Copper is used in catalysts for a wide variety of reactions, including CO oxidation, preferential CO oxidation under H₂-rich steam, water-gas shift, methanol synthesis, and steam reforming [1-6]. CeO₂, in view of its high oxygen storage capacity (OSC), ready redox cycle, and the ability to firmly anchor copper species from sintering [7-9], has been widely used as the support, and the resulting CuO_x/CeO₂ catalysts, in which CuO_x represents all likely copper species (metallic Cu, Cu ions, Cu₂O and CuO) and their mixtures, have exhibited excellent catalytic performances in catalyzing various oxidation reactions of CO [10-16]. For examples, Avgouropoulos et al. [10] reported that the catalytic performance of CuO/CeO₂ catalyzed the preferential CO oxidation in excess H₂ was comparable to those of Pt/Al₂O₃ and Au/Fe₂O₃ catalysts. CuO_x/CeO₂ catalysts have been demonstrated to be structurally sensitive, particularly in CO oxidation [17-27]. Liu et al. [28] firstly proposed that CO tended to chemisorb at the surface Cu(I) sites stabilized by the CuO-CeO₂ interaction during CO oxidation catalyzed by CuO_x/CeO₂ catalysts. Since then, various factors, such as compositions [22], preparation methods [29, 30] and cerium precursors [31], have been investigated in detail to understand structure-activity relationship of CuO/CeO₂ catalysts in CO oxidation.

Recently the morphology of oxide nanocrystals has been demonstrated to strongly affect the oxide-involved catalysis [32-35]. According to the Wulff's rule [36], the morphology of a crystalline particle determines the crystal planes exposed on the surface, subsequently affects the surface compositions and geometric structures. CeO₂ nanocrystals with various types of morphologies, including cubes, octahedra, rods, wires, tubes, and polyhedra [37-42], have been successfully synthesized and used as catalysts and catalyst supports, in which morphology-dependent CeO₂ catalysis, including CuO_x/CeO₂ catalysis, has been well established [17, 18, 32-34, 43-47]. Wang et al. [17] reported that sub-nanometer CuO_x clusters supported on CeO₂{110} facets were less active in catalyzing CO oxidation than those supported on CeO₂{111} facets due to a strongly bound Cu-[O_x]-Ce structure on CeO₂{110} facets. Lykaki et al. [18] employed CeO₂ rods, cubes and polyhedra to investigate the CeO₂ morphology effect on the structure and catalytic activity in CO oxidation of CuO/CeO₂ catalysts with a ~7.5% Cu loading. CuO/CeO₂ rods catalyst exhibited the highest catalytic activity, which was related with its abundance of weakly bound oxygen species, relative population of Cu⁺/Ce³⁺ redox pairs, and relative abundance of defects and oxygen vacancies.

In spite of a lot of previous studies, a comprehensive understanding of CuO_x-CeO₂ interactions still lacks. In this paper, we have prepared CuO_x/CeO₂ catalysts with Cu loadings of 0.025%‒5% employing CeO₂ rods, cubes and polyhedra by the incipient wetness impregnation method. CeO₂ morphology-dependent structural evolutions of CuO_x species in various CuO_x/CeO₂ catalysts have been identified and correlated to the catalytic activity in CO oxidation. These results comprehensively elucidate the CuO_x-CeO₂ interactions and structure-activity relationship of CuO_x/CeO₂ catalysts.

All chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd. and used without further purification. Commercial CeO₂ polyhedral nanoparticles (denoted as p-CeO₂) were purchased from Sigma Aldrich. 1% CO/Air, 5% H₂/Ar, 1% CO/Ar, Ar (99.999%), and 1% CO/Ar were purchased from Nanjing Shangyuan Industrial Factory and used without further purification. Ultrapure water (resistance > 18.5 MΩ) was used.

CeO₂ cubes (denoted as c-CeO₂) and rods (denoted as r-CeO₂) were synthesized followed the Mai et al.'s procedure [16]. In the typical method, 1.96 g Ce(NO₃)₃·6H₂O and 16.88 g NaOH were respectively dissolved in 40 and 30 mL ultrapure water. Next the NaOH solution was added dropwise into the Ce(NO₃)₃ solution under continual stirring at room temperature. The acquired solution was stirred at room temperature for an additional 0.5 h and then transferred into a 100-mL Teflon bottle. The Teflon bottle was tightly sealed and hydrothermally treated in a stainless-steel autoclave at a desired temperature for 24 h (100 ℃ for cubes and 180 ℃ for rods). After cooling down to room temperature, the acquired precipitate was collected and washed with ultrapure water for 5 times and then dried in vacuum at 80 ℃ for 16 h. The final solid powder was calcined in a muffle oven at 500 ℃ for 4 h to synthesis CeO₂ cubes and rods.

