Manganese dioxides (MnO₂) have been widely used as heterogeneous catalysts for aerobic oxidations because of their high availability, tunable crystal structures, mixed Mn valence and high mobility of lattice oxygen [1, 2]. Diverse MnO₂ materials are formed by octahedral MnO₆ units shared with corners or edges, which lead to various tunnel and layered structures (such as α-, β-, γ- and δ-MnO₂, etc.) [3]. For gas-phase oxidations, the crystal structure of MnO₂ is suggested to play a crucial role in determining the catalytic performance, with the optimal crystal form varying by the different oxidations [4-7]. However, most studies for the gas-phase oxidation with crystalline MnO₂ are focusing on the catalytic combustion of CO [4], formaldehyde [5, 8] and aromatics [7]; the reports on the aerobic selective oxidation are scarce. Therefore, it is highly desired to develop efficient and practical MnO₂ catalysts for gas-phase selective oxidation.

Recently, gas-phase aerobic oxidation of ethanol to acetaldehyde (AC) has attracted wide attention, because this process is increasingly competitive to the traditional AC production by ethylene-based Wacker oxidation [9]. Although supported noble metal catalysts can achieve moderate to high AC yield [10-13], noble-metal-free oxide catalysts are more promising for practical application given their lower cost and easier regeneration [14]. MnO₂ catalysts have been used for gas-phase aerobic oxidation of ethanol [15-22]. However, most of reports focused on the catalytic combustion of ethanol with excessive oxygen [17-22]. Only silver-doped α-MnO₂ and the supported catalysts derived thereof were effective for the selective ethanol oxidation [16, 23]. We have found that the unreducible metal-doped α-MnO₂ (M-OMS-2), especially Na-OMS-2, could act as stable and efficient catalysts for the selective ethanol oxidation, achieving a comparable AC yield (~66% at 200 ℃) to the Ag-α-MnO₂ (~71% at 230 ℃) [24]. However, besides the α-MnO₂, other crystalline MnO₂ materials have not yet been studied in the selective ethanol oxidation. Moreover, it is still unknown whether the phase structure effect of MnO₂ on the catalytic selective oxidation is different from that on the catalytic complete oxidation. Undoubtedly, clarifying these points will provide rational guidance for designing improved MnO₂ catalysts for selective oxidation.

In the present work, four types (α, β, γ and δ) of MnO₂ catalysts were initially prepared and evaluated in the gas-phase aerobic oxidation of ethanol and CO with one equivalent of O₂. Interestingly, γ-MnO₂ achieved the highest AC yield (~65%) and CO conversion (~100%) at 200 ℃, which are apparently higher than the results obtained by the other three MnO₂ catalysts. Because the catalytic oxidation capability of MnO₂ can be modulated by metal doping [3, 15, 24-27], there are potential opportunities to further improve the catalytic activity of γ-MnO₂ by framework modification with metal dopants. Catalytic CO oxidation can be employed as a complete oxidation to compare with the selective ethanol oxidation. To the best of our knowledge, there are no reports on the aerobic selective oxidation of ethanol or CO oxidation by using metal-doped γ-MnO₂ materials [28]. In this sense, it would be interesting to clarify the difference in the structural and component requirements of γ-MnO₂ for the selective ethanol oxidation and CO oxidation. Therefore, various M-γ-MnO₂ (M = Cu²⁺, Zn²⁺, Mg²⁺, Co²⁺, Ni²⁺, Ca²⁺, Al³⁺, Fe³⁺, La³⁺) catalysts were prepared by a one-pot hydrothermal method and evaluated in the two oxidations. The results strongly evidenced that Zn-γ-MnO₂ with the highest molar ratio of surface base/acid sites obtained the best activity in the CO oxidation, whereas Cu-γ-MnO₂ with the highest surface reducibility achieved the highest AC yield (~75% at 200 ℃) in the selective ethanol oxidation. Cu-γ-MnO₂ can show higher activity than previously reported noble-metal-con- taining catalysts at lower temperature. Their kinetic behavior were also studied.

Various salts (KMnO₄, MnSO₄·H₂O, (NH₄)₂S₂O₈, Cu(NO₃)₂·3H₂O, Zn(NO₃)₂·6H₂O, Mg(NO₃)₂·6H₂O, Co(NO₃)₂·6H₂O, Ni(NO₃)₂·6H₂O, Ca(NO₃)₂·4H₂O, Al(NO₃)₃·9H₂O, Fe(NO₃)₃·9H₂O, La(NO₃)₃·6H₂O) with > 99% purity, acetaldehyde (99%) and ethanol (> 99.8%) were provided by Sinopharm Chemical Reagent Co., Ltd. CH₃CH₂OD (99 atom% D) was purchased from Sigma-Aldrich.

