HCHO is a harmful indoor volatile organic compound because long-term exposure to HCHO atmosphere even at a low concentration can cause severe health problems, such as eye irritation, skin irritation, headache, and even cancer [1-3]. HCHO has extensive sources available across our daily lives. For example, it is released from wood-based building products, decorative materials, coatings, and even motor vehicles. It has become a major indoor pollutant in recent years [4]. Furthermore, methods to mitigate this major pollutant has gained significant attention. Among all the existing removal methods, catalytic oxidation is regarded as the most effective method because of its high removal efficiency, harmless final products, and long-time effectiveness [4-7].

Transition metal oxides, especially Co-based oxides, usually exhibit excellent HCHO catalytic ability because of their multiple valence states and wide range of sources. Much research has been conducted to synthesize and modify Co-based materials to develop a promising material. González-Prior et al. [8] enlarged the specific surface area of Co₃O₄ using the SBA-15 hard template. Zhou et al. [9] prepared Cu-substituted Co₃O₄-Cu_x by replacing a certain amount of Co²⁺ with Cu²⁺. The modified materials exhibited enhanced CO oxidation ability. Bai et al. [4] prepared nano-Co₃O₄, 2D-Co₃O₄, and 3D-Co₃O₄ catalysts and observed that these types of Co₃O₄ could influence the HCHO catalytic oxidation ability, with 3D-Co₃O₄ exhibiting the best performance. Besides, research on exploring the reason for the high catalytic ability of Co-based materials has also been conducted, and remarkable achievements have been obtained. Omata et al. [10] observed that during the CO oxidation process, Co³⁺ significantly influences the catalytic performance while Co²⁺ is useless. Mo et al. [11] discovered that high-valence Co ions are the main active sites during the benzene hydrocarbon combustion process, because they can induce more surface Lewis acidic sites and enhance the reducibility of the catalyst. Some studies reported that Co₃O₄ facets, mainly composed of Co³⁺ on the surface of Co₃O₄, are the main active sites for HCHO catalytic oxidation [12]. The amount of surface oxygen vacancies and different exposed crystal planes have also been explored and considered as essential factors affecting the catalytic process. Yu et al. [13, 14] desired Co₃O₄ treated at a moderate temperature. They believed that the high CO catalytic performance and relatively long time stability are related to the abundant surface oxygen vacancies. Hu et al. [15] prepared several Co₃O₄ nanocrystals with different exposed planes that exhibited different activity order for methane in the order of {112}>{011}>{001}.These results imply that Co-based catalysts are promising materials and Co-based oxides can be developed further in the field of catalytic oxidation.

Previous studies proved that different calcination atmospheres greatly influence the surface chemical properties of the catalysts. Ku's et al. [16] operated MgO in different atmospheres, such as nitrogen, oxygen, nitrogen-oxygen mixture, and helium. They discovered that the catalysts operating in oxygen-enriched atmosphere exhibited lower surface basicity, which gave lower methane conversion than the catalysts operating in the inert atmosphere. Galetti et al. [17] prepared a series of Ni/ZnAl₂O₄ catalysts calcined in reductive and oxidative atmospheres. They observed that the reductive atmosphere enhanced the ethanol steam reforming activity and stability. Zou et al. [18] evaluated the catalytic effects of different 3Fe8Ni/PG products operated under different calcination conditions including different calcination temperatures, times, and atmospheres. These results proved that the catalyst reduced in H₂ exhibited better catalytic performance than the catalyst calcined in air. The characterization results indicated that different managing processes could create different compositions, thereby leading to different catalytic performances. It can be concluded that the calcination method plays a key role in adjusting the properties of the catalyst.

In this study, we prepared a series of Co-based oxides calcined in different atmospheres. The HCHO oxidization mechanism is presented according to the various characterization methods. From the X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, and H₂-TPR results, the Co²⁺ and oxygen vacancies can greatly influence its activity in this system. This is because the N₂ calcination process is conducive to the generation of CoO, which is related to the more octahedrally coordinated Co²⁺ species. Besides, the N₂ calcined catalysts also enhanced the surface O₂^– and O^– content. Based on the XPS and DRIFT experiments, the HCHO molecule in an oxygen-deficit environment will be adsorbed on the surface of the catalyst. When abundant oxygen was introduced over the catalyst, HCHO will be oxidized into dioxymethylene (DOM) and formate by O₂^– or O^–. When the oxygen species get consumed, the vacancy would be replenished by dissociating the oxygen gas. Thus, it can be proposed that both octahedrally coordinated Co²⁺ species and surface oxygen species are the key factors that influence the HCHO degradation ability.

