CO oxidation to CO₂ has been broadly investigated for decades due to its great value not only in fundamental catalytic research as a probe reaction, but also in practical applications, such as air purification, automotive exhaust emissions control and preferential oxidation in proton-exchange-membrane fuel cells [1-5]. Among these applications, especially in controlling automobile exhaust, large amounts of CO are generally produced during the cold starting, which lead to serious environmental problems. In order to regulate CO emission, developing high performance catalysts that can convert CO into CO₂ at low temperature is significant. Although supported noble metal catalysts show superior performance for CO oxidation and have been thoroughly investigated [6-10], their applications have been restrained by high cost and limited sources. As alternative, catalyst systems, with decent catalytic performance and high availability but low cost, have been extensively studied.

Among these catalyst systems, transition metal oxides are mostly studied for CO oxidation [11-15]. Co₃O₄ with spinel structure has been considered to be the most promising alternative to noble metals due to its high catalytic activity at relatively low temperature, which comes from its suitable strength of adsorbing CO, low barrier to activate oxygen and superior redox capacity [16]. Considerable work has been done towards CO oxidation over cobalt-based materials [17-24]. Thormählen et al. prepared a pre-oxidized Al₂O₃-supported Co₃O₄ catalysts, which showed a light-off temperature (T₅₀) of −63 ℃ [25]. Xie et al. reported a Co₃O₄ nanorods with exposed (100) planes rich in Co³⁺ could completely catalyze CO oxidation at −77 ℃ [26]. Jia et al. synthesized Co₃O₄/SiO₂ nanocomposite with enriching silica and Co²⁺ species on surface, which exhibited a high activity for total conversion of CO at −76 ℃ [27]. Ordered mesoporous CoO with octahedrally coordinated Co²⁺ species displayed unexpectedly high activity for CO oxidation due to its easy oxidation of surface Co²⁺ to Co³⁺, suggesting the importance of coordination environment of the active sites [28]. Considering the reactant diffusion process is a crucial factor affecting the catalytic performance, a porous structure is expected to increase the amount of accessible active sites and benefit the reactant-catalyst contact, thus enhancing catalytic activity. For example, Ren et al. used a series of mesoporous metal oxides (i.e., Co₃O₄, CuO) as catalysts for CO oxidation, which exhibited higher activities than their bulk counterparts [29]. Template methods have been widely used in materials synthesis [30-35]. Hard-templating method (nanocasting), by using ordered mesoporous solids as nanoreactors, can create nanosized porous metal oxides with high surface areas and long-range ordering [36-40]. Additionally, the prepared mesoporous metal oxides own similar texture and structure with their templates. However, the specific surface area of nanocast metal oxides is confined, since primary particles metal oxides are prone to sinter during thermal treatment. For the sake of attaining mesoporous metal oxides with high specific surface area providing abundant reactive sites to enhance their catalytic performance, traditional nanocasting method needs improvement.

Herein, we report a dual-template method to synthesize mesoporous Co₃O₄ (Co₃O₄-DT), which is rich in defects and possesses very high specific surface area. A modification to the traditional nanocasting method was used to create defects inside the nanorods of the nanocast mesoporous Co₃O₄. Both mesoporous silica SBA-15 and in situ generated SiO₂ are used as hard templates at the same time. After the removal of templates by NaOH, a secondary pore was generated, which created more exposed surface. The specific surface area of the obtained Co₃O₄ is as high as 169 m²/g, which is more than two times higher than that of Co₃O₄-NC that synthesized by normal nanocasting method. The Co₃O₄-DT catalyst exhibits prominent catalytic activity for CO oxidation, superior to Co₃O₄-NC. Further investigations reveal the factors that affecting the performance of CO oxidation. What is more, the dual-template method can be extended to synthesize other mesoporous metals oxides including iron, nickel and cerium oxides, etc., providing scientific insights into rational design of meso-structures with high specific surface area for superior catalytic activity.

