The catalytic combustion of hydrocarbons is of great importance in the control of volatile organic compounds (VOCs) and automobile exhaust gases [1-6]. At present, most commercial catalysts are noble-metal-based materials with low combustion temperatures and long catalyst lifetimes [1, 7-11]. In contrast, transition metal oxide catalysts are inexpensive and based on earth-abundant elements, but their activities and stabilities are somewhat low [4, 12-15], obstructing their path toward commercialization. In particular, cobalt-manganese oxides (Co_xMn_1-xO_y) with multi-redox cycles are promising candidates in the catalytic oxidation of hydrocarbons [16-19]. Several synthetic methods have been developed for these catalysts. Among the parameters that determine the catalytic performance of combustion catalysts, the porosity and surface active oxygen species appear to be most critical [11, 20, 21].

The traditional strategies for Co_xMn_1-xO_y synthesis are the co-precipitation (CP) method, sol-gel (SG) method [22] and redox method [23], all of which are solution-based processes (SBPs). From the viewpoint of process intensification, there is still some room to modify the SBP. For example, the scope of metal precursors in SBPs is limited to liquid or soluble solid materials. Several operation units, such as the dissolution of metal precursors and liquid-liquid mixing, are required. Meanwhile, good dispersion of cobalt and manganese species is a key factor in Co_xMn_1-xO_y catalysts, since active oxygen species are expected to form at the Co_xMn_1-xO_y interfaces. Because of the different equilibrium constants for the Co²⁺ (1.6 × 10^-15) and Mn²⁺ (1.9 × 10^-13) ions involved, local separation of cobalt and manganese oxides is often observed when the CP method is used. Thus, the development of efficient methods for the synthesis of Co_xMn_1-xO_y catalysts is of great interest.

Recently, mechanochemistry has been revisited; mechanochemical methods could enable the synthesis of porous catalytic materials (e.g., zeolites, ordered mesoporous polymers and carbons, porous metal oxides, metal-organic frameworks, MgAl layered double hydroxide), and are attractive due to their solvent-free nature. Herein, we demonstrate a mechanochemical redox process (MRP) for the facile synthesis of a Co_xMn_1-xO_y catalyst via ball milling. The solution-based redox reaction involving the oxidation of a cobalt (II) salt with potassium permanganate (KMnO₄) is a well-known method for the synthesis of Co_xMn_1-xO_y catalysts [23, 24]. However, in solution processes, the oxidative ability of KMnO₄ depends on its concentration, which may affect the formation of high valence species and high porosity within the Co_xMn_1-xO_y. Unexpectedly, the Co_xMn_1-xO_y catalyst prepared via MRP exhibited a surface area of 479 m² g^-1, which was higher than that of control samples prepared via CP (34 m² g^-1), SG (72 m² g^-1), and liquid redox (131 m² g^-1) methods. The physical effects of the mechanical action continually reduced the particle size during redox, leading to the formation of interstitial porosity.

In the aerobic combustion of propylene, propylene was completely oxidized at 200 ℃ using the Co_xMn_1-xO_y prepared by MRP; this temperature was much lower than those of the control Co_xMn_1-xO_y catalysts fabricated via CP (450 ℃) and SG (400 ℃). In addition, the Co_xMn_1-xO_y obtained by MRP also showed good resistance to water vapor (4.2%, > 65 h) and sulfur dioxide (20-100 ppm SO₂). At the same time, this catalyst can be extended to other substrates, such as carbon monoxide and methane, to give low combustion temperatures (T₉₀ = 120 ℃ for CO, T₉₀ = 150 ℃ for ethanol, T₉₀ = 225 ℃ for acetone, T₉₀ = 250 ℃ for toluene, T₉₀ = 540 ℃ for CH₄). Furthermore, this mechanochemical redox process is scalable, and the kilogram-scale preparation is already complete, making it an attentive alternative to the traditional methods.

