Catalytic oxidation of CO over mesoporous copper-doped ceria catalysts via a facile CTAB-assisted synthesis

Nanosized copper-doped ceria CuCe catalysts with a large surface area and well-developed mesoporosity were synthesized by a surfactant-assisted co-precipitation method. The prepared catalysts with different Cu doping concentrations were characterized by XRD, DLS analysis, TEM, BET, Raman, H2-TPR and in situ DRIFTS techniques. The influence of Cu content on their catalytic performance for CO oxidation was also studied. The XRD results indicate that at a lower content, the Cu partially incorporates into the CeO2 lattice to form a CuCe solid solution, whereas a higher Cu doping causes the formation of bulk CuO. Copper doping favors an increase in the surface area of the CuCe catalysts and the formation of oxygen vacancies, thereby improving the redox properties. The CuCe samples exhibit higher catalytic performance compared to bare CeO2 and CuO catalysts. This is ascribed to the synergistic interaction between copper oxide and ceria. In particular, the Cu0.1Ce catalyst shows the highest catalytic performance (T50 = 59 °C), as well as excellent stability. The in situ DRIFTS results show that CO adsorbed on surface Cu+ (Cu+–CO species) can easily react with the active oxygen, while stronger adsorption of carbonate-like species causes catalyst deactivation during the reaction.


Introduction
Carbon monoxide (CO) is a toxic gaseous pollutant produced due to the incomplete combustion of carbon-containing fuels and in other application elds. [1][2][3] The majority of CO emissions come from modern automobiles; in particular, up to 80% of CO is still emitted during the cold-start stage of a vehicle, namely the rst 30 s of starting a car. 4,5 The catalytic oxidation of CO with energy-efficient and environmentally friendly features has been the most widely used control technology to remediate CO emissions. 6 Although noble metal catalysts offer excellent catalytic properties and desirable stability, their high cost and low abundance limit their application.
Over the last few decades, ceria has been applied as an efficient three-way catalyst (TWC) for the elimination of toxic autoexhaust gases due to its unique redox properties and high oxygen storage capacity (OCS) due to the facile redox cycle between the Ce 3+ and Ce 4+ in the structure. [7][8][9] In particular, much attention has been paid to copper-doped ceria-based catalysts because of their unique activity for ambient CO oxidation. [10][11][12][13][14] It should be noted that synergistic redox properties are produced upon the formation of CuO-CeO 2 interfacial sites, which are considered to be the active sites for the CO oxidation reaction. 15 Moreover, the substitution of Ce 4+ cations with Cu 2+ promotes the formation of structural defects and oxygen vacancies on the catalyst surface, which can favor the generation and mobility of charged species, such as electrons or oxygen anions in the solid catalyst. 16 Thus, defect sites and oxygen vacancies are among the most desirable active species for oxidation catalysis. 17,18 Nevertheless, some recent work has suggested that more surface oxygen vacancies may actually decrease the intrinsic activity of pure CeO 2 . 16,19 On the other hand, the catalytic activities of nanocrystalline mesoporous materials are reported to be superior to those of their nonporous counterparts of the same material, due to their tunable porosity, surface oxygen vacancies, and large surface areas, which promote lattice oxygen mobility. 20,21 For a gas-solid reaction, the large surface area and high mesoporosity of the catalyst favors easy access of the reactants to the active sites and diffusion of the products, 22 thus improving the reaction efficiency on the material surface and resulting in enhanced catalytic performance. It should be noted that the preparation method could exert a basic inuence on the structural properties of the catalysts, like surface area, component dispersion and strength of interaction, which in turn determine the redox properties and reactivity of the nal catalysts. 23 In order to synthesize a mesoporous material, the cationic surfactant CTAB is used as a so template in the precipitation process, which may serve as a growth controller and an agglomeration inhibitor, by forming a covering lm on the newly formed oxide crystal. In addition, the cationic surfactants could reduce the surface tension of the solution and readily be adsorbed onto the surface of the oxide through strong electrostatic interactions. Ma et al. have synthesized CuO/CeO 2 bimetal oxides by using CTAB as the structure directing agent, which possesses a high surface area and disordered worm-like mesoporous structure. 24 Wu et al. have reported that FeMnTiO x mixed oxide catalysts prepared by a CTAB-assisted process 25,26 show obvious advantages in contrast to those prepared by other methods. In our previous work, we have also successfully synthesized mesoporous ceria-based catalysts doped with different transition metals using CTAB as a so template. 27 Though a number of Cu-doped ceria-based catalysts have already been reported, a comparative study is necessary to determine the dependence of the mesoporous structure on the amount of dopant and the inuence of the dopant on the catalytic activity, which have not been well explored yet. In addition, a detailed investigation of the mechanism and catalytic application of doped ceria-based oxides for enhanced CO oxidation activity is still needed.
