain the OSC and enhance the catalytic activity and thermal stability of CeO 2 , CeO 2 -based composite oxides by combining ceria with other low-valence metal oxides have been widely studied. Importantly, due to the synergistic effect between the two components, CeO 2 -based composite oxides exhibit a remarkable catalytic activity that is comparable with or even superior to that of noble metal-based catalysts in some catalytic reactions [11,12].
So far, a remarkable process has been developed for the synthesis of CeO 2 -based composite oxides, including CeO 2 -CuO x [13], CeO 2 -ZnCo 2 O 4 [14], CeO 2 -CoO x [15], CeO 2 -MnO x [16], CeO 2 -ZnO [17], CeO 2 -Fe 2 O 3 [18], and CeO 2 -ZrO 2 systems [19]. Taking the CeO 2 -CuO catalyst as a typical example, the improved catalytic activity is closely related to the synergistic interaction between copper and ceria, which promotes the exchange of charges between Ce 4+ /Ce 3+ and Cu 2+ /Cu + and leads to faster oxidation and reduction than that of the corresponding independent forms. The formation of highly-dispersed copper species promotes the adsorption of CO molecules, while the presence of oxygen vacancies provided by CeO 2 can in turn create active oxygen in the oxidation reactions [12,13]. Therefore, the creation of two-phase interfaces as numerous as possible and, thus, the facilitation of synergistic interaction between two components are necessary to optimize the catalytic performances. The unique structure and texture of CeO 2 -based catalysts is also associated with high activity and stability in the catalytic reaction. For instance, rod-like CeO 2 -CuO catalysts with highly dispersed copper oxide clusters as active species had been reported to exhibit superior activity toward CO oxidation in contrast with commonly used CeO 2 /CuO composite catalysts [13]. Consequently, the construction of ceria-based composite oxides with pore features, hollow structure or/and hierarchical architecture, which possess excellent redox properties and abundant oxygen vacancies, will be favorable for the enhancement of catalytic activity toward CO oxidation.
CeO 2 -based hybrid oxides with hollow structure can be synthesized by a sacrificial-template method based on interfacial oxidation-reduction under mild conditions. For example, Mn 3 O 4 / CeO 2 hybrid nanotubes were created by a template-based process involving a redox reaction between the cryptomelane-type manganese oxide nanowire template and Ce(NO 3 ) 3 [20]. Ce-Mn nanotubes were also fabricated by treating Ce(OH)CO 3 templates with aqueous KMnO 4 solution and subsequent selective washing with HNO 3 to remove the residual Ce(OH)CO 3 [21]. In another case, well-dispersed MnO 2 @CeO 2 -MnO 2 and CeO 2 -CuO x composite hollow spheres were synthesized through a facile reflux method using carbon spheres as sacrificial templates. The obtained material showed a high catalytic activity for CO oxidation [16,22]. In addition, porous/hollow CeO 2 -based composite oxides with high surface area can be prepared through heat treatment of suitable cerium-containing precursors. Typically, uniform porous Ce 1−x Zn x O 2−δ solid-solution nanodisks were synthesized by thermal decomposition of the as-prepared Ce-Zn precursor and exhibited excellent activity for removing CO [17]. CeO 2 -ZnO composite hollow microspheres were fabricated via annealing of a precursor of amorphous zinc-cerium citrate hollow microspheres and presented excellent catalytic activity in CO oxidation [23]. Porous CeO 2 :Cu 2+ materials with a tunable surface area were prepared through the thermolysis of a nanosized CeCu(BTC)(H 2 O) 6 precursor [24]. Impressively, Ce 2 (SO 4 ) 3 was employed as the precursor to synthesize mesoporous CeO 2 -CuO bimetal oxide nanorods without the need for additional heat treatment, and the resultant sample exhibited enhanced catalytic activity in the oxidation of CO [25]. Otherwise, as a special composite structure, heterogeneous core@shell structures are believed to integrate the function of individual nanocrystals and induce the unique synergetic catalytic activities. For instance, ZnCo 2 O 4 @CeO 2 core@shell spheres [14] and Co 3 O 4 @CeO 2 core@shell cubes [26] with tunable CeO 2 shell thickness were prepared by a facile self-assembly method and exhibited promising performance in the catalytic oxidation of CO. Consequently, a suitable choice of templates or cerium-containing precursors and the control of the various factors that govern the morphology, texture and physico-chemical properties, provide a promising approach to fabricate CeO 2 -based mixed oxide with various nanostructures. However, compared with pure CeO 2 , it remains a challenge to synthesize CeO 2 -based mixed oxides with well-defined morphology, tunable chemical composition and distribution, and desirable physico-chemical properties.
