The catalytic oxidation of CO has wide applications in CO₂ laser exhaust abatement, trace amount CO removal in the enclosed atmospheres and the automotive emission control, and so on [1]. CO oxidation on cobalt oxide has been well investigated recently, due to its low cost and high activity [2-5]. Ceria is presently used in a large number of industrial processes, and it accounts for a large part of the rare earth oxide market [6, 7]. Thereby, ceria plays an important role in catalysis, arising from its ability of storing and releasing oxygen [8, 9]. In our previous work [10], ceria doping was found to have a significant promoting effect on Co₃O₄ catalyst. Ceria can increase not only the surface area of catalyst, but also the reaction speed of molecules on catalyst surface. On the other hand, the characterization of catalysts can help researchers understand the structure and property of catalysts, as well as the reaction process. While the operation of most characterizations is usually under normal conditions, where is not sensitive enough to get the information if the reaction happens. For example, H₂-temperature programmed reduction (H₂-TPR) is an informative way, and usually employed to characterize the redox property of catalysts. However, when doped or loaded on different supports, the reduction peaks may become complicated, making it difficult to get an intuitive conclusion of the redox property. There is thereby an urgent need, but it is still a significant challenge to achieve a fundamental understanding of catalytic mechanisms at a molecular level by establishing a correlation between the observed catalytic performance and the corresponding surface chemistry during the CO oxidation over Co₃O₄ catalyst.

Herein, the in situ investigations are carried out to get more intuitive observation of catalysts under reaction conditions. In situ analysis plays an important role in catalyst characterization, which allows the study of catalyst phase, as well as the formation of surface complexes and intermediates under given conditions. In this article, it is concentrated on the in situ X-ray Diffraction (in situ-XRD) and in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) analysis of the catalysts. In order to get an in-situ view of the phase transformation during H₂ reduction process, it is carried out in situ XRD in H₂ atmosphere. A detailed analysis of the change of Co₃O₄ in different samples is given to evaluate the redox property of samples. It is important to highlight that the in situ DRIFTS experiment is employed to study the carbonate species on the ceria doped Co₃O₄ catalysts' surface. According to literatures [11, 12], carbonate species have been considered as an important intermediate during the CO oxidation reaction, or a reason causing the deactivation of catalysts. Thereby, we believe that carbonate species would give important information of the catalyst's surface, which is really difficult to obtain from the former studies that concentrated on the characterization of CO adsorption on Co₃O₄ surface [3].

The catalysts (Co₃O₄, CeO₂-Co₃O₄ and CeO₂) were prepared by precipitation-oxidation method. The details of preparation process can be referred from our previous work [10].

The catalytic activities of the catalysts for CO oxidation were tested in a continuous flow quartz tube microreactor (φ 8 mm × 23 cm). 200 mg catalyst (40–60 mesh) and 600 mg silica sand were mixed and filled into the reactor. The flow rate of feed gases (consisted of 1% CO, 10% O₂ and 89% N₂) was 20 mL/min, and the trace amount of water in the feed gas was eliminated by passing the feed gas through a cooled trap. Before activity test, the catalysts were pretreated in N₂ flow at 500 ℃ for 30 min and then cooled to room temperature. The temperature under 0 ℃ for the activity test was achieved by putting the quartz tube microreactor in an insulation cup filled with liquid nitrogen, and the different temperature was obtained by adjusting the distance between the microreactor and the liquid level of the liquid nitrogen.

The microstructures of catalysts were characterized by transmission electron microscopy instrument (TEM, JEM-2100EX) and scanning electron microscope (SEM, JEOL JSM-6360 LV). An instrument of the X-ray photoelectron spectroscopy (XPS) with the Perkin-Elmer PHI 5000C ESCA system was conducted at a condition of Al Kα irradiation. Raman measurements were performed at room temperature with a Via+ Reflex Raman spectrometer and an excitation wavelength of 514 nm.