In this work, a set of CuCe catalysts with different molar ratios were prepared using a facile CTAB-assisted method. The Cu/Ce atomic ratio is optimized by using CO oxidation as a model reaction, to probe the reaction of more complex oxidation processes. Systematic techniques were employed to determine the structural/textural properties of the synthesized catalysts. Moreover, the critical factors for the superior catalytic activity were thoroughly investigated in order to correlate the catalytic behavior with the nature of the catalysts.

Catalyst preparation
The catalysts were synthesized using the co-precipitation method applying CTAB as a so template. 27 Cu(NO 3 ) 2 $3H 2 O (AR) and Ce(NO 3 ) 3 $6H 2 O (AR) were added to deionized water at molar ratios of Cu/(Cu + Ce) ¼ (0, 0.05, 0.1, 0.2, 0.3, and 1). Then, the salt solution was added to the CTAB solution (with a 1 : 1 ratio of CTAB to metal ions) under vigorous stirring for 30 min, giving a clear homogeneous solution. Aerwards, a 1 M NaOH aqueous solution was added dropwise into the mixed solution under vigorous stirring until pH ¼ 11. Aer stirring in the mother liquor for 3 h at 80 C, the obtained precipitate was then ltered and repeatedly washed with sufficient deionized water and alcohol. Finally, the solid product was dried at 110 C for 12 h in an electrical oven and then calcined at 550 C for 6 h. The catalysts were denoted as CeO 2 , Cu 0.05 Ce, Cu 0.1 Ce, Cu 0.2 Ce, Cu 0.3 Ce, and CuO, respectively. For comparison, the reference CeO 2 and Cu 0.1 Ce samples were synthesized by the coprecipitation method without CTAB (denoted as CeO 2 -co and Cu 0.1 Ce-co), and the mechanical mixture CuCe was prepared with a 1 : 9 ratio of CuO to CeO 2 in order to compare it with the Cu 0.1 Ce sample (denoted as Cu 0.1 Ce-m).

Catalyst characterization
Powder XRD patterns were acquired using a BRUKER-AXS D8 Advance X-ray Diffractometer using Cu Ka radiation (l ¼ 0.15418 nm), operated at 40 kV and 40 mA. The diffractograms were recorded in the scanning angle (2q) range of 10-80 with a step size of 0.02 and a scanning speed of 6 min À1 .
The average particle size and size distribution of the catalysts were observed by using the dynamic light scattering (DLS) technique on a Zetasizer Nano-S system (Malvern 3690 Instruments, UK). These measurements throw light on the stability and particle size distribution in uids.
The morphology of the samples was analyzed using a transmission electron microscope (TEM, HT7700, Hitachi, Japan) operated at an acceleration voltage of 100 kV.
The textural properties of the as-prepared samples were determined from N 2 adsorption-desorption isotherms at À196 C on a Micrometrics ASAP 2020 analyzer. Prior to each measurement, the samples were degassed under vacuum at 300 C for 5 h to remove the residual moisture. The specic surface areas were calculated using the Brunauer-Emmett-Teller (BET) method. The pore volumes and sizes were estimated from the pore size distribution curves determined from the adsorption isotherms using the Barrett-Joyner-Halenda (BJH) method.