Typical procedure for the synthesis of Co_xMn_1-xO_y by MRP. Co_xMn_1-xO_y was synthesized by MRP based on the redox reaction of cobalt chloride, sodium hydroxide, and potassium permanganate without any templates. In a typical process, 2.8600 g of CoCl₂·6H₂O and 0.8000 g of NaOH were added to a stainless-steel reactor (25 mL) along with six stainless steel balls (two with a diameter of 1.0 cm, and four with a diameter 0.6 cm). The reactor was placed in a high-speed ball milling apparatus (Focucy Planetary Ball Mill F-P50) and the reactants were ball-milled for 30 min at a frequency of 50 Hz. After the first round of ball milling, 0.6321 g of KMnO₄ was added to the reactor. The mixture was then ball-milled for a further 30 min, and the resulting solid was washed in 80 mL deionized water three times and dried in a vacuum oven at 100 ℃ for 12 h. This method allowed the Co_xMn_1-xO_y-M catalyst to be obtained without thermal treatment.

Typical procedure for the synthesis of Co_xMn_1-xO_y by the SG method. In a typical process, 1.4940 g of cobalt acetate tetrahydrate, 0.4900 g of manganese acetate tetrahydrate, and 1.000 g of citric acid were added to 60 mL of deionized water. The mixture was stirred for 1 h at room temperature and then dried at 80 ℃ for 12 h to obtain a sticky solid. Finally, the resulting solid was calcined at 350 ℃ for 2 h under air.

Typical procedure for the synthesis of Co_xMn_1-xO_y by the CP method. In a typical process, 0.2520 g of MnCl₂ and 1.4280 g of CoCl₂·6H₂O were added to 60 mL of deionized water. Separately, 0.6400 g of NaOH was added to 20 mL of deionized water. The NaOH solution was then added to the 60 mL mixture. The final mixture was stirred for 1 h at room temperature and then aged for 6 h. After centrifugation, the solid was dried at 80 ℃ for 12 h. Finally, the resulting solid was calcined at 350 ℃ for 2 h under air.

Typical procedure for the synthesis of Co_xMn_1-xO_y by the solution redox process. In a typical process, 0.6321 g of KMnO₄ and 2.8600 g of CoCl₂·6H₂O were dissolved in 200 mL of deionized water under vigorous stirring. After being stirred at 80 ℃ for 30 min, the catalyst was obtained by filtration and dried at 80 ℃ for 12 h.

X-ray diffraction (XRD) patterns of the catalyst were obtained using a D8 Advance Diffractometer with Cu K_α (-10°-168°) radiation. Each sample was scanned at a rate of 6° min^-1 over the 2θ range 10°-70°.

The specific surface areas and pore structure of the catalysts were determined using N₂ adsorption-desorption isotherms measured at -196 ℃ using a Micromeritics TriStar II 3020 Version 3.02 analyzer. The catalysts were degassed under vacuum at 180 ℃ for 12 h prior to testing. The specific surface areas were calculated from the isotherms using the Brunauer-Emmett-Teller (BET) method. The pore size distribution was obtained using the Barrett-Joyner-Halenda (BJH) method.

The thermal behavior of the catalysts was analyzed using a thermogravimetry analyzer (TGA) at a heating rate of 10 ℃ min^-1 from 50 to 800 ℃ in air.

The performance of the catalysts was evaluated in a fixed-bed reactor at atmospheric pressure. A detailed experimental flow chart is shown in Fig. S1. A 30 mg of the catalyst mixed with 10 mg of quartz sand was used for each measurement; the flow rate was 10 mL min^-1, which corresponds to a gas hourly space velocity (GHSV) of 20000 mL h^-1 g^-1. The feed gas was composed of 1 vol% C₃H₆, 19.9 vol% O₂, and 79.1 vol% N₂. The effluent gas was analyzed using an online gas chromatograph (GC2060). The feed gas C₃H₆ and product CO₂ could be detected in real time.

In the catalyst water resistance experiment, the feed gas was passed through a constant temperature (30 ℃) bubbling device before being introduced into the reactor; the space velocity was kept unchanged.

In the catalyst sulfur resistance experiment, the feed gas was first mixed with 1000 ppm of SO₂ in different volume ratios and then introduced into the reactor; the space velocity was kept unchanged.