Herein, we report a general approach to fabricate uniform CeO 2 -MO x (M = Cu, Co, Ni) composite yolk-shell nanospheres with highly dispersed MO x species, in which CeO 2 yolk-shell nanospheres were first constructed by non-equilibrium heat treatment of a Ce-CPCSs precursor and subsequent treatment with M(CH 3 COO) 2 in a solvothermal process. Due to the well dispersion of MO x and the close contact between CeO 2 and MO x , the resultant nanospheres exhibited improved catalytic activity in the oxidation of CO.

Formation and characterization
The synthetic mechanism for the CeO 2 -MO x yolk-shell nanospheres is illustrated in Figure 1. First, according to the reported method [27], well-dispersed Ce-CPCSs with uniform size were synthesized on the basis of the integration of coordination chemistry with anti-solvent effects for synchronized precipitation. Second, the resultant Ce-CPCSs served as the precursor to produce CeO 2 yolk-shell nanospheres by non-equilibrium heat treatment with a proper heating rate. Heterogeneous contraction, caused by non-equilibrium heat treatment process, is regarded as a promising and effective approach to controllably design hollow structures with single and multilevel shells. The complexity of the shell structures is generally determined by the difference between the cohesive force (F c ) and the adhesive force (F a ) created by a proper temperature ramping rate [28][29][30]. In the early stages of heat treatment, at a high heating rate, a temperature gradient (ΔT) along the radial direction leads to the generation of a dense CeO 2 shell at the surface of the Ce-CPCS core (stage I). Then, heterogeneous contraction takes place because the balance between the opposite forces F c and F a is disturbed. When F c > F a at a high ΔT, the inner core contracts inward and detaches from the preformed outer shell (stage II). During further annealing, the inner core shrinks into a solid structure and consequently CeO 2 yolk-shell spheres form (stage III). Upon solvothermal treatment, M(CH 3 COO) 2 is hydrolyzed to M(OH) 2 and subsequently transformed into MO x and finally deposited onto the surface of CeO 2 yolk-shell spheres, leading to the formation of CeO 2 −MO x composite yolk-shell spheres. The morphology of the obtained samples was investigated by using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The as-synthesized Ce-CPCSs pre- cursor exhibits a well-dispersed solid spherical structure in the range of 400-500 nm (Figure 2a). It has been reported that, during the heating process, the calcination temperature and time is correlated with the crystalline structure of the as-obtained oxides, while a proper heating rate is necessary for the controllable synthesis of hollow structures with different multilevel interiors [30]. In our present work, when the Ce-CPCSs precursor was calcined at 500 °C with a heating rate of 8 °C/min, most of the obtained CeO 2 nanostructures present a yolk-shell spherical morphology with a shell thickness of about 50 nm (Figure 2b-d). In contrast, at a low heating rate of 2 °C/min, the Ce-CPCS spheres were almost homogeneously heated from the surface to the center and resulted in the formation of porous solid spheres along with some hollow structures (Figure 2e-g).
After further reaction between the yolk-shell structured CeO 2 and Cu(CH 3 COO) 2 at 120 °C, the resultant CeO 2 -CuO x com-posites still maintained the yolk-shell morphology without any obvious change after the harsh solvothermal treatment. The morphology of the CeO 2 -CuO x yolk-shell spheres can be observed from TEM images in Figure 3. The strong contrast between outer shell and inside cavity indicates the hollow structure of the spheres. Besides, the size of CeO 2 -CuO x composite yolk-shell nanospheres ranges from 400 to 500 nm with a shell thickness of approximately 50 nm (Figure 3b). The high-resolution TEM image is displayed in Figure 3c. The lattice fringe spacing of 0.31 nm and 0.27 nm can be indexed to the (111) and (200) crystal planes of face-centered-cubic fluorite-type CeO 2 , respectively. No lattice spacing can be corresponded to Cu species. In the corresponding selected-area electron diffraction (SAED) pattern in Figure 3d only the diffraction rings belonging to CeO 2 , indicative of a polycrystalline structure, can be recognized. This result is similar to the previously reported literature [22]. Energy-dispersive X-ray (EDX) analysis was employed to further investigate the elemental distribution of Ce and Cu species in the yolk-shell nanospheres. In Figure 4, the elements Ce, Cu, and O are represented by yellow, red, and green colors, respectively. A uniform distribution of Ce, Cu, and O in each yolk-shell sphere, which is favorable for the synergistic interaction between CeO 2 and CuO x , can clearly be seen.