α-MnO₂ was prepared by oxidation of MnSO₄·H₂O with KMnO₄ under refluxing, as describing in our previous paper [24]. β-MnO₂ and γ-MnO₂ were prepared by the hydrothermal reaction of MnSO₄·H₂O (3.38 g) and (NH₄)₂S₂O₈ (4.56 g) in deionized water (100 mL), which was transferred into an autoclave (200 mL), then heated at 140 ℃ for 12 h and at 90 ℃ for 24 h, respectively [27]. δ-MnO₂ was synthesized from the oxidation of MnSO₄·H₂O (0.22 g) by KMnO₄ (1.2 g) in deionized water (60 mL), which was hydrothermally heated at 220 ℃ for 24 h [4]. The resulting solid was filtered, washed and dried at 110 ℃ overnight. All metal-doped M-γ-MnO₂ (M = Cu²⁺, Zn²⁺, Mg²⁺, Co²⁺, Ni²⁺, Ca²⁺, Al³⁺, Fe³⁺, La³⁺) catalysts were synthesized by using the similar procedure for the undoped γ-MnO₂. The metal ion dopant with M/(Mn+M) ratio of 10 mol% was initially added with MnSO₄·H₂O to form a homogeneous mixture, followed by the above procedure. All catalysts were calcined at 300 ℃ in air before use.

XRD patterns were collected on an Empyrean apparatus with Cu K_α radiation (40 mA and 40 kV). The metal ratio of M-γ-MnO₂ was evaluated by X-ray fluorescence spectrometry (XRF-1800). The surface areas were analyzed by the BET method using a Micromeritics ASAP 2420-4MP apparatus. SEM micrographs were taken using a FEI Nova NanoSEM 450 microscope. TEM micrographs were collected on a FEI Tecnai G2 F30 electron microscope at an acceleration voltage of 300 kV. XPS analysis was performed with an AXIS-ULTRA DLD-600W spectrometer with Al K_α irradiation. The binding energies were adjusted by setting the C 1s peak of adventitious carbon at 284.5 eV.

The surface acidity and basicity of M-γ-MnO₂ catalysts were studied with the temperature-programmed desorption (TPD) of NH₃ and CO₂ on a Micrometrics AutoChem 2920II instrument, respectively. Typically, 50 mg sample was loaded in a quartz tube and heated in Ar at 350 ℃ for 2 h to diminish the release of O₂ during TPD. After the sample was cooled to 100 ℃, 0.3% NH₃ in Ar or 10% CO₂ in Ar with a flow rate of 50 mL/min were passed through the sample for 0.5 h. Then pure Ar was used to flush the sample at 100 ℃ for 1 h, which can remove the physically adsorbed NH₃ or CO₂. After that, the sample was heated from 100 to 400 ℃ with a ramp rate of 10 ℃/min and a flow rate of 20 mL/min Ar. The NH₃ or CO₂ evolution was monitored by a thermal conductivity detector (TCD). The amount of desorbed NH₃ or CO₂ was calculated by using the peak areas and a calibration coefficient. H₂-TPR experiments were conducted on the same instrument as TPD. After the sample (20 mg) was pretreated at 300 ℃ in Ar for 1 h and cooled to room temperature, it was treated with 10% H₂ in Ar at a flow rate of 10 mL/min and heated to 600 ℃ at a ramp rate of 10 ℃/min. The H₂ consumptions were calculated by the peak area which was recorded by TCD.

Gas-phase aerobic oxidation of ethanol was performed on a fixed-bed reactor at atmospheric pressure. The detailed procedure is similar to our previous report [24]. Briefly, 60 mg catalyst (80-100 mesh) mixed with 600 mg α-Al₂O₃ (60-80 mesh) was pretreated with O₂ at 300 ℃ for 1 h. Then the preheated gas mixture with 2.5 mL/min of ethanol in the gas phase and a molar ratio of ethanol/O₂/N₂ = 1/1/38 was passed by the catalyst bed with the gas-hourly-space-velocity (GHSV) of about 100000 mL g_cat^-1 h^-1. GC-FID and GC-TCD were simultaneously employed to analyze the reaction products online. In all cases, the carbon balances closed at 100 ± 2%. Other catalyst particle sizes and weights were checked to exclude the presence of diffusion and mass transfer limitations. Blank experiments were also performed to exclude the activity contribution by the empty quartz reactor and α-Al₂O₃.