In situ XRD patterns were collected on a PANalytical PW 3040/60 X'Pert Pro powder diffractometer equipped with an in situ cell that allows heating and introduction of gases. The diffractometer was operated at 40 kV and 40 mA and a scanning speed was 5°/min using Cu K_α radiation. To trace the structural changes of the catalyst, the catalyst sample was loaded into the cell and heated up to 300, 400, and 500 ℃ respectively (staying at each temperature for 30 min), at a rate of 10 ℃/min and under a flow of 10 vol% H₂/N₂ (total flow: 20 mL/min). The XRD patterns were recorded under 300, 400, and 500 ℃ respectively.

In situ DRIFTS analysis was carried out with a Nicolet 6700 spectrometer fitted with an MCT detector. The sample in the cell was pretreated in N₂ (30 mL/min) at 350 ℃ for 30 min, and then cooled to room temperature. After the cell was outgassed in vacuum to lower than 10^‒3 Pa, the background was scanned. Then pure CO gas (P = 8×10³ Pa) was inlet, the IR spectra was recorded after CO inlet at different times. The spectral resolution was 4 cm^‒1 and the number of scans was 64. The gas used here was dried by passing through a cooled trap.

SEM images of CeO₂-Co₃O₄ composite are shown in Fig. 1(a) and (b). It is observed that the composite is assembled from small nanoparticles. TEM image of composite in Fig. 1(c) suggests that the CeO₂ nanoparticles with 10–20 nm are dispersed on the surface of Co₃O₄ nanosheet, which is the unit for the assembly of CeO₂-Co₃O₄ composite. It can be observed in the HRTEM image of Fig. 1(d) that the lattice fringes of d = 0.295 nm are ascribed to the (220) facet of Co₃O₄, and the lattice fringes of d = 0.315 nm are assigned to the (111) facet of CeO₂ [13, 14]. The corresponding elements EDS mappings are shown in Fig. 1(e) and (f). Obviously, it can be detected the Ce, Co and O elements over the catalyst of CeO₂-Co₃O₄. The formation of a cubic structure in the as-synthesized Co₃O₄ is supported by the Raman spectroscopy in Fig. 2(a). The peaks at 193.6, 522.0, and 617.5 cm^‒1 can be attributed to F_2g mode, at 418.6 cm^‒1 to Eg mode, and at 689.7 cm^‒1 to A_1g mode [15]. After loading of CeO₂, there is a new characteristic band appearing at 473.9 cm^‒1 (Fig. 2(b)), which can be assigned to the F_2g mode of CeO₂ [16, 17]. All the above mentioned results imply the successful loading of CeO₂ nanoparticles on the surface of Co₃O₄.

To study the chemical forces between Co₃O₄ and CeO₂, the surface composition and chemical states of the atoms of CeO₂-Co₃O₄ catalyst were analyzed by using XPS, as shown in Fig. 3. The Ce 3d XPS spectra of reduced CeO₂ and CeO₂-Co₃O₄ are shown in Fig. 3(a), which indicates the co-existence of Ce³⁺ and Ce⁴⁺ ions in the composite. The peaks at 881.3, 885.0, 898.6 and 903.0 eV are the characteristic peaks of Ce³⁺, and the remaining peaks belong to the characteristic peaks of Ce⁴⁺. The CeO₂-Co₃O₄ exhibits an enhanced concentration of Ce³⁺ in Ce 3d XPS spectra, compared with that of CeO₂. The high concentration of Ce³⁺ in CeO₂-Co₃O₄ is likely caused by the transfer of electrons from cobalt to ceria [16], indicating that there is an interaction between CeO₂ and Co₃O₄. Seen from the Co 2p core level XPS peaks of Fig. 3(b), suggesting that the Co atom in the as-synthesized Co₃O₄ has two valence states: tetrahedral Co²⁺ and octahedral Co³⁺. Those of Co³⁺ 2 p_3/2 and 2 p_1/2 are located at 783.8 and 799.2 eV, respectively [18, 19]. And the peaks at 785.7 and 800.7 eV are ascribed to Co²⁺ 2p_3/2 and 2p_1/2, respectively. Meanwhile, some satellite peaks are also generated at 790.5 and 809.1 eV. Interestingly, after the reduction treatment in H₂ atmosphere, there is no generation of Co⁰ on CeO₂-Co₃O₄, confirming the absence of the reduction of Co²⁺ to Co⁰. It is because of the oxygen storage capacity of CeO₂, making it possible continuously provide oxygen to Co₃O₄ during H₂ reduction.