Ce-Co and Ce-Ni composite oxide nanostructures were prepared by a similar solvothermal process in which CeO 2 yolk-shell nanospheres were mixed with Co(CH 3 COO) 2 or Ni(CH 3 COO) 2 in ethanol solution. Spherical yolk-shell structures in the range of 400-500 nm are clearly observed in the TEM images (Figure 5a,b,d,e). The coexistence of Ce and Co or Ni is further confirmed by the EDX spectra (Figure 5c,f). These results indicate the versatility of this synthesis approach in the preparation of CeO 2 -based transition-metal mixed-oxide nanostructures. Based on this synthesis method, other CeO 2 -based composite oxides with various nanostructures can be expected to be fabricated through the pre-formation of CeO 2 or CeO 2based solid solution with different morphologies and the subsequent decoration of highly dispersed transition-metal oxide cluster species.
Information about crystallinity and phases of the samples were obtained from X-ray diffraction (XRD) analysis. Figure 6a displays the XRD patterns of the as-synthesized CeO 2 -MO x  The Raman spectra of the CeO 2 -MO x yolk-shell spheres are illustrated in Figure 6b. All of the spectra reveal a main band at 450-465 cm −1 , which can be assigned to the F 2g mode of the fluorite structure of CeO 2 . Compared to pure CeO 2 , the peak intensity of the CeO 2 -MO x composite spheres decreases and the peak position shifts from 463 to 453 cm −1 . The peak shift depends on various parameters, including the crystal defects, oxygen vacancies, phonon confinement, and inhomogeneous strain related to the reduced ceria [31,32]. Herein, the shift can be related to the interaction between the MO x species and CeO 2 surface, which leads to lengthening and weakening of the M-O bond by sharing oxygen at the interface [29,30]. In addition, the incorporation of dopants can also result in shifts of the peak positions. This is, e.g., because the ion radius of M 2+ (Cu 2+ = 0.72 Å, Co 2+ = 0.79 Å, Ni 2+ = 0.83 Å) is much smaller than that of Ce 4+ (1.01 Å). Also, when a small number of M 2+ is embedded into the CeO 2 lattice and replaces Ce 4+ , the Raman band shifts to lower wavenumbers because additional oxygen vacancies form to compensate for the valence mismatch between M 2+ and Ce 4+ ions [31]. Additionally, the weak peaks of pure CeO 2 at 592 and 1177 cm −1 can be assigned to the D (defect-induced mode) and 2LO (second-order longitudinal optical mode) bands, respectively, which indicate the amount of surface oxygen vacancies [17]. The bands become more pronounced and shifts to lower wavenumbers for CeO 2 -MO x samples. It appears that CeO 2 -MO x samples, especially CeO 2 -CuO x and CeO 2 -NiO, generate more surface oxygen vacancies than pure CeO 2 .