The CuO_x/CeO₂ catalysts were prepared by a conventional incipient wetness impregnation method. Typically, the desired amount of Cu(NO₃)₂ solution was added dropwise to 0.3 g CeO₂ powders and ultrasonicated for 10 min at room temperature. Afterwards, the sample was kept at room temperature for 24 h and then dried at 80 ℃ for 12 h to obtain the catalyst precursor. The acquired catalyst precursor was calcined in a muffle oven at 500 ℃ for 2 h to synthesis the CuO_x/CeO₂ catalyst. The Cu loading (0.025-5 wt%) is adjusted by changing the concentration of added Cu(NO₃)₂ solution, and the acquired CuO_x/CeO₂ catalysts are denoted as x-CuO_x/b-CeO₂, in which 'x' represents the calculated Cu loading.

The Cu loadings of various CuO_x/CeO₂ catalysts were determined by using a PerkinElmer Optima 7300 DV inductively coupled plasma atomic emission spectrometer (ICP-AES). BET specific surface areas of various CeO₂ morphologies were measured by using a Beckman Coulter SA3100 surface area analyzer. Before the measurement, the sample was first degassed at 300 ℃ for 5 h in a N₂ atmosphere. Power X-ray diffraction (XRD) patterns were conducted by using a Philips X'Pert PRO diffractometer using a nickel-filtered Cu K_α (λ = 0.15418 nm) radiation source with the operation voltage of 40 kV and operation current of 40 mA. X-ray photoelectron spectroscopy (XPS) measurements were performed by using an ESCALAB 250 high-performance electron spectrometer using monochromatized Al K_α (hν = 1486.7 eV) as the excitation source. The likely charging of samples was corrected by setting the binding energy of the adventitious carbon (C 1s) to 284.8 eV. Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) images were obtained on a JEM-2100F high resolution transmission electron microscope with the electron acceleration energy of 200 kV. Visible Raman spectra were recorded in back-scattering configuration by using a LABRAM-HR Confocal Laser Raman spectrometer. Ar⁺ (514.5 nm) laser was employed as the excitation source.

H₂-TPR and CO-TPR experiments were performed by using a Micromeritics ChemiSorb 2750 instrument, in which 50 mg fresh sample was put into the U-shaped quartz microreactor and then heated at a rate of 10 ℃·min^‒1 with a flow rate of 20 mL·min^‒1 in 5% H₂/Ar and 1% CO/Ar, respectively. The consumptions of H₂ and CO were recorded by TCD and mass spectra (MS), respectively. X-ray absorption spectra (XAS) was conducted on the 14W1-XAFS beamline at Shanghai Synchrotron Radiation Facility (SSRF) operating at 3.5 GeV under "top-up" mode with a constant current of 220 mA.

In-situ diffuse reflectance infrared Fourier transformed spectroscopy (DRIFTS) of CO adsorption experiments were conducted on a Nicolet 6700 FT-IR spectrometer equipped with an in situ high-temperature reaction cell (Harrick Scientific Products, Inc.) in series mode with 256 scans and a resolution of 4 cm^‒1 with an MCT/A detector. 100 mg fresh samples were loaded on the sample stage of the reaction cell and purged with Ar at room temperature for 0.5 h. The backgrounds at different temperatures were first recorded under Ar atmosphere. For CO adsorptions, 1% CO/Ar was purged into the reaction cell at different temperatures and kept for 0.5 h to reach the stable state. Subsequently, the DRIFTS spectrum was recorded.

The catalytic activities of various CeO₂ and CuO_x/CeO₂ catalysts for the CO oxidation were evaluated with a fixed-bed flow reactor. 50 mg as-synthesized sample diluted with 50 mg SiO₂ was used and the catalytic reaction was conducted with the reaction gas that consisted of 1% CO/Air fed at a rate of 30 mL·min^‒1. The sample was heated to the desired temperatures at a rate of 2 ℃·min^‒1 and kept at this temperature for 0.5 h until the catalytic reaction reached the steady state. The composition of the effluent gas was tested by an online GC-14 gas chromatograph. The CO conversion was calculated from the change in CO concentrations of the inlet and outlet gases.