Different from the result of Co 2p XPS, the O 1s XPS spectrum of Co₃O₄ changes greatly after the loading of CeO₂, as shown in Fig. 3(c). The fitting peak at 530.9 eV in the O 1s spectrum of bulk cobalt oxide is assigned to the lattice oxygen attached to cobalt (Co–O). The peaks at 532.8 and 534.1 eV belong to the oxygen present in the surface adsorbed -OH groups and the multiplicity of physisorbed and chemisorbed water on or near the surface of Co₃O₄ [20], respectively. It is worth noting that the peak binding energies of CeO₂-Co₃O₄ shift to a low value compared to that of Co₃O₄, which implies the strong interaction between CeO₂ and Co₃O₄ and the enhanced oxygen availability at the reaction site [16]. In addition, after the loading of CeO₂, the peak intensity of the surface adsorbed -OH has a great decrease, suggesting the possibility of the formation of Ce-O-Co bond between HO-CeO₂ and HO-Co₃O₄ by the consuming of surface -OH groups during the calcination process. The strong interaction between components and the increased oxygen availability at the reaction site are beneficial for the enhancement of the catalytic reaction rate over CeO₂-Co₃O₄ composite.

Fig. 4 shows the catalytic performances of Co₃O₄, CeO₂-Co₃O₄ and CeO₂ for CO oxidation. Among them, CeO₂ shows the lowest activity with T₁₀₀ (the reaction temperature for 100% CO conversion) as high as 323 ℃. The catalytic activity of Co₃O₄ for the CO oxidation is obviously higher than that of CeO₂, and its T₁₀₀ is 153 ℃. The CeO₂-Co₃O₄ catalyst exhibits the highest activity and its T₁₀₀ is only –60 ℃.

The redox property plays crucial role in the activity of catalysts for CO oxidation. Usually, H₂ reduction can be employed to evaluate the redox property of catalysts by investigating the reduction behavior of catalyst in H₂ atmosphere under different temperatures. According to our previous work [10], the addition of ceria has a significant influence on the reduction of Co₃O₄, leading to a more difficult reduction of Co²⁺. In order to trace the phase transformation during H₂ reduction process, in-situ XRD experiments were carried out in H₂ atmosphere under elevated temperatures. The results are shown in Figs. 5 and 6, and the corresponding XRD peak positions of Co₃O₄, CeO₂, CoO and Co are listed in Table 1. According to the results, Co₃O₄ was first reduced to CoO, then forming metallic cobalt as the temperature increasing [21]. The peaks of Co₃O₄ (36.8^o for example) disappeared at almost the same temperature in Co₃O₄ and CeO₂-Co₃O₄, while the peaks of CoO (42.4° for example) disappeared at higher temperature in CeO₂-Co₃O₄ than in Co₃O₄. At 400 ℃, the peak of CoO almost disappeared, and the peak of metallic cobalt can be observed in Co₃O₄ obviously, while the peak of CoO can still be observed even at 500 ℃ in CeO₂-Co₃O₄. These results correspond to our previous H₂-TPR experiments [10]. The reduction of Co²⁺ to Co⁰ becomes more difficult, which is beneficial for the stability of Co²⁺. The result is consistent with the result of XPS in Fig. 3(b). The oxygen storage capacity of CeO₂ can continuously provide oxygen to Co₃O₄ during H₂ reduction. Seen from Scheme 1, it has been proposed that CO oxidation on Co₃O₄ follows a redox cycle, where gas-phase CO adsorbed on a cobalt site, and the adsorbed CO reacts with a lattice oxygen to form CO₂ (g) and an oxygen vacancy, the oxygen vacancy is subsequently replenished by oxygen from gas phase [22, 23]. In this process, Co³⁺ is reduced to Co²⁺ firstly, then Co²⁺ is oxidized to Co³⁺ by oxygen. Thus, the Co³⁺—Co²⁺—Co³⁺ cycle plays a crucial role during CO oxidation on the Co₃O₄ surface. In our case, considering the result of in situ XRD, the introduction of CeO₂ can significantly improve the stability of Co²⁺, owing to the oxygen replenish property of ceria. In the other words, it can enhance the reduction ability of Co²⁺ and benefit the salability of the Co³⁺-Co²⁺-Co³⁺ cycle by lowering the oxidation ability of Co²⁺, resulting in making the conversion of Co²⁺ to Co³⁺ easier.