The synthesis schematic of Co₃O₄-DT and Co₃O₄-NC is shown in Fig. 1A and detailed process is elaborated in the Supporting information. According to the different amounts of tetraethoxysilane (TEOS), the obtained Co₃O₄-DT by dual-template were denoted as Co₃O₄-DT1 (low amount of TEOS) and Co₃O₄-DT2 (high amount of TEOS), respectively. XRD patterns of all replica materials (Fig. 1B) show well-resolved reflection peaks, which can be assigned to spinel Co₃O₄ phase (PDF#42-1467) with a Fd3m symmetry. No other diffraction peaks were detected, indicating high purity phase. With increasing TEOS amount, the diffraction peaks of the replica Co₃O₄ obviously become weaker and broader, revealing a lower level of crystallization and/or smaller crystalline grain size. The grain sizes were calculated based on the (311) facets of Co₃O₄ by using Scherrer equation, as listed in Table 1. It is shown that the grain sizes of Co₃O₄-DT2 (8.9 nm) are smaller than that of Co₃O₄-DT1 (10.9 nm) and Co₃O₄-NC (12.7 nm), indicating that the introduction of TEOS can effectively suppress Co₃O₄ nanocrystals sintering during the calcination process. And this phenomenon on cobalt catalysts also have been observed in literature [41,42]. The N₂ adsorption-desorption isotherms and corresponding pore size distributions curves of the replica samples are showed in Figs. 1C and D. And the relevant parameters are summarized in Table 1. For Co₃O₄-NC, the specific surface area and pore volume are calculated to be 81 m²/g and 0.13 cm³/g, respectively. Interestingly, the specific surface area of Co₃O₄-DT1 and Co₃O₄-DT2 prepared by dual-template method are significantly increased to 122 and 169 m²/g, and the corresponding pore volumes are also increased remarkably to 0.52 and 0.39 cm³/g, respectively. The pore sizes distribution of this catalyst has a maximum around 3–4 nm, revealing a uniform mesoporous structure. The morphological and structural features were further analyzed by transmission electron microscopy (TEM). It can be seen that Co₃O₄-NC presents a rod-like morphology (Figs. 2A and B), and the rod diameter is consistent with the pore size of SBA-15, suggesting a good replication of the SBA-15 template. In addition, all the nanorods are quite dense without obvious cracks or grain boundaries. Lattice fringes show a d-spacing of 0.24 nm in high-resolution TEM (HR-TEM) image (Fig. 2C), which is in well accordance with the (311) crystal planes of spinel Co₃O₄ [43,44]. Co₃O₄-DT1 prepared by dual-template method also exhibits a rod-like morphology (Figs. 2D and E). However, compared to traditional Co₃O₄-NC, plenty of randomly oriented cracks/pores are observed in the nanorod arrays of Co₃O₄-DT1, and these cracks/pores can connect the meso-channels to build an interconnected mesopore networks. When further increase the amount of TEOS, the cracks/pores inside the nanorods are increases remarkably (Figs. 2G and H). The crystals lattice fringes for both Co₃O₄-DT1 and Co₃O₄-DT2 catalysts can be clearly observed in the HR-TEM images (Figs. 2F and I). An average d-spacing is measured to be 0.29 nm, which can be assigned to the (220) reflection of Co₃O₄ spinel structure [45,46]. The existence of the inner nanorods pores are contributed to the increase of the specific surface area. These results reveal that the introduction of TEOS during synthesis process gives rise to numerous cracks, whose density can be controllably adjusted while maintaining structural integrity. The more open porous structures leading by cracks are expected to favor the facile accessibility of active sites. These results indicate the successful synthesis of Co₃O₄-DT with high specific surface area and high porosity by dual-template method. Moreover, this strategy can be extended to the synthesis of other mesoporous metal oxides with high specific surface area. Mesoporous Fe₂O₃ (Fe₂O₃-DT), NiO (NiO-DT), and CeO₂ (CeO₂-DT) with similar structure and enhanced specific surface areas can also be synthesized. The related structure characterizations are showed in Fig. S1 (Supporting information) and Table 1.