In order to obtain detailed information about the chemical bonding states of the as-prepared CeO 2 -MO x yolk-shell nanospheres, X-ray photoelectron spectroscopy (XPS) analyses were carried out and the results are shown in Figure 7. As shown in Figure 7a, all samples exhibit similar profiles in the Ce 3d spectral region, which clearly represents a typical Ce(IV) 3d spectrum. Peaks centered at 882.1, 888.6, and 898.1 eV can be assigned to Ce(IV) 3d 5/2 , while peaks located at 900.5, 907.1, and 916.3 eV can be indexed to Ce(IV) 3d 3/2 , respectively [12,20]. In the O 1s spectra in Figure 7b, two peaks marked as O α and O β are clearly identified. According to the literature, the main peak O α at a binding energy of 529 eV is corresponding to lattice oxygen, while the shoulder peak O β at a binding energy of 531.3 eV is ascribed to defective or adsorptive oxygen species, such as low-coordinated oxygen atoms, chemisorbed oxygen, or hydroxy groups [12]. It is well-accepted that O β species are more active than O α species due to their higher mobility. With the stronger O β peak, the CeO 2 -MO x samples are expected to be richer in surface-active oxygen and have a better capacity for oxygen storage. In the Cu 2p spectrum (Figure 7c), two pronounced peaks at about 932.6 and 952.3 eV can be attributed to Cu 2p 3/2 and Cu 2p 1/2 of Cu(I), respectively, whereas a broad shake-up peak observed at around 946 eV can be assigned to the presence of Cu(II) [33]. It is therefore clear that the CeO 2 -CuO x sample contains coexisting Cu + /Cu 2+ oxidation states. In the Co 2p spectrum (Figure 7d), two strong peaks at 780.3 eV assigned to Co 2p 3/2 and 796.3 eV to Co 2p 1/2 , along with two shake-up peaks at around 785.6 and 791 eV, are observed, indicating the coexistence of Co(II) and Co(III) in the CeO 2 -CoO x sample. It has been reported that the oxidation state of Co can be distinguished by the binding energy and the intensity of shake-up satellites of the Co 2p photo-peak. Generally, Co 2+ shows two obvious shake-up satellites at around 785 and 802 eV, while pure Co 3+ displays only a very weak shake-up peak at about 791 eV. If Co 2+ /Co 3+ oxidation states coexist, a plateau in the range of 783-792 eV will be observed instead of two distinct shake-up peaks [31,34]. In Ni 2p spectrum of the CeO 2 -NiO sample (Figure 7e), the binding energies at 855.6 and 873.4 eV are ascribed to the characteristics Ni 2p 3/2 and Ni 2p 1/2 signals of Ni(II), while the peak at around 861.1 eV is the shake-up peak at the high-energy side of the Ni 2p 3/2 edge [31].
Additionally, direct quantification of surface concentration ratio of M has been estimated by the integrated intensities of the M 2p and Ce 3d XPS peaks. As shown in Table 1, the surface M/(M + Ce) ratio calculated from XPS data is 37%, 22% and 28% for CeO 2 -CuO x , CeO 2 -CoO x and CeO 2 -NiO, respectively. Compared with 11.3%, 8.3% and 6.6% for the actual concentration measured by inductively coupled plasma mass spec-trometry (ICP-MS) analysis, the surface concentration is higher than the actual concentration, which indicates the surface enrichment of MO x species. The high concentration of MO x species on the CeO 2 surface may result in the enhanced catalytic activity of the as-synthesized CeO 2 -MO x samples.
In order to further understand the effect of the interaction between CeO 2 and MO x on the redox properties of these binary oxides, hydrogen temperature-programmed reduction (H 2 -TPR) analysis was performed on the various CeO 2 -MO x composite yolk-shell nanospheres. As displayed in Figure 8, pristine CeO 2 shows a clear reduction peak at 540 °C, which is attributed to the reduction of surface-oxygen species (capping oxygen). After the introduction of the CuO x phase, the surface-reduction peak is changed significantly and new peaks appear below 300 °C. For the CeO 2 -CuO x composite yolk-shell nanospheres, two major reduction peaks are observed. The wide low-temperature band centered at 179 °C originates from the highly dispersed CuO x clusters, while the sharp high-temperature peak located at 200 °C is due to the strong interactions in the Cu-[O]-Ce structure [13]. In general, commercial Cu 2 O and pure CuO synthesized using a conventional precipitation method exhibit a H 2 consumption peak in the range of 240-300 °C, related to the reduction in the pure bulk CuO x phase [25,35]. For comparison, the binary Ce-Cu oxide shows relatively low reducibility compared to pure ceria and copper oxides due to the significant interaction between the two phases. In the case of CeO 2 -CoO x sample, three reduction peaks are clearly displayed. Two weak peaks are located at relatively low temperatures of 260 and 326 °C, while the other board one is centered at 450 °C. According to