The Raman spectra were recorded at room temperature (RT) on a LabRAM HR Evolution Raman spectrophotometer (Horiba Scientic, America) equipped with a TE-cooled charge coupled device (CCD) detector, using monochromatic radiation emitted by a laser (l ¼ 532 nm).
The temperature-programmed reduction with H 2 (H 2 -TPR) experiments were performed in a xed-bed quartz reactor with a thermal conductivity detector (TCD) to monitor the H 2 consumption. Prior to the test, a 50 mg sample (40-80 mesh) was pretreated in situ at 500 C for 30 min in a ow of He and cooled to RT. Finally, the reactor temperature was raised up to 900 C at a heating rate of 10 C min À1 in 5 vol% H 2 /N 2 (50 mL min À1 ).

In situ DRIFTS analysis
The diffuse reectance infrared Fourier transform spectra (DRIFTS) were recorded on a Nicolet iS50 spectrometer equipped with a temperature-controllable diffuse reection chamber and a high sensitivity MCT detector. The catalysts were pretreated in a He stream at 500 C for 30 min to clean the surface of the catalysts and then cooled down to RT. The background spectrum was taken at each temperature of interest. The sample wafers were exposed to a controlled stream of 2000 ppm CO or/ and 5 vol% O 2 /He at a ow rate of 50 mL min À1 . All the spectra were determined by accumulating 32 scans at a resolution of 4 cm À1 as a function of adsorption temperature.

Evaluation of catalytic activity
The catalytic activity of CO oxidation was evaluated in a xedbed reactor (i.d ¼ 6 mm) using 100 mg of catalyst (40-80 mesh). The catalysts were pretreated in He (100 mL min À1 ) at 500 C for 30 min to remove surface-adsorbed species. Aer cooling down to RT, the inlet gas was switched to 2000 ppm CO/ 5 vol% O 2 in He with a ow rate of 50 mL min À1 and then a TPO test was carried out in the temperature range from RT to 700 C at a heating rate of 4 C min À1 . The outlet gas from the reactor was analyzed online by a gas chromatograph (GC) (SP-6890, Shandong Lunan Ruihong Chemical Instrument Corporation, China). The values for the conversion of CO in the CO oxidation process are dened as follows: where F is the (inlet or outlet) molar ow of CO. Temperatures corresponding to 50% and 90% CO conversion (denoted as T 50 and T 90 , respectively) were taken as indices of the catalytic activity.

XRD and DLS analysis
The XRD patterns of the samples with different molar ratios of Cu to Ce are shown in Fig. 1. It can be observed that all of the diffraction peaks of the samples are characteristic of the typical cubic uorite phase of ceria (JCPDS no. 65-5923). Additionally, it can be noted that the diffraction peaks of these samples are slightly shied to higher Bragg angles due to the incorporation of Cu ions into the ceria lattice. This interesting observation may lead to the decrease in the lattice parameter due to the smaller size of the copper ions (Table 1). Furthermore, the changes in the lattice parameter could be directly related to the concentration of oxygen vacancies and the strong interaction of CuO with the ceria phase. 28 The changes in these distinct features, including peak shi and lattice parameter, compared to cubic CeO 2 provide evidence of the formation of CuCe solid solutions with Cu doped into the CeO 2 lattice, or/and the existence of some well dispersed CuO species on the surface of CeO 2 . 29-31 However, at a higher Cu content (namely Cu 0.3 Ce), very weak reections at 35.5 and 38.6 attributed to CuO can be detected, which indicates that when the copper doping is higher than the limit of Cu x+ incorporated into the CeO 2 lattice, the remainder could reside on the CeO 2 surface in the form of bulk CuO crystallites. On the other hand, the diffraction peaks of these CuCe samples are broadened resulting from the crystallite nano dimension effect compared with CeO 2 , 32 which is supported by the calculated crystallite size of the samples based on the 111 plane using the Debye-Scherrer equation (Table 1). The smaller crystallite size of the CuCe mixed oxides may be due to the introduction of Cu x+ into the lattice of CeO 2 , which thereby suppresses the crystallite growth. 