The Co(Ⅱ)Co(Ⅲ)LDH precursor was first prepared by a hydrothermal method stated in previous reports [11, 18]. In a typical synthetic process, 20 mL 1.0 mol L^–1 Co(NO₃)₂·6H₂O aqueous solution was slowly injected into 100 mL 0.5 mol L^–1 ammonia solution with vigorous magnetic stirring at 40 ℃. The mixture was kept at this temperature for 48 h with purified oxygen gas bubbling. The resultant product was then washed several times with deionized water, followed by drying in an oven at 70 ℃. Before being used in the catalytic experiments, the powders were calcined at 400 ℃ under N₂ or air for 4 h. The resultant products were denoted as Co-LDO(N) or Co-LDO, respectively. A similar method was used to obtain CoMg-LDO(N) and CoMg-LDO, although an appropriate amount of Mg(NO₃)₂·6H₂O was added [19]. CoMg-LDO(N) and CoMg-LDO were prepared to obtain a type of contrast material with larger area [20, 21]. A transition metal with multiple valence states exhibits better catalytic abilities, while the Mg element is considered to be inactive. It is difficult for Mg to show better redox ability because of the single valence. Thus, the addition of Mg could increase the BET surface area without introducing other active sites.

The X-ray diffraction (XRD) patterns of the powder were recorded on a Rigaku D/MAX 2500 using monochromatized Cu K_α radiation from 3° to 80° at a scanning speed of 4° min^–1. The morphologies and sizes were determined by a Hitachi S-4800 scanning electron microscope (SEM). High-resolution transmission electron microscopy (HRTEM) was conducted to investigate the microstructure. To detect the specific surface areas and pore sizes of the samples, BET experiments were characterized on an AUTOSORB-IQ instrument. Prior to N₂ adsorption-desorption measurement, each sample was degassed with N₂-purged at 100 ℃ for 10 h. XPS measurements were performed on a VG ESCALAB250XI. The binding energies were corrected by setting the binding energy of the C 1s to 284.8 eV. The Raman spectra of the samples were obtained on a LabRAM Aramis Raman spectrometer (HORIBA Jobin Yvon S. A. S.) equipped with a 532 nm laser. For the H₂-TPR analysis, 100 mg samples were pretreated in N₂ at 300 ℃ for 1 h and then cooled to room temperature. Further, the temperature was set from room temperature to 700 ℃ by heating it at a rate of 10 ℃ min^{– 1}. The reducing gas flow was 5% H₂ in N₂. The in situ DRIFTS spectra were collected using a Nicolte iS50. In a typical process, 100 ppm HCHO and 20 vol% O₂ balanced with N₂ were fed over a certain amount of sample.

The activities of the catalysts for HCHO catalytic oxidation were tested in a fixed-bed quartz tube reactor (inner diameter = 4 mm), and the reaction conditions were controlled as follows. A steady gas reactant mixture containing 100 ppm HCHO and 20.0 vol% O₂ balanced with N₂ flowing at 100 mL min^–1 was passed through an 80 mg fresh catalyst. The HCHO gas was generated by passing N₂ through a paraformaldehyde incubator at 37.5 ℃. All the outlet gases were analyzed using an online Agilent 7890B gas chromatograph equipped with a TCD detector, and the type of chromatographic column used was HP-PLOT/Q+PT from Agilent. The catalytic activity for HCHO oxidation was evaluated in terms of CO₂ concentration according to the carbon balance; the specific formula was HCHO conversion = C_CO2/C_HCHO × 100%, where C_CO2 is the outlet concentration of CO₂ and C_HCHO refers to the inlet concentration of HCHO.