the previous reports [11,36], the reduction peaks of CeO 2 -Co 3 O 4 catalysts in the H 2 -TPR profiles can be generally divided into four temperature regions: α (< 240 °C), β (240-320 °C), γ (320-480 °C), and δ (480-700 °C). The peaks α and β are assigned to the reduction of adsorptive oxygen species on the surface and Co 3+ at the CeO 2 -Co 3 O 4 contact interface to Co 2+ , respectively; while the peaks γ and δ are attributed to the reduction of Co 3 O 4 and Co 2+ which interacted with the CeO 2 support to metal Co, respectively. On basis of these reduction patterns, the H 2 consumption peaks at 260 and 326 °C of the CeO 2 -CoO x sample in our present work could be ascribed to the reduction of Co 3+ interacting with CeO 2 at the interface to Co 2+ and the reduction of CoO x weakly interacting with CeO 2 directly to metal Co, respectively. The broad reduction peak at 450 °C can be related with the reduction of Co 2+ interacting with CeO 2 to metal Co and Ce 4+ cations at the interface between CoO x and CeO 2 . In addition, the pure Co 3 O 4 bulk gets reduced at a high temperature of above 280 °C in the literature [6,11]. For the H 2 -TPR profile of CeO 2 -NiO sample, a sharp peak is clearly observed at around 250 °C, which can be due to the reduction of NiO cluster species weakly interacting with CeO 2 [37]. Comparatively, Ni 2+ ions were reduced to Ni 0 at the temperature of 330-430°C in pure NiO sample in the literature [38]. These results reveal that the strong interaction between the MO x species and CeO 2 is created at the two-phase interface of CeO 2 −MO x composite yolk-shell nanospheres.

Catalytic performance
CO oxidation as a model reaction was carried out to evaluate the catalytic performance of the CeO 2 -MO x composite yolk-shell nanospheres and pristine CeO 2 for comparison. Figure 9 presents the catalytic activities for the above samples.  process, CO molecules may easily diffuse into the cavities of the CeO 2 -CuO x yolk-shell structure and then contact more active interface sites, thus enhancing the CO oxidation activity. Figure 9b shows the TEM image of the Ce-Cu binary oxides yolk-shell spheres after the catalytic test. The yolk-shell nanosphere morphology is largely retained during the catalytic oxidation, suggesting an excellent structural stability. To further explore the durability of the CeO 2 -MO x catalyst, the CeO 2 -CuO x sample was employed as a typical example and a cycling test was performed. As shown in Figure 9c, the CeO 2 -CuO x sample still maintained 100% CO conversion at 165 °C after ten successive cycles. Interestingly, the catalytic activity in the 2nd to 10th run was not reduced, and even slightly higher than that in the first run. The possible reason can be attributed to an improvement of the oxidizability of CuO x species during the high-temperature treatment in the first run.
The initial concentration of M(CH 3 COO) 2 ·xH 2 O in the synthesis process played a significant role in catalytic perfor-mance for CO oxidation. By simply varying the amount of M(CH 3 COO) 2 ·xH 2 O, a series of composite nanospheres was prepared to investigate the effects on the catalytic activity (the detailed synthesis is given in Experimental section). The CO conversion curves are shown in Figure 10.  However, an excessive amount of MO x introduced into CeO 2 may block the pore channels generated during the heating process and cover the active sites of CeO 2 , leading to a decline of catalytic activity.
In principle, the catalytic process mainly involves the adsorption and desorption of gas molecules at the interface or on the surface during CO oxidation over metal-oxide catalysts. According to previously reported literature, the possible CO oxidation reaction mechanism over the ceria-based mixed oxides could be tentatively proposed as follows: -Ce denotes a metal species incorporated into CeO 2 , and □ vac denotes an oxygen vacancy [39]. Essentially, the interaction between CeO 2 and MO x is due to the lengthening and weakening of the M-O bond by sharing oxygen at the interface [40]. Therefore, the molecular oxygen activation and extraction preferentially takes place at the interface site between MO x and CeO 2 . During CO oxidation over the CeO 2 -MO x catalyst, MO x provides chemisorption sites for CO molecules, while CeO 2 traps oxygen molecules with oxygen vacancies and increases the concentration of surface oxygen due to its excellent oxygen storage capacity. In the CO oxidation process, oxygen atoms will transfer from molecular oxygen to the MO x clusters through the CeO 2 , which is in close contact with MO x at the two-phase interface, remarkably promoting oxygen mobility on the surface of CeO 2 -MO x composite and effectively facilitating molecular oxygen activation [41,42].