The large specific surface area and high porosity of Co₃O₄-DT can provide abundant active sites, making them promising candidates for catalysis. CO oxidation catalytic performances of these samples were evaluated under both normal and dry feed gas conditions. Meanwhile, the catalytic performances of Co₃O₄-NC and commercial Co₃O₄ were also explored for comparison. The catalytic activities of the catalysts for CO oxidation were measured in a plug flow reactor using a gas mixture of 1 vol% CO, 20 vol% O₂ and 79 vol% He. The results are shown in Fig. 3 and summarized in Table 1. Among these catalysts, Co₃O₄-DT2 shows the optimum catalytic activity with a T₅₀ value of as low as −73 ℃ at a high space velocity of 80 000 mL h^-1 g_cat^-1, and a full CO conversion at −43 ℃. The T₅₀ value of Co₃O₄-DT1 slightly increases to −65 ℃. Co₃O₄-NC prepared by conventional nanocasting method could only reach 50% CO conversion at −50 ℃. In addition, all these mesoporous materials show much higher catalytic activities than that of the commercial Co₃O₄, which is only unable to fully convert CO to CO₂ until the temperature reaches 225 ℃. Apparently, the T₅₀ of CO conversion markedly decreases with increasing specific surface area of catalysts, indicating that catalytic activity can be improved by higher specific surface area promoting the exposure of abundant reactive sites. The excellent stability of the dual-template Co₃O₄ catalysts is generally required in practical applications, so long-term durability performances were further investigated. The obtained results are showed in Fig. 3B. Co₃O₄-DT2 catalyst maintains full CO conversion for up to 32 consecutive hours at 30 ℃ under dry feed gas condition. When the test was performed under normal feed gas condition with about 3 ppm water, complete CO conversion remains during the initial 24 h, much longer than that reported in literature [43]. After that, the CO conversion suddenly drops to 85% and then slightly to 72% after 32 h. Such a deactivation can be explained by the poisoning effects of H₂O/OH⁻ species or carbonate intermediates on the catalyst surface, which block the catalytic active sites [47-49]. With the aim to provide deeper insights into CO oxidation catalytic properties of the as-prepared catalysts, the kinetic experiments were conducted to determine the apparent activation energy (E_a) for CO oxidation (Fig. 3C), which are calculated from the Arrhenius plots and listed in Table 1. Co₃O₄-DT2 displays enhanced reaction kinetic with the lowest apparent activation energy (33.9 kJ/mol), followed by Co₃O₄-DT1 (41.6 kJ/mol) and Co₃O₄-NC (50.3 kJ/mol). The lowest apparent activation energy of Co₃O₄-DT2 is in good according with the light-off results.

The above results show that high-surface-area mesoporous Co₃O₄ catalysts prepared by dual-template can effectively improve the catalytic activity for low temperature CO oxidation. Since the catalytic activity is affected by many factors such as metal valent states and absorbed species states, therefore, further investigation of the origin of the high catalytic performance is needed. The chemical state and surface properties of the catalysts were investigated by X-ray photoelectron spectroscopy (XPS). The oxidation states of cobalt were evaluated by Co 2p spectra which were split into Co²⁺ and Co³⁺ peaks, as shown in Fig. 4A. The Co 2p_3/2,1/2 peaks with binding energies at 780.1 and 795.5 eV can be assigned to Co³⁺ species, while the peaks at 781.8 and 797.2 eV are ascribed to Co²⁺species [43]. The surface Co²⁺/Co³⁺ molar ratios of samples are calculated based on the fitting of Co 2p spectra and listed in Table 1. As one can see, when the specific surface areas increase, surface Co²⁺/Co³⁺ molar ratios are increased in the order of Co₃O₄-NC (0.38) < Co₃O₄-DT1 (0.43) < Co₃O₄-DT2 (0.55), implying the relative enrichment of Co²⁺ species on Co₃O₄-DT2. For the O 1s spectra in Fig. 4B, it can be mainly deconvoluted into two peaks standing for the existence of different surface oxygen species. The peaks with banding energy at ~529.7 and 531.2 eV are attributed to surface lattice oxygen (O_lat) and adsorbed oxygen (O_ads, e.g., O⁻, O₂⁻), respectively [50]. The O_ads species with higher binding energy are generally supposed to be more active than O_lat and may play a key role in determining the low-temperature performance of CO oxidation [49,51,52]. The O_ads/O_lat molar ratios of these catalysts are calculated according to quantitative analysis of corresponding peaks areas, as shown in Table 1. The molar ratios of O_ads/O_lat follow the order of Co₃O₄-NC (0.59) < Co₃O₄-DT1 (0.73) < Co₃O₄-DT2 (1.10), which is identical to that of Co²⁺/Co³⁺ molar ratio. This order is reasonable in light of the overall surface charge neutrality. Assuming adsorbed oxygen species are produced by oxygen vacancies, it is suggested that the number density of oxygen vacancies on Co₃O₄-DT2 is higher than that of others. The spectra of Co 2p and O 1s suggest that both higher Co²⁺/Co³⁺molar ratio and abundant oxygen vacancies are important in enhancing the catalytic activity of Co₃O₄ catalyst.