In several studies, the binding of carbonate on cobalt has been examined with in situ IR spectroscopy [24, 25]. Non-coordinated CO₃^2‒ has a symmetric stretching frequency ν₁ at 1010 cm^‒1 [26, 27] and an asymmetric frequency ν₃ at 1390 cm^‒1 [28]. According to literature [24], when carbonate coordinated with metal, the symmetry of the carbonate changed, leading to the split of ν₃. The splitting degree (△ν₃) varies with different types of coordination [24]. There are three types of coordination: monodentate (M) complexes have a small △ν₃ about 80‒120 cm^‒1, mononuclear bidentate (MB) carbonate complexes have a large △ν₃ about 300‒340 cm^‒1 [28, 29], and binuclear bidentate (BB) complexes have a lower △ν₃ than MB about 100‒200 cm^‒1 [30]. The △ν₃ is also related to the polarization of the coordinating metal ions [31]. In order to further investigate the intermediates and adsorption species on the pretreated catalyst surface after CO inlet, in situ DRIFTS experiments were carried out on Co₃O₄ and CeO₂-Co₃O₄ samples, respectively. The spectra of Co₃O₄ and CeO₂-Co₃O₄ are shown in Figs. 7 and 8. As can be observed in Fig. 7(a), the IR peaks are mainly in the region of 1700‒1000 cm^‒1, so the surface species should be carbonate species. As mentioned above, CO oxidation on Co₃O₄ follows a redox cycle, where gas-phase CO adsorbs on a cobalt site, then adsorbed CO reacts with a lattice oxygen to form CO₂ (g) and an oxygen vacancy, the oxygen vacancy is subsequently replenished by oxygen from gas phase. Under the IR experimental conditions, the IR cell was inclosed, and only CO was inlet, so the redox cycle cannot be well maintained without oxygen, leading to the surface concentration of carbonate species. Different types of carbonate species on surface might contain certain information about the property of the surface.

It can be observed from Fig. 7(a) that several weak bands located at 1060, 1360, and 1560 cm^‒1 initially after CO inlet, and several wide and overlapped absorption bands in 900‒1800 cm^‒1 region, such as bands at 1000, 1320, 1496, 1618, and 1672 cm^‒1, grew in with time on stream. On the basis of literature data [24, 28-30] and the analysis of spectra, the band at 1060 cm^‒1 can be attributed to ν₁ vibration of monodentate carbonate species, the bands at 1360 and 1560 cm^‒1 to the split of ν₃ vibration of monodentate carbonate species (M), the bands at 1320 and 1618 cm^‒1 to the split of ν₃ vibration of mononuclear bidentate carbonate species (MB), and the bands at 1496 and 1672 cm^‒1 to the split of ν₃ vibration of binuclear bidentate carbonate species (BB), respectively. The identification of the IR peaks can also be verified by the spectra of heating samples after CO adsorption under different temperatures in Fig. 7(b). As the stability of carbonate species follows BB > MB > M, the M species first diminished by heating, and then MB and BB species diminished as the temperature increasing. These results indicate that at the beginning of CO exposing to Co₃O₄, monodentate carbonate (M) species formed, and then the surface species became complicated, and bidentate carbonate species grew in with time on stream.