33 For the comparison samples, the XRD peaks of the CeO 2 -co and Cu 0.1 Ce-co samples also match well the standard diffraction data for uorite ceria; however, a larger crystallite size for these samples can be observed compared with the corresponding CeO 2 and Cu 0.1 Ce samples, which illustrates that CTAB may serve as a growth controller of crystallites in the precipitation process. The particle sizes of the samples were also estimated by performing the DLS technique, which gives the particle size in the form of hydrodynamic radius, considering each particle as a separate sphere in Brownian motion. The size distribution is shown in Fig. 2 in the form of a histogram. The pure CuO sample has a larger particle size with an average value of about 811 nm (calculated based on Fig. 2). A narrow range of particle size distribution can be observed in CeO 2 , Cu 0.05 Ce, Cu 0.1 Ce, Cu 0.2 Ce and Cu 0.3 Ce, whose average particle sizes are 327 nm, 388 nm, 307 nm, 224 nm and 301 nm, respectively. These results reveal that an appropriate amount of copper doping can inuence the particle growth and inhibit the agglomeration of particles due to the synergistic interaction between copper oxide and cerium oxide. Furthermore, these size values are also much larger than those calculated from the XRD results, which is because the crystallite size from XRD is the average grain size while the particle size from DLS analysis is the size of the particles composed of multiple grains. 34 TEM was undertaken to further evaluate the morphological properties of the synthesized representative catalysts, as shown in Fig. 3. A disordered wormhole-like mesostructure, aggregated by metal oxide nanoparticles with a diameter of several nanometers, can be directly observed for the samples. The accessible pores are connected randomly, lacking discernible long-range order in the pore arrangement among the nanoparticles. Furthermore, the results also conrmed that the particle size from the DLS analysis is the size of the particles composed of multiple grains.

Textural characteristics
The nitrogen adsorption/desorption isotherms of the CuCe catalysts are plotted in Fig. 4(a). All the samples present classical type IV isotherms with well-dened H2-type hysteresis loops, conrming the formation of mesoporous materials according to IUPAC classication. The H2 type of hysteresis loop reects a complex pore structure, including typical inkpotshaped pores, tubular holes with an asymmetrical distribution of pore diameter and the mesopores of closely packed spherical particle, etc. As can be seen in Fig. 4(b), the mesopores in the Cebased samples are very similar and uniform, and the most probable pore diameter of the catalyst is in the range of 4-6 nm. This result is consistent with the TEM result. The corresponding results of textural properties are summarized in Table 1. It can be seen that the specic surface area of pure CeO 2 is 105 m 2 g À1 , and the BET surface areas of the catalysts increase from 105 to 131 m 2 g À1 with increasing Cu content, indicating that the introduction of Cu ions into the ceria lattice diminished the crystallite size, resulting in an increase in the surface area, which is in accordance with the XRD results. 35 Moreover, as the Cu content increased, the pore size and pore volume of the samples remained stable at around 0.194 cm 3 g À1 and 6 nm, respectively. For comparison, the surface area of CeO 2 -co and Cu 0.1 Ce-co is only 57 m 2 g À1 and 69 m 2 g À1 , respectively, which is smaller than that of CeO 2 (105 m 2 g À1 ) and Cu 0.1 Ce prepared with CTAB (121 m 2 g À1 ). Moreover, the samples prepared without CTAB also possess a smaller pore size and pore volume compared with the corresponding sample using CTAB. These results indicate that the use of CTAB in the precipitation process favors the synthesis of mesoporous materials with a large surface area. The large surface area and high mesoporosity of the catalyst allows easy access of reactants to the active sites and diffusion of products from the catalyst surface.