The CO-TPD experiments were conducted to investigate the surface oxygen activity of catalysts for CO oxidation. During the heating process, the majority of CO adsorbed on catalysts surface desorbed as CO₂, which were produced by the reaction of adsorbed CO with surface lattice oxygen [53]. As shown in Fig. 4C, the CO-TPD profile of Co₃O₄-NC shows three peaks located at 118, 434, and 557 ℃, indicating three kinds of lattice oxygen with different environments. All catalysts display an identical CO₂ desorption peak at around 118 ℃, and other two broad peaks for Co₃O₄-DT1 and Co₃O₄-DT2 are located at 360 and 422 ℃. Compared with Co₃O₄-NC, the CO₂ desorbed temperatures of Co₃O₄-DT1 and Co₃O₄-DT2 shift to lower value, implying that adsorbed CO are more easily oxidized by surface lattice oxygen. Since oxygen mobility on the metal oxides plays a crucial role in oxygen-involving catalytic reactions [54-57], O₂-TPD experiments were performed to study the mobility of different oxygen species on catalysts. Generally, oxygen species adsorbed on surface can go through succeeding transformation with electron gain ability: O₂(ad) → O₂⁻(ad) → O⁻(ad) → O²⁻(ad/lattice). O₂(ad) is ascribed to the physically adsorbed oxygen, which can be removed by purging before test. The adsorbed species of O₂⁻(ad) and O⁻(ad) are poorly bonded on the catalysts surface, while O²⁻(ad) species are usually regarded as a kind of surface lattice oxygen, which are difficult to be removed. Based on the literature results [51,58], the peaks at temperature below 300 ℃ are related to the desorption of surface adsorbed O₂⁻(ad) and O⁻(ad) species, while the peaks above 350 ℃ are attributed to surface lattice oxygen and lattice oxygen in bulk phase (> 700 ℃). As shown in Fig. 4D, a clear moderating trend in desorption temperature of surface lattice oxygen species can be observed: Co₃O₄-DT2 (415 ℃) < Co₃O₄-DT1 (465 ℃) < Co₃O₄-NC (485 ℃), indicating Co₃O₄-DT2 possesses the most active surface lattice oxygen. Increasing surface areas of samples can alleviate Co-O bond and thus promote the desorption of lattice oxygen from Co₃O₄ [59]. Hence, the increased mobility of surface lattice oxygen on Co₃O₄-DT2 can be relevant to surface area effect. Based on the CO-TPD and O₂-TPD results, higher specific surface area accelerates lattice oxygen mobility, leading to much easier CO oxidation.