Representative TEM images of r-CeO₂, c-CeO₂ and p-CeO₂ (Fig. 1 (A1)(C1)-) are similar to those previously reported [43-47] in which r-CeO₂, c-CeO₂ and p-CeO₂ were proposed to {110}+{100}, {100} and {111} facets, respectively. Corresponding XRD patterns of r-CeO₂, c-CeO₂, and p-CeO₂ (Fig. 2(A)) display typical diffraction peaks arising from the cubic fluorite phase structure of CeO₂ (JCPDS card No. 34-0394). The specific surface area was measured to be 69, 23 and 60 m²/g for r-CeO₂, c-CeO₂, and p-CeO₂, respectively.

The actual Cu loadings of CuO_x/CeO₂ catalysts were determined by ICP-AES (Table S1) to be very close to the calculated values. Figs. 1(A2)-(C2) and (A3)-(C3) show representative TEM and HETEM images of various 5%-CuO_x/CeO₂ catalysts. CeO₂ nanocrystals in various 5%-CuO_x/CeO₂ catalysts retain their original morphologies quite well. The identified lattice fringes of 0.19, 0.27, and 0.31 nm in the HRTEM images of all the 5%-CuO_x/CeO₂ catalysts match with the spacing of CeO₂ {220}, {200} and {111} crystal planes, respectively, while no Cu-associated lattice fringes can be observed. This could be due to the similar contrasts of CuO_x and CeO₂ under HRTEM mode and/or the high dispersion of CuO_x species on CeO₂ even at a Cu loading of 5 wt%.

Fig. 2(A) shows XRD patterns of various CuO_x/CeO₂ catalysts. No diffraction pattern of any likely CuO_x species could be observed in all the CuO_x/CeO₂ catalysts with the Cu loadings below 5%, suggesting the high dispersions of supported CuO_x species. The enlarged XRD patterns show weak CuO(-111) and (111) diffraction peaks in the 5%-CuO_x/c-CeO₂ and 5%-CuO_x/p-CeO₂ catalysts but not in the 5%-CuO_x/r-CeO₂ catalyst, indicating a higher dispersion of CuO_x species on r-CeO₂. The poorer dispersion of CuO_x on c-CeO₂ and p-CeO₂ can be likely attributed to the small specific surface area of c-CeO₂ and the relatively weak CuO_x interaction with the CeO₂{111} facets exposed on p-CeO₂, respectively.

The nature of supported CuO_x species in various CuO_x/CeO₂ catalysts was examined by XAS. As shown in Fig. 2(B), the Cu K-edge XANES spectra of the CuO_x species supported on CeO₂ with different morphologies are quite similar. However, their corresponding differentiated spectra (Fig. 2(C)) show obvious differences. At the lowest Cu loading of 0.025%, 0.025%-CuO_x/c-CeO₂ catalyst exhibits a dominant Cu⁺ feature while 0.025%-CuO_x/r-CeO₂ and 0.025%-CuO_x/p-CeO₂ catalysts exhibit dominant isolated Cu²⁺ species. These results demonstrate the dominance of isolated Cu ions in 0.025%-CuO_x/CeO₂ catalysts, but the Cu²⁺ precursor gets reduced into Cu⁺ when supported on c-CeO₂ but not on r-CeO₂ and p-CeO₂. This indicates the preferential occurrence of redox reaction between the Cu²⁺ precursor and the Ce³⁺ species associated with surface oxygen vacancies on c-CeO₂ during the catalyst preparation process. Similar redox reactions were observed to occur between the Nb⁵⁺ precursor and the Ce³⁺ species on c-CeO₂ and r-CeO₂ to produce the Nb⁴⁺ species [45, 46] and between the Pd²⁺ precursor and the Ce³⁺ species created by surface reduction on p-CeO₂ to produce the Pd(0) species [48]. Thus, c-CeO₂ exhibits much more Ce³⁺ species and associated surface oxygen vacancies that are capable of reducing the Cu²⁺ precursor than r-CeO₂ and p-CeO₂, which should be related to the exclusive O-terminated CeO₂{100} facets exposed on c-CeO₂. The feature of isolated Cu²⁺ species dominates in the spectra of all three 0.05%-CuO_x/CeO₂ catalysts, and the CuO signal appears in the spectra of 0.1%-CuO_x/CeO₂ catalysts and further grows in the spectra of 0.5%-CuO_x/CeO₂ catalysts. These Cu K-edge XANES results demonstrate that as the Cu loading increases, the CuO_x species evolves from the isolated Cu²⁺ ions to the CuO species in CuO_x/r-CeO₂ and CuO_x/p-CeO₂, and from the isolated Cu⁺ ions to the isolated Cu²⁺ ions and then the CuO species in CuO_x/c-CeO₂.