Raman spectroscopy
Raman spectroscopy allows us to probe the presence of oxygen defects in ceria-based materials. It can be seen from Fig. 5 that all samples exhibit a prominent and symmetrical band at  Paper 464 cm À1 corresponding to the triply degenerated F 2g Raman active mode of uorite-type structure CeO 2 and a weak band at 595 cm À1 attributed to the presence of Frenkel-type anion defects in the ceria lattice. 16,36 In the latter case, an oxygen ion is displaced from its lattice position to an interstitial position, thus leading to a vacancy at its initial position and a defect at the interstitial site. In addition, the ratio between the areas of the peaks at 595 cm À1 and 464 cm À1 (noted as A 595 /A 464 ) is the most appropriate way to compare the concentration of oxygen vacancies in the solid solutions: the higher the value of A 595 / A 464 , the higher the amount of oxygen vacancies. 33,36 The calculated results are summarized in

H 2 -TPR
The redox properties of the catalysts were characterized by H 2 -TPR. As can be seen in Fig. 6, pure CeO 2 exhibits two reduction peaks due to the surface and bulk reduction, which are centred at 500 C and 800 C, respectively. The difference in these two successive reduction temperatures is due to the different binding energy of oxygen to cerium cations in the lattice. 37 A well-dened two-step reduction prole (designated as a and b) is observed for all CuCe catalysts in the range of 100-250 C, which indicates that there are at least two copper species in the catalysts. It has been reported that the a and b peaks can be ascribed to the reduction of nely dispersed CuO species strongly interacting with the CeO 2 and the reduction of larger CuO particles weakly associated with CeO 2 , respectively. 12, 24 The a and b peaks coexist in all TPR proles even at 5 mol% CuO content, and the area of peak b is always larger than that of peak a. With increasing CuO content from 5 to 30 mol%, the intensities of the a and b peaks increase. When the CuO content increases up to 30 mol%, the intensity of the b peak continually increases and the a and b peaks coalesce and overlap to some extent, which could be due to some larger CuO particles occurring, corresponding to the fact that a separate CuO phase is obviously detected by XRD over the Cu 0.3 Ce sample. It can be clearly seen that with an increase in Cu amount from 5 to 10 mol%, peak a moves from 175 C to 154 C. It should be noted that the reduction peak temperature of Cu 0.1 Ce is much lower than that of Cu 0.1 Ce-co, which suggests that the redox property of Cu 0.1 Ce with more oxygen vacancies is stronger than that of Cu 0.1 Ce-co. On a further increase in the Cu amount, it stabilizes at around 154 C. Furthermore, no reduction peak can be seen for CeO 2 at a temperature less than 400 C. As is well known, the reduction prole of pure CuO is characterized by a single peak at about 300 C. The H 2 -TPR peak temperature of the CuCe catalysts is much lower than that of pure CuO and CeO 2 . This nding suggests that a strong synergistic interaction between ceria and copper oxide takes place, namely, the Ce-O bond is weakened by the substitution of ceria and Cu-O-Ce  structures are formed on the surface, leading to an easier reduction of both CeO 2 and CuO. Quantitative analysis of the H 2 consumption from RT to 350 C was carried out ( Table 2) and the results show that the H 2 -consumption progressively increases with the Cu-content, thus conrming the benecial role of copper towards the reducibility of ceria-based catalysts. On the other hand, for Cu 0.05 Ce and Cu 0.1 Ce samples, the ratio of actual amount of H 2 consumption to the theoretical amount of H 2 consumption is larger than 1, suggesting that the reduction peak areas come from the reduction of CuO, but also CeO 2 takes part in the reaction in the range of lower temperature due to the strong synergistic interaction. However, the H 2 /Cu ratio is less than 1 for the Cu 0.2 Ce and Cu 0.3 Ce, which implies that more large CuO particles weakly interact with CeO 2 , conrming that the reduction of surface oxygen species could be hindered due to the formation of excess CuO crystallites with higher Cu-content, which reduces the interaction of Cu-O-Ce structures. In general, the high value of H 2 consumption implies that the samples contain more reducible oxygen species. As discussed in the XRD section, the oxygen vacancies would be generated once the Cu ions are incorporated into the CeO 2 lattice, which is also conrmed by Raman analyses. The oxygen vacancies would result in charge imbalance and lattice distortion of CeO 2 , thereby creating more active sites on the surface that adsorb more oxygen molecules. The adsorbed oxygen molecules are very reactive oxygen species and are readily reduced at relatively low temperatures.

