Introduction

Carbon monoxide (CO) is generally produced and released from fossil fuel combustion, and its removal is highly necessary in industrial applications. Recently, preferential CO oxidation (PROX) [1–3] has triggered a lot of interests in the application of PEMFC. In the past decades, both noble metals and transition metal oxides have been widely studied for CO oxidation. Although it is generally recognized that noble metals (i.e. Au, Pt, Pd and Rh) [4–7] are the most effective catalysts for CO oxidation, their applications might be limited by the high costs and easy deactivation. Alternatively, CuO/CeO2 catalysts were found to be excellent candidates due to their low costs and high activities [8], and, therefore, they have been widely applied in CO oxidation or CO-PROX [1, 9–12]. The catalytic performances of the CuO–CeO2 catalysts are mainly attributed to the high oxygen storage/release capability of CeO2 via the redox couple of Ce4+/Ce3+ and the strong interaction between copper oxides and ceria.

Continuous efforts have been made to improve the efficiencies of the CuO/CeO2 catalysts. A general approach is the addition of other active components to the catalyst. For example, the addition of Zr4+ [13, 14] and Sn4+ [15] in the lattice of CeO2 could increase the mobility of lattice oxygen and enhance the CO selective oxidation activity. Besides, the addition of α-Fe2O3 could also promote the reactivity of Cu–Ce–Fe–O catalysts [16]. Recently, CuO/CeO2 catalysts modified with manganese have been reported for CO-PROX [17–21]. Li et al. [17, 18] reported that doping proper amount of Mn into CuO–CeO2 catalysts would improve the catalytic performance for CO-PROX reaction, owing to the formation of a more stable Cu–Mn–Ce–O solid solution and the enhancement of the redox properties of catalysts. Tang et al. [22] showed that the addition of copper significantly improved the catalytic activity of MnOx–CeO2 mixed oxide in the complete oxidation of benzene. Recently, Gong et al. [21] reported a series of CuO/CeO2–MnOx with different atomic Mn/(Ce + Mn) ratios, among which the catalyst with a Mn/Ce + Mn atomic ratio of 0.4 was the most active due to a larger BET surface area and more highly dispersed copper oxides on the surface. However, detailed kinetic studies on these catalysts were absent, and thus some essential information on the intrinsic properties of these catalysts could not be provided.

In addition, the activity of the catalyst strongly depends on not only chemical composition of catalysts, but also preparation methods. Even changing the impregnation sequence of components can make an effect on the catalytic performance. For example, Zhang et al. [23] reported that the impregnation sequence of Ir and Fe species on the Al2O3 support had a remarkable effect on the catalytic performance for CO oxidation. In the current work, three types of CuO–MnOx–CeO2 catalysts with different impregnation sequences of CuO and MnOx were prepared. It was found that the impregnation sequence of CuO and MnOx obviously affected the overall CO oxidation activities of the CuO–MnOx–CeO2 catalysts. Detailed kinetic studies were performed on these catalysts to investigate their intrinsic behavior. Based on these results, a correlation between the properties of the catalysts (such as CuO dispersion, the amount of oxygen vacancy, reducibility and the Cu+ content) and the catalytic performance was established.

Experimental

Catalyst preparation

The CeO2 support was prepared using a precipitation method. In a typical synthesis, 25.23 g of Ce(NO3)3·6H2O (58.1 mmol) was completely dissolved in deionized water (500 ml), and then a certain quantity of NH3·H2O solution was gradually added to the solution until the pH of solution was adjusted to about 11.0 to form a precipitate. The precipitate was stirred at 40 °C for 8 h and then aged at 100 °C overnight. Then it was filtered, repeatedly washed with deionized water and dried at 100 °C overnight, followed by calcination at 400 °C for 4 h. The resulting CeO2 powder had a surface area of 58 m2 g−1, as determined by N2 sorption at 77 K.