In the extended experiment, the feed gas was changed to CO (1 vol% CO, 20 vol% O₂, and 79 vol% N₂) or CH₄ (1 vol% CH₄, 20 vol% O₂, and 79 vol% N₂), or bubbling by air; the space velocity was kept unchanged.

The C₃H₆ conversion was calculated using the following expression:

C₃H₆ conversion(%) = ([C₃H₆]_in- [C₃H₆]_out)/[C₃H₆]_in×100

[C₃H₆]_in and [C₃H₆]_out represent the concentrations of C₃H₆ in the inlet and outlet gases.

The chemistry in the MRP mainly followed the redox equation below:

MnO₄^- + 3Co²⁺ +5OH^- = MnO₂ +3CoOOH + H₂O

The synthesis of Co_xMn_1-xO_y by the mechanochemical redox process was conducted in two steps, as shown in Scheme 1. First, cobalt chloride and sodium hydroxide were added to the ball-milling apparatus, and the solid mixture was ball-milled for 30 min, resulting in the formation of cobalt ions in a basic environment. Next, the oxidant KMnO₄ was added to the reactor, and the mixture was ball-milled for a further 30 min; it was expected that the redox process would take place in the solid state. The primary catalyst was obtained after the removal of alkali metal salts by washing the product in water (Co_xMn_1-xO_y-M). At the same time, control Co_xMn_1-xO_y catalysts were synthesized by the co-precipitation method, sol-gel method, and solution redox method; these catalysts are referred to Co_xMn_1-xO_y-P, Co_xMn_1-xO_y-SG, and Co_xMn_1-xO_y-SR. All samples were ground to a powder between 80 and 100 mesh for catalytic performance tests.

The XRD patterns of the as-synthesized catalysts are shown in Fig. 1. As shown in Fig. 1(a), the Co_xMn_1-xO_y-M catalyst obtained via MRP had an amorphous structure and presented two broad peaks. No obvious peaks were detected for the Co_xMn_1-xO_y-SR sample. The lack of sharp features indicates that both catalysts obtained via redox reactions were poorly crystallized. The Co_xMn_1-xO_y catalysts obtained by the other two methods were then compared; the results are shown in Fig. 1(b). Both Co_xMn_1-xO_y-P and Co_xMn_1-xO_y-SG exhibited peaks at 31°, 36°, 45°, 60°, and 65°. The crystal structure of these two catalysts was similar to that of Co₂MnO_4.5.

The N₂ adsorption-desorption isotherms of Co_xMn_1-xO_y-M and the corresponding pore size distribution are shown in Fig. 2(a) and (b). Significant N₂ adsorption was observed at low relative pressure (e.g., < 0.05), which was attributed to the presence of abundant micropores. The N₂ uptake continued to increase from P/P₀ > 0.05, and a clear hysteresis loop was observed, indicating its mesoporous structure. Its BET specific surface area was 479 m² g^-1, which was a higher value compared to those of other reported cobalt manganese oxides (e.g., 227 m² g^-1 for Mn_xCo_3-xO₄ [24], 59 m² g^-1 for CoOOH [25], 205 m² g^-1 for Mn₂O₃ [26], 150 m² g^-1 for MnO₂ [27], and 55 m² g^-1 for Co₃O₄ [25]). The pore size distribution (PSD) calculated using the BJH model was quite narrow and centered around 3.6 nm, as shown in Fig. 2(b). Figure 2(c) compares the N₂ adsorption-desorption curves of the three control samples. The results demonstrate that the mechanochemical method (479 m² g^-1) significantly promoted the formation of a porous structure (131 m² g^-1 by solution redox, 72 m² g^-1 by sol-gel, and 34 m² g^-1 by co-precipitation). The mechanical action, mainly in the form of kinetic energy and frictional heating, may generate a high degree of interstitial porosity during the solid-state redox reaction, which encouraged us to carry out further characterization.