Catalytic CO oxidation was carried out in a fixed-bed reactor with stainless steel tube (i.d. = 8 mm) under atmospheric pressure. Typically, 0.1 g M-γ-MnO₂ (80-100 mesh) diluted with 0.5 g quartz sand (60-80 mesh) were loaded into the reactor. After the catalyst was pretreated with dry air at 300 ℃ for 1 h and cooled to the room temperature, a mixture of 1 vol% CO and 1 vol% O₂ in N₂ was introduced as the reactants at a flow rate of 50 mL/min. The resulting GHSV is about 30000 mL g_cat^-1 h^-1. The reaction temperature was monitored by a thermal couple inside the catalyst bed and raised from room temperature to 250 ℃ by steps of 10-30 ℃. The reactor was kept at each temperature for 15 min to obtain experimental data at steady state. The composition of the outlet gas was analyzed by GC (Fuli GC 9070) with a TDX-01 packed column. The carbon balance was near 100% in all cases. Similarly, control and blank experiments were performed to exclude the presence of diffusional limitations and background activity, respectively.

The XRD patterns of α-MnO₂, β-MnO₂, γ-MnO₂ and δ-MnO₂ (Fig. S1) are in good agreement with those of cryptomelane-type (JCPDS 42-1348), pyrolusite-type (JCPDS 24-0735), nsutite-type (JCPDS 14-0644), and birnessite-type (JCPDS 80-1098) MnO₂, respectively. The average crystallite sizes and surface areas of the four types of MnO₂ are listed in Table 1. The layered δ-MnO₂ has the smallest crystallite size (~15.1 nm) and the highest surface area (97 m²/g), which is consistent with previous reports [5, 27]. As shown in Fig. 1, the SEM images of α-, β- and γ-MnO₂ show nanorod-like morphologies with widths of 10-200 nm and lengths of 200-2000 nm, whilst the δ-MnO₂ shows a distinct nanosheet-like shape.

Since various metal-doped M-γ-MnO₂ (M = Cu²⁺, Zn²⁺, Mg²⁺, Co²⁺, Ni²⁺, Ca²⁺, Al³⁺, Fe³⁺, La³⁺) catalysts were prepared by the same "one-pot" hydrothermal method as the undoped γ-MnO₂, the urchin-like nanosphere morphology was largely retained (see Fig. 1 and Fig. S2), with different lengths and widths of the nanorods. The average nanorod size of Fe-γ-MnO₂ is below one-fifth that of the γ-MnO₂, resulting in that the surface area of the former (125 m²/g) is much higher than the latter (55 m²/g). The surface areas of the other M-γ-MnO₂ catalysts decreased from 71 m²/g for Co-γ-MnO₂ to 47 m²/g for Ca-γ-MnO₂. As seen in Table 1, all the resulting M/(Mn + M) ratios were lower than the theoretical 10 mol%, indicating that the dopant metal ions incapable of entering the framework of γ-MnO₂ were removed by washing with water. It is known that the substituting degree of the octahedral MnO₂ framework depends on the coordination and the crystal radius (CR) of dopant ions [29, 30]. Only six-coordinated metal ions with similar CR sizes to those of Mn³⁺ (0.072 nm for low spin species, 0.078 for high spin species) and Mn⁴⁺ (0.067 nm) can be readily incorporated in the octahedral MnO₂ framework [27]. In this study, various metal dopants with different CR sizes (Table 1) allow six-coordination, thus different degree of substitution could be expected. Evidently, doping with Fe³⁺, whose CR sizes (0.069 nm for low spin species and 0.079 nm for high spin species) are more close to the Mn³⁺/Mn⁴⁺ species, resulted in the highest Fe/(Mn + Fe) molar ratio (9.7 mol%). In contrast, doping La³⁺ rendered the lowest La/(Mn + La) ratio (0.3 mol%), as La³⁺ has the largest CR size (0.117 nm).

The XRD patterns of all M-γ-MnO₂ are shown in Fig. 2. Clearly, the nsutite structure was basically retained after metal doping, but some peaks for the doped M-γ-MnO₂ shifted to lower angles. The Ca-γ-MnO₂ with low doping content (0.3 mol%) showed similar 2θ shift as the Fe-γ-MnO₂, suggesting that the presence of dopant in both the mixture solution and the final framework may render partial loss of nsutite structure symmetry of γ-MnO₂ and some lattice expansion [31, 32]. Furthermore, as shown in Table 1, the crystallize sizes of M-γ-MnO₂ estimated from the Scherrer equation using the strongest (131) peak (2θ = 37.1^o) varied by the metal dopants, with Cu-γ-MnO₂ showing the biggest average crystallize size (22.8 nm). The Fe-γ-MnO₂ showed the smallest crystallize size (13.8 nm), which is in good agreement with its highest surface area and the smallest nanorod observed by SEM.