The XRD patterns of the as-prepared catalysts are shown in Fig. 1. All the samples without calcination show typical reflections of hydrotalcites, and no other crystalline phases can be detected, thereby indicating the pure phase of the product [22, 23]. After calcination, the diffraction peaks of all the products in Fig. 1(b) can be indexed as a spinel structure with 2θ values of 18.8°, 31.0°, 36.6°, 44.5°, 59.0°, and 65.0°, corresponding to (111), (220), (311), (400), (511), and (440) crystal planes of Co₃O₄, respectively [19, 24]. The patterns of the pure Co oxides exhibit more sharp and symmetric peaks, which strongly suggest that these phases were better crystallized. Besides, the interfusion of Mg doesn't induce additional peaks, indicating the relatively low amount of Mg or the amorphous MgO phase. Although there are no changes in the crystalline phase, the specific surface areas increase with the incorporation of Mg, as presented in Table 1. A previous study discovered that the incorporation of second metal ions can influence the crystallization process, which can decrease the particle size and increase the BET value [21]. Besides, low-temperature calcination may also help prevent the sample from recrystallizing. All these may explain the relatively high BET values of CoMg hydrotalcite derivative [19].

Figure 2 shows the SEM and HRTEM images. The hydrotalcite derivates exhibit uniform lamellar structure. In contrast, CoMg-LDO(N) and CoMg-LDO exhibit smaller particle sizes because of the inhibition effect caused by the above-mentioned incorporation of Mg during the crystal growth. The detailed lattice fringes information is given by the HRTEM images. As shown in Figs. 2(e)–2(h), the dominant exposed planes of the samples are calculated to be {001}, which are the only planes normal to the set of (220) planes with a lattice space of around 0.28 nm [15, 25].

The H₂-TPR experiments were performed to investigate the redox ability. As shown in Fig. 3, a two-step reduction process can be observed, which can be attributed to the path of Co³⁺→Co²⁺ and Co²⁺→Co⁰, respectively [11]. The samples calcined in different atmospheres exhibit conspicuous distinction. Co-LDO(N) and CoMg-LDO(N) samples have a relatively lower reduction temperature than Co-LDO and CoMg-LDO samples, indicating the better redox ability of samples calcined in the inert atmosphere. A previous study proved that the shifts toward the lower temperature are caused by the presence of more active oxygen species [1]. Thus, it can be reasonably summarized that the N₂ atmosphere is beneficial to the generation of surface active oxygen species. Furthermore, it is noted that in the inert atmosphere the incorporation of Mg does not deteriorate the redox ability of the samples, while the oxygen atmosphere evokes an increase in the reduction temperature from 348 ℃ to 372 ℃ by the introduction of Mg.

To investigate the surface information of the samples, Raman signals, shown in Fig. 4, can be observed. Typically, five bands located at 198, 484, 522, 622, and 694 cm^–1, which can be assigned to the F_2g¹, E_g, F_2g², and A_1g modes of spinel Co₃O₄, respectively, are obtained [26, 27]. Accordingly, the peaks banding at 667 cm^–1 can be assigned to the octahedral sites, while those at 189 cm^–1 can be ascribed to the tetrahedral sites in the standard Co₃O₄ types [28]. Comparing the samples calcined in N₂ with those in O₂, a slight shift toward the lower frequency can be observed for Co-LDO(N) and CoMg-LDO(N). The presence of this red shift can be attributed to the relative abundance of Co²⁺ and oxygen vacancies [11, 26, 29]. Besides, a wide and weak band at around 1100 cm^–1, related to the oxygen vacancies can be observed in both Co-LDO(N) and CoMg-LDO(N) [30]. According to a previous study, the dissociation of gaseous oxygen usually occurs at the oxygen vacancies. The oxygen vacancies influence the improvement of the oxygen mobility and acquirement of the facial method for the active oxygen species [31]. Thus, an oxygen vacancy can be considered as an important oxygen supplements source during the catalytic process [32]. Accordingly, the concentration of oxygen vacancies is related to the oxidation ability of the catalysts. In Fig. 3, it can be observed that the catalysts calcined in N₂ possess more oxygen vacancies, demonstrating that Co-LDO(N) and CoMg-LDO(N) have better catalytic ability.