XPS was also used to characterize the CuO_x species in CuO_x/CeO₂ catalysts. Only the CuO_x/CeO₂ catalysts with the Cu loadings not lower than 0.5% show visible Cu 2p XPS spectra (Fig. 3). The Cu 2p_3/2 binding energy of the CuO_x species locates at 934 eV in 0.5%-CuO_x/CeO₂ and 1%-CuO_x/CeO₂ catalysts and then shifts to 933.5 eV in 5%-CuO_x/CeO₂ catalysts. The CuO_x species with the Cu 2p_3/2 binding energies at 934 and 933.5 eV can be assigned to the supported highly dispersed CuO clusters and large CuO nanoparticles, respectively [49-51]. These XPS results demonstrate the evolutions of the CuO_x species from highly dispersed CuO clusters to large CuO nanoparticles with the increase of Cu loading from 0.5% to 5% in CuO_x/CeO₂ catalysts.

Fig. 4 shows Raman spectra of various CeO₂ and CuO_x/CeO₂ catalysts excited by lasers with the wavelength of 514.5 nm. A strong peak at 463‒464 cm^‒1 and two weak peaks at 596 and 1176‒1182 cm^‒1 were observed, corresponding to the F_2g, defect-induced mode (D), and second-order longitudinal modes of the cubic CeO₂ fluoride phase, respectively [52-54]. The I_D/I_F2g ratios were calculated to estimate the defect concentrations of CeO₂ in the CuO_x/CeO₂ catalysts. Fig. 4D shows the variations of I_D/I_F2g ratios of CeO₂ in the CuO_x/CeO₂ catalysts as a function of Cu loadings. The I_D/I_F2g ratio follows the order of r-CeO₂ ≈ c-CeO₂ > p-CeO₂, consistent with the previously reported results [44]. The supported CuO_x species strongly affect the I_D/I_F2g ratio of CeO₂ in CuO_x/CeO₂ catalysts. The I_D/I_F2g ratios of r-CeO₂ in CuO_x/r-CeO₂ catalysts and p-CeO₂ in CuO_x/p-CeO₂ catalysts decrease with the increase of Cu loading up to 0.05%, but then keep growing with the further increase of Cu loading. The I_D/I_F2g ratio of c-CeO₂ in CuO_x/c-CeO₂ catalysts decrease with the increase of Cu loading up to 0.1%, but then keep growing with the further increase of Cu loading. Comparing the evolution of the CuO_x species in CuO_x/CeO₂ catalysts, it can be deduced that the loading of isolated Cu ions on CeO₂ should decrease the density of defects while the loading of CuO species should create more defects. Furthermore, for various 5%-CuO_x/CeO₂ catalysts, the surface concentrations of oxygen vacancies are all increase significantly, indicating that big CuO nanoparticles, which have been proved spectroscopically, interacting with CeO₂ tremendously promotes the generation of surface defects. Thus, the isolated Cu ions seem to preferentially deposit at the defective sites of CeO₂, likely the surface oxygen vacancies.

Fig. 5 shows H₂-TPR profiles of various CeO₂ and CuO_x/CeO₂ catalysts. The CeO₂ nanocrystals (Fig. 5(A)) show typical surface reduction peaks (denoted as α peak) whose peak areas follow an order of r-CeO₂ > p-CeO₂ > c-CeO₂, agreeing with their specific BET surface areas. The loading of CuO_x species greatly promotes the reduction of CeO₂, as demonstrated by the shift of the CeO₂ surface reduction peaks to the low temperatures, and such promotion effects depend on the nature of CuO_x species [55]. Meanwhile, additional reduction features related to the reduction of supported CuO_x species appear. Based on the reduction behaviors in the H₂-TPR profiles, five types of supported CuO_x species on CeO₂ have been proposed [18, 29, 56-59], including isolated Cu ions (denoted as β peak), CuO_x clusters weakly interacting with the support (denoted as γ peak), CuO_x clusters strongly interacting with the support (denoted as δ peak), highly dispersed CuO nanoparticles (denoted as η peak), and large CuO nanoparticles (denoted as σ peak).