The reducibility of various M-γ-MnO₂ samples was evaluated by H₂-TPR (Fig. 3A). There are two main peaks between 250-500 ℃ attributed to reduction of (ⅰ) MnO₂ to Mn₃O₄ (via Mn₂O₃) and (ⅱ) Mn₃O₄ to MnO [2, 33], with the initial reduction starting from 125-200 ℃. Evidently, except for La-γ-MnO₂, all M-γ-MnO₂ samples showed lower temperature for the first reduction peak than the undoped γ-MnO₂, which may be ascribed to the weakening effect of the metal dopants (other than La³⁺) on Mn-O bonds [26, 32]. These results point to the improved reducibility of M-γ-MnO₂ materials after metal doping. Especially for Cu-γ-MnO₂, the first reduction peak was shifted to 285 ℃ with the initial reduction starting as low as 125 ℃, which can be due to the surface reduction of adsorbed oxygen species to form oxygen vacancies without framework degradation. This result suggests that doping with Cu²⁺ can greatly facilitate the reduction of γ-MnO₂. On the contrary, the reduction of γ-MnO₂ is markedly retarded by doping with La³⁺. As shown in Table 2, the measured H₂ consumptions for all M-γ-MnO₂ samples (7.7-10.7 mmol/g) are lower than the theoretical value (11.5 mmol/g) for the complete reduction of MnO₂ to MnO, pointing to the mixed Mn valence in the framework.

The NH₃-TPD and CO₂-TPD profiles of various M-γ-MnO₂ are shown in Fig. 3B and 3C, respectively. The corresponding amount of surface acid/base sites is listed in Table 2. Clearly, the NH₃/CO₂-TPD profiles of the M-γ-MnO₂ samples varied with the metal dopants. The peak at lower temperatures (~130 ℃) could be ascribed to desorption at surface Brönsted acid sites (Mn-OH and M-OH groups) for NH₃-TPD and surface weak base sites (such as O₂^-, OH^-) species for CO₂-TPD. The higher temperature peaks in the NH₃-TPD and CO₂-TPD curves are attributed to the stronger Lewis acid sites derived from the defective metal atoms [32, 34] and the stronger base sites derived from lattice oxygen species (such as metal-O^2- pairs and defective O^2- anions) [35], respectively. Compared with the undoped γ-MnO₂, doping with Fe³⁺, La³⁺, Co²⁺ and Al³⁺ resulted in a considerable increase of surface acidity and basicity, especially for stronger acid/base sites. Fe-γ-MnO₂ with the highest specific surface area possessed the highest surface acidity (87 μmol/g) and basicity (875 μmol/g). For all samples, the surface basic sites are more than the acidic sites by about an order of magnitude. This suggests that the basic oxygen species are predominant on the surface [24], which leads to the alkaline nature of the M-γ-MnO₂ materials.

We initially investigated the influence of crystal structure of MnO₂ on the catalytic performance for the aerobic ethanol oxidation and CO oxidation with one equivalent of O₂ (Fig. 4). Interestingly, the activity order in ethanol oxidation varied with reaction temperature, but even so, γ-MnO₂ exhibited the highest catalytic activity in the range of 125-225 ℃. As listed in Table 2, the turnover frequency (TOF) and space-time-yield (STY) at 200 ℃ increased in the order of δ-MnO₂ < β-MnO₂ < α-MnO₂ < γ-MnO₂, and the AC selectivity of various catalysts maintained > 99% except for δ-MnO₂, which yielded CO₂ (selectivity ~40%) and ethyl acetate (selectivity ~10%) as by-products. Above 200 ℃, CO₂ became the main by-product for all catalysts. This activity order is totally different from previous report on the catalytic ethanol combustion [36], in which the optimal catalyst was α-MnO₂ followed by δ-, γ- and β-MnO₂. This confirms that there are different structure effects of MnO₂ on the aerobic selective oxidation and complete oxidation of ethanol. Furthermore, a different activity order was also observed for CO oxidation, increasing with β-MnO₂ < α-MnO₂ < δ-MnO₂ ~ γ-MnO₂. Clearly, γ-MnO₂ and β-MnO₂ showed the lowest (185 ℃) and highest (~275 ℃) temperature for reaching 90% CO conversion, respectively. The different activity orders in the selective ethanol oxidation and CO oxidation further point to the different structural requirements of MnO₂ in the two oxidations. It is noteworthy that α- and δ-MnO₂ were reported to show higher activity than γ- and β-MnO₂ in the previous report on MnO₂-catalyzed CO oxidation [4], which may be due to the different catalyst preparation, pretreatment and reaction conditions. Undoubtedly, γ-MnO₂ showed the highest activity not only in the selective ethanol oxidation but also in the CO oxidation in this study.