Surface element analysis tests were conducted to detect the surface valence states, as shown in Fig. 5. According to the binding energy and the intensities of the shake-up peaks, a plateau can be observed, indicating the existence of both CoO and Co₃O₄ [33]. In addition, by deconvoluting the Co 2_p3/2 signals, the materials were well fitted with the mixture of Co₃O₄ (pink curve) and CoO (blue curve) [34], and the specific CoO/Co₃O₄ values are calculated and listed in Table 1. It can be seen that the CoO/Co₃O₄ values of Co-LDO(N) and CoMg-LDO(N) are 0.16 and 0.18, respectively, which are higher than those of Co-LDO and CoMg-LDO. The N₂ calcination process created more surface CoO species, which means that more Co²⁺ ions exist on the surface [35-38]. Co²⁺ ions are generally considered to be inactive during the catalytic performance. However, Schüth et al. [33] discovered that the rock-salt structure of CoO with a single Co²⁺ ion octahedrally coordinated by lattice oxygen can exhibit very high catalytic abilities for CO. In contrast to the inactive tetrahedrally coordinated Co²⁺ ion, the octahedrally coordinated Co²⁺ ion is more open and more easy to oxidize. The octahedrally Co²⁺ sites in CoO are similar to the Co³⁺ sites in Co₃O₄, and may have higher surface density if Co²⁺ is oxidized.

The XPS spectra of O 1s in Fig. 6 consist of three peaks at 529.7, 531.7, and 534.4 eV, corresponding to lattice oxygen atoms (O_latt) in the bulk Co₃O₄, surface chemisorbed oxygen including O^– and O₂^– (O_ads), and oxygen-containing (hydro)carbons, respectively [34, 39-41]. It can be seen in Table 1 that the samples calcined in N₂ exhibit relatively higher content of surface adsorbed oxygen species. Combining this with the Co 2p XPS results, it can be deduced that a relatively high surface oxygen species should be produced by the oxygen vacancies, because oxygen vacancies would be generated to maintain the electrostatic balance when the reduced element states exist as discussed earlier [42]. The O XPS results are consistent well with the Raman results [40]. Thus, it can be speculated that a relatively high concentration of surface adsorbed oxygen may enhance the catalytic performance greatly.

HCHO catalytic oxidation experiments were conducted to evaluate the catalytic ability. In Fig. 7(a), it is clear that Co-LDO(N) and CoMg-LDO(N) exhibit the best catalytic performance. The complete conversion of HCHO occurs at 125 ℃, about 10 ℃ and 30 ℃ higher than that of Co-LDO and CoMg-LDO, respectively. Combining the Raman and Co 2p XPS results, the calcination in the N₂ atmosphere can provide the CoO species and possess more vacancies. A previous study discovered that Co₃O₄ pretreated in the inert atmosphere could enhance the formation of oxygen vacancies [13]. The abundant oxygen vacancies can provide more surface oxygen species, such as Co³⁺-O₂^–-Co²⁺, and the O₂^– species can further dissociate into O^– to form Co³⁺-O^–-Co²⁺, which is the key active oxygen species during the oxidation reaction [13]. This agrees well with our O 1s XPS and Raman results, in which more oxygen species are observed on the surface of Co-LDO(N) and CoMg-LDO(N). The surface oxygen species play important roles in the oxidation reactions because they can react with the adsorbed reactants directly, and their mobilities are better than that of the lattice oxygen species [42-44]. Thus, the relatively high surface oxygen content created by the Co²⁺ ions is the key factor related to the high activity.

The stability of the catalyst was verified and the result is shown in Fig. 7(b). Typically, the test was conducted at 115 ℃ to bring the conversion under 50%. Accordingly, the HCHO conversion was maintained at around 50% for 30 h. It can be seen that there is less than 10% decrease during the entire test. The slight decrease may be because of the accumulation of carbonates. In general, the Co-LDO(N) catalyst exhibits excellent stability and shows potential application value.