The morphology of the Co_xMn_1-xO_y-M was directly observed using scanning TEM-high angle angular dark field (STEM-HAADF) images (Figs. 3 and S2). As shown in Fig. 3, the Co_xMn_1-xO_y-M sample exhibited a honeycomb-like structure with irregular porous aggregates. Abundant pores with sizes of ~2-12 nm from the surface to the bulk were observed, in agreement with the pore size distribution determined from the N₂ sorption isotherm. The interstitial porosity side by side was the main factor contributing to its high specific surface area.

The electron energy loss spectroscopy (EELS) images obtained using Co-K_α and Mn-K_α are shown in Fig. S3, and demonstrated the presence of cobalt and manganese in the Co_xMn_1-xO_y-M sample. The STEM-EDX mapping images were collected in Fig. 4. The Co and Mn species were uniformly dispersed throughout the porous structure. (Figs. 4(b-d)). This result demonstrated that the Co_xMn_1-xO_y catalyst synthesized by the mechanochemical redox method showed good dispersion of its constituent elements.

In the HR-TEM images, the visible interlayer spacings of 0.35 and 0.32 nm were attributed to the (220) plane of MnO₂ and the (120) plane of CoOOH, respectively (Figs. 5(a) and (b)). These results further suggested that the Co_xMn_1-xO_y-M sample might exist in the form MnO₂·CoOOH. Interestingly, most of the MnO₂ or CoOOH NPs were smaller than 5 nm and poorly crystallized, which was consistent with the broad reflections in the XRD patterns.

The elemental composition of the Co_xMn_1-xO_y-M sample was then studied using inductively coupled plasma-atomic emission spectroscopy (ICP-AES). In the mechanochemical redox process, the theoretical molar ratio of Co to Mn in Co_xMn_1-xO_y-M should be 3:1. Interestingly, the weight contents of Co and Mn as determined by ICP-AES were 44.5 and 14.0 wt%, respectively. The corresponding molar ratio of Co to Mn is 2.93:1, which is close to the theoretical molar ratio. Additionally, the ICP analysis of the catalyst confirmed the removal of most of the alkali metal salts by the washing procedure (0.40 wt% Na and 0.47 wt% K in the Co_xMn_1-xO_y-M sample). Therefore, the specific molecular formula of the Co_xMn_1-xO_y-M sample was MnO₂·(CoOOH)_2.93. This result proved that the mechanochemical redox reaction between KMnO₄, NaOH and CoCl₂ took place with the formation of MnO₂ and CoOOH.

TG analysis of the Co_xMn_1-xO_y-M catalyst was then carried out to further explore the elemental formula of MnO₂·(CoOOH)_2.93. As shown in Fig. S4, the mass loss in the TG profile below 100 ℃ was about 4%, which was attributed to the evaporation of the moisture adsorbed due to its high specific surface area. Between 200 and 350 ℃, the actual weight loss was about 12.0%, which was close to the theoretical mass loss (~12.7%) estimated for the pyrolysis of CoOOH to Co₃O₄. Above 350 ℃, there was no obvious mass loss. These results further demonstrated that the Co species existed as CoOOH in the Co_xMn_1-xO_y-M catalyst.

Subsequently, we studied the chemical states of Co and Mn using XPS. The XPS spectrum of Co 2p for the Co_xMn_1-xO_y-M catalyst is shown in Fig. 6(a). The Co 2p spectrum of the Co_xMn_1-xO_y-M catalyst contained two peaks at 795.5 and 780.4 eV, corresponding to the Co 2p_1/2 and Co 2p_3/2 spin orbits, respectively. The spin-orbit splitting value was 15.1 eV. The spin-orbit splitting value of Co³⁺ compounds is known to be ~15.0 eV [28]. Hence, the presence of trivalent Co on the Co_xMn_1-xO_y-M surface was suggested. By performing peak fitting, the Mn 2p band could be reasonably divided into two characteristic peaks, which were assigned as Mn³⁺ and Mn⁴⁺. The fitting results are shown in Fig. 6(b) and indicated that the proportion of tetravalent manganese was 53% in the Co_xMn_1-xO_y-M catalyst. The XPS spectra thus revealed that the redox reaction did occur during the ball-milling process.