Catalytic performance
The activity results corresponding to various samples are represented in Fig. 7(a). For comparison, CO oxidation was also performed with CeO 2 -co, Cu 0.1 Ce-co and Cu 0.1 Ce-m samples. It is obvious that pure CeO 2 exhibits almost no activity in this temperature window and CuO shows modest catalytic activity. The CeO 2 -co sample exhibits lower catalytic performance compared with the CeO 2 prepared with CTAB as expected. The increased catalytic performance of the CeO 2 sample can be ascribed to the mesostructures and high surface area. In contrast, when the content of Cu reaches 5 mol%, its catalytic activity is improved to a certain extent, suggesting that appropriate doping of CuO into CeO 2 contributes to an increase in catalytic performance for CO. The CO conversions gradually increase with an increase in the content of Cu until 10 mol% (Cu 0.1 Ce). The mechanical mixture Cu 0.1 Ce-m also exhibits enhanced catalytic performance compared with single CeO 2 and CuO, which conrms the synergistic interaction between CuO and CeO 2 . For comparison purposes, we have calculated the temperature at which 50% conversion occurs, denoted as T 50 . The Cu 0.1 Ce catalyst exhibits the best catalytic activity and the T 50 is 59 C, which is also superior to that of the Cu 0.1 Ce-co sample. With further increasing CuO content, the catalytic performance of the samples decreases slightly. The catalytic activity of CO oxidation in terms of T 50 is in the order: Cu 0.1 Ce > Cu 0.2 Ce ¼ Cu 0.3 Ce > Cu 0.05 Ce > CuO > CeO 2 .
The catalytic oxidation of CO on an active metal supported ceria catalyst is proposed to involve reductive and oxidative steps, which was interpreted by the classical Mars-van Krevelen and Langmuir-Hinshelwood mechanisms. In the former mechanism, the adsorbed CO (CO ad ) is oxidized by lattice oxygen (O L ). In the latter case, the CO ad is oxidized mainly by adsorbed oxygen (O ad ), which is the rate-determining process. Therefore, the mobility and/or reducibility of the active oxygen (O ad and/or O L ) is of crucial importance for the CO oxidation reaction. Following the previous characterizations, some reasonable analyses under the current conditions are concluded a The area ratio of the Raman bands at 595 and 464 cm À1 . b The total H 2 consumption of the reduction peaks from RT to 350 C. c The ratio of actual amount of H 2 consumption to theoretical amount of H 2 consumption. d Temperature for 50% and 90% conversion for CO oxidation. Paper to further illustrate factors that affect the catalytic activity. The results of XRD suggest that the incorporation of Cu ions with smaller ionic radii into the ceria lattice results in defective solid solutions. The synergistic interaction between CuO and CeO 2 has a positive effect on the generation of surface oxygen vacancies, which has also been conrmed by the Raman spectroscopy study. Gaseous oxygen in the feed gases could be absorbed at surface oxygen vacancies and sequentially activated as active oxygen (O 2 À ) by activation of the catalysts. In addition, the generated oxygen vacancies acting as active sites could also improve the mobility of oxygen species from the bulk to the surface, which is well shown by the results of the H 2 -TPR experiment. The large surface area and high mesoporosity of the doped catalyst allows easy access of reactants (CO) to the active sites and diffusion of products (CO 2 ) from the catalyst surface. Thus, the Cu-doped ceria-based catalysts exhibit better catalytic performance than pure oxides. According to the results of structure and Raman characterization, the BET surface areas and oxygen vacancies of the doped catalysts increase with the Cu content, but the catalytic activity of Cu 0.2 Ce and Cu 0.3 Ce is slightly decreased compared with Cu 0.1 Ce. In this sense, the surface area is not the main factor inuencing the catalytic activity. Furthermore, more surface oxygen vacancies may even decrease the intrinsic activity of the catalysts. 16 These results indicate that the role of ceria surface oxygen vacancies in CO oxidation may be more complex than ever expected. Given that the H 2 /Cu ratio of Cu 0.1 Ce (1.264) is larger than 1, the interface between CuO and CeO 2 has a stronger synergistic interaction for Cu 0.1 Ce, and the interaction decreases with increasing Cu content. It should be noted that the active sites of the Cu-doped ceria for CO oxidation are located in the interface between CuO and the CeO 2 substrate. 38 When the doping amount of CuO increases, some bulk CuO clusters could appear and cover the surface active sites, 39 because a separate CuO phase is obviously detected by XRD over the Cu 0.3 Ce sample.