The supported CuO–MnOx–CeO2 catalysts were prepared by impregnating aqueous solutions of Cu(NO3)2 and Mn(NO3)2 with the CeO2 support. In the preparation procedure of the MnOx(5)/CuO(5)/CeO2 catalyst, an aqueous solution of Cu(NO3)2 was impregnated onto the CeO2 support. The slurry was heated at 90 °C under stirring for 15 min to remove the water, followed by calcination at 400 °C for 4 h in air. Then the obtained CuO/CeO2 powder was immersed in a certain quantity of Mn(NO3)2 solution, and followed the same procedure as described. The obtained catalyst was designated as MnOx(5)/CuO(5)/CeO2 with the number referring to the weight percentage of MnOx (nominally MnO2) or CuO in the catalyst. For the CuO(5)/MnOx(5)/CeO2 catalyst, the preparation process is similar to that of the MnOx(5)/CuO(5)/CeO2, but with the impregnation of Mn(NO3)2 first and then the Cu (NO3)2. For the CuO(5)–MnOx(5)/CeO2 catalyst, a mixed aqueous solution of Cu(NO3)2 and Mn(NO3)2 was impregnated onto the CeO2 support, which was heated at 90 °C under stirring for 15 min to remove the water, followed by calcination at 400 °C for 4 h in air. In addition, two monometallic catalysts CuO(10)/CeO2 and MnOx(10)/CeO2 were also prepared with a similar manner.

Characterization

X-ray diffraction (XRD) patterns were collected on a PANalytic X’Pert PRO MPD powder diffractometer using Cu Kα radiation (λ = 0.1542 nm). The working voltage was 40 kV and the working current was 40 mA. The patterns were collected in a 2θ range from 10° to 110° with a scan rate of 0.15° s−1. The lattice parameter was determined by the Rietveld method using JADE 6.5 software.

Raman measurements were performed on a Renishaw RM1000 with a confocal microprobe, using excitation laser line of 514 nm. Data acquisition was carried out at 25 °C with the scanning range from 200 to 2000 cm−1.

The reducibility of the catalyst was measured by H2 temperature-programmed reduction (H2-TPR) technique. 50 mg of the catalyst was placed in a quartz reactor and heated from 50 to 450 °C with a heating rate of 10 °C min−1 in a gas mixture (5 % H2 in N2, 30 ml min−1). In order to remove adsorbed oxygen and water in the catalyst, the catalyst was subjected to a flow of He (30 ml min−1) at 300 °C for 0.5 h before the experiment. The amount of H2 consumption was determined by a gas chromatograph with a TCD detector, calibrated by the reduction of a known amount of CuO powder.

The X-ray photoelectron spectroscopy (XPS) measurements were performed on an ESCALAB 250Xi high performance electron spectrometer using Mg Kα excitation source (hν = 1253.6 eV) of the twin anode in this work. The pass energies of 150.0 and 10.0 eV were used for recording survey and high-resolution spectra, respectively. Binding energies were calibrated by using the contaminant carbon (C1s = 284.8 eV). The XPS analysis was done at room temperature and pressures on the order of 10−9 mbar. The spectra were deconvoluted using the Avantage software 5.926.

In situ Fourier transform infrared (FTIR) spectra of CO chemisorption over the catalysts were recorded on a NEXUS670 spectrometer equipped with a MCT detector. Self-supported sample wafers (diameter = 1 cm) were prepared from 30 mg of catalyst by pressing at about 3 MPa. The sample was transferred to a quartz IR cell connected to the closed circulation systems and then pretreated under a He flow (30 ml min−1) at 300 °C for 1 h in order to remove water and carbonate in the catalyst. After the pretreatment, the sample was cooled down to 30 °C and the gas mixture (1 % CO in N2, 10 ml min−1) was introduced to the IR cell and the sample was heated from 30 to 300 °C at a ramp of 10 °C min−1. The catalyst was purged with pure He for 30 min to remove the gaseous and physisorbed CO and then a spectrum was collected.

Catalytic testing

The reaction was carried out in a quartz tubular (6 mm i.d.) fixed-bed reactor under atmospheric pressure. 100 mg of catalyst (100–120 mesh) was loaded in the reactor and the reaction temperature was monitored by a thermocouple placed in the middle of the catalyst bed. A feed gas consisting of 1 % CO and 1 % O2 in N2 with a total flow rate of 80 ml min−1, corresponding to a space velocity of 48,000 ml g −1cat h−1, was introduced as the reactants. The catalyst was directly exposed to reaction gas without any pretreatment. The CO concentration of the reactor effluent was analyzed using an Agilent 6850 gas chromatograph equipped with a TCD detector attached to an HP PLOT (30 m × 0.32 mm × 12 μm) column.