The catalytic oxidation of hydrocarbons is attracting increasing attention because of the severe pollution caused by volatile organic compounds (VOCs) and vehicle exhaust [29-33]. Initially, catalytic combustion of the model molecule propylene was investigated using the different catalysts [34-37].

The light-off curves of the oxidation of propylene by the catalysts obtained using the three different synthetic processes are shown in Fig. 7. In the presence of the Co_xMn_1-xO_y-P, Co_xMn_1-xO_y-SR, and Co_xMn_1-xO_y-SG catalysts, which were prepared by co-precipitation, solution redox, and sol-gel methods, respectively, propylene was oxidized over a relatively high and wide temperature range (200-450 ℃). Compared with these three catalysts, the Co_xMn_1-xO_y-M catalyst, which was obtained by MRP, afforded better activity. Propylene was completely converted over a low and narrow temperature range (150-200 ℃), with oxidation beginning at 150 ℃ (T₁₀, temperature at which propylene reaches 10% conversion). Using Co_xMn_1-xO_y-M, Co_xMn_1-xO_y-SR, Co_xMn_1-xO_y-SG, and Co_xMn_1-xO_y-P, the T₁₀₀ values (temperature at which propylene was completely converted) were 200, 300, 325, and 450 ℃, respectively. Therefore, in the presence of Co_xMn_1-xO_y-M, both the initial combustion temperature and the total oxidation temperature were lower. Additionally, Co_xMn_1-xO_y-M showed comparable activity to state-of-art catalysts (e.g., T₉₀ = 360 ℃ over Pt/Al₂O₃, T₅₀ = 325 ℃ over CeO₂, T₅₀ = 261 ℃ over 10% Ag/CeO₂-DP, T₉₀ = 200 ℃ over 10% Ag/CeO₂-Imp) [10, 38].

Since the reaction temperature generally fluctuates in practical conditions such as vehicle exhaust [39], we then moved to investigate the catalytic response of Co_xMn_1-xO_y-M during temperature change (Fig. 8). During four cycles in which the temperature was varied between 25 and 250 ℃, the C₃H₆ conversion was closely related to the temperature. After reusing the catalyst four times, only a small decrease in the activity in the low temperature region (25-150 ℃) was observed. The T₅₀ and T₉₀ values of these four light-off curves did not exhibit obvious differences. Overall, there was almost no loss of the initial activity of the Co_xMn_1-xO_y-M catalyst, revealing its reversible activity in the presence of changing temperature.

Encouraged by the good activity of Co_xMn_1-xO_y-M at varied temperatures, water- and sulfur-resistance and thermal stability tests were carried out, as shown in Fig. 9. The light-off curve of C₃H₆ in the presence of water vapor (4.2%) is shown in Fig. 9(a), C₃H₆ was completely converted at ~200 ℃, with activity similar to that of Co_xMn_1-xO_y-M in dry conditions. We then studied the hydrothermal stability of Co_xMn_1-xO_y-M at T₉₀ during long-term combustion. The results are shown in Fig. 9(b). After 65 h of continuous oxidation, the catalyst still showed good activity without any clear loss of catalytic performance. Hence, the current Co_xMn_1-xO_y-M catalyst works well in the presence of water vapor.

In addition to water vapor, vehicle exhaust and industrial VOCs usually contain SO₂ [40]. Thus, the light-off curves for the Co_xMn_1-xO_y-M catalyst in the presence of different SO₂ concentrations were investigated (Fig. 9(c)). Surprisingly, no significant decrease in catalyst activity was observed at SO₂ concentrations of 20 and 100 ppm. When the SO₂ concentration was further increased to 500 ppm, clear stagnation occurred when the conversion rate reached 60%. In this situation, some of the active sites might be occupied by SO₂, causing the oxidation rate of C₃H₆ to decrease. It can be seen that Co_xMn_1-xO_y-M catalyst exhibits competitive performance at lower concentrations of SO₂ (e.g., < 100 ppm).