As shown in Fig. 5 B-H, the H₂-TPR profiles of CuO_x/CeO₂ catalysts depend on the CeO₂ morphology. In CuO_x/c-CeO₂ catalysts, the reduction peaks are composed of α+β peak and shift gradually to the low temperature with the Cu loading increasing up to 0.1%. A sharp reduction peak emerges at 270 ℃ at 0.1% Cu loading and is composed of α+δ peak since it is accompanied by the obvious weakening of α peak. This indicates that supported CuO_x clusters strongly interacting with CeO₂ are more capable of promoting the surface reduction of CeO₂ than supported isolated Cu ions. Notably, a very weak reduction peak locating at 150 ℃ also appears and is assigned to the γ peak [59] in which the involved Cu_xO clusters were proposed to weakly interact with surface oxygen species on CeO₂. When the Cu loading reaches 0.2%, the α+δ reduction peak further grows while the original α reduction peak at high temperature completely disappears. The α+δ reduction peak shifts to low temperature and saturates at 0.5% Cu loading, and the η peak corresponding to the reduction of highly dispersed CuO nanoparticles appears at lower temperatures than the α+δ reduction peak and grows with the Cu loading, while the σ peak corresponding to the reduction of large CuO nanoparticles appears at higher temperatures than the α+δ reduction peak at 5% Cu loading. Interestingly, a negative peak appears on CuO_x/CeO₂ catalysts with the Cu loading no more than 0.5%, especially on the catalysts with p-CeO₂ as support, which can be reasonably ascribed the reaction of hydroxyl groups on CeO₂ surface to produce H₂ at high temperature [60].

Similar reduction peaks were also observed in the H₂-TPR profiles of CuO_x/r-CeO₂ and CuO_x/p-CeO₂ catalysts, but evolved in different ways. The most obvious difference is that the α reduction peak can remain in CuO_x/p-CeO₂ and CuO_x/r-CeO₂ catalysts up to the Cu loading of 0.2% but in CuO_x/c-CeO₂ catalysts up to 0.05%. This indicates that supported CuO_x species on c-CeO₂ should be more extensively promote the surface reduction of CeO₂ than those on r-CeO₂ and p-CeO₂. This could be likely related with the much smaller surface area of c-CeO₂ than r-CeO₂ and p-CeO₂. Another noteworthy difference is that the very weak γ reduction observed firstly in 0.1%-CuO_x/c-CeO₂ does not appear for all CuO_x/p-CeO₂ catalysts and for CuO_x/r-CeO₂ catalysts till the Cu loading of 1%. On one hand, this indicates the formation of Cu_xO clusters weakly interacting with surface oxygen species mainly on O-terminated CeO₂{100} facets exposed on c-CeO₂ but barely on the (Ce, O)-terminated CeO₂ {111} and {110} facets. On the other hand, this indicates that the CuO_x species is preferentially deposited on the {110} facets of r-CeO₂ and then on the {100} facets.

Various CeO₂ and CuO_x/CeO₂ catalysts were also characterized by CO-TPR (Fig. 6). It can be seen that the reduction of CeO₂ by CO is easier than by H₂. This can be attributed to the fact that CO only needs molecular adsorption to reduce CeO₂ while H₂ needs to dissociatively adsorb to reduce CeO₂. Three main reduction peaks appear at 350, 475, and 700 ℃, corresponding to the reductions of surface CeO₂ (denoted as α peak), surface hydroxyl groups on CeO₂, and bulk CeO₂, respectively [61]. In addition, p-CeO₂ displays a sharp reduction peak at 180 ℃, which could be related with the reduction of oxygen species adsorbed on p-CeO₂ surface. This suggests that although mainly exposing the {111} facets, p-CeO₂ has certain an amount of defective sites capable of adsorbing oxygen, likely relevant to the presence of particles with irregular shapes. The loading of CuO_x species promotes the reduction of CeO₂ by CO, and the reduction peaks of various CuO_x species, including isolated Cu ions (denoted as β peak), CuO_x clusters weakly interacting with the support (denoted as γ peak), CuO_x clusters strongly interacting with the support (denoted as δ peak), highly dispersed and large CuO nanoparticles (denoted as η peak), appear in order in the CO-TPR profiles [62, 63].

One interesting observation is that the initial reduction rates of CuO_x/r or c or p-CeO₂ catalysts vary with the Cu loadings. They are highest for CuO_x/p-CeO₂ catalysts with the Cu loadings up to 0.1%, due to the weakest interaction between dominant isolated Cu ions with the CeO₂{111} facets exposed on p-CeO₂; and they are highest for CuO_x/c-CeO₂ catalysts of the Cu loadings with 0.2% and 0.5%, consistent with the appearance of CuO_x clusters weakly interacting with the support in CuO_x/c-CeO₂ catalysts; and they are highest for CuO_x/r-CeO₂ catalysts of the Cu loadings with 1% and 5%, suggesting that the CuO nanoparticles are most facilely reduced on r-CeO₂.