Reaction kinetics

The kinetic study was performed on the same fixed bed reactor as mentioned in the catalytic testing. The reaction conversion was controlled by reaction temperature since the loading of the catalyst in each experiment was kept at 50 mg, diluted with same weight of quartz sand. The diluted catalyst was embedded with glass wool on both sides. A thermocouple was inserted into the middle of the catalyst bed to monitor the reaction temperature. In this kinetic experiment, the concentration of CO or O2 in the feed was varied between 0.3 and 3 %. Each measurement was taken after stable rate was achieved, which took about 1 h. Separate experimental tests showed that both external and internal mass transfer could be eliminated using a gas flow rate of 120 ml min−1 and catalyst particles smaller than 100–120 mesh, which therefore were employed for the kinetic study. The absence of mass and heat transfer limitation was verified by checking the Weisz–Prater criterion and Mears’ criterion [24] as applied in our previous work [9]. The kinetic study was conducted at 80 °C for all these catalysts based on that their CO conversion was not exceeding 10 % under these temperatures. CO2 concentration in the outlet gas stream was also analyzed by the same Agilent 6850 gas chromatograph mentioned above. The calculation of CO conversion and the power-rate law expressions were described elsewhere [9].

Results and discussion

Catalyst characterizations

X-ray diffraction (XRD)

The XRD patterns of the CuO–MnOx–CeO2 catalysts are shown in Fig. S1 (see Fig. S1 in Supplementary Information) and their lattice parameters extracted from the patterns are listed in Table 1. All these catalysts have characteristic peaks of fluorite-type oxide structure of CeO2. In addition, two peaks characteristic of CuO phase at 35.5° and 38.8° are detected in the CuO-containing catalysts (except for the CuO(5)–MnOx(5)/CeO2), arising from the aggregation of copper species on the surface of ceria. For the CuO(5)–MnOx(5)/CeO2 catalyst, no crystalline peaks of copper species are observed, which suggests that copper species are highly dispersed on the ceria surface in an amorphous phase and/or incorporated into ceria lattice to form ceria-based solid solution. In other words, the co-impregnation of MnOx with CuO leads to an increase of the CuO dispersion on the surface of supports as confirmed by the patterns. On the other hand, no diffraction peaks corresponding to manganese oxides are found, suggesting that the manganese oxides are highly dispersed or poorly crystallized [25] or combined with the finely dispersed copper oxide and ceria to form solid solution [26]. It can be seen in Table 1 that the lattice parameters of CuO–MnOx–CeO2 catalyst decrease slightly compared to that of pure CeO2. This finding suggests that a small amount of Cu–Mn–Ce–O solid solution [27, 28] could be formed in these catalysts due to the partial substitution of Ce4+ by Cu2+ and/or Mnx+ ions since smaller ionic radii (the radii of Cu2+, Mn4+, Mn3+ and Mn2+ are 0.073, 0.056, 0.062 and 0.067 nm) than that of Ce4+ (0.097 nm). Such a contraction of the lattice leads to a bigger strain, which enhances the generation of oxygen vacancies in the samples and thus improves the redox properties of CeO2 [29].

Table 1 Lattice parameters, H2 consumption, molar contents of O, Mn and Cu species and surface compositions in various catalysts
Full size table