The CeO 2 -CuO x catalyst exhibited much better activity than the other samples, indicating that the catalytic activity may be related with the intrinsic property of the doped metal and the metal-oxide interactions. For MO x -decorated CeO 2 nanostructures, the enhanced redox properties at the interface sites play a key role in the superior catalytic efficiency in the oxidation reaction [33]. The charge balance of redox couples of Cu 2+ /Cu + and Ce 4+ /Ce 3+ (Ce 3+ + Cu 2+ ↔ Ce 4+ + Cu + ) and the lowering of redox potentials of Cu species interacting with CeO 2 support could be responsible for the improved catalytic performance of the CeO 2 -CuO x catalyst [43]. For Ce-Co and Ce-Ni binary oxides, Co 3+ /Co 2+ and Ni 3+ /Ni 2+ charge pairs will also be balanced with Ce 4+ /Ce 3+ pair at the two-phase interface [44]. Nevertheless, since the redox potentials follow the sequence of Cu 2+ /Cu + << Co 3+ /Co 2+ ≈ Ni 3+ /Ni 2+ , Cu will be more effective than Co or Ni to achieve a charge balance with Ce 4+ /Ce 3+ , leading to more pronounced catalytic activity for CO oxidation [39,44]. In addition, the adsorption capacity for CO molecules is another important factor that influences the catalytic activities. Previous studies have shown that Ni and Co cations are less efficient for CO chemisorption than Cu cations, resulting in inferior CO oxidation activity [44,45]. As previously reported, the catalytic activities of 5% doped M-CeO 2 (M = Cu, Co and Ni) catalyst were evaluated, and the order of reaction rates in CO oxidation were Cu > Co > Ni > undoped [39,43], which is in accordance with the results of our present study.

Conclusion
In

Synthesis of CeO 2 yolk-shell nanospheres
The Ce-CPCSs were synthesized according to [27]. In a typical procedure, 0.

Characterization
Crystallographic phases and purity were investigated by X-ray diffraction (XRD) on a Bruker D8-Advance powder X-ray diffractometer with Cu Kα radiation (λ = 0.15418 nm). The morphologies and structures were examined by scanning elec-tron microscopy (SEM) using a FEI Quanta TM 250 and transmission electron microscopy (TEM) using a FEI Tecnai G2 F20, equipped with an energy dispersive X-ray spectrometer (EDX) for elemental mapping. Raman spectra were carried out using a Bruker Senterra Raman spectrometer with an excitation laser of 532 nm. Surface analysis was obtained by an X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250Xi) with Al Kα radiation. All binding energies were corrected for surface charging by use of the C 1s peak (284.8 eV) of adventitious carbon as reference. The M contents in CeO 2 -MO x samples were analyzed by inductively coupled plasma mass spectrometry (ICP-MS, Agilent NWR 213-7900). Hydrogen temperature programmed reduction (H 2 -TPR) experiment was performed on a PCA-1200 instrument, equipped with a thermal conductivity detector (TCD) to detect H 2 consumption. Typically, 50 mg of the sample was heated (10 °C/min) from room temperature to 700 °C in a 5 vol % H 2 /Ar gaseous mixture with a flow rate of 30 mL/min.

Catalytic tests
The activity measurements were carried out in a continuous flow fixed-bed microreactor at atmospheric pressure. In a typical experiment, 50 mg of catalyst with 500 mg silica sand was loaded into a stainless steel tube. A gas mixture of CO/O 2 / N 2 (1:10:89) with a total flow rate of 50 mL/min flowed through the reactor, equivalent to a weight hourly space velocity (WHSV) of 60000 mL·g −1 ·h −1 . The composition of the gas exiting from the reactor was monitored with an online infrared gas analyzer (Gasboard-3100, China Wuhan Cubic Co.) that can simultaneously detect CO, CO 2 and O 2 . The CO conversion ratio was calculated based on the CO consumption and CO 2 formation.