Another interesting observation is that the low-temperature reduction peak areas of CuO_x/c-CeO₂ catalysts below 200 ℃ are much smaller than those of CuO_x/r-CeO₂ and CuO_x/p-CeO₂ catalysts with similar Cu loadings. This can be attributed to the dominant Cu⁺ species in CuO_x/c-CeO₂ catalysts with low Cu loadings whose consumption of CO is less than that of Cu²⁺ dominant in CuO_x/r-CeO₂ and CuO_x/p-CeO₂ catalysts, and as demonstrated below, to the low-temperature reduction of CuO nanoparticles mainly to Cu⁺ species in CuO_x/c-CeO₂ catalysts with high Cu loadings but mainly to Cu(0) in CuO_x/r-CeO₂ and CuO_x/p-CeO₂ catalysts. The Cu⁺ intermediate in CuO_x/c-CeO₂ catalysts is further reduced by CO at elevated temperatures. This indicates that the Cu⁺ species interacts more strongly with the CeO₂{100} facets exposed on c-CeO₂ and is subsequently more stable than with the CeO₂ {110} and {111} facets respectively exposed on r-CeO₂ and p-CeO₂.

Fig. 7 shows in situ DRIFTS spectra of CO adsorption on various CuO_x/CeO₂ catalysts at different temperatures. It is well established that the CO only adsorb at the Cu⁺ site but barely at the Cu²⁺ and Cu(0) sites at room temperature and above [64]. No CO adsorption could be observed on the CuO_x/CeO₂ catalysts with the Cu loadings below 0.05%, likely due to the very weak signals beyond the detection sensitivity of the employed IR spectrometer. For the CuO_x/CeO₂ catalysts with the Cu loadings no less than 0.05%, reductions of supported CuO_x species occur during CO adsorption processes, giving the vibrational features of adsorbed CO dependent both on the CeO₂ morphologies and adsorption temperatures. Totally two vibrational features of adsorbed CO were observed at 2099‒2103 and 2091‒2095 cm^‒1, which were previously assigned to CO adsorbed at the Cu⁺ species derived from the CuO_x clusters strongly interacting with CeO₂ or highly dispersed CuO nanoparticles and at the Cu⁺ species derived from the CuO_x clusters weakly interacting with CeO₂, previously [23, 50, 65-67].

At a Cu loading of 0.05%, CO adsorption peak appears at 2009‒2013 cm^‒1 only at 150 ℃, indicating the occurrence of reduction of CuO_x clusters strongly interacting with CeO₂ into Cu⁺ at this temperature, consistent with the CO-TPR results. At a Cu loading of 0.1%, this peak appears obviously at 100 ℃ for CuO_x/c-CeO₂, much weaker for CuO_x/p-CeO₂ and does not appear for CuO_x/r-CeO₂. The CO adsorption peak significantly grows at 150 ℃, demonstrating a more extensive reduction of supported CuO_x species into Cu⁺. These observations suggest that the interaction of CuO_x clusters with various CeO₂ surfaces follows an order of CeO₂{110} facets exposed on r-CeO₂ > CeO₂{111} facets exposed on p-CeO₂ > CeO₂{100} facets exposed on c-CeO₂.

An obvious CO adsorption peak at 2095 cm^‒1 appears for CO adsorption on 0.2%-CuO_x/c-CeO₂ at RT but not on 0.2%-CuO_x/r-CeO₂ and 0.2%-CuO_x/p-CeO₂, corresponding to the facile reduction of CuO_x clusters weakly interacting with the support that only forms in CuO_x/c-CeO₂ catalysts. An additional much stronger CO adsorption peak appears at 2103 cm^‒1 for CO adsorption on 0.2%-CuO_x/c-CeO₂ at 100 ℃. This feature also appears for CO adsorption on 0.2%-CuO_x/r-CeO₂ and 0.2%-CuO_x/p-CeO₂ at 100 ℃, but with much weaker intensities.