To better understand the specific reaction path of the enhanced catalytic ability over the Co-LDO(N) catalyst, the in situ DRIFT experiments were performed to detect the mechanism.As shown in Fig. 8(a), two main peaks at 1581 and 1359 cm^–1 corresponding to the asymmetric Vas(COO) and symmetric Vs(COO) stretch of formate, respectively, can be observed after the exposure to the flow of O₂+HCHO+N₂ [45]. The band at 2840 cm^–1 corresponding to the CH stretch of formate can also be observed [46]. With time, the bands at and 1211 cm^–1, which can be ascribed to the carbonate species, appeared [39], indicating the accumulation of carbonate during the oxidation process. It can be inferred that formate and carbonate are two important species during the HCHO oxidation. To detect more details, Co-LDO(N) was exposed to HCHO+N₂, as shown in Fig. 8(b). The peak at 1755 cm^–1 assigned to HCHO is observed, suggesting that the HCHO would be adsorbed on the surface of the catalyst directly in the anaerobic atmosphere [47, 48]. Once oxygen is induced into the reaction system, adequate surface O₂^– or O^– are supplied, which promotes the fast occurrence of the HCHO oxidization process. Therefore, we cannot observe HCHO in the flow of O₂+HCHO+N₂. The peak at around 2986, 1443, 1323, and 1048 cm^–1 can be ascribed to DOM [7], which is an important intermediate in the previous discoveries; however, it can be hardly observed in Fig. 8(a). This demonstrates that DOM could be oxidized in a very short time in the oxygen-rich environment. Bands corresponding to formate can also be seen; however, the intensity is relatively weaker as compared to the figure exposed to the oxygen-rich environment. Additionally, no peaks of carbonate can be seen, indicating that only adequate or excess oxygen species can oxidize the intermediate carbonate.

According to the above observation, the XPS and Raman results confirmed that the catalyst calcined in N₂ presents relatively higher oxygen vacancy sites. These vacancies are the major active sites for oxygen dissociation, indicating that there can be abundant oxygen species especially active oxygen species (O₂^– and O^–) on the surface of Co-LDO(N). The excellent redox ability comparing to the sample calcined in O₂ is proved by the H₂-TPR results. This phenomenon gives the evidence that the N₂ calcination is beneficial for the enhancement of the mobility of the surface active oxygen atoms. Combining with the above results, the mechanisms of the enhanced HCHO oxidization ability by O₂^– and O^– are shown in Fig. 9. In Fig. 9(a), HCHO is first adsorbed on the surface of the octahedrally coordinated Co²⁺ sites. Further, O₂^– would oxidize HCHO into the DOM intermediate. The excellent oxygen mobility would move one oxygen ion toward the surface of the active Co²⁺ ion to obtain the formate. Subsequently, the formate will be attacked by other oxygen atoms. CO₂ and H₂O will be obtained during this process. The consumption of O₂^– would result in unsaturated sites, which are the active sites for O₂ dissociation [49]. Figure 9(b) shows a similar mechanism, except the O₂^– species are replaced by surface O^– species. Accordingly, HCHO was oxidized into DOM first, then transformed into the formate via the oxygen migration. Both the processes were quick steps. After the consumption, the vacancies were formed and gaseous oxygen would dissociate at the vacancies resulting in O₂^– species. Subsequently, DOM would transform into formate and one proton would migrate to O forming –OH. The formate would react with –OH to obtain CO₂ and H₂O. Finally, Co³⁺-O^–-Co²⁺ is recovered to its initial form. During the entire process, the decomposition of the formate is regarded as the rate-determining step according to the DRIFT. In summary, it can be speculated that the enhanced HCHO conversion ability was related to Co²⁺-O^–-Co³⁺ and Co²⁺-O₂^–-Co³⁺. Both octahedrally coordinated Co²⁺ and active oxygen species at the vacancies play important roles in this system. In addition, extra oxygen may over-oxidize the formate into carbonate, causing inactivation of the catalyst after long-time use.

A Co-LDO(N) catalyst, rich in extra octahedrally coordinated Co²⁺ ions and vacancies, was prepared by a hydrothermal method, and calcined in N₂. The N₂ calcined catalyst exhibited better HCHO oxidation ability than that calcined in O₂. Relative characterizations proved that the octahedrally coordinated Co²⁺ ions are more open than the tetrahedrally coordinated Co²⁺ ions and more easy to oxidize. Furthermore, the N₂ calcination process greatly increased the amount of active oxygen species and their mobility. The in situ DRIFTS indicated that HCHO would be adsorbed on the surface of the catalyst and O^– and O₂^– would oxidize the adsorbed HCHO molecule into DOM. Subsequently, DOM transformed into formate. The formate would finally decompose into CO₂ and H₂O. The entire catalytic process occurred at Co³⁺-O₂^–-Co²⁺ and Co³⁺-O^–-Co²⁺.