For practical application purposes, catalytic stability is an important feature for materials used in industrial applications. Therefore, CO oxidation cycling tests have been performed over the Cu 0.1 Ce catalyst as a function of temperature. An additional long-term stability test in normal feed gas was conducted at constant temperature. As shown in Fig. 7(b), the Cu 0.1 Ce catalyst exhibited high stability, and no deactivation occurred for a time-on-stream of 72 h at 120 C (inset to Fig. 7(b)).

In situ DRIFTS analysis
In order to investigate the inuencing factors of the CO oxidation activity and explore the reaction mechanism further, the in situ DRIFTS technique was carried out over the CuCe catalysts and the results are shown in Fig. 8. Fig. 8(a) shows the spectra of CO adsorption over Cu 0.1 Ce with the purpose of probing the surface species. CO can absorb on either Cu surfaces to form adsorbed CO or on CeO 2 surfaces to form various carbonate, bicarbonate and formate species in the 1700-1000 cm À1 range. 13,16,32,40 It is widely reported that the bands at 1575, 1525 and 1299 cm À1 are ascribed to bidentate carbonate species. The bands attributed to bridged carbonate and monodentate carbonates appear at 1399 and 1473 cm À1 . Furthermore, the intensity of the bands attributed to these carbonates enhances on further raising the temperature. In addition, bridged formate (1356 cm À1 ) is observed at low temperature and disappears when the temperature is increased up to 100 C. In principle, the CO molecule is an electron-donor probe during adsorption, so the active interface oxygen species may capture the CO molecule, forming interface carbonate species. The peak at about 2101 cm À1 in the spectra should be attributed to the vibration of Cu + -CO, which indicates the presence of some Cu + species in the sample. Generally, at ambient temperature, the adsorption of Cu + -CO is the most stable. 41,42 Meanwhile, the intensity of the band for Cu + -CO shows a maximum at 125 C and a disappearance at 200 C, which may be due to the interaction of CO with Cu 2+ and Cu + ions with unsaturated coordination leading to their reduction, i.e., from Cu 2+ to Cu + to Cu 0 .
The DRIFTS spectra of CO and O 2 co-adsorption over Cu 0.1 Ce are shown in Fig. 8(b). It can be seen that the intensity of Cu + -CO at ambient temperature is slightly weaker than that of only CO adsorption, which could be due to the preferential adsorption of O 2 molecules on the surface of the catalyst to form O 2 À species in the oxygen vacancies, restraining the adsorption of CO. 42 The Cu + -CO species weakens gradually with the increase in temperature and disappears completely at 100 C, which suggests that gaseous CO may be completely oxidized to CO 2 by excess O 2 . This is conrmed by the catalytic activity of CO oxidation. The bands that appeared between 1200 and 1700 cm À1 increase gradually and reach the maximum at 50 C, then weaken gradually when the temperature exceeds 100 C. This may be due to the carbonate-like species rst being produced through the conversion of CO 2 or other processes and then desorbing from the catalyst surface at higher temperatures. On the other hand, the CO conversion tend to grow slowly when the reaction temperature exceed 100 C. It is reasonable that the carbonate species would limit the reaction rate and lead to low activity due to the produced carbonate-like species blocking some of the active sites, resulting in surface likepoisoning. 13,43 For comparison purposes, CO and O 2 co-adsorption in situ DRIFTS spectra of the Cu 0.2 Ce sample were recorded so as to explore the mechanism and better understand the difference in catalytic activity for the CO oxidation reaction. It can be seen from Fig. 8(c) that the change of the adsorbed species with temperature is similar to that of Cu 0.1 Ce. It should be noted that the band of Cu + -CO shows a lower intensity at RT and obviously exists at 100 C compared to Cu 0.1 Ce. The respective intensities of these Cu + -carbonyls provide a measure to evaluate the potential of CO oxidation for each catalyst. 