Raman spectroscopy

To obtain further information on the structural features, the catalysts were examined by Raman spectroscopy, as shown in Fig. 1. All the catalysts have two distinct bands at 463 and 612 cm−1. The band at 463 cm−1 is ascribed to the F2g vibration mode of fluorite structure of CeO2 [30]. A slight red shift is also observed in the fluorite F2g band for the CuO–MnOx–CeO2 catalysts, which is related to the incorporation of a small amount of Cu2+ and/or Mnx+ ions to the fluorite lattice [31]. The bands at 612 cm−1 are related to oxygen vacancies in the fluorite ceria lattice [32], which is generated from the formation of the Cu–Mn–Ce–O solution and charge compensation for the partial replacement of Ce4+ ions by the Cu2+ [33] and/or Mnx+. However, it should be noted that the feature band of MnO2 is located at 588 cm−1 [34], the observed peak at 612 cm−1 may contain some contribution from the MnO2. In order to compare the relative amount of oxygen vacancies present in the catalysts, the peak areas of the bands at 463 and 612 cm−1 are calculated based on the results shown in Fig. 1, which are denoted as A463 and A612, respectively. The ratios of A612/A463, which reflect the relative concentrations of oxygen vacancies in the catalysts [29], are also shown in Fig. 1. It can be seen that the pure CeO2 has a value of 0.01, while the supported catalysts have higher ratios. The higher A612/A463 ratios on the supported catalysts strongly suggest that Cu2+ and/or Mnn+ ions penetrate into the CeO2 matrix, leading to the formation of extra oxygen vacancies in these catalysts. In addition, note that the MnOx(5)/CuO(5)/CeO2(0.74) catalyst has the largest ratio compared to the CuO/MnOx(5)/CeO2 (0.51) and the CuO(5)–MnOx(5)/CeO2 (0.71) catalysts, implying that the MnOx(5)/CuO(5)/CeO2 catalyst has the largest amount of oxygen vacancies.

Fig. 1
figure 1

Raman spectra of various catalysts

Full size image

Reducibility

Fig. S2 in the Supplementary Information presents the H2-TPR profiles of various catalysts and their H2 consumptions (calibrated by known amount of pure CuO powder) are listed in Table 1. The pure CeO2 is barely reduced below 450 °C (with a H2 consumption of 20 μmol g−1), which is assigned to the reduction of surface CeO2. The H2-TPR profile of the MnOx(10)/CeO2 exhibits two overlapping reduction peaks at 275 and 344 °C, which is similar to that of pure MnOx [20]. This reduction is assigned to the two-step reduction of MnOx (MnO2/Mn2O3) to Mn3O4, and Mn3O4 to MnO. However, the reduction temperatures of the MnOx/CeO2 are much lower than those of pure MnOx, which indicates synergistic promotion between MnOx and CeO2 in the reduction of the catalysts. As for the CuO–MnOx–CeO2 catalysts, two reduction peaks shift to lower temperature at 190 and 210 °C due to the spillover process promoted by CuO–CeO2 interfacial sites, which are attributed to the overlapped reduction of CuO and MnOx together with Cu2+ and/or Mnx+ ions in the CeO2 lattice [9]. However, these reduction temperatures are slightly higher than those of the CuO(10)/CeO2, which shows three overlapping reduction peaks at 169, 189 and 222 °C. The reduction peak at 169 °C is associated to copper oxide in strict contact with ceria, i.e. to Cu–O–Ce sites [35]. The other two peaks are due to reduction of bulk CuO with different sizes [9]. The H2 consumption of each catalyst is given in Table 1. For all the catalysts, the total hydrogen consumption for CuO–MnOx–CeO2 reduction exceeds the theoretical values, which imply a simultaneous reduction of metal oxide species and support by means of hydrogen spillover and that incorporating Cu and Mn into the CeO2 lattice to form a solid solution facilitated the release of the lattice oxygen. According to the results in Table 1, CuO(5)–MnOx(5)/CeO2 has the highest amount of H2 consumption, the CuO(5)/MnOx(5)/CeO2 has the medium while the MnOx(5)/CuO(5)/CeO2 has the lowest. These results suggest that the CuO(5)–MnOx(5)/CeO2 possesses the best reducibility or oxygen mobility, probably due to its highly dispersed surface species (CuO and MnOx) and strong interaction between the metal oxides and and the CeO2 support.