The CO adsorption behavior on 0.5%-CuO_x/c-CeO₂ is similar to that on 0.2%-CuO_x/c-CeO₂. On 1%-CuO_x/c-CeO₂, CO adsorption at RT gives a strong peak at 2095 cm^‒1. This strong peak should arise from CO adsorption at the Cu⁺ species derived from the highly dispersed CuO nanoparticles, instead of from the CuO_x clusters weakly interacting with CeO₂ observed in the case of 0.2%-CuO_x/c-CeO₂ and 0.5%-CuO_x/c-CeO₂. Thus we tend to assign the feature at 2095 cm^-1 herein to CO adsorbed at the Cu⁺ site coexisting with CuO, i.e., the partially reduced CuO. Our assignment is supported by the observation that the vibrational feature shifts to 2103 cm^‒1 upon CO adsorption at 100 ℃ at which temperature supported CuO nanoparticles are more extensively reduced than at RT. The vibrational feature at 2095 cm^‒1 also appear for CO adsorption at RT on 0.5% & 1%-CuO_x/r-CeO₂ and 0.5% & 1%-CuO_x/p-CeO₂ that contain few CuO_x clusters weakly interacting with the support. When the adsorption temperature increases to 100 ℃, this feature almost disappears and a strong peak appears at 2009-2012 cm^‒1, indicating the reduction of more extensive reduction of the highly dispersed CuO nanoparticles.

It can be seen from Fig. 7 that the intensity of CO vibrational peaks for CO adsorption at adopted temperatures always follow an order of CuO_x/c-CeO₂ > > CuO_x/p-CeO₂ > > CuO_x/r-CeO₂; however, the reduction peak at the same temperature ranges in the CO-TPR profiles is weakest for CuO_x/c-CeO₂ (Fig. 6). These results suggest that the fraction of the reduction of supported CuO_x species into Cu⁺ below 150 ℃ follows an order of CuO_x/c-CeO₂ > > CuO_x/p-CeO₂ > > CuO_x/r-CeO₂ while the fraction of the reduction of supported CuO_x species into Cu(0) follows a reverse order of CuO_x/c-CeO₂ < < CuO_x/p-CeO₂ < < CuO_x/r-CeO₂. These results suggest that the Cu²⁺-CeO₂(100) (c-CeO₂) interaction is weaker than the Cu⁺-CeO₂(100) (c-CeO₂) interaction so that the Cu⁺ intermediate can be stabilized during the reduction of Cu²⁺ by CO, whereas the Cu²⁺-CeO₂(110) (r-CeO₂) interaction is stronger than the Cu⁺-CeO₂(100) (c-CeO₂) interaction so that the Cu²⁺ is reduced to Cu(0) by CO.

CO oxidation was used as a probe reaction to investigate the catalytic performance of various CeO₂ and CuO_x/CeO₂ catalysts. As shown in Fig. 8, the relative catalytic activities among CeO₂ and CuO_x/CeO₂ catalysts vary with both the Cu loading and the CeO₂ morphology. p-CeO₂ and CuO_x/p-CeO₂ catalysts are most active among bare CeO₂ and CuO_x/CeO₂ catalysts with Cu loadings up to 0.1%; CuO_x/c-CeO₂ catalysts are most active among CuO_x/CeO₂ catalysts with Cu loadings up of 0.2% and 0.5%; CuO_x/r-CeO₂ catalysts are most active among CuO_x/CeO₂ catalysts with Cu loadings no less than 1%. Similar results were also previously reported [18, 67]. We found that the orders of catalytic activities of various catalysts with different CeO₂ morphologies follow very well with those of their corresponding initial reduction rates in CO-TPR profiles (Fig. 7). Thus CO oxidation over CeO₂ and CuO_x/CeO₂ catalysts follows the Mvk mechanism, which has been well documented in the literatures [17-27]. The catalytic activities of 0.025%-CuO_x/CeO₂ and 0.05%-CuO_x/CeO₂ catalysts are relatively lower than those of bare CeO₂, ascribed to the decrease of surface oxygen defects (Fig. 4). These results also demonstrate that CO oxidation catalyzed by CuO_x/CeO₂ catalysts is sensitive to both Cu and CeO₂ structures.

The Arrhenius plots of CO oxidation catalyzed by various CeO₂ and CuO_x/CeO₂ catalysts were plotted in order to gain insight into the active structure of various CuO_x/CeO₂ catalysts (Figs. S1-S3), from which the apparent activation energy (E_a) and apparent pre-exponential factor (ln(A)) values, respectively relating to the intrinsic activity and the density of active sites, were derived and summarized (Tables S2-S4). The ln(A) increases with the Cu loading, particularly, it increases are linearly related with the Cu loading for CuO_x/CeO₂ catalysts with similar E_a (Fig. S4). However, the E_a changes in a complex way along with the nature of supported CuO_x species. Fig. 9 shows the variations of E_a as a function of the Cu loadings and corresponding supported CuO_x species. It can be seen that the E_a of CuO_x/CeO₂ catalysts with supported isolated Cu ions, except 0.05%-CuO_x/c-CeO₂, does not vary much from that of corresponding CeO₂ catalysts, implying that their active structures contributing to the catalytic activity remain as CeO₂ while the supported isolated Cu ions should act as spectators.