Given the moderate surface area of γ-MnO₂, it could be deduced that the surface area was not the main factor determining the catalytic activity of MnO₂ catalysts. According to the H₂-TPR results of the four catalysts (Fig. 4D), β-MnO₂ and δ-MnO₂ showed lower temperature for the first reduction peak (~300 ℃), but γ-MnO₂ exhibited the lowest starting reduction temperature (~150 ℃). These results imply that the catalytic activity in ethanol oxidation depends not on the bulk reducibility but on the surface reducibility of MnO₂. The higher surface reducibility can render more oxygen vacancies for O₂ activation at lower temperature and benefit the oxidations [24, 37]. Therefore, the drastic activity increase observed by β-MnO₂ above 175 ℃ could be primarily due to the formation of more surface oxygen vacancies during ethanol oxidation. Although δ-MnO₂ also showed lower starting reduction temperature (~175 ℃), its selectivity to CO₂ dramatically increased above 175 ℃. This may be ascribed to the presence of stronger surface acid sites in δ-MnO₂ (Table 2). Evidently, the sample with the highest surface acidity and basicity (δ-MnO₂) showed much lower activity than that with moderate surface acidity and basicity (γ-MnO₂) in the ethanol oxidation, whereas the former showed comparable activity to the latter in the CO oxidation. These results imply that both the surface reducibility and the surface acid-base property can influence the catalytic performance of MnO₂ in the oxidation of ethanol and CO. Since the surface acid-base and redox properties of the optimal γ-MnO₂ catalyst can be tuned by metal-doping, there are great opportunity to further improve the catalytic performance and to clarify the different structural and component requirements of M-γ-MnO₂ for the selective ethanol oxidation and CO oxidation, which would provide an interesting example for the comparative study on these two oxidations using the same catalyst series [38].

The CO conversions as a function of temperature over various M-γ-MnO₂ catalysts are shown in Fig. 5A and 5B. Almost all catalysts achieved ~100% CO conversion below 250 ℃. Obviously, all M-γ-MnO₂ (except for La-γ-MnO₂) showed higher activity than the undoped γ-MnO₂ below 150 ℃, plausibly due to the presence of more surface defect sites after metal doping as suggested by the H₂-TPR results (Fig. 3A). To better compare the activity, T₉₀ was used to identify the temperature at which CO conversion reaches 90% over various catalysts. As listed in Table 2, the T₉₀ value decreased in the order of metal dopant La > Co > Ni > Al > γ ~ Mg > Fe > Ca > Cu > Zn, with Zn-γ-MnO₂ and La-γ-MnO₂ showing the lowest (160 ℃) and highest (240 ℃) T₉₀, respectively. Fig. 5C and 5D show the Arrhenius plots and the estimated apparent activation energy (E_a) between room temperature and 100 ℃. The listed E_a values (13.7-26.0 kJ/mol) are in good agreement with the previously reported E_a values (12-85 kJ/mol) for CO oxidation over metal doped α-MnO₂ catalysts [3]. Interestingly, the decreasing order of the E_a value is very similar to that of the T₉₀, with Zn-γ-MnO₂ showing the lowest E_a and La-γ-MnO₂ showing the highest E_a. Therefore, among various M-γ-MnO₂, Zn-γ-MnO₂ is the most efficient catalyst for CO oxidation and enables O₂ and CO activation at lower temperatures.

To establish the structure-performance relationship in the M-γ-MnO₂-catalyzed CO oxidation, we attempted to relate the catalytic activity with the structural and chemical properties of the catalysts. However, the optimal Zn-γ-MnO₂ is not the catalyst with the highest surface area, reducibility, acidity or basicity, implying that the CO oxidation activity is not determined by one factor but influenced by the factors comprehensively. As listed in Table 2, the Zn-γ-MnO₂ showed the lowest surface acid sites (40 μmmol/g), a moderate surface base sites (529 μmmol/g) but the highest molar ratio of surface base/acid sites (13.2). In contrast, the La-γ-MnO₂ showed the lowest surface base/acid sites ratio (8.0). In an attempt to correlate the ratio of surface base/acid sites with the catalyst activity, we compared the orders of the ratio of surface base/acid sites and the E_a of CO oxidation (Fig. 6). Interestingly, the E_a values decrease with increasing the molar ratio of surface base/acid sites, suggesting that the less acid sites and more base sites in the M-γ-MnO₂ would benefit the CO oxidation. Doping with Zn²⁺ can significantly decrease the surface acid sites and increase the base sites in the γ-MnO₂, and thus may facilitate the CO adsorption and activation.