Oxidation state analysis

Oxidation states of the surface species are analyzed by XPS, as shown in Fig. S3 and Fig. 2. Fig. S3 in the Supplementary Information shows the XPS spectra of the Ce3d (Fig. S3a) and O1s (Fig. S3b) of the CuO(5)–MnOx(5)–CeO2 catalysts. The Ce3d XPS spectra show multiple states arising from different Ce4f level occupancies in the final state [36]. According to the literature [37], the curves of Ce 3d spectra are composed of six peaks corresponding to three pairs of spin–orbit doublets. Letters U and V refer to the 3d3/2 and 3d5/2 spin–orbit doublets, respectively. The peaks labeled as U′ (ca.900.8 eV), U″ (ca.908.0 eV) and U′′′ (ca.916.4 eV) result from Ce4+ 3d3/2 while V′ (ca.882.4 eV), V″ (ca.888.2 eV) and V′′′ (ca.898.3 eV) result from Ce4+ 3d5/2. Unfortunately, no peak ascribed to the Ce3+ ion was observed probably due to the reason that the signal of Ce3+ is too weak and overlapped by the peak of Ce4+. In other words, the Ce species in all CeO2 based catalysts are mostly Ce4+. Fig. S3b displays the O1s XPS spectra of CuO(5)–MnOx(5)–CeO2 catalysts, in which two oxygen species can be distinguished. The lower binding energy at 529.0–529.4 eV was attributed to the lattice oxygen (denoted as Olatt), while the higher binding energy at 530.7–531.3 eV was assigned to surface adsorbed oxygen (denoted as Oads) with low coordination [22]. Compared with the CuO(10)/CeO2, MnOx(10)/CeO2 and CeO2 support, the bimetallic CuO–MnOx–CeO2 catalysts has relatively higher concentration of Oads as shown in Table 1, which are favorable for the mobility of oxygen species and transferring to lattice oxygen via oxygen vacancy [19]. Among these, the MnOx(5)/CuO(5)/CeO2 has the highest concentration of adsorbed surface oxygen (Oads), while the CuO–MnOx(5)/CeO2 has the medium and CuO(5)/MnOx(5)/CeO2 has the lowest, following the same sequence of A612/A463 as displayed in Fig. 1.

Fig. 2
figure 2

a Mn2p and b Cu2p XPS spectra of various catalysts

Full size image

For the Mn 2p spectra (Fig. 2a), the main peaks of the Mn 2p3/2 could be deconvoluted into two components centered at 641.5 and 642.8 eV, which can be assigned to Mn3+ and Mn4+, respectively. As listed in Fig. 2a, it is found that the binding energy (BE) of Mn2p3/2 (Mn3+ and Mn4+) in the CuO(5)–MnOx(5)/CeO2 is 0.3 eV higher than that of MnOx(5)/CuO(5)/CeO2 or the CuO(5)/MnOx(5)/CeO2 catalyst, which are essentially identical to that of the MnOx(10)/CeO2 catalyst, indicating an stronger interaction between the MnOx and CuO or CeO2 in the CuO(5)–MnOx(5)/CeO2 catalyst. Fig. 2b shows the Cu 2p XPS of the CuO–MnOx–CeO2 catalysts. The main peaks of Cu 2p3/2 core level could be deconvoluted into two components located at 932.2 eV attributed to Cu+ species and 934.8 eV assigned to Cu2+ species, along with the shake-up peak above Cu2+ species at 942.1 eV [38], suggesting the presence of copper species in forms of Cu2+ and Cu+. Also, the Cu2p2/3 peak shifts to higher energy compared with that of the CuO(10)/CeO2 catalyst, which implies a strong interaction between CuO and MnOx or the CeO2 support. Surface concentrations of Cu and Mn species listed in Table 1 reveal that the CuO(5)/MnOx(5)/CeO2 contains the highest amount of Cu+ (76 %) and the lowest amount of Mn4+ (about 43.5 %), indicating that an electron transfer process may exist between Mn and Cu species. In the CuO(5)/MnOx(5)/CeO2 catalyst, the Cu2+ are probably most facile to accept electrons provided by Mn species, resulting in the formation of the highest amount Cu+. These results indicate that the capability of the redox couple Cu2+–Cu+ in the catalyst may be strongly enhanced by a large number of Mn4+ ions existing on the surface of CeO2 [39]. As reported by Liu et al. [40], there was a strong interaction between copper, manganese and the support via charge transfer.

The surface contents of Cu and Mn as well as surface composition ratios derived by XPS are also summarized in Table 1. The ratios verify that the CuO(5)/MnOx(5)/CeO2 catalysts possesses the highest amount of exposed surface CuO species among three CuO–MnOx–CeO2 catalysts, while the MnOx(5)/CuO(5)/CeO2 has the least. The surface Cu/(Cu + Mn + Ce) atomic ratios for the MnOx(5)/CuO(5)/CeO2, CuO(5)/MnOx(5)/CeO2 and CuO(5)–MnOx(5)/CeO2 catalysts are 0.31, 0.44 and 0.45, in order. This implies that the CuO(5)–MnOx(5)/CeO2 has the highest dispersion of Cu species, which is consistent with the XRD results.