The E_a of CuO_x/CeO₂ catalysts with supported CuO_x clusters and CuO particles is obviously larger than that of CeO₂, indicating that their active structures contributing to the catalytic activity should be the CuO_x-CeO₂ interface. In CuO_x/r-CeO₂ catalysts, the E_a keeps increasing as the supported Cu species evolves from supported CuO_x clusters to supported highly-dispersed CuO nanoparticles and then supported CuO nanoparticles. In CuO_x/c-CeO₂ catalysts, the E_a remains similar for the Cu loadings of 0.05%‒0.2%, and then increases to similar values for the Cu loadings of 0.5%‒5%. These observations indicate that although with different supported CuO_x species, the dominant active structures in 0.05-0.2%-CuO_x/c-CeO₂ catalysts and in 0.5-5%-CuO_x/c-CeO₂ catalysts are the same, respectively as supported CuO_x clusters-CeO₂ interface and supported CuO nanoparticles-CeO₂ interface. Therefore, the isolated Cu⁺ species dominant in as-prepared 0.5%-CuO_x/c-CeO₂ catalyst likely restructures into supported CuO_x clusters during CO oxidation. In CuO_x/p-CeO₂ catalysts, the E_a keeps increasing as the supported Cu species evolves from supported CuO_x clusters to supported highly-dispersed CuO nanoparticles but then decreases with the formation of supported CuO nanoparticles. These results reveal the complex CuO_x-CeO₂ interfacial catalysis in CO oxidation.

In the CuO_x-CeO₂ interfacial catalysis in CO oxidation, a key step is the formation of CO₂ by the reaction of CO adsorbs at the Cu⁺ sites derived from the reduction of CuO_x species with surface lattice oxygen of CeO₂ [19]. The E_a is mainly determined the reactivity of surface lattice oxygen of CeO₂, i.e., the reducibility of CeO₂, while the lnA is mainly determined by the density of Cu⁺-CeO₂ interface. The lnA reasonably increases with the Cu loading; however, the general observations that the E_a increases with the Cu loading suggests that the intrinsic reactivity of surface lattice oxygen of CeO₂ should be suppressed by the loading of CuO_x. Such a suppressing effect not only depends on the nature of various CuO_x species that follows an order of as CuO nanoparticles > CuO_x clusters > isolated Cu ions, but also vary with the morphology and surface structures of CeO₂.

In summary, employing CuO_x/CeO₂ catalysts with different CeO₂ morphologies (rods, cubes and polyhedra) and Cu loadings (0.025%‒5%) prepared by incipient wetness impregnation method, we have successfully identified morphology-dependent CuO_x-CeO₂ interaction and catalysis in CO oxidation. The surface defects of CeO₂ are preferentially occupied by the CuO_x species with the formation of isolated Cu ions. As the Cu loading increases, the CuO_x clusters strongly/weakly interacting with the CeO₂, highly dispersed CuO nanoparticles, and large CuO nanoparticles form sequentially in CuO_x/CeO₂ catalysts. However, the isolated Cu⁺ ions and CuO_x clusters weakly interacting with the CeO₂ are formed mainly on c-CeO₂ due to the exposed unique O-terminated {100} facets. The interaction of CuO_x clusters with various CeO₂ surfaces follows an order of CeO₂{110} facets exposed on r-CeO₂ > CeO₂{111} facets exposed on p-CeO₂ > CeO₂{100} facets exposed on c-CeO₂. The Cu²⁺-c-CeO₂ interaction is weaker than the Cu⁺-c-CeO₂ interaction so that the Cu⁺ intermediate can be stabilized during the reduction of Cu²⁺ by CO, whereas the Cu²⁺-r-CeO₂ interaction is stronger than the Cu⁺-r-CeO₂ interaction so that the Cu²⁺ is reduced to Cu(0) by CO. CO oxidation catalyzed by CuO_x/CeO₂ catalysts follows the Mvk mechanism and is sensitive to both Cu and CeO₂ structures. These results not only comprehensively elucidated the morphology-dependent CuO_x-CeO₂ interaction but also greatly deepen the fundamental understanding of CO oxidation reactions catalyzed by CuO_x/CeO₂ catalysts.