The catalytic performances of various M-γ-MnO₂ catalysts in the gas-phase ethanol oxidation as the function of temperature are shown in Fig. 7. Obviously, all M-γ-MnO₂ catalysts showed high AC selectivity (up to 99%) below 200 ℃ and a dramatic selectivity decrease above 200 ℃ with CO₂ as the predominant by-product. It is evident that both reducible metals (such as Cu²⁺, Ni²⁺, Co²⁺) and unreducible metals (such as Ca²⁺, Zn²⁺) doped M-γ-MnO₂ catalysts can render higher activity than the undoped γ-MnO₂. As shown in Table 2, the ethanol conversion at 200 ℃ increased in the order of metal dopant La < Fe < Al < Mg < γ < Zn < Co < Ca < Ni < Cu, with Cu-γ-MnO₂ and La-γ-MnO₂ achieving the highest (75%) and lowest (49%) conversion, respectively. The maximum AC yield at 200 ℃ was achieved by Cu-γ-MnO₂ (~75%), accompanied by the highest TOF (5.6 mol_ethanol mol_Mn^-1 h^-1) and STY (2.8 g_AC g_cat^-1 h^-1). The STY value of Cu-γ-MnO₂ was compared with that of the previously reported catalysts for the gas-phase selective ethanol oxidation (see Table S1). Evidently, the Cu-γ-MnO₂ shows higher STY of AC than the previously reported Na-OMS-2 (2.5 g_AC g_cat^-1 h^-1 at 200 ℃) [24] and the noble-metal-containing Ag-α-MnO₂ (2.5 g_AC g_cat^-1 h^-1 at 230 ℃) [16] and Au/MgCuCr₂O₄ (2.8 g_AC g_cat^-1 h^-1 at 250 ℃) [12], confirming the enhancement effect of metal doping in γ-MnO₂ on gas-phase selective ethanol oxidation. As expected, the E_a values (40.0-96.7 kJ/mol), evaluated from the Arrhenius plots in the range of 100-150 ℃ (Fig. 7E and 7F), are in inverse proportion to the catalytic activity of M-γ-MnO₂. The Cu-γ-MnO₂ showed the lowest E_a (40.0 kJ/mol), which is much lower than the undoped γ-MnO₂ (58.8 kJ/mol) and the Na-OMS-2 (63.7 kJ/mol) [24], suggesting that ethanol and oxygen are more prone to activation on Cu-γ-MnO₂ at lower temperatures.

The activity order of M-γ-MnO₂ catalysts in ethanol oxidation is clearly different from that in CO oxidation, confirming that MnO₂-catalyzed different oxidations have different structure-performance relationships. Although the Fe-γ-MnO₂ possesses the highest surface area, acidity and basicity among various M-γ-MnO₂ catalysts, it shows even lower activity than the undoped γ-MnO₂. Given that the Cu-γ-MnO₂ shows the highest reducibility and activity, whereas the La-γ-MnO₂ shows the lowest reducibility and activity, it is reasonable to propose that the surface reducibility is more important than the other properties of M-γ-MnO₂ catalysts in the gas-phase selective oxidation of ethanol. As shown in Fig. 3A, the surface reduction of Cu-γ-MnO₂ started as low as 125 ℃, implying that more surface adsorbed oxygen species or oxygen vacancies are present in the catalyst. It is known that the higher MnO₂ reducibility and more surface oxygen vacancies would benefit the O₂ and ethanol activations [22, 37]. Accordingly, the enhanced catalytic performance after metal (Cu²⁺, Ni²⁺, Ca²⁺, Co²⁺, Zn²⁺) doping could be mainly attributed to the increased surface oxygen vacancies in the doped γ-MnO₂.

The TEM images of the Zn-γ-MnO₂ catalyst before and after use in the CO oxidation are shown in Fig. S3. Clearly, the nanorod-like morphology remained unchanged. The recycled catalyst displayed the lattice fringes of (120) planes and (131) planes with the d spacing distances of 0.39 nm and 0.24 nm, respectively, which are in good agreement with the XRD results for the fresh Zn-γ-MnO₂ sample. It is worth noting that there are many defect sites at the surface and the edge of the nanorods, which would be the active sites for the CO oxidation. To clarify the redox property of M-γ-MnO₂ in the CO oxidation, the oxidation states of Mn and O in the fresh and spent Zn-γ-MnO₂ and γ-MnO₂ catalysts were studied by XPS (Fig. 8A and 8B). The Mn 2p_3/2 spectra can be deconvoluted into four peaks locating at ca. 640.3, 641.6, 642.9 and 644.3 eV, which can be ascribed to Mn²⁺, Mn³⁺, Mn⁴⁺ and a satellite, respectively [24, 39, 40]. Due to the partial decomposition of MnO₂ to Mn₂O₃ and Mn₃O₄ upon calcination at 300 ℃, the fraction of Mn²⁺ species (Mn²⁺/Mn ratio) in the fresh samples reached about 10%. After the CO oxidation, the fractions of surface Mn²⁺ (~ 10%), Mn³⁺ (~51%) and Mn⁴⁺ (~25%) species remained almost unchanged. In line with this, the fractions of surface lattice oxygen (~529.2 eV) and adsorbed oxygen (~530.7 eV) species [33, 41] kept ~ 69% and ~ 30% before and after the reaction, respectively. However, the surface O/Mn ratio increased notably from 1.65 to 1.96 for γ-MnO₂ and from 2.07 to 2.37 for Zn-γ-MnO₂, pointing to the enhanced ratio of surface base/acid sites in the recycled catalysts. The higher surface O/Mn ratio for the fresh Zn-γ-MnO₂ is consistent with the NH₃/CO₂-TPD results and can account for its much higher activity than the undoped γ-MnO₂ at lower temperatures (< 200 ℃). Based on the above characterization and activity results, we can conclude that surface acidity-basicity is more important than the reducibility of M-γ-MnO₂ in the CO oxidation. The M-γ-MnO₂ catalysts can maintain their defect-enriched structure without framework degradation during CO oxidation, and the reaction may follow the Mars-van Krevelen (MVK) mechanism starting from CO activation on surface lattice oxygen with the consumed lattice oxygen being replenished by O₂ rapidly [3, 42].