FTIR spectra of CO chemisorption

Fig. 3 presents the FTIR spectra of CO chemisorption on the catalysts at 80 °C. It is found that the pure CeO2 support and MnOx(10)/CeO2 cannot chemisorb CO. For all the CuO-containing catalysts, bands at 2110 cm−1 are clearly observed. This characteristic band could be assigned to linear chemisorption of CO on Cu+ (Cu+-carbonyl) [11]. The formation of the Cu+ species may be caused by three facts. First, Cu+ species naturally coexists with Cu2+ in the CuO–MnOx–CeO2 catalysts as demonstrated by our XPS results listed in Table 1. Second, Cu+ species are formed as a result of reduction of Cu2+ by CO when the CuO–MnOx–CeO2 catalysts are subjected to a mixture of CO and O2. Martínez-Arias et al. [41] reported that for a fully oxidized CuO–CeO2 catalyst, around 70 % of Cu2+ was reduced to Cu+ by 1 Torr CO even at 300 K and the Cu+ state remains unchanged under 1 Torr O2 at 373 K, indicating a stabilized Cu+ valance state on the catalyst. Third, the formation of another part of Cu+ species is related to the redox process between Cu and Mn species at interfacial position of CuO–MnOx, which could transfer electrons from Mn3+ to Cu2+ to form Cu+, as confirmed by the XPS results (Table 1).

Fig. 3
figure 3

FTIR spectra for CO chemisorption on various catalysts at 80 °C

Full size image

Kinetic study

The kinetic study of CO oxidation was performed over the three types of CuO–MnOx–CeO2 catalysts. By changing the partial pressures of CO and O2, a series of CO conversions were obtained and consequently reaction rates in unit of molCO g −1cat s−1 were calculated. Fig. 4 shows the variation of CO oxidation rates over the CuO–MnOx(5)/CeO2, MnOx(5)/CuO(5)/CeO2 and CuO(5)–MnOx(5)/CeO2 catalysts. For all the catalysts, the reaction rate increases nonlinearly with PCO under constant \({\text{P}}_{{{\text{O}}_{2} }}\) (Fig. 4a), but hardly changes with \({\text{P}}_{{{\text{O}}_{2} }}\) (Fig. 4b) under constant PCO. The results of the non-linear regression using the POLYMATH 5.1 program for the reaction rates of CO oxidation over CuO–MnOx(5)/CeO2, MnOx(5)/CuO(5)/CeO2 and CuO(5)–MnOx(5)/CeO2 catalysts are given in Table 2. The CO pressure dependency of 0.48–0.72 of these catalysts indicates that the reaction probably goes through an intermediate formed by adsorption of CO on the catalyst surface, which is consistent with the FTIR results (Fig. 3), while the O2 pressure dependency of very close to 0 (0.004–0.043) indicates that gaseous O2 is not directly involved in the reaction. These power orders of reactants are comparable to the findings of the CuO/Ce1−xCuxO2−δ catalyst in our previous work [9], except that the CuO(5)/MnOx(5)/CeO2 catalyst has a lower CO pressure dependency (0.48) than that of MnOx(5)/CuO(5)/CeO2 (0.72) and CuO(5)–MnOx(5)/CeO2 catalyst (0.69). In other words, the CuO(5)/MnOx(5)/CeO2 possesses stronger CO adsorption ability than that of the MnOx(5)/CuO(5)/CeO2 and the CuO(5)–MnOx(5)/CeO2 catalysts, which is closely related to its largest amount of Cu+ as shown in Table 1 derived from XPS result. It should be noted that the kinetic study was carried out at 80 °C, which implies a limited temperature window. The reaction orders (particularly for CO) may change if the kinetic study could be conducted at a wider temperature range (e.g. 50–120 °C), by which more general rate expressions would be obtained.