It is well-known that the adsorption and activation of molecular oxygen are essential to Co-based catalysts for CO oxidation. The generation of adsorbed reaction oxygen species, such as superoxide ions (O₂⁻), can be related to the presence of surface oxygen vacancies on metal oxides supports or at the interfaces between metal and the support. A wide variety of superoxide species have been reported to exist between −196 ℃ and 25 ℃ on the surface of CoO-MgO solid solutions [60], Co₃O₄/SiO₂ composite materials [27], and reduced Co₃O₄ (100) [61]. It is thus hypothesized that such superoxide species may play a crucial role in CO oxidation and also over the mesoporous Co₃O₄ studied here at low temperature. For this consideration, CO titration experiments in the absence of O₂ (1 vol% CO in N₂) were conducted to evaluate the availability of oxygen stored on the catalysts. The transient responses of CO₂ and CO evolution over these three pretreated Co₃O₄ catalysts at different temperatures are showed in Fig. S2 (Supporting information). For Co₃O₄-DT2, a maximum (CO₂ concentration 0.08%) of CO₂ evolution is observed at 108 s after exposure to CO at −50 ℃, indicating a high capacity of reactive oxygen. The prompt CO₂ responses show a fast reaction proceeding via CO rapid adsorption, followed by the reaction of CO with reactive oxygen species on the catalyst surface. At 30 and 120 ℃, strong CO₂ evolutions are noticed after 84 s with CO₂ concentration of 0.32% and 0.63%, respectively, higher than that observed at −50 ℃. Similar trends of CO₂ respond at different temperature are also observed for Co₃O₄-NC and Co₃O₄-DT1. The capacity of these catalysts to provide reactive oxygen is enhanced as increasing temperature, consistent well with the temperature dependence of CO oxidation activity. Moreover, the maxima of CO₂ evolution for Co₃O₄-DT2 at corresponding temperature are integrally higher than that of others, suggesting the strongest ability of Co₃O₄-DT2 to provide reactive oxygen. The integral under the curves showed in Fig. S2 allow to calculate the total amount of active oxygen on surface of catalysts. To assess whether bulk oxygen contributes to CO oxidation in the titration experiments, the maximum amount of oxygen only present on the surface was calculated assuming that the surface of catalyst only consists of closely-packed (111) planes of Co₃O₄. For all the catalysts at different temperatures, the total amounts of consumed oxygen are lower than the theoretical maximum amount (Table S1 in Supporting information). It can be attributed to the presence of other planes such as (100) or (110) with lower oxygen density. The calculated oxygen consumption in 1200 s shows higher values for Co₃O₄-DT2 than that for Co₃O₄-DT1 and Co₃O₄-NC at −50, 30 and 120 ℃, suggesting Co₃O₄-DT2 contains more reactive oxygen. A high ability to supply reactive oxygen would be responsible for the extraordinary activity of CO oxidation at low temperatures, which explains the different activity of these catalysts in a plausible way.

To further investigate the reaction mechanism, in situ DRIFT experiments were performed on Co₃O₄-NC and Co₃O₄-DT2. The obtained DRIFT spectra were collected to get the direct information about the reaction intermediates and side products during the exposure to CO and O₂ with a ramping temperature (20–310 ℃). The signals of two peaks at 2171 and 2110 cm^-1 (Figs. S3A and C in Supporting information) attributed to CO characteristic bands are very weak due to the rapid CO conversion. And CO₂ characteristic peaks centered at around 2360 and 2337 cm^-1 are obviously detected at the very beginning of reaction. Both catalysts show a set of bands in the range of 1300–1700 cm^-1 (Figs. S3B and D in Supporting information). These bands have been assigned to various carbonate vibrations, which become stronger with increasing temperature, indicating the accumulation of carbonates on the catalysts surface. These carbonate species are kept stable on the catalysts surface even at 310 ℃. In view of the oppositional results in literature, it is hard to precisely assign these bands. However, during the reaction process, Co₃O₄-NC and Co₃O₄-DT2 have definitely different carbonate vibrations, which may also affect the catalytic reaction process.

In summary, a dual-template method was developed for the successful synthesis of ordered mesoporous Co₃O₄ with enhanced defects and high specific surface area by using SBA-15 and in situ generate SiO₂ as hard templates. The obtained samples were proved to be significantly active for low-temperature CO oxidation, superior to that prepared by traditional nanocasting method. The enhanced catalytic performance is mainly associated with high specific surface area, abundant Co²⁺ species and active oxygen species. Both high specific surface area and abundant oxygen vacancies are conducive to accelerate the mobility of lattice oxygen. Such a controllable dual-template method can be extended to construct other mesoporous metal-oxides-based materials with high specific surface areas for various applications.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors acknowledge funding from the National Key R&D Program of China (No. 2018YFE0201703), the "1000-Youth Talents Plan", and the Fundamental Research Funds for the Central Universities (No. 2042019kf0230).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.107777.