44 That is to say, the intensity of the Cu + -carbonyl band may be proportional to the CO oxidation activity. The bands attributed to these carbonates and carboxylates with higher intensity reach the maximum at 100 C, and decrease gradually with increasing temperature aer 125 C for Cu 0.2 Ce. Two new bands ascribed to bridged formate and carboxylates are observed at 1348 and 1550 cm À1 . These observations indicate that an increase in Cu doping may improve the adsorption capacity for carbonaceous species, which could result from its larger surface area and more oxygen vacancies. In turn, the carbon deposition phenomenon could easily cover the catalyst surface, blocking some active sites and further causing catalyst deactivation during the reaction.
In order to further determine the inuence of the pretreatment conditions on the catalytic activity, the representative sample Cu 0.1 Ce was pretreated in different atmospheres (He, O 2 and H 2 ) before the CO oxidation reaction, as shown in Fig. 9. The observed T 50 temperatures of the catalysts pretreated with He, O 2 , and H 2 are 59 C, 82 C and 87 C, respectively. Taking into account the H 2 -TPR results, in an oxidizing or reducing atmosphere, Cu x+ species could be oxidized to Cu 2+ or reduced to Cu 0 . Thus, it can be concluded that the presence of more Cu + species could result in higher activity, which conrmed the DRIFTS results. Furthermore, this consequence is in accordance with the literature, 45 which reports partially reduced copper (Cu + ) as the most active species.
Based on the above discussion and previous reports, the proposed cyclic schematic representation of CO oxidation involves a combination of steps as shown in Fig. 10. In the rst step, the surface Cu-O-Ce species could be reduced by CO in the mixed atmosphere to form more Cu + species and surface synergetic oxygen vacancies. This process of electron transfer activates the lattice oxygen nearby the copper species. In the second step, the exposed Cu + and oxygen vacancy could provide an adsorption site for CO and an activation site for O 2 , respectively. Finally, this kind of active oxygen (O 2 À ) can easily react with CO molecules adsorbed on Cu + (Cu + -CO) to form CO 2 via intermediate carbonate-like species. In short, both the lattice oxygen derived from ceria and the active oxygen (O 2 À ) generated on the CuCe interface can react with the adsorbing CO to form CO 2 . The consumed surface oxygen can be relled from bulk diffusion in the catalyst lattice and gaseous oxygen by the oxygen vacancies near to the copper species.

Conclusions
A series of mesostructured CuCe catalysts were successfully prepared via a CTAB-assisted co-precipitation method and tested for low-temperature CO oxidation. In these catalysts, Cu could enter the CeO 2 matrix with a uorite-like structure to form a solid solution, but a higher Cu content causes the formation of bulk CuO. A small crystallite size (5-7 nm) and high surface area (105-131 m 2 g À1 ) of the nanocatalysts can be obtained. The synergistic interaction of Cu and Ce species can produce large amounts of oxygen vacancies and improve the redox ability, thereby favoring the catalytic activity. Among the obtained catalysts, Cu 0.1 Ce exhibits the highest catalytic activity for CO oxidation with the lowest T 50 ¼ 59 C and high catalytic stability. Active oxygen species from oxygen vacancies can easily react with Cu + -CO to produce CO 2 via intermediate carbonatelike species. Furthermore, the stronger adsorption of carbonate-like species causes catalyst deactivation during the CO oxidation reaction due to the fact that the carbon deposition may block some active sites.

Conflicts of interest
There are no conicts to declare.