Fig. 4
figure 4

Variation of CO oxidation rate and Arrhenius plots of CO reaction rates over various catalysts with partial pressures of CO under 1 % O2 or O2 under 1 % CO (R2 is the correlation coefficient). Reaction condition: 50 mg catalyst diluted with the same weight of quartz sand (100–120 mesh), the concentration of CO or O2 was varied between 0.3 and 3 %, T = 80 °C

Full size image
Table 2 Summary of kinetics results for CO oxidation over various catalysts
Full size table

The Arrhenius plots of CO reaction rates over all the CuO–MnOx–CeO2 catalysts are also shown in Fig. 4c, from which their reaction activation energies (Ea) are calculated. The activation energies are 51.8 (R2 = 0.998), 73.8 (R2 = 0.980) and 49.5 kJ mol−1 (R2 = 0.992) for the MnOx(5)/CuO(5)/CeO2, CuO(5)/MnOx(5)/CeO2, and CuO(5)–MnOx(5)/CeO2 catalysts, respectively. The highest activation energies on the CuO(5)/MnOx(5)/CeO2 suggest that it is less active than the MnOx(5)/CuO(5)/CeO2 and CuO(5)–MnOx(5)/CeO2 catalysts, which seems to contradict the fact that the CuO(5)/MnOx(5)/CeO2 possesses the highest CO adsorption ability. Nevertheless, it reflects the fact that the adsorption bond energy between CO and catalyst surface is very strong in the CuO(5)/MnOx(5)/CeO2, which is disadvantageous to CO transfer and surface reaction thus resulting in its highest activation energy [13].

By comparing the rate expressions over these three catalysts, it is believed that the CO oxidation over these three catalysts follows the same pathways. These power rate expressions, particularly the zeroth order dependence on O2 pressure, suggests that Mars–van Krevelen (M–K) reaction model is more favorable over these catalysts, similar to the findings obtained by our previous study over a CuO/Ce1−xCuxO2−δ catalyst for CO oxidation [9]. However, the classical Langmuir–Hinshelwood (L–H) model was also applied for CO oxidation over Cu-based catalysts [10, 42] with a reaction order of O2 pressure of near zero (0–0.12). Actually, the rate equation based on the L–H model could also lead to a zeroth reaction order with respect to O2 pressure if one assumes that the adsorption of O2 on the catalyst surface is dominant and near-saturated. Therefore, L–H reaction model could not be ruled out at present and the question remains open.