As the combustion temperature varies widely under practical conditions [39], we tested the thermal stability of Co_xMn_1-xO_y-M at 400 ℃. The results are shown in Fig. 9(d). During the 60 h test, the conversion of C₃H₆ was maintained at 100%. All these results demonstrated the good performance of the Co_xMn_1-xO_y-M catalyst obtained via the mechanochemical redox process.

In view of its good stability, we further tested the catalytic performance of Co_xMn_1-xO_y-M under different space velocity conditions (Fig. 10). At a space velocity of 10000 mL h^-1 g^-1, the conversion rate reached 10% below 130 ℃. As the space velocity was further increased to 50000 mL h^-1 g^-1, the activity of the catalyst did not decrease significantly, and the conversion rate was ~100% at 200 ℃. A delay in the propylene conversion was observed at a space velocity at 75000 mL h^-1 g^-1, with the Co_xMn_1-xO_y-M catalyst affording 100% C₃H₆ conversion at 280 ℃. In theory, the space velocity affects the external diffusion process. At the same time, the higher the space velocity, the shorter the residence time of the reactants, resulting in a lower conversion rate. However, the Co_xMn_1-xO_y-M catalyst obtained via the mechanochemical redox process had a high surface area with abundant surface oxygen species, possibly allowing the reactants to react within a short residence time. Therefore, the Co_xMn_1-xO_y-M catalyst still performed well over a wide range of space velocities.

Since the Co_xMn_1-xO_y-M catalyst offered significant activity and stability in propylene combustion, the scope of this material was then studied. A variety of gaseous pollutants, such as carbon monoxide and low concentrations of methane, are also found under various conditions [41, 42]. Therefore, we conducted a series of tests on the activity of the Co_xMn_1-xO_y-M catalyst toward two different substrates, carbon monoxide and methane. As shown in Fig. 11(a) and (b), when the temperature reached 150 ℃, CO was completely oxidized. At 200 ℃, Co_xMn_1-xO_y still had a strong catalytic activity toward CO, and the conversion rate of CO remained around 100% after 100 h, demonstrating its good stability. In addition, several typical VOCs were selected as combustion substrates, and Co_xMn_1-xO_y-M was found to show competitive activity (T₉₀ = 150, 225, 250 ℃ for ethanol, for acetone, and for toluene, respectively); the results are shown in Figs. 11(c)-(e).

Even in the case of methane, an inert molecule that is difficult to oxidize, when Co_xMn_1-xO_y-M was used as a catalyst, the reaction started at 300 ℃ and the methane was completely oxidized at 600 ℃, as shown in Fig. 11(f). These results demonstrate that the Co_xMn_1-xO_y-M catalyst synthesized via the mechanochemical redox process exhibited good activity and stability with a variety of substrates.

In summary, we used a mechanochemical redox process to prepare the catalyst Co_xMn_1-xO_y. XRD, XPS, HRTEM, ICP-AES, and TGA results proved that Co_xMn_1-xO_y existed as MnO₂ and CoOOH with a Co:Mn molar ratio of 2.97:1, which was close to the theoretical value (3:1). The obtained Co_xMn_1-xO_y catalyst had a specific surface area of 479 m² g^-1, which was higher than that of the control catalysts prepared by the co-precipitation method (34 m² g^-1), sol-gel method (72 m² g^-1), and solution redox process (131 m² g^-1). Moreover, Co_xMn_1-xO_y-M showed good activity in the complete oxidation of propylene; its T₁₀₀ value (200 ℃) was lower than that of Co_xMn_1-xO_y-SG (325 ℃) and Co_xMn_1-xO_y-P (450 ℃). A series of stability tests also proved that Co_xMn_1-xO_y-M had good hydrothermal stability and sulfur resistance. The mechanochemical action may contribute to the formation of a porous structure, while the redox reaction between Co²⁺ and MnO₄^- promotes good dispersion of the cobalt and manganese species at the atomic scale. This mechanochemical redox process for Co_xMn_1-xO_y-M synthesis in kilogram scale is already complete, which may provide a potential strategy for processing porous transition metal oxide catalysts.