Continuous gas-phase oxidation of ethanol was carried out at 200 ℃ to evaluate the stability of M-γ-MnO₂ catalysts (Fig. 9). To our delight, the four representative catalysts can keep stable without significant decrease in activity and selectivity for 30 h on-stream. The AC selectivity was maintained over 99% for all catalysts. The ethanol conversion of γ-MnO₂ and Cu-γ-MnO₂ decreased slightly from 65% to 64% and from 75% to 74%, respectively. These results suggest that Cu-γ-MnO₂ can serve as a stable and efficient catalyst for the gas-phase selective oxidation of ethanol. The TEM images of the fresh and recycled Cu-γ-MnO₂ catalysts (see Fig. S3) point to the unchanged nanorod-like morphology with clear lattice fringes of (120) planes and (131) planes, but many defects occur at the surface and edge. The XPS spectra of the fresh and spent Cu-γ-MnO₂ and γ-MnO₂ catalysts (Fig. 8C and 8D) verify that the surface Mn²⁺ fraction significantly increased from 10% to 44% for the spent γ-MnO₂ and from 17% to 49% for the spent Cu-γ-MnO₂. Moreover, the fractions of Mn³⁺ and Mn⁴⁺ species decreased somewhat, pointing to the partial reduction of surface Mn³⁺ and Mn⁴⁺ species to Mn²⁺ during ethanol oxidation. In line with this, the surface lattice oxygen (~529.3 eV) fraction decreased from 70% to 64% for the spent γ-MnO₂ and from 68% to 65% for the spent Cu-γ-MnO₂, whilst the fraction of surface oxygen vacancies (~531.0 eV) increased from 24% to 28% for the spent γ-MnO₂ and from 25% to 30% for the spent Cu-γ-MnO₂. The more Mn²⁺ species and oxygen vacancies observed in the spent Cu-γ-MnO₂ is in good agreement with its higher surface reducibility. Moreover, these observations confirm that increasing the defective Mn²⁺ sites and oxygen vacancies in the γ-MnO₂ framework could enhance the catalytic oxidation activity plausibly by promoting the reactivity and mobility of active oxygen species [24, 37].

To further understand the reaction kinetics and mechanism for the selective ethanol oxidation over M-γ-MnO₂, the influence of reaction conditions including reactant concentration and GHSV on the catalytic performance of Cu-γ-MnO₂ was studied. Fig. 10A shows the effect of O₂ concentration at 2.5 vol% ethanol concentration and a GHSV of 100000 mL g_cat^-1 h^-1. Evidently, Cu-γ-MnO₂ achieved 47% ethanol conversion at 200 ℃ in the absence of O₂, which is much higher than the previously reported Na-OMS-2 (< 5% at 200 ℃) [24], pointing to the higher reactivity and mobility of the lattice oxygen in the former case. The reaction rate and AC selectivity were almost not influenced by O₂ concentrations (1.25-12.5 vol%) when O₂ was fed at 125 ℃. These results suggest that the number of surface oxygen vacancies for O₂ activation keeps about the same level at lower reaction temperature [43]. When more than 2.5 vol% O₂ was fed at 200 ℃, the ethanol conversion increased steadily, but the AC selectivity decreased drastically. This can be explained by the presence of more Mn²⁺ defect sites and oxygen vacancies at 200 ℃, which would enhance the reactivity of surface oxygen species and lead to complete oxidation of ethanol to CO₂.