Catalytic performance

A comparison of CO oxidation performance over three types of CuO–MnOx–CeO2 catalysts as well as CuO/CeO2, MnOx/CeO2 and CeO2 support is shown in Fig. 5. The MnOx/CeO2 shows relatively higher CO conversion compared to pure CeO2 in the temperature range investigated, which indicates that the MnOx species may promote the release of oxygen species from CeO2. The CuO-containing catalysts exhibit much higher activities for CO oxidation than other CuO-free catalysts, suggesting that CuO species play a much more important role in CO oxidation since CuO can provide the Cu+ sites for CO adsorption as shown in Fig. 3. In order to obtain more intrinsic reactivity of these catalysts, the catalytic performances of these catalysts are normalized to specific reaction rate in unit of μmol g Cu s−1. As shown in Table 3, the specific reaction rate of CuO-based catalyst decreased following the sequence of CuO(5)–MnOx(5)/CeO2 > MnOx(5)/CuO(5)/CeO2 ≈ CuO(10)/CeO2 > CuO(5)/MnOx(5)/CeO2. It could be concluded that changing the impregnation sequence of CuO and MnOx species on CeO2 support has obvious influence on the catalytic performance. Interestingly, even though CuO(5)/MnOx(5)/CeO2 has the highest Cu+ contents revealed by XPS (Fig. 2b; Table 1), its CO oxidation activity is lowest among three types of CuO–MnOx–CeO2 catalyst. This observation indicates that the amount of Cu+ in CuO–MnOx–CeO2 catalysts is not the only factor influencing the catalytic activity of CO oxidation. Based on the results summarized in Table 3, there are probably four main factors that influence the catalytic activity. The first one is the CuO dispersion. As reported by Avgouropoulos et al. [43, 44], the best activity of CuO–CeO2 catalyst prepared by urea nitrates combustion method compared to those prepared by the co-precipitation and impregnation method is related to the formation of well-dispersed copper oxide species in strong interaction with ceria particles and the absence of bulk CuO in weak interaction with the ceria support. As evidenced by the XRD results (Fig. S1), the aggregation of copper species on the surface of ceria may be disadvantageous to the CO oxidation of the MnOx(5)/CuO(5)/CeO2 and CuO(5)/MnOx(5)/CeO2 catalyst while the CuO species in CuO(5)–MnOx(5)/CeO2 are highly dispersed. The second one is the amount of oxygen vacancies created in these catalysts. Liu et al. [45] investigated the influence of four preparation methods and obtained the best activity from the chelating preparation that favors the formation of defects into the ceria structure and enhances redox cycle between Cu and Ce. As confirmed by Raman spectroscopy (Fig. 1) and the amount of Oads derived from XPS results (Table 1), MnOx (5)/CuO(5)/CeO2 has the highest amount of oxygen vacancies, while the CuO(5)/MnOx(5)/CeO2 catalyst has the lowest. It could be deduced that the high amount of oxygen vacancy in MnOx(5)/CuO(5)/CeO2 could compensate the weakness of its low CuO dispersion and poor reducibility, result in only a little higher apparent activation energy (51.8 kJ mol−1) than that of CuO(5)–MnOx(5)/CeO2 (49.5 kJ mol−1) catalyst. The third one is the reducibility properties as evidenced by H2-TPR result (Table 2). The co-impregnation of CuO and MnOx species in CuO(5)–MnOx(5)/CeO2 catalyst enhances the reactivity of lattice oxygen which is expected to play an important role in determining the redox behavior and are more facile to extract the oxygen from CeO2 supported than MnOx(5)/CuO(5)/CeO2 and CuO(5)–MnOx(5)/CeO2 as demonstrated by H2-TPR results. Last but not least, the amount of Cu+ species, as confirmed by XPS, is also a factor. The CuO(5)/MnOx(5)/CeO2 has the highest amount of Cu+ adsorption sites, which is a very important factor contributed to CO oxidation. It was confirmed that Cu+ species are the main adsorptive and active centers of CO which is very important for CO oxidation, as the activation barrier of CO molecules could be significantly lowered [9]. However, the CuO(5)/MnOx(5)/CeO2 catalyst has a relatively suppressed activity compared to the other two CuO–MnOx–CeO2 catalysts, which could be related to the lowest amount of oxygen vacancy, poor reducibility and strongest CO adsorption ability, thus resulting in the lowest reaction order of 0.47. In summary, all these factors such as CuO dispersion, the amount of oxygen vacancy, reducibility and Cu+ content could be collectively dedicated to the catalytic activity for CO oxidation, which may account for the best catalytic performance of CuO(5)–MnOx(5)/CeO2 compared to MnOx(5)/CuO(5)/CeO2 and CuO(5)/MnOx(5)/CeO2 catalyst. Based on these results, the CuO dispersion and the amount of oxygen vacancy seems to be more important in the catalytic activities for CuO–MnOx–CeO2 since the highly dispersed CuO species could produce more active sites, accelerating the mobility of lattice oxygen, which was advantageous to enhance its catalytic activity.

Fig. 5
figure 5

CO oxidation over various catalysts. CO oxidation experiment conditions: catalyst weight 100 mg (100–120 mesh), feed gas 1 % CO + 1 % O2 + 98 % N2, flow rate 80 ml min−1, space velocity 48,000 ml g −1cat h−1

Full size image
Table 3 Summary of characterization, kinetic and catalytic results over CuO–MnOx–CeO2 catalysts
Full size table

Conclusions

This work demonstrates that different impregnation sequences of CuO and MnOx have a remarkable effect on the CO oxidation activity of CuO–MnOx–CeO2 catalysts. The different ways of MnOx doping in the catalysts lead to significant modifications in the CuO dispersion, reducibility properties, the amount of oxygen vacancy and Cu+ content in the catalysts. Among these catalysts, CuO–MnOx/CeO2 has the best CO oxidation performance, which could be ascribed to the highest CuO dispersion and the best reducibility or oxygen mobility along with moderate amount of oxygen vacancy and Cu+ on the catalyst surface. The kinetic study for CO oxidation performed over representative catalysts shows that the partial pressure dependency is 0–1 for CO and 0 for O2, which could be interpreted by a Mars–van Krevelen type mechanism over these catalysts. This work demonstrates a relationship between the structural property (such as CuO dispersion, the amount of oxygen vacancy, reducibility, the Cu+ content) and kinetics as well as catalytic behavior for CO oxidation over CuO–MnOx–CeO2 catalyst.