In the case of CeO₂-Co₃O₄ (Fig. 8), the IR bands appeared very fast in 900‒1800 cm^‒1 region. The positions of these bands are different from that of Co₃O₄. The assignments of these bands can be found in Table 2. There are some points worth noting by comparing the results of Co₃O₄ and CeO₂-Co₃O₄. Firstly, the bands of carbonate species are of higher intensity, and grew much faster on CeO₂-Co₃O₄ than on Co₃O₄, indicating that much more active sites for adsorption on CeO₂-Co₃O₄ surface. At the same time, the adsorption took place in a faster mode. Secondly, the peaks on CeO₂-Co₃O₄ shifted to lower frequency than on Co₃O₄, indicating that the carbonate species bond weakly on CeO₂-Co₃O₄. Thirdly, the band at 1398 cm^‒1 on CeO₂-Co₃O₄ is similar to the symmetric stretching frequency of noncoordinated CO₃^2- in aqueous solution, implying that carbonate species on CeO₂-Co₃O₄ is to some extent in free state with weak bonding to catalyst surface. This type of carbonate species is more active, and not deactivating the surface as those strongly bonding one. The difference of carbonate species on CeO₂-Co₃O₄ from Co₃O₄ is attributed to the addition of CeO₂'s modifying the surface. After pretreated under N₂ at high temperature, the surface is clean and some oxygen vacancies formed, the special oxygen storage capacity of CeO₂ enriches the surface with weakly adsorbed oxygen species.

The IR experiments over CeO₂ were also carried out to check CO adsorption on CeO₂ surface. Fig. 9 shows the spectra of CO fully adsorbed on Co₃O₄, CeO₂ and CeO₂-Co₃O₄. As can be observed, there are several peaks on CeO₂-Co₃O₄, such as peaks at 1292 and 1608 cm^‒1 arising from CO absorbance on CeO₂. The peak at 1398 cm^‒1 is attributed to a highly active intermediate species during CO oxidation, which is of high symmetry and plays crucial role in reaction. While the peak at 1398 cm^‒1 cannot be found on either pure CeO₂ or Co₃O₄, so it should be caused by the interaction between CeO₂ and Co₃O₄. In CO₂ region, on CeO₂-Co₃O₄, when CO was inlet for the third time, CO₂ can be generated, while on Co₃O₄, even since the second time of CO inlet the carbonate bands did not change much and no CO₂ can be generated regardless of the subsequent CO inlet. After adequate CO has been introduced, the peak intensity of carbonate species on Co₃O₄ is much lower than on CeO₂-Co₃O₄. These results are in perfect coincidence with the above mentioned surface property of the samples.

The ceria promoted cobalt oxide was prepared and evaluated for CO oxidation. The results showed that ceria adding can significantly improve the catalytic activity of Co₃O₄ for CO oxidation. The effect of ceria on the redox property of Co₃O₄ was investigated by in situ XRD. By comparing the phase change during H₂ reduction, it can be inferred that the reduction of Co²⁺ to Co⁰ became difficult in CeO₂-Co₃O₄, to some extent, increasing its reducibility. So the oxidation of Co²⁺ to Co³⁺ became easier, and the Co³⁺-Co²⁺-Co³⁺ cycle was promoted. As the redox cycle plays a crucial role in the oxidation of CO, the activity of the Co₃O₄ catalyst was promoted. In situ DRIFTS showed that the carbonate species on CeO₂-Co₃O₄ exhibited much more sites for adsorption, and the adsorption of them is in a weaker bonding mode. Meanwhile, the carbonate species on CeO₂-Co₃O₄ surface is similar to that in the free state, which effectively inhibits catalyst deactivation. These results verify that after modifying by CeO₂, the catalyst surface is more active and suitable for CO oxidation to take place.