The effect of ethanol concentration on the catalytic performance of Cu-γ-MnO₂ at 2.5 vol% O₂ concentration and a GHSV of 100000 mL g_cat^-1 h^-1 is shown in Fig. 10B. Clearly, the ethanol conversion decreased steadily from 18% to 6% at 125 ℃ and from 90% to 52% at 200 ℃ with the ethanol concentration increasing from 1 vol% to 5 vol%, whilst the AC selectivity kept stable (up to 99%). The reaction rates were nearly independent of ethanol concentration at 125 ℃. This kinetic behavior suggest that the active sites at the catalyst surface can be almost saturated with ethanol-derived intermediates (e.g., C₂H₅O-) at lower ethanol concentration (~2 vol%). Fig. 10C shows the effect of GHSV on the catalytic behavior of the Cu-γ-MnO₂ catalyst at 200 ℃. The GHSV was changed by the amount of catalyst, while keeping ethanol (2.5 vol%) and O₂ (2.5 vol%) and the total flow rate (100 mL h^-1). With the GHSV increasing from 50000 to 300000 mL g_cat^-1 h^-1, the ethanol conversion decreased from 92% to 48%. Interestingly, the STY increased with GHSV and achieved up to 5.4 g_AC g_cat^-1 h^-1, which is much higher than that obtained by Na-OMS-2 (4.7 g_AC g_cat^-1 h^-1) under similar conditions [24] and is unequivocally the highest value for the gas-phase selective ethanol oxidation over noble-metal-free catalysts at 200 ℃ [14, 44]. It needs to point out that further optimization of the reaction conditions can improve the STY value to higher level.

As we proposed previously [12, 24, 45], the initial steps of ethanol oxidation over reducible metal oxides may involve both the ethanol activation on surface lattice oxygen and the O₂ activation on the reduced metal (e.g. Cu⁺, Mn^2+/3+)-derived oxygen vacancies. Superoxide-type species (O₂^-) is thought to serve as the activated oxygen species, which can dissociate the hydroxyl group of ethanol by a endothermic process [46, 47]. To verify the possible O-H activation by O₂^- species, ethanol molecule deuterated at the hydroxyl group (C₂H₅OD) was employed to probe the kinetic relevance of the elementary step involving O-H bond cleavage to form ethoxide during ethanol oxidation on Cu-γ-MnO₂. The ratio of oxidative dehydrogenation rates for undeuterated and deuterated ethanol (KIE) for C₂H₅OD at 125 ℃ was measured to be 1.58, which confirms that the O-H bond activation is a kinetically relevant step for ethanol oxidation over Cu-γ-MnO₂. This result is opposite to the literatures on aerobic oxidation of ethanol over VO_x/Al₂O₃ [43] and MoO₂/TiO₂ [48], in which the O-H bond cleavage is not kinetically relevant but quasi-equilibrated. The normal KIE value obtained by Cu-γ-MnO₂ is comparable to that observed by Na-OMS-2-catalyzed ethanol oxidation [24]. Therefore, the reaction process of ethanol oxidation over Cu-γ-MnO₂ not only follows the MVK mechanism [19] but also involves the O₂^--mediated O-H and α-C-H activations. Finally, we attribute the higher activity of Cu-γ-MnO₂ than Na-OMS-2 and other metal doped γ-MnO₂ catalysts in the gas-phase selective ethanol oxidation to the higher reducibility, which results in more surface Mn²⁺/Mn³⁺ defects and oxygen vacancies thus benefit the activation of oxygen and ethanol.

In summary, we have shown that γ-MnO₂ is the most efficient catalyst for gas-phase selective ethanol oxidation and CO oxidation among four types (α, β, γ and δ) of MnO₂ catalysts, and metal doping has a strong influence on the surface properties and catalytic performance of M-γ-MnO₂ catalysts. The doped M-γ-MnO₂ can be readily synthesized by a "one-pot" hydrothermal method. The resulting catalysts show different activity orders in CO and ethanol oxidations. For CO oxidation, Zn-γ-MnO₂ shows the highest catalytic activity, which has been associated with its highest molar ratio of surface base/acid sites. The stable oxidation state of surface Mn and oxygen species suggests that CO oxidation over M-γ-MnO₂ follows the lattice oxygen-based Mars-van Krevelen mechanism. In contrast, Cu-γ-MnO₂ shows the best catalytic activity and high stability in selective ethanol oxidation, which have been attributed to its highest surface reducibility and the presence of more Mn²⁺/Mn³⁺ defects and oxygen vacancies for O₂ and ethanol activation at lower temperature. The different structure-performance relationships for M-γ-MnO₂-catalyzed selective ethanol oxidation and CO oxidation would provide useful guidelines for the design of improved MnO₂-based catalysts for selective gas-phase oxidation.

Acknowledgments

The authors thank the Analytical and Testing Center of Huazhong University of Science and Technology, and the Key Laboratory of Material Chemistry for Energy Conversion and Storage (Ministry of